1 2018-12-07 Jeff Law <law@redhat.com>
4 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Add
5 m_update_global_ranges member. Add corresponding argument to ctor.
6 * gimple-ssa-evrp-analyze.c
7 (evrp_range_analyzer::evrp_range_analyzer): Add new argument and
8 initialize m_update_global_ranges.
9 (evrp_range_analyzer::set_ssa_range_info): Assert that we are
10 updating global ranges.
11 (evrp_range_analyzer::record_ranges_from_incoming_edge): Only
12 update global ranges if explicitly requested.
13 (evrp_range_analyzer::record_ranges_from_phis): Similarly.
14 (evrp_range_analyzer::record_ranges_from_stmt): Similarly.
15 * gimple-ssa-evrp.c (evrp_dom_walker): Pass new argument to
16 evrp_range_analyzer ctor.
17 * gimple-ssa-sprintf.c (sprintf_dom_walker): Similarly.
18 * tree-ssa-dom.c (dom_opt_dom_walker): Similarly.
20 2018-12-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum):
23 Add SVE_NOT_IMPLEMENTED value.
24 * config/aarch64/aarch64-protos.h (struct tune_params): Add sve_width
26 * config/aarch64/aarch64.c (generic_tunings,cortexa35_tunings,
27 cortexa53_tunings, cortexa57_tunings, cortexa72_tunings,
28 cortexa73_tunings, exynosm1_tunings, thunderx_tunings,
29 thunderx_tunings, tsv110_tunings, xgene1_tunings, qdf24xx_tunings,
30 saphira_tunings, thunderx2t99_tunings, emag_tunings):
32 (aarch64_estimated_poly_value): Define.
33 (TARGET_ESTIMATED_POLY_VALUE): Define.
35 2018-12-07 Paul A. Clarke <pc@us.ibm.com>
38 * config/rs6000/mmintrin.h (_mm_packs_pu16): Correctly use "__vector".
40 2018-12-07 Vladimir Makarov <vmakarov@redhat.com>
42 PR rtl-optimization/88349
43 * ira-costs.c (record_operand_costs): Check bigger reg class on
46 2018-12-07 Richard Sandiford <richard.sandiford@arm.com>
48 * config/aarch64/aarch64-sve.md (*mul<mode>3, *v<optab><mode>3):
49 Split the patterns after reload if we don't need the predicate
51 (*post_ra_mul<mode>3, *post_ra_v<optab><mode>3): New patterns.
53 2018-12-07 Richard Sandiford <richard.sandiford@arm.com>
55 * config/aarch64/iterators.md (SVE_UNPRED_FP_BINARY): New code
57 (sve_fp_op): Handle minus and mult.
58 * config/aarch64/aarch64-sve.md (*add<mode>3, *sub<mode>3)
59 (*mul<mode>3): Split the patterns after reload if we don't
60 need the predicate operand.
61 (*post_ra_<sve_fp_op><mode>3): New pattern.
63 2018-12-07 Bin Cheng <bin.cheng@linux.alibaba.com>
65 * profile-count.h (profile_count::oeprator>=): Fix typo by inverting
66 return condition when *this is precise zero.
68 2018-12-07 Jakub Jelinek <jakub@redhat.com>
71 * final.c (rest_of_handle_final): Don't call collect_fn_hard_reg_usage
72 for functions with naked attribute.
75 * tree-vrp.c (extract_range_from_binary_expr): For POINTER_PLUS_EXPR
76 with -fno-delete-null-pointer-checks, set_nonnull only if the pointer
77 is non-NULL and offset is known to have most significant bit clear.
78 * vr-values.c (vr_values::vrp_stmt_computes_nonzero): For ADDR_EXPR
79 of MEM_EXPR, return true if the MEM_EXPR has non-zero offset with
80 most significant bit clear. If offset does have most significant bit
81 set and -fno-delete-null-pointer-checks, don't return true even if
82 the base pointer is non-NULL.
84 2018-12-06 Alexandre Oliva <aoliva@redhat.com>
86 * cselib.c (cselib_record_sets): Skip strict low part sets
89 2018-12-06 Paul A. Clarke <pc@us.ibm.com>
92 * config/rs6000/smmintrin.h: New file.
93 * config.gcc: Add smmintrin.h to extra_headers for powerpc*-*-*.
95 2018-12-06 Paul A. Clarke <pc@us.ibm.com>
98 * config/rs6000/mmintrin.h (_mm_unpackhi_pi8): Fix for big-endian.
99 (_mm_unpacklo_pi8): Likewise.
100 (_mm_mulhi_pi16): Likewise.
101 (_mm_packs_pi16): Fix for big-endian. Use preferred API.
102 (_mm_packs_pi32): Likewise.
103 (_mm_packs_pu16): Likewise.
104 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix for big-endian.
105 (_mm_cvtss_si64): Likewise.
106 (_mm_cvtpi32x2_ps): Likewise.
107 (_mm_shuffle_ps): Likewise.
108 (_mm_movemask_pi8): Likewise.
109 (_mm_mulhi_pu16): Likewise.
110 (_mm_sad_pu8): Likewise.
111 (_mm_sad_pu8): Likewise.
112 (_mm_cvtpu16_ps): Fix for big-endian. Use preferred API.
113 (_mm_cvtpu8_ps): Likewise.
114 (_mm_movemask_ps): Better #else case for big-endian (no functional
116 (_mm_shuffle_pi16): Likewise.
117 * config/rs6000/emmintrin.h (_mm_movemask_pd): Fix for big-endian.
118 Better #else case for big-endian (no functional change).
119 (_mm_movemask_epi8): Likewise.
120 (_mm_shufflehi_epi16): Likewise.
121 (_mm_shufflelo_epi16): Likewise.
122 (_mm_shuffle_epi32): Likewise.
123 (_mm_mul_epu32): Fix for big-endian.
124 (_mm_bsrli_si128): Likewise.
125 (_mm_cvtps_pd): Better #else case for big endian.
126 (_mm_mulhi_epi16): Likewise.
127 (_mm_mul_epu32): Likewise.
128 (_mm_slli_si128): Likewise.
129 (_mm_sll_epi16): Likewise.
130 (_mm_sll_epi32): Likewise.
131 (_mm_sra_epi16): Likewise.
132 (_mm_sra_epi32): Likewise.
133 (_mm_srl_epi16): Likewise.
134 (_mm_srl_epi32): Likewise.
135 (_mm_mulhi_epu16): Likewise.
136 (_mm_sad_epu8): Likewise.
137 * config/rs6000/pmmintrin.h (_mm_hadd_ps): Fix for big-endian.
138 (_mm_sub_ps): Likewise.
139 * config/rs6000/mmintrin.h (_mm_cmpeq_pi8): Fix for 32-bit mode.
140 * gcc/config/rs6000/tmmintrin.h (_mm_alignr_epi8): Use ENDIAN
141 macros consistently (no functional changes).
142 (_mm_alignr_pi8): Likewise.
144 2018-12-06 Iain Sandoe <iain@sandoe.co.uk>
147 * config/darwin.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC) Remove, use the
149 * config/rs6000/darwin7.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC): New.
151 2018-12-06 Vladimir Makarov <vmakarov@redhat.com>
154 * ira.c (ira_init_register_move_cost): Use info from
155 hard_regno_mode_ok instead of contains_reg_of_mode.
156 * ira-costs.c (contains_reg_of_mode): Don't use cost from bigger
157 hard register class for some fixed hard registers.
159 2018-12-06 Segher Boessenkool <segher@kernel.crashing.org>
161 * doc/extend.texi (Using Assembly Language with C): Document asm inline.
162 (Size of an asm): Fix typo. Document asm inline.
163 * gimple-pretty-print.c (dump_gimple_asm): Handle asm inline.
164 * gimple.h (enum gf_mask): Add GF_ASM_INLINE.
165 (gimple_asm_set_volatile): Fix typo.
166 (gimple_asm_inline_p): New.
167 (gimple_asm_set_inline): New.
168 * gimplify.c (gimplify_asm_expr): Propagate the asm inline flag from
170 * ipa-icf-gimple.c (func_checker::compare_gimple_asm): Compare the
171 gimple_asm_inline_p flag, too.
172 * tree-core.h (tree_base): Document that protected_flag is ASM_INLINE_P
174 * tree-inline.c (estimate_num_insns): If gimple_asm_inline_p return
175 a minimum size for an asm.
176 * tree.h (ASM_INLINE_P): New.
178 2018-12-06 Segher Boessenkool <segher@kernel.crashing.org>
181 * doc/extend.texi (Basic Asm): Update grammar.
182 (Extended Asm): Update grammar.
184 2018-12-06 Iain Sandoe <iain@sandoe.co.uk>
187 * config/i386/darwin.h (STACK_BOUNDARY): Remove macro.
188 * config/i386/i386.c (ix86_compute_frame_layout): Ensure at least 128b
189 stack alignment in non-leaf functions.
191 2018-12-06 Jakub Jelinek <jakub@redhat.com>
194 * config/aarch64/aarch64.c (aarch64_print_address_internal): Don't
195 call output_operand_lossage on VOIDmode CONST_INTs. After
196 output_operand_lossage do return false.
198 2018-12-06 Richard Biener <rguenther@suse.de>
200 * df-problems.c (df_rd_local_compute): Use bitmap_release.
201 (df_live_free): Likewise.
202 (df_md_local_compute): Likewise.
203 (df_md_free): Release df_md_scratch bitmap.
204 * loop-invariant.c (calculate_loop_reg_pressure): Use
206 * sched-deps.c (true_dependency_cache, output_dependency_cache,
207 anti_dependency_cache, control_dependency_cache,
208 spec_dependency_cache): Use bitmap instead of bitmap_head *.
209 * sched-ebb.c (schedule_ebbs_init): Initialize non-GTY
210 dont_calc_deps as bitmap allocated from obstack not GC.
211 (schedule_ebbs_finish): Use bitmap_release.
212 * sched-rgn.c (schedule_insns): Initialize non-GTY
213 not_in_df as bitmap allocated from obstack not GC.
215 * sel-sched.c (_forced_ebb_heads): Remove premature optimization.
216 (sel_region_init): Allocate forced_ebb_heads.
217 (sel_region_finish): Free forced_ebb_heads.
219 2018-12-06 Richard Biener <rguenther@suse.de>
221 * bitmap.c (bitmap_head::crashme): Define.
222 * bitmap.h (bitmap_head): Add constexpr default constructor
223 poisoning the obstack member.
224 (bitmap_head::crashme): Declare.
225 (bitmap_release): New function clearing a bitmap and poisoning
227 * gengtype.c (main): Make it recognize CONSTEXPR.
228 * lra-constraints.c (lra_inheritance): Use bitmap_release
229 instead of bitmap_clear.
230 * ira.c (ira): Work around class-memaccess warning.
231 * regrename.c (create_new_chain): Likewise.
233 2018-12-06 Jakub Jelinek <jakub@redhat.com>
235 PR tree-optimization/85726
236 * generic-match-head.c (optimize_successive_divisions_p): New function.
237 * gimple-match-head.c (optimize_successive_divisions_p): Likewise.
238 * match.pd: Don't combine successive divisions if they aren't exact
239 and optimize_successive_divisions_p is false.
242 * tree-nested.c (convert_nonlocal_reference_stmt): Remove clobbers
243 for non-local automatic decls.
245 2018-12-05 David Edelsohn <dje.gcc@gmail.com>
247 * config/rs6000/aix72.h (ASM_DEFAULT_SPEC): Match Power7 processor
250 2018-12-05 Uros Bizjak <ubizjak@gmail.com>
252 * dwarf2out.c (field_byte_offset): Remove
253 #ifdef PCC_BITFIELD_TYPE_MATTERS check.
255 2018-12-05 Richard Biener <rguenther@suse.de>
256 Martin Sebor <msebor@redhat.com>
259 * calls.c (get_attr_nonstring_decl): Avoid setting *REF to
261 * gcc/gimple-low.c (lower_stmt): Fold builtin calls here.
262 * gimplify (maybe_fold_stmt): Avoid folding builtin calls.
264 2018-12-05 Iain Sandoe <iain@sandoe.co.uk>
266 * configure.ac (gcc_cv_otool): Set.
267 * configure: Regenerate.
269 2018-12-05 Uros Bizjak <ubizjak@gmail.com>
271 * config/i386/cygming.h (PCC_BITFIELD_TYPE_MATTERS): Remove.
272 (GROUP_BITFIELDS_BY_ALIGN): Ditto.
274 2018-12-05 Richard Sandiford <richard.sandiford@arm.com>
276 PR tree-optimization/88064
277 * tree-vect-patterns.c (vect_convert_input): Convert the result of
278 an existing cast if it has the right width but the wrong sign.
279 Do not test the signedness of the required result when
280 considering whether to split an existing cast; instead split to
281 a type with the same signedness as the source of the cast, then
282 convert it to the opposite signedness where necessary.
283 (vect_recog_over_widening_pattern): Handle sign changes between
284 the final PLUS_EXPR and the RSHIFT_EXPR.
285 (vect_recog_average_pattern): Use an unsigned operation when
286 truncating an addition, subtraction or multiplication. Cast the
287 result back to the "real" signedness before promoting.
289 2018-12-05 Richard Biener <rguenther@suse.de>
291 PR tree-optimization/86637
292 * tree-vectorizer.c (pass_slp_vectorize::execute): Reset
293 vect_location at the end.
295 2018-12-05 Jakub Jelinek <jakub@redhat.com>
297 PR tree-optimization/87360
298 * gimple-loop-jam.c (tree_loop_unroll_and_jam): On failure to analyze
299 data dependencies, don't return false, just continue. Formatting
301 (merge_loop_tree, bb_prevents_fusion_p, unroll_jam_possible_p,
302 fuse_loops): Formatting fixes.
305 * cfgexpand.c (expand_stack_vars): If asan_vec is empty, start with
306 aligning frame offset to ASAN_RED_ZONE_SIZE bytes.
308 2018-12-05 Xianmiao Qu <xianmiao_qu@c-sky.com>
310 * config.gcc (csky-*-linux-gnu*): Force .init_array support.
312 2018-12-05 Segher Boessenkool <segher@kernel.crashing.org>
314 * config/rs6000/rs6000.c (rs6000_function_arg): Only do the special
315 aggregate handling on actual AIX, not on somewhat similar systems.
316 (rs6000_arg_partial_bytes): Ditto.
318 2018-12-04 Vladimir Makarov <vmakarov@redhat.com>
320 PR rtl-optimization/88317
321 * lra-constraints.c (split_reg): Don't set up check_only_regs if
322 we are outside the inheritance pass.
324 2018-12-04 Jan Hubicka <hubicka@ucw.cz>
326 * ipa-prop.c (jump_function_useful_p): New.
327 (ipa_read_node_info): Do not allocated useless jump functions.
329 2018-12-04 David Malcolm <dmalcolm@redhat.com>
332 * spellcheck.c (selftest::test_suggestions): Add another assertion.
334 2018-12-04 Martin Sebor <msebor@redhat.com>
336 * doc/extend.texi (__builtin_has_attribute): Fix typo.
338 2018-12-04 Jakub Jelinek <jakub@redhat.com>
341 * config/rs6000/rs6000.c (ccr_bit): Return -1 instead of assertion
343 (print_operand): Use REG_P instead of GET_CODE == REG.
344 <case 'D'>: Also check CR_REGNO_P (REGNO (x)).
345 <case 't'>: Likewise. Remove GET_MODE check.
346 * config/rs6000/rs6000.md (scc patterns): Assert ccr_bit didn't
349 2018-12-04 Martin Sebor <msebor@redhat.com>
351 * doc/extend.texi (attribute aligned): Expand.
353 2018-12-04 David Edelsohn <dje.gcc@gmail.com>
356 * config/rs6000/rs6000.c (rs6000_function_arg): Don't pass aggregates
358 (rs6000_arg_partial_bytes): Same.
360 2018-12-04 Vladimir Makarov <vmakarov@redhat.com>
363 * ira-costs.c (exec): Try bigger class to use smaller register
366 2018-12-04 Michael Ploujnikov <michael.ploujnikov@oracle.com>
369 * ipa-cp.c (create_specialized_node): Track clone counters by
370 node assembler names.
371 (ipcp_driver): Change type of clone_num_suffixes key to const char*.
373 2018-12-04 Richard Biener <rguenther@suse.de>
375 * tree-vect-loop.c (vect_transform_loop): Fix epilogue iterations
376 for epilogue vectorization.
378 2018-12-04 Richard Biener <rguenther@suse.de>
380 PR tree-optimization/88301
381 * tree-vrp.c (register_edge_assert_for_2): Fix sign-conversion
382 issues in last commit.
384 2018-12-04 Richard Biener <rguenther@suse.de>
386 PR tree-optimization/88315
387 * tree-vect-loop.c (get_initial_defs_for_reduction): Simplify
388 and fix initialization vector for SAD and DOT_PROD SLP reductions.
390 2018-12-03 Sandra Loosemore <sandra@codesourcery.com>
393 * doc/extend.texi (Nonlocal gotos): New section.
395 2018-12-03 Uros Bizjak <ubizjak@gmail.com>
397 * config/i386/sse.md (vec_concatv2df): Change (v,xm,C)
398 alternative to (v,vm,c).
400 2018-12-03 Segher Boessenkool <segher@kernel.crashing.org>
402 * gcc/config.gcc (Unsupported targets): Add powerpc*-*-linux*paired*.
403 (powerpc*-*-linux*): Don't handle it here.
405 2018-12-03 Jeff Law <law@redhat.com>
407 * config/h8300/h8300.md (<code>qi3_1): Add gcc_unreachable to keep
410 * config/h8300/h8300.md (low byte masking peepholes): Only accept
411 nonimmediate_operand for the to-be-masked operand.
413 2018-12-03 Richard Biener <rguenther@suse.de>
415 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
416 allow expected function-pointer cast re-instantiation.
418 2018-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
420 * common/config/s390/s390-common.c (s390_option_init_struct):
421 Use section anchors by default.
422 * config/s390/s390.c (s390_check_symref_alignment): Handle
424 (TARGET_MAX_ANCHOR_OFFSET): Use short displacement.
425 * output.h (assemble_align): Pass `align' as unsigned int, so
426 that the value 0x80000000, which corresponds to `aligned(1 <<
427 28)', would pass the `align > BITS_PER_UNIT' check.
428 * varasm.c (assemble_align): Likewise.
430 2018-12-03 Julian Brown <julian@codesourcery.com>
432 * tree-pretty-print.c (dump_omp_clause): Make default case
435 2018-12-03 Richard Biener <rguenther@suse.de>
437 PR tree-optimization/88301
438 * tree-vrp.c (register_edge_assert_for_2): Handle conversions
439 that do not change the value by registering the same assert
442 2018-12-03 Richard Biener <rguenther@suse.de>
444 * tree-vect-loop.c (vect_transform_loop): Properly compute
445 upper bound for the epilogue when doing epilogue vectorization.
447 2018-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
450 * cfgcleanup.c (class pass_postreload_jump): New pass.
451 (pass_postreload_jump::execute): Likewise.
452 (make_pass_postreload_jump): Likewise.
453 * passes.def: Add pass_postreload_jump before
455 * tree-pass.h (make_pass_postreload_jump): New pass.
457 2018-12-02 Uros Bizjak <ubizjak@gmail.com>
459 * config/i386/i386.c (inline_memory_move_cost): Correctly
460 compute number of TFmode GPR moves.
462 2018-12-02 Jakub Jelinek <jakub@redhat.com>
465 * config/i386/sse.md (*vec_concatv4sf_0, *vec_concatv4si_0): New insns.
467 2018-12-02 Jeff Law <law@redhat.com>
469 * config/h8300/h8300.md (call, call_value): Drop mode from
470 operand containing number of bytes pushed.
471 (call_insn, call_value_insn): Use mode iterator for number
472 of bytes pushed operand. Make pattern name vary appropriately.
474 2018-12-02 Jan Hubicka <jh@suse.cz>
476 * bb-reorer.c (find_rarely_executed_basic_blocks_and_crossin_edges):
477 trust precise profiles.
479 2018-12-02 Jakub Jelinek <jakub@redhat.com>
481 * tree-nested.c (convert_nonlocal_omp_clauses,
482 convert_local_omp_clauses): Handle OMP_CLAUSE_IN_REDUCTION,
483 OMP_CLAUSE_TASK_REDUCTION and OMP_CLAUSE__SIMT_ clauses.
484 (convert_nonlocal_reference_stmt, convert_local_reference_stmt):
485 Convert clauses for GIMPLE_OMP_TASKGROUP.
487 * omp-low.c (check_omp_nesting_restrictions): Allow cancel or
488 cancellation point with taskgroup clause inside of taskloop. Consider
489 a taskloop construct without nogroup clause as implicit taskgroup for
490 diagnostics if cancel/cancellation point with taskgroup clause is
491 closely nested inside of taskgroup region.
494 * asan.c (asan_clear_shadow): Move assert that len is multiple of 4
495 to the start of the function.
496 (asan_emit_stack_protection): When emitting clearing sequence for
497 epilogue, align offset down to ASAN_RED_ZONE_SIZE granularity,
498 add last_size_aligned which is last_size padded to multiples of
499 ASAN_RED_ZONE_SIZE and emit asan_clear_shadow always on 4 byte
502 2018-12-01 Jakub Jelinek <jakub@redhat.com>
505 * asan.c (asan_redzone_buffer::flush_redzone_payload): Fix up
506 an off-by-one for BYTES_BIG_ENDIAN.
509 * combine.c (find_split_point): For invalid memory address
510 nonobj + obj + const, if reg + obj + const is valid addressing
511 mode, split at nonobj. Use if rather than else if for the
512 fallback. Comment fixes.
514 2018-11-30 Indu Bhagat <indu.bhagat@oracle.com>
516 * coverage.c (get_coverage_counts): Use from_function_decl for precise
518 * profile-count.c (profile_count::dump): Add handling for precise
520 * profile.c (compute_branch_probabilities): Rely on exec_counts instead
521 of profile_info to set profile_status of function.
522 (branch_prob): Do not set profile_status of function based on
523 profile_info. Done above based on exec_counts.
525 2018-11-30 Jakub Jelinek <jakub@redhat.com>
527 PR tree-optimization/88274
528 * tree-ssa-reassoc.c (optimize_range_tests_xor,
529 optimize_range_tests_diff): If type has smaller precision than
530 corresponding mode or if it has non-standard min/max, compute
531 everything in a standard type for the precision.
534 * params.def (PARAM_LOGICAL_OP_NON_SHORT_CIRCUIT): New param.
535 * tree-ssa-ifcombine.c (ifcombine_ifandif): If
536 --param logical-op-non-short-circuit is present, override
537 LOGICAL_OP_NON_SHORT_CIRCUIT value from the param.
538 * fold-const.c (fold_range_test, fold_truth_andor): Likewise.
540 2018-11-30 Jeff Law <law@redhat.com>
542 * optabs.c (expand_binop): Use "machine_mode" rather than
543 "enum machine mode" in most recent change.
545 2018-11-30 Wilco Dijkstra <wdijkstr@arm.com>
548 * builtins.c (expand_builtin_longjmp): Use a temporary when restoring
550 (expand_builtin_nonlocal_goto): Likewise.
552 2018-11-30 David Malcolm <dmalcolm@redhat.com>
554 * diagnostic-core.h (emit_diagnostic): New decl.
555 * diagnostic.c (emit_diagnostic): New overload, taking a
558 2018-11-30 David Malcolm <dmalcolm@redhat.com>
560 * pretty-print.c (class selftest::test_pretty_printer): New
561 subclass of pretty_printer.
562 (selftest::test_prefixes_and_wrapping): New test.
563 (selftest::pretty_print_c_tests): Call it.
565 2018-11-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
567 Minimize clone counter memory usage in create_virtual_clone.
568 * cgraph.h (clone_function_name): Add a variant that takes a
570 * cgraph.h (cgraph_node::create_virtual_clone): Add a new
571 argument: num_suffix.
572 * cgraphclones.c (cgraph_node::create_virtual_clone): Pass
573 num_suffix to clone_function_name.
574 (clone_function_name): Add a variant that takes a tree decl.
575 * ipa-cp.c (create_specialized_node): Keep track of clone
576 counters in clone_num_suffixes hash map.
577 (ipcp_driver): Free the counter hash map.
578 * ipa-hsa.c (process_hsa_functions): Creates at most one hsa
581 2018-11-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
582 Make function assembly more independent.
584 This is achieved by having clone_function_name assign unique clone
585 numbers for each function independently.
587 * cgraphclones.c: Replaced clone_fn_id_num with clone_fn_ids;
589 (clone_function_name_numbered): Use clone_fn_ids.
591 2018-11-30 Vladimir Makarov <vmakarov@redhat.com>
593 PR rtl-optimization/88179
594 * lra-constraints.c (address_eliminator::address_eliminator):
595 Don't eleminate regs in illegitimate address.
597 2018-11-30 David Malcolm <dmalcolm@redhat.com>
599 PR preprocessor/88257
600 * input.c (get_substring_ranges_for_loc): Fix indentation.
601 Bulletproof against getting a different files back from
602 linemap_resolve_location and expand_location_to_spelling_point.
604 2018-11-30 Alexander Monakov <amonakov@ispras.ru>
606 PR gcov-profile/88279
607 * tree-profile.c (init_ic_make_global_vars): Call
608 decl_default_tls_model for variable and not it's type.
610 2018-11-30 Sam Tebbs <sam.tebbs@arm.com>
612 * config/arm/arm.h (TARGET_DOTPROD): Add TARGET_VFP5 constraint.
614 2018-11-30 Martin Liska <mliska@suse.cz>
617 * asan.c (asan_shadow_cst): Remove, partially transform
618 into flush_redzone_payload.
619 (RZ_BUFFER_SIZE): New.
620 (struct asan_redzone_buffer): New.
621 (asan_redzone_buffer::emit_redzone_byte): Likewise.
622 (asan_redzone_buffer::flush_redzone_payload): Likewise.
623 (asan_redzone_buffer::flush_if_full): Likewise.
624 (asan_emit_stack_protection): Use asan_redzone_buffer class
625 that is responsible for proper aligned stores and flushing
626 of shadow memory payload.
627 * asan.h (ASAN_MIN_RED_ZONE_SIZE): New.
628 (asan_var_and_redzone_size): Likewise.
629 * cfgexpand.c (expand_stack_vars): Use smaller alignment
630 (ASAN_MIN_RED_ZONE_SIZE) in order to make shadow memory
631 for automatic variables more compact.
633 2018-11-30 Alan Modra <amodra@gmail.com>
635 * config/rs6000/predicates.md (easy_fp_constant): Avoid long
636 dependent insn sequences.
637 * config/rs6000/rs6000.c (num_insns_constant): Support long
639 * config/rs6000/rs6000.md (mov<mode>_softfloat <FMOVE128>): Adjust
642 2018-11-30 Alan Modra <amodra@gmail.com>
644 * config/rs6000/rs6000.c (num_insns_constant_gpr): Renamed from
645 num_insns_constant_wide. Make static. Revise comment.
646 (num_insns_constant_multi): New function.
647 (num_insns_constant): Formatting. Correct CONST_WIDE_INT
648 calculation. Simplify and extract code common to both
649 CONST_INT and CONST_DOUBLE. Add gcc_unreachable for unhandled
651 * config/rs6000/rs6000-protos.h (num_insns_const_wide): Delete.
653 2018-11-30 Richard Biener <rguenther@suse.de>
655 PR tree-optimization/88274
656 * tree-vrp.c (ranges_from_anti_range): Fix handling of
659 2018-11-29 Andi Kleen <ak@linux.intel.com>
661 * config/i386/i386.c (current_fentry_section): Add.
662 (x86_function_profiler): Handle fentry section.
663 (ix86_attribute_table): Add fentry section.
664 * config/i386/i386.opt: Add -mfentry-section.
665 * doc/extend.texi: Document fentry_section attribute.
666 * doc/invoke.texi: Document -mfentry-section.
668 2018-11-29 Andi Kleen <ak@linux.intel.com>
670 * config/i386/i386.c (x86_print_call_or_nop): Handle nop name.
671 (current_fentry_name): Add.
672 (x86_function_profiler): Handle fentry_name attribute.
673 (ix86_handle_fentry_name): Add.
674 (ix86_attribute_table): Add fentry_name.
675 * config/i386/i386.opt: Add -mfentry-name
676 * doc/extend.texi: Document fentry_name.
677 * doc/invoke.texi: Document minstrument-return.
679 2018-11-29 Andi Kleen <ak@linux.intel.com>
681 * config/i386/i386-opts.h (enum instrument_return): Add.
682 * config/i386/i386.c (output_return_instrumentation): Add.
683 (ix86_output_function_return): Call output_return_instrumentation.
684 (ix86_output_call_insn): Call output_return_instrumentation.
685 * config/i386/i386.opt: Add -minstrument-return=.
686 * doc/invoke.texi (-minstrument-return): Document.
688 2018-11-29 Eric Botcazou <ebotcazou@adacore.com>
691 * config/sparc/sparc-modes.def: Minor tweak.
692 * config/sparc/sparc.c: Minor reordering.
693 (sparc_pass_by_reference): Move around.
694 (traverse_record_type): Change offset from HOST_WIDE_INT to int.
695 (classify_registers): Likewise for bitpos.
696 (function_arg_slotno): Remove dead test and tweak comments.
697 <MODE_RANDOM>: Remove useless assertion and test whether the
698 parameter is named in order to pass it in FP registers. Return
699 the regno for floating-point vector types.
700 (compute_int_layout): Change bitpos from HOST_WIDE_INT to int.
701 (compute_fp_layout): Likewise.
702 (count_registers): Likewise.
703 (assign_int_registers): Likewise.
704 (assign_fp_registers): Likewise.
705 (assign_registers): Likewise.
706 (function_arg_record_value): Change size from HOST_WIDE_INT to int
707 and use CEIL_NWORDS to compute the number of registers.
708 (function_arg_union_value): Minor tweaks.
709 (function_arg_vector_value): Add slotno and named parameters, use
710 CEIL_NWORDS to compute the number of registers.
711 (sparc_function_arg_1): Rework handling of vector types. Change
712 size from HOST_WIDE_INT to int.
713 (sparc_arg_partial_bytes): Rework handling of 32-bit ABI and deal
714 with vector types for the 64-bt ABI.
715 (sparc_function_arg_advance): Likewise.
716 (sparc_return_in_memory): Add reference to -fpcc-struct-return.
717 (sparc_struct_value_rtx): Return NULL_RTX instead of 0.
718 (sparc_function_value_1): Rework handling of vector types. Change
719 size from HOST_WIDE_INT to int.
721 2018-11-29 Jakub Jelinek <jakub@redhat.com>
724 * config/i386/sse.md (*<sse>_movmsk<ssemodesuffix><avxsizesuffix>_lt,
725 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_lt,
726 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_shift,
727 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_shift,
728 *<sse2_avx2>_pmovmskb_lt, *<sse2_avx2>_pmovmskb_zext_lt): New
729 define_insn_and_split patterns.
733 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
734 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
735 *<sse4_1_avx2>_pblendvb_lt): Change define_insn into
736 define_insn_and_split.
738 2018-11-29 Peter Bergner <bergner@linux.ibm.com>
741 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
742 -mabi=ieeelongdouble without both -mpopcntd and -mvsx.
744 2018-11-29 Uros Bizjak <ubizjak@gmail.com>
746 * config/i386/i386.c (inline_memory_move_cost):
747 Check "in" for 2 in MMX_CLASS_P case.
748 * config/i386/mmx.md (*mov<mode>_internal): Correct
749 TARGET_INTER_UNIT_MOVES_FROM_VEC and TARGET_INTER_UNIT_MOVES_TO_VEC
750 alternatives in preferred_for_speed attribute calculation.
752 2018-11-29 Martin Sebor <msebor@redhat.com>
756 * doc/extend.texi (attribute constructor): Clarify.
758 2018-11-29 Martin Liska <mliska@suse.cz>
761 * builtins.c (expand_movstr): Fix thinko introduced
762 when switching to the new enum.
764 2018-11-29 qing zhao <qing.zhao@oracle.com>
766 * cif-code.def (EXTERN_LIVE_ONLY_STATIC): New CIF code.
767 * common.opt: Add -flive-patching flag.
768 * doc/invoke.texi: Document -flive-patching.
769 * flag-types.h (enum live_patching_level): New enum.
770 * ipa-inline.c (can_inline_edge_p): Disable external functions from
771 inlining when flag_live_patching is LIVE_PATCHING_INLINE_ONLY_STATIC.
772 * opts.c (control_options_for_live_patching): New function.
773 (finish_options): Make flag_live_patching incompatible with flag_lto.
774 Control IPA optimizations based on different levels of
777 2018-11-29 Giuliano Belinassi <giuliano.belinassi@usp.br>
779 * match.pd (sinh (atanh (x))): New simplification rules.
780 (cosh (atanh (x))): Likewise.
782 2018-11-29 Jakub Jelinek <jakub@redhat.com>
785 * config/i386/sse.md (ssebytemode): Add V16SI, V8SI and V4SI entries.
786 (ssefltmodesuffix, ssefltvecmode): New define_mode_attrs.
787 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
788 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
789 *<sse4_1_avx2>_pblendvb_lt): New define_insns.
792 * tree.h (build_uniform_cst, uniform_integer_cst_p): Declare.
793 * tree.c (build_uniform_cst, uniform_integer_cst_p): New functions.
794 * match.pd (define_predicates): Add uniform_integer_cst_p.
795 (cmp @0 INTEGER_CST@1, cmp (convert?@2 @0) INTEGER_CST@1): Adjust
796 so that it works also for vector comparisons with uniform constants
797 with INTEGER_CST element.
800 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): For
801 vec_add and vec_sub builtins, perform PLUS_EXPR or MINUS_EXPR
802 in unsigned_type_for instead of vector integral type where overflow
805 2018-11-29 Michael Ploujnikov <michael.ploujnikov@oracle.com>
807 There can be at most one .resolver clone per function
808 * config/rs6000/rs6000.c (make_resolver_func): Generate
809 resolver symbol with clone_function_name instead of
810 clone_function_name_numbered.
812 2018-11-29 Richard Biener <rguenther@suse.de>
814 PR tree-optimization/88243
815 * tree-vect-patterns.c (vect_mark_pattern_stmts): Set the def
816 type of all pattern-sequence stmts to vect_internal_def.
818 2018-11-29 Andre Vieira <andre.simoesdiasvieira@arm.com>
821 * config/arm/arm-cpus.in (armv7-r): Add FP16conv configurations.
822 (cortex-r7, cortex-r8): Update default and add new configuration.
823 * doc/invoke.texi (armv7-r): Add two new vfp options.
824 (nofp.dp): Add cortex-r7 and cortex-r8 to the list of targets that
827 2018-11-29 Alan Modra <amodra@gmail.com>
829 * config/rs6000/rs6000.c (rs6000_emit_move): Disable long
830 double split for targets other than Darwin.
832 2018-11-29 Alan Modra <amodra@gmail.com>
834 * config/rs6000/rs6000.md (movdi_internal64): Correct lengths.
836 2018-11-29 Alan Modra <amodra@gmail.com>
838 * config/rs6000/predicates.md (easy_fp_constant): Remove code
839 dealing with integers in const_double. Assert on unexpected
840 modes. Delete superfluous ABI_V4 test.
841 * config/rs6000/rs6000.c (rs6000_emit_move): Comment fixes.
842 Don't call easy_fp_constant for Pmode.
843 (rs6000_hash_constant): Remove code dealing with integers in
845 (rs6000_legitimate_constant_p): Likewise.
846 (output_toc): Formatting, use CONST_DOUBLE_P.
848 2018-11-29 Alan Modra <amodra@gmail.com>
850 * config/rs6000/constraints.md (G, H): Comment on purpose of
851 constraint. Correct mode comments and "G" mode test.
852 * config/rs6000/rs6000.md (movdi_internal32): Remove "GH" from
853 alternative handling "F". Add length attr. Formatting.
854 (mov<mode>_softfloat <FMOVE128>): Delete "GH" from
855 alternative, and move "F" to separate alternative. Correct
858 2018-11-29 Alan Modra <amodra@gmail.com>
860 * config/rs6000/predicates.md (easy_vector_constant): Don't call
863 2018-11-29 Alan Modra <amodra@gmail.com>
865 * config.in (HAVE_AS_PLTSEQ): Add.
866 * config/rs6000/predicates.md (indirect_call_operand): New.
867 * config/rs6000/rs6000-protos.h (rs6000_pltseq_template),
868 (rs6000_sibcall_sysv): Declare.
869 * config/rs6000/rs6000.c (init_cumulative_args): Set cookie
870 CALL_LONG for -fno-plt.
871 (print_operand <T, z, 0>): Handle UNSPEC_PLTSEQ.
872 (rs6000_indirect_call_template_1): Emit .reloc directives for
874 (rs6000_pltseq_template): New function.
875 (rs6000_longcall_ref): Add arg parameter. Use PLT16 insns if
876 relocs supported by assembler. Move SYMBOL_REF test to callers.
877 (rs6000_call_aix): Adjust rs6000_longcall_ref call. Package
878 insns in UNSPEC_PLTSEQ, preserving original func_desc.
879 (rs6000_call_sysv): Likewise.
880 (rs6000_sibcall_sysv): New function.
881 * config/rs6000/rs6000.h (HAVE_AS_PLTSEQ): Provide default.
882 * config/rs6000/rs6000.md (UNSPEC_PLTSEQ, UNSPEC_PLT16_HA,
883 UNSPEC_PLT16_LO): New.
884 (pltseq_tocsave, pltseq_plt16_ha, pltseq_plt16_lo, pltseq_mtctr): New.
885 (call_indirect_nonlocal_sysv): Don't differentiate zero from non-zero
886 cookie in constraints. Test explicitly for flags in length attr.
887 Handle unspec operand 1.
888 (call_value_indirect_nonlocal_sysv): Likewise.
889 (call_indirect_aix, call_value_indirect_aix): Handle unspec operand 1.
890 (call_indirect_elfv2, call_value_indirect_elfv2): Likewise.
891 (sibcall, sibcall_value): Use rs6000_sibcall_sysv.
892 (sibcall_indirect_nonlocal_sysv): New pattern.
893 (sibcall_value_indirect_nonlocal_sysv): Likewise.
894 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv): Remove indirect
896 * configure.ac: Check for gas plt sequence marker support.
897 * configure: Regenerate.
899 2018-11-29 Alan Modra <amodra@gmail.com>
901 * config/rs6000/predicates.md (unspec_tls): New.
902 * config/rs6000/rs6000-protos.h (rs6000_call_template),
903 (rs6000_sibcall_template): Update prototype.
904 (rs6000_longcall_ref): Delete.
905 (rs6000_call_sysv): Declare.
906 * config/rs6000/rs6000.c (edit_tls_call_insn): New function.
907 (global_tlsarg): New variable.
908 (rs6000_legitimize_tls_address): Rewrite __tls_get_addr call
910 (print_operand): Extract UNSPEC_TLSGD address operand.
911 (rs6000_call_template, rs6000_sibcall_template): Remove arg
912 parameter, extract from second call operand instead.
913 (rs6000_longcall_ref): Make static, localize vars.
914 (rs6000_call_aix): Rename parameter to reflect new usage. Take
915 tlsarg from global_tlsarg. Don't create unused rtl or nop insns.
916 (rs6000_sibcall_aix): Rename parameter to reflect new usage. Take
917 tlsarg from global_tlsarg.
918 (rs6000_call_sysv): New function.
919 * config/rs6000/rs6000.md: Adjust rs6000_call_template and
920 rs6000_sibcall_template throughout.
921 (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix, tls_gd_call_sysv): Delete.
922 (tls_ld_aix, tls_ld_sysv, tls_ld_call_aix, tls_ld_call_sysv): Delete.
923 (tls_gdld_nomark): New insn.
924 (tls_gd): Swap operand order. Simplify mode selection.
925 (tls_gd_high, tls_gd_low): Swap operand order.
926 (tls_ld): Remove const_int 0 vector element from UNSPEC_TLSLD.
927 Simplify mode selection.
928 (tls_ld_high, tls_ld_low): Similarly adjust UNSPEC_TLSLD.
929 (call, call_value): Don't assert for second call operand.
930 Use rs6000_call_sysv.
932 2018-11-29 Alan Modra <amodra@gmail.com>
934 * config/rs6000/darwin.md (call_indirect_nonlocal_darwin64),
935 (call_nonlocal_darwin64, call_value_indirect_nonlocal_darwin64),
936 (call_value_nonlocal_darwin64): Remove constraints from second call
937 arg, the rounded_stack_size_rtx arg.
938 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix),
939 (tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv, tls_ld_call_aix),
940 (tls_ld_call_sysv, call_local32, call_local64, call_value_local32),
941 (call_value_local64, call_indirect_nonlocal_sysv),
942 (call_nonlocal_sysv, call_nonlocal_sysv_secure),
943 (call_value_indirect_nonlocal_sysv, call_value_nonlocal_sysv),
944 (call_value_nonlocal_sysv_secure, call_local_aix),
945 (call_value_local_aix, call_nonlocal_aix, call_value_nonlocal_aix),
946 (call_indirect_aix, call_value_indirect_aix, call_indirect_elfv2),
947 (call_value_indirect_elfv2, sibcall_local32, sibcall_local64),
948 (sibcall_value_local32, sibcall_value_local64, sibcall_aix),
949 (sibcall_value_aix): Likewise.
951 2018-11-29 Alan Modra <amodra@gmail.com>
953 * config/rs6000/rs6000.md (TLSmode): Delete mode iterator. Replace
954 with P throughout except for call mems which should use SI.
955 (tls_abi_suffix, tls_sysv_suffix, tls_insn_suffix): Delete mode
956 attributes. Replace with bits, mode and ptrload respectively.
958 2018-11-29 Alan Modra <amodra@gmail.com>
960 * config/rs6000/rs6000-protos.h (rs6000_indirect_call_template),
961 (rs6000_indirect_sibcall_template): Declare.
962 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
963 (rs6000_indirect_call_template, rs6000_indirect_sibcall_template):
965 * config/rs6000/rs6000.md (call_indirect_nonlocal_sysv),
966 (call_value_indirect_nonlocal_sysv, sibcall_nonlocal_sysv),
967 (call_indirect_aix, call_value_indirect_aix): Use
968 rs6000_indirect_call_template and rs6000_indirect_sibcall_template.
969 call_indirect_elfv2, call_value_indirect_elfv2): Likewise, and
970 handle both speculation and non-speculation cases.
971 (call_indirect_aix_nospec, call_value_indirect_aix_nospec): Delete.
972 (call_indirect_elfv2_nospec, call_value_indirect_elfv2_nospec): Delete.
974 2018-11-29 Alan Modra <amodra@gmail.com>
976 * config/rs6000/rs6000-protos.h (rs6000_call_template): Declare.
977 (rs6000_sibcall_template): Declare.
978 (macho_call_template): Rename from output_call.
979 * config/rs6000/rs6000.c (rs6000_call_template_1): New function.
980 (rs6000_call_template, rs6000_sibcall_template): Likewise.
981 (macho_call_template): Rename from output_call.
982 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv),
983 (tls_gd_call_aix, tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv),
984 (tls_ld_call_aix, tls_ld_call_sysv, call_nonlocal_sysv),
985 (call_nonlocal_sysv_secure, call_value_nonlocal_sysv),
986 (call_value_nonlocal_sysv_secure, call_nonlocal_aix),
987 (call_value_nonlocal_aix): Use rs6000_call_template and update
988 occurrences of output_call to macho_call_template.
989 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv, sibcall_aix),
990 (sibcall_value_aix): Use rs6000_sibcall_template.
992 2018-11-28 Aaron Sawdey <acsawdey@linux.ibm.com>
994 * config/rs6000/rs6000-string.c (expand_block_clear): Change how
995 we determine if unaligned vsx is ok.
997 2018-11-28 Jan Hubicka <jh@suse.cz>
999 * predict.c (determine_unlikely_bbs): Forward declare; also determine
1000 edges that are always known to be taken because it is only likely
1001 edge out of given BB.
1002 (tree_estimate_probability): Call before profile guessing to get bit
1005 2018-11-28 Jan Hubicka <jh@suse.cz>
1007 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Handle
1008 profile_probability::always better.
1010 2018-11-28 Jan Hubicka <jh@suse.cz>
1012 * profile-count.h (profile_count::split): Give better result when
1013 splitting profile_probability::always.
1015 2018-11-28 Vladimir Makarov <vmakarov@redhat.com>
1018 * ira-costs.c (scan_one_insn): Process subregs when updating costs
1019 for pseudos and allocnos from insn.
1021 2018-11-28 David Edelsohn <dje.gcc@gmail.com>
1023 * config/rs6000/aix72.h: Update to match aix71.h changes.
1025 2018-11-28 Jeff Law <law@redhat.com>
1027 * config/h8300/constraints.md: Add "C" constraint for call insns.
1028 * config/h8300/h8300.md (call, call_value): Turn into a define_expand
1029 and define_insn pair. Move invalid call targets into a register in
1030 the expander and fix constraints in the matching pattern.
1031 * config/h8300/predicates.md (call_expander_operand): Renamed from
1032 call_insn_operand. Reject things we shouldn't be trying to handle.
1033 (call_insn_operand): New predicate for use by the call/call_value
1035 (small_call_insn_operand): Update appropriately.
1037 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
1039 * config/aarch64/aarch64.c (aarch64_process_target_attr): Replace
1040 calls to strtok with strtok_r.
1042 2018-11-28 Richard Biener <rguenther@suse.de>
1044 PR tree-optimization/88223
1045 * tree-ssa-sccvn.c (vn_reference_lookup_3): When skipping
1046 over a stored-same value may-alias store make sure to consider
1047 partial overlaps which are valid when TBAA reasonings do not
1048 apply and byte-granular overlaps are possible at all.
1050 2018-11-28 Richard Biener <rguenther@suse.de>
1052 PR tree-optimization/88217
1053 * vr-values.c (vr_values::extract_range_from_phi_node): Make
1054 sure to handle results > +INF and < -INF correctly when
1055 trying to drop down to +INF - 1 or -INF + 1.
1057 2018-11-28 Alan Modra <amodra@gmail.com>
1059 * xcoffout.c (do_block): Signed/unsigned warning fix.
1061 2018-11-28 Richard Biener <rguenther@suse.de>
1063 PR tree-optimization/79351
1064 * tree-ssa-sccvn.c (vn_reference_lookup_3): For assignments from
1065 empty CONSTRUCTORs ensure the store is at a constant position.
1067 2018-11-28 Richard Biener <rguenther@suse.de>
1069 PR tree-optimization/88229
1070 * tree-ssa.c (non_rewritable_mem_ref_base): Check DECL_SIZE_UNIT
1071 is an INTEGER_CST before accessing it so.
1073 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
1075 * doc/invoke.texi (-mtune=): Remove obsolete CPU names.
1077 2018-11-28 Jakub Jelinek <jakub@redhat.com>
1080 * config/i386/i386.c (ix86_expand_sse_movcc): Handle DFmode and
1081 SFmode using sse4_1_blendvs[sd] with TARGET_SSE4_1. Formatting fixes.
1082 * config/i386/sse.md (sse4_1_blendv<ssemodesuffix>): New pattern.
1084 2018-11-27 Martin Liska <mliska@suse.cz>
1086 * gcov.c (generate_results): Append current_working_directory
1089 2018-11-27 Mike Gulick <mgulick@mathworks.com>
1091 PR preprocessor/83173
1092 * input.c (dump_location_info): Dump reason and included_from
1093 fields from line_map_ordinary struct. Fix indentation when
1094 location > 5 digits.
1095 * diagnostic-show-locus.c (num_digits, num_digits): Move to
1096 diagnostic.c to allow it to be utilized by input.c.
1097 * diagnostic.c (num_digits, selftest::test_num_digits): Moved
1099 (selftest::diagnostic_c_tests): Run selftest::test_num_digits.
1100 * diagnostic.h (num_digits): Add extern definition.
1102 2018-11-27 Fredrik Noring <noring@nocrew.org>
1104 * config/mips/mips.c (mips_reorg_process_insns)
1105 (mips_option_override): Handle `-mfix-r5900'.
1106 * config/mips/mips.h (ASM_SPEC): Add `mfix-r5900' and
1108 * config/mips/mips.opt (mfix-r5900): New option.
1109 * doc/invoke.texi: Document the `r5900' processor name, and
1110 `-mfix-r5900' and `-mno-fix-r5900' options.
1112 2018-11-27 Jakub Jelinek <jakub@redhat.com>
1115 * config/rs6000/rs6000.c (print_operand) <case 'D'>: Use
1116 output_operand_lossage instead of gcc_assert.
1117 <case 't'>: Likewise.
1118 <case 'z'>: Likewise.
1119 <case 'V'>: Use output_operand_lossage instead of gcc_unreachable.
1121 2018-11-27 Jeff Law <law@redhat.com>
1123 * config/riscv/riscv (riscv_block_mvoe_straight): Use RETURN_BEGIN
1124 in call to move_by_pieces.
1126 * config/sh/sh-mem.c (expand_block_move): Use RETURN_BEGIN in call
1129 * config/lm32/lm32.c (lm32_block_move_inline): Use RETURN_BEGIN in
1130 call to move_by_pieces.
1132 * config/mips/mips.c (mips_block_move_straight): Use RETURN_BEGIN
1133 in call to move_by_pieces.
1135 * config/microblaze/microblaze.c (microblaze_block_move_straight): Use
1136 RETURN_BEGIN in call to move_by_pieces.
1137 (microblaze_expand_block_move): Likewise.
1139 2018-11-27 Tamar Christina <tamar.christina@arm.com>
1141 * doc/md.texi (xorsign): Document it.
1143 2018-11-27 Martin Liska <mliska@suse.cz>
1145 * config/gnu-user.h (TARGET_F951_OPTIONS): New.
1146 * gcc.c (find_fortran_preinclude_file): New function
1147 to handle Fortran pre-include.
1149 2018-11-27 Martin Liska <mliska@suse.cz>
1151 * asan.c (asan_emit_stack_protection): Use new enum values
1152 instead of int constants.
1153 * builtins.c (expand_builtin_memory_copy_args): Replace int
1154 type with memop_ret enum type.
1155 (expand_builtin_mempcpy_args): Likewise.
1156 (expand_builtin_memcpy): Use new enum values
1157 instead of int constants. Likewise.
1158 (expand_builtin_mempcpy): Likewise.
1159 (expand_movstr): Likewise.
1160 (expand_builtin_strcpy_args): Likewise.
1161 (expand_builtin_stpcpy_1): Likewise.
1162 (expand_builtin_strncpy): Likewise.
1163 (expand_builtin_memset_args): Likewise.
1164 * expr.c (move_by_pieces_d::finish_endp): Rename to ...
1165 (move_by_pieces_d::finish_retmode): ... this.
1166 (move_by_pieces): Change last argument type to memop_ret.
1167 (store_by_pieces): Use new enum values
1168 instead of int constants.
1169 (emit_block_move_hints): Likewise.
1170 (emit_push_insn): Likewise.
1171 (store_expr): Likewise.
1172 * expr.h (store_by_pieces): Change int to newly added enum
1174 * rtl.h (enum memop_ret): Define.
1175 (move_by_pieces): Use the enum type.
1177 2018-11-27 Alan Modra <amodra@gmail.com>
1179 * config/rs6000/aix71.h (ASM_SPEC): Don't select default -maix64
1181 (ASM_CPU_SPEC): Do so here. Rewrite using if .. else if .. specs
1182 form. Error on missing -mcpu case.
1183 * config/rs6000/driver-rs6000.c (asm_names <_AIX>): Update NULL case.
1184 (asm_names <!_AIX>): Add missing cpus. Update NULL case. Apply
1185 PR63177 fix for -mcpu=power8 and -mcpu=powerpc64le.
1186 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Rewrite using if ..
1187 else if .. specs form. Error on missing -mcpu case. Don't output
1188 duplicate -maltivec. Apply PR63177 fix for -mcpu=powerpc64le.
1190 2018-11-26 David Malcolm <dmalcolm@redhat.com>
1192 * dump-context.h (dump_context::dump_loc): Convert 1st param from
1193 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
1194 const dump_location_t & to const dump_user_location_t &.
1195 (dump_context::dump_loc_immediate): Convert 2nd param from
1196 const dump_location_t & to const dump_user_location_t &.
1197 (dump_context::dump_gimple_stmt): Convert 1st param from
1198 dump_flags_t to const dump_metadata_t &.
1199 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
1200 2nd param from const dump_location_t & to
1201 const dump_user_location_t &.
1202 (dump_context::dump_gimple_expr): Convert 1st param from
1203 dump_flags_t to const dump_metadata_t &.
1204 (dump_context::dump_gimple_expr_loc): Likewise; convert
1205 2nd param from const dump_location_t & to
1206 const dump_user_location_t &.
1207 (dump_context::dump_generic_expr): Convert 1st param from
1208 dump_flags_t to const dump_metadata_t &.
1209 (dump_context::dump_generic_expr_loc): Likewise; convert
1210 2nd param from const dump_location_t & to
1211 const dump_user_location_t &.
1212 (dump_context::dump_printf_va): Convert 1st param from
1213 dump_flags_t to const dump_metadata_t &.
1214 (dump_context::dump_printf_loc_va): Likewise; convert
1215 2nd param from const dump_location_t & to
1216 const dump_user_location_t &.
1217 (dump_context::dump_dec): Convert 1st param from
1218 dump_flags_t to const dump_metadata_t &.
1219 (dump_context::dump_symtab_node): Likewise.
1220 (dump_context::begin_scope): Split out 2nd param into
1221 user and impl locations.
1222 (dump_context::ensure_pending_optinfo): Add metadata param.
1223 (dump_context::begin_next_optinfo): Replace dump_location_t param
1224 with metadata and user location.
1225 * dumpfile.c (dump_context::dump_loc): Convert 1st param from
1226 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
1227 const dump_location_t & to const dump_user_location_t &.
1228 (dump_context::dump_loc_immediate): Convert 2nd param from
1229 const dump_location_t & to const dump_user_location_t &.
1230 (dump_context::dump_gimple_stmt): Convert 1st param from
1231 dump_flags_t to const dump_metadata_t &.
1232 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
1233 2nd param from const dump_location_t & to
1234 const dump_user_location_t &.
1235 (dump_context::dump_gimple_expr): Convert 1st param from
1236 dump_flags_t to const dump_metadata_t &.
1237 (dump_context::dump_gimple_expr_loc): Likewise; convert
1238 2nd param from const dump_location_t & to
1239 const dump_user_location_t &.
1240 (dump_context::dump_generic_expr): Convert 1st param from
1241 dump_flags_t to const dump_metadata_t &.
1242 (dump_context::dump_generic_expr_loc): Likewise; convert
1243 2nd param from const dump_location_t & to
1244 const dump_user_location_t &.
1245 (dump_context::dump_printf_va): Convert 1st param from
1246 dump_flags_t to const dump_metadata_t &.
1247 (dump_context::dump_printf_loc_va): Likewise; convert
1248 2nd param from const dump_location_t & to
1249 const dump_user_location_t &.
1250 (dump_context::dump_dec): Convert 1st param from
1251 dump_flags_t to const dump_metadata_t &.
1252 (dump_context::dump_symtab_node): Likewise.
1253 (dump_context::begin_scope): Split out 2nd param into
1254 user and impl locations.
1255 (dump_context::ensure_pending_optinfo): Add metadata param.
1256 (dump_context::begin_next_optinfo): Replace dump_location_t param
1257 with metadata and user location.
1258 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
1259 const dump_metadata_t &.
1260 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
1261 const dump_location_t & to const dump_user_location_t &.
1262 (dump_gimple_expr): Convert 1st param from dump_flags_t to
1263 const dump_metadata_t &.
1264 (dump_gimple_expr_loc): Likewise; convert 2nd param from
1265 const dump_location_t & to const dump_user_location_t &.
1266 (dump_generic_expr): Convert 1st param from dump_flags_t to
1267 const dump_metadata_t &.
1268 (dump_generic_expr_loc): Likewise; convert 2nd param from
1269 const dump_location_t & to const dump_user_location_t &.
1270 (dump_printf): Convert 1st param from dump_flags_t to
1271 const dump_metadata_t &.
1272 (dump_printf_loc): Likewise; convert 2nd param from
1273 const dump_location_t & to const dump_user_location_t &.
1274 (dump_dec): Convert 1st param from dump_flags_t to
1275 const dump_metadata_t &.
1276 (dump_symtab_node): Likewise.
1277 (dump_begin_scope): Split out 2nd param into user and impl
1279 (selftest::assert_impl_location_eq): New function.
1280 (ASSERT_IMPL_LOCATION_EQ): New macro.
1281 (selftest::test_impl_location): Update to use
1282 ASSERT_IMPL_LOCATION_EQ.
1283 (selftest::test_capture_of_dump_calls): Convert "loc" to
1284 dump_user_location_t. Add ASSERT_IMPL_LOCATION_EQ throughout,
1285 verifying line numbers of dump emissions.
1286 * dumpfile.h (class dump_metadata_t): New class.
1287 (dump_printf): Convert 1st param from dump_flags_t to
1288 const dump_metadata_t &.
1289 (dump_printf_loc): Likewise; convert 2nd param from
1290 const dump_location_t & to const dump_user_location_t &.
1291 (dump_generic_expr_loc): Likewise.
1292 (dump_generic_expr): Convert 1st param from dump_flags_t to
1293 const dump_metadata_t &.
1294 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
1295 const dump_location_t & to const dump_user_location_t &.
1296 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
1297 const dump_metadata_t &.
1298 (dump_gimple_expr_loc): Likewise; convert 2nd param from
1299 const dump_location_t & to const dump_user_location_t &.
1300 (dump_gimple_expr): Convert 1st param from dump_flags_t to
1301 const dump_metadata_t &.
1302 (dump_symtab_node): Likewise.
1303 (dump_dec): Likewise.
1304 (dump_begin_scope): Split out 2nd param into user and impl
1306 (auto_dump_scope::auto_dump_scope): Split "loc" param into a user
1307 location and impl_location, and capture the impl_location.
1308 (AUTO_DUMP_SCOPE): Rename param from LOC to USER_LOC.
1309 * loop-unroll.c (report_unroll): Update for changes to
1310 dump_printf_loc and dump_printf.
1311 * opt-problem.cc (opt_problem::opt_problem): Update for change to
1313 * optinfo-emit-json.cc
1314 (selftest::test_building_json_from_dump_calls): Convert "loc" from
1315 dump_location_t to dump_user_location_t.
1316 * optinfo.cc (optinfo::emit_for_opt_problem): Update for change in
1318 * profile.c (compute_branch_probabilities): Update for change to
1320 * selftest.h (ASSERT_STR_CONTAINS_AT): New macro.
1321 * tree-vect-slp.c (vect_print_slp_tree): Update for change to
1324 2018-11-27 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1326 * doc/sourcebuild.texi: Document check_effective_target_int_eq_float
1327 and check_effective_target_ptr_eq_long.
1329 2018-11-27 Alan Modra <amodra@gmail.com>
1331 * config.gcc (powerpc*-*-freebsd*, powerpc-*-netbsd*),
1332 (powerpc-*-eabisimaltivec*, powerpc-*-eabisim*, powerpc-*-elf*),
1333 (powerpc-*-eabialtivec*, powerpc-*-eabi*, powerpc-*-rtems*),
1334 (powerpc-wrs-vxworks*, powerpc-*-lynxos*, powerpcle-*-elf*),
1335 (powerpcle-*-eabisim*, powerpcle-*-eabi*): Add gnu-user.h to tm_file.
1336 * config/rs6000/freebsd.h (CPLUSPLUS_CPP_SPEC),
1337 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
1338 (ASM_APP_ON, ASM_APP_OFF): Don't define.
1339 * config/rs6000/freebsd64.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
1340 * config/rs6000/lynx.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
1341 * config/rs6000/linux64.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
1342 * config/rs6000/netbsd.h (CPLUSPLUS_CPP_SPEC),
1343 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
1344 * config/rs6000/rtems.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
1345 * config/rs6000/sysv4.h (GNU_USER_TARGET_CRTI): Redefine.
1346 (GNU_USER_TARGET_CRTN): Redefine.
1347 (CC1_SPEC): Use GNU_USER_TARGET_CC1_SPEC.
1348 (LIB_LINUX_SPEC): Use GNU_USER_TARGET_LIB_SPEC.
1349 (CRTOFFLOADBEGIN, CRTOFFLOADEND): Don't define.
1350 (STARTFILE_LINUX_SPEC): Define as GNU_USER_TARGET_STARTFILE_SPEC.
1351 (ENDFILE_LINUX_SPEC): Define as GNU_USER_TARGET_ENDFILE_SPEC.
1352 (UCLIBC_DYNAMIC_LINKER, CHOOSE_DYNAMIC_LINKER): Don't define.
1353 (LINK_EH_SPEC): Don't define.
1355 2018-11-26 Martin Sebor <msebor@redhat.com>
1358 * expr.c (string_constant): Handle top-level decls of all character
1359 types and subobjects of narrow character type.
1361 2018-11-27 Alan Modra <amodra@gmail.com>
1363 * config.gcc (powerpc*-*-linux*): Add linux.h to tm_file.
1364 * config/rs6000/linux.h (TARGET_OS_CPP_BUILTINS): Use
1365 GNU_USER_TARGET_OS_CPP_BUILTINS.
1366 (RS6000_ABI_NAME): Define.
1367 * config/rs6000/linux64.h (TARGET_OS_CPP_BUILTINS): Use
1368 GNU_USER_TARGET_OS_CPP_BUILTINS.
1369 (MUSL_DYNAMIC_LINKER32): Undef before defining.
1370 (UCLIBC_DYNAMIC_LINKER32, UCLIBC_DYNAMIC_LINKER64): Don't define.
1371 (CHOOSE_DYNAMIC_LINKER): Don't define.
1372 (GNU_USER_DYNAMIC_LINKER32, GNU_USER_DYNAMIC_LINKER64): Don't define.
1373 * config/rs6000/sysv4.h (MUSL_DYNAMIC_LINKER): Undef before defining.
1374 (CHOOSE_DYNAMIC_LINKER, GNU_USER_DYNAMIC_LINKER): Only define when
1375 not already defined.
1376 (CPP_OS_LINUX_SPEC): Remove defines and asserts handled by
1377 TARGET_OS_CPP_BUILTINS.
1379 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1381 * doc/sourcebuild.texi: Document check_effective_target_newlib_nano_io.
1383 2018-11-26 Jakub Jelinek <jakub@redhat.com>
1386 * config/i386/i386.c (def_builtin2): If tcode == VOID_FTYPE_UINT64
1387 and !TARGET_64BIT, return NULL_TREE.
1390 * dwarf2out.c (secname_for_decl): For functions with
1391 DECL_SECTION_NAME if in_cold_section_p, try to return
1392 current_function_section's name if it is a named section.
1394 2018-11-26 Maya Rashish <coypu@sdf.org>
1397 * config.host (*-*-netbsd*): Use x-netbsd and host-netbsd.o.
1398 * config/x-netbsd: New file.
1399 * config/host-netbsd: New file.
1401 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1403 * doc/sourcebuild.texi: Document check_effective_target_size20plus.
1404 Clarify documentation for check_effective_target_size32plus.
1406 2018-11-26 Aaron Sawdey <acsawdey@linux.ibm.com>
1408 * config/rs6000/rs6000-string.c (expand_cmp_vec_sequence): Rename
1409 and modify expand_strncmp_vec_sequence.
1410 (emit_final_compare_vec): Rename and modify emit_final_str_compare_vec.
1411 (generate_6432_conversion): New function.
1412 (expand_block_compare): Add support for vsx.
1413 (expand_block_compare_gpr): New function.
1414 * config/rs6000/rs6000.opt (rs6000_block_compare_inline_limit): Increase
1415 default limit to 63 because of more compact vsx code.
1417 2018-11-26 Uros Bizjak <ubizjak@gmail.com>
1420 * config/i386/i386.c (dbx_register_map): Use IGNORED_DWARF_REGNUM
1421 for registers for which no debug information can be generated.
1422 Use INVALID_REGNUM for non-existent registers.
1423 (dbx64_register_map): Ditto.
1424 (svr4_dbx_register_map): Ditto.
1426 2018-11-26 Sandra Loosemore <sandra@codesourcery.com>
1429 * doc/invoke.texi (Optimize Options): Document that
1430 -freorder-functions also uses "hot" and "cold" attributes.
1432 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1434 * config/aarch64/aarch64.c (aarch64_mangle_type): Fix typo in comment.
1436 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1438 * config/arm/types.md (mul64): Rename to...
1439 (widen_mul64): ... This.
1440 * config/arm/arm-generic.md: Rename mul64 to widen_mul64.
1441 * config/arm/cortex-a15.md: Likewise.
1442 * config/arm/cortex-a5.md: Likewise.
1443 * config/arm/cortex-a53.md: Likewise.
1444 * config/arm/cortex-a57.md: Likewise.
1445 * config/arm/cortex-a7.md: Likewise.
1446 * config/arm/cortex-m4.md: Likewise.
1447 * config/arm/exynos-m1.md: Likewise.
1448 * config/arm/marvell-pj4.md: Likewise.
1449 * config/arm/xgene1.md: Likewise.
1451 2018-11-26 Richard Biener <rguenther@suse.de>
1453 PR tree-optimization/88182
1454 * tree-vect-loop.c (vectorizable_reduction): Pick up single
1455 correct reduc_def_info.
1456 * tree-vect-slp.c (vect_analyze_slp_instance): Set
1457 STMT_VINFO_REDUC_DEF of the first stmt.
1459 2018-11-26 Andreas Krebbel <krebbel@linux.ibm.com>
1461 * doc/invoke.texi: Document z14/arch12 -march option.
1463 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1466 * doc/invoke.texi (Option Summary): Add -Wmissing-noreturn.
1467 (Warning Options): Likewise.
1469 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1472 * doc/extend.texi (Common Function Attributes): Clarify that
1473 functions with "const" attribute can read const global variables.
1475 2018-11-26 Alan Modra <amodra@gmail.com>
1477 * config/rs6000/rs6000.h (OBJECT_PEF, TARGET_MACOS): Delete.
1479 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1482 * doc/extend.texi (Common Variable Attributes): Use preferred
1483 placement of type attributes in examples, plus whitespace fixes.
1484 (Type Attributes): Clarify why placement of attributes
1485 immediately after struct/union/enum keyword is preferred.
1486 (Common Type Attributes): Use preferred placement of type
1487 attributes in examples, plus more whitespace fixes.
1489 2018-11-25 Paul Koning <ni1d@arrl.net>
1491 * config/pdp11/pdp11.h (TARGET_HAS_NO_HW_DIVIDE): Define.
1493 2018-11-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1495 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
1496 attributes): Document d_runtime.
1498 2018-11-25 Vladimir Makarov <vmakarov@redhat.com>
1501 * ira-costs.c (record_operand_costs): Initiate register move cost
1504 2018-11-23 Jeff Law <law@redhat.com>
1506 PR rtl-optimization/87468
1507 * tree-ssa-threadupdate.c (create_block_for_threading): Clear
1508 EDGE_IGNORE on all outgoing edges of the duplicate block.
1510 2018-11-23 Vladimir Makarov <vmakarov@redhat.com>
1513 * ira-costs.c (record_operand_costs): Use bigger hard reg class if
1514 its mode does not fit to the original class.
1516 2018-11-23 Martin Sebor <msebor@redhat.com>
1518 PR tree-optimization/87756
1519 * expr.c (string_constant): Synthesize a string literal from
1520 the address of a constant character.
1521 * tree.c (build_string_literal): Add an argument.
1522 * tree.h (build_string_literal): Same.
1524 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
1526 * config/aarch64/aarch64-cores.def: Define emag.
1527 * config/aarch64/aarch64-tune.md: Regenerated with emag.
1528 * config/aarch64/aarch64.c (emag_tunings): New struct.
1529 * doc/invoke.texi: Document mtune value.
1531 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
1532 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
1534 * config/aarch64/aarch64.c (xgene1_tunings): Optimize Xgene1 tunings
1536 * config/aarch64/aarch64.c (xgene1_tunings): Add Xgene1 specific
1538 * config/aarch64/aarch64.c (xgene1_addrcost_table): Correct the post-modify
1540 * config/arm/aarch-cost-tables.h (xgene1_extra_costs): Update the cost table
1543 2018-11-23 Richard Biener <rguenther@suse.de>
1545 PR tree-optimization/88149
1546 * tree-vect-slp.c (vect_slp_analyze_node_operations): Detect
1547 the case where there are two different def types for the
1548 same operand at different operand position in the same stmt.
1550 2018-23-11 Mihail Ionescu <mihail.ionescu@arm.com>
1552 * config/arm/arm.c (arm_expand_compare_and_swap): Simplify and call
1553 gen_atomic_compare_swap_1.
1554 (arm_evpc_neon_vuzp): Likewise gen_neon_vuzp_internal.
1555 (arm_evpc_neon_vtrn): Likewise gen_neon_vtrn_internal.
1556 (arm_evpc_neon_vext): Likewise gen_neon_vext_internal.
1557 (arm_evpc_neon_vzip): Likewise gen_neon_vzip_internal.
1558 (arm_evpc_neon_vrev): Replace the function pointer and simplify the mode
1560 * config/arm/arm.md (neon_vext<mode>),
1561 (neon_vrev64<mode>, neon_vrev32<mode>),
1562 (neon_vrev16<mode>, neon_vtrn<mode>_internal),
1563 (neon_vzip<mode>_internal, neon_vuzp<mode>_internal): Add an '@'character
1564 before the pattern name.
1565 * config/arm/sync.md:
1566 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1),
1567 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): Likewise.
1569 2018-11-23 Jakub Jelinek <jakub@redhat.com>
1571 PR tree-optimization/86614
1572 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return early
1573 if TREE_NO_WARNING is set on ref.ref.
1575 2018-11-22 Sandra Loosemore <sandra@codesourcery.com>
1576 Alan Coopersmith <alan.coopersmith@oracle.com>
1579 * doc/extend.texi (Designated Inits): Clarify handling of multiple
1580 initializers for unions.
1582 2018-11-22 Jan Hubicka <jh@suse.cz>
1585 * ipa-devirt.c (type_variants_equivalent_p): Drop warn and warned
1586 parameters; do not warn here.
1587 (odr_subtypes_equivalent_p): Likewise.
1588 (warn_odr): Fix typo.
1589 (warn_types_mismatch): Do not output confused warnings on integer types.
1590 (odr_types_equivalent_p): Update.
1592 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1594 * config/i386/i386.c (ix86_option_override_internal,
1595 ix86_can_inline_p, classify_argument, construct_container,
1596 ix86_expand_prologue, ix86_expand_split_stack_prologue,
1597 ix86_expand_carry_flag_compare, expand_set_or_movmem_via_loop,
1598 expand_setmem_epilogue_via_loop, promote_duplicated_reg,
1599 ix86_expand_set_or_movmem, ix86_init_builtins_va_builtins_abi):
1602 * config/i386/i386.c (ix86_option_override_internal): For
1603 stack_protector_guard related options, use opts_set->x_ instead
1604 of global_options_set. and prefix options with opts->x_ . Move
1605 defaults for offset and reg into else block.
1607 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
1609 * config/i386/i386.c (ix86_check_avx_upper_register):
1610 Return true for all SSE registers with mode bitsize > 128.
1612 2018-11-22 Vladimir Makarov <vmakarov@redhat.com>
1614 PR rtl-optimization/87718
1615 * ira-costs.c: Remove trailing white-spaces.
1616 (record_operand_costs): Add a special treatment for moves
1617 involving a hard register.
1619 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
1621 * config/i386/i386.c (ix86_avx_emit_vzeroupper): Remove.
1622 (ix86_emit_mode_set) <case AVX_U128>: Emit vzeroupper here.
1624 2018-11-22 Martin Liska <mliska@suse.cz>
1626 * common/config/i386/i386-common.c (processor_names): Add
1627 static assert and add missing "znver2".
1628 (ix86_get_valid_option_values): Add checking assert for null
1629 values and add "native" value if feasible.
1630 * config/i386/i386.h: Do not declare size of processor_names.
1631 * common/config/i386/i386-common.c:
1632 * config/i386/i386.c: Add static assert for size
1633 of processor_cost_table.
1635 2018-11-22 Thomas Preud'homme <thomas.preudhomme@linaro.org>
1637 * target-insns.def (stack_protect_combined_set): Define new standard
1639 (stack_protect_combined_test): Likewise.
1640 * cfgexpand.c (stack_protect_prologue): Try new
1641 stack_protect_combined_set pattern first.
1642 * function.c (stack_protect_epilogue): Try new
1643 stack_protect_combined_test pattern first.
1644 * config/arm/arm.c (require_pic_register): Add pic_reg and compute_now
1645 parameters to control which register to use as PIC register and force
1646 reloading PIC register respectively. Insert in the stream of insns if
1648 (legitimize_pic_address): Expose above new parameters in prototype and
1649 adapt recursive calls accordingly. Use pic_reg if non null instead of
1651 (arm_load_pic_register): Add pic_reg parameter and use it if non null.
1652 (arm_legitimize_address): Adapt to new legitimize_pic_address
1654 (thumb_legitimize_address): Likewise.
1655 (arm_emit_call_insn): Adapt to require_pic_register prototype change.
1656 (arm_expand_prologue): Adapt to arm_load_pic_register prototype change.
1657 (thumb1_expand_prologue): Likewise.
1658 * config/arm/arm-protos.h (legitimize_pic_address): Adapt to prototype
1660 (arm_load_pic_register): Likewise.
1661 * config/arm/predicated.md (guard_addr_operand): New predicate.
1662 (guard_operand): New predicate.
1663 * config/arm/arm.md (movsi expander): Adapt to legitimize_pic_address
1665 (builtin_setjmp_receiver expander): Adapt to thumb1_expand_prologue
1667 (stack_protect_combined_set): New expander..
1668 (stack_protect_combined_set_insn): New insn_and_split pattern.
1669 (stack_protect_set_insn): New insn pattern.
1670 (stack_protect_combined_test): New expander.
1671 (stack_protect_combined_test_insn): New insn_and_split pattern.
1672 (arm_stack_protect_test_insn): New insn pattern.
1673 * config/arm/thumb1.md (thumb1_stack_protect_test_insn): New insn pattern.
1674 * config/arm/unspecs.md (UNSPEC_SP_SET): New unspec.
1675 (UNSPEC_SP_TEST): Likewise.
1676 * doc/md.texi (stack_protect_combined_set): Document new standard
1678 (stack_protect_set): Clarify that the operand for guard's address is
1680 (stack_protect_combined_test): Document new standard pattern name.
1681 (stack_protect_test): Clarify that the operand for guard's address is
1684 2018-11-22 Richard Biener <rguenther@suse.de>
1686 PR tree-optimization/88148
1687 * tree-ssa-loop-niter.c (simplify_replace_tree): Get optional
1688 valueization callback parameter and handle it.
1689 * tree-ssa-loop-niter.h (simplify_replace_tree): Export.
1690 * tree-ssa-sccvn.c (process_bb): Eliminate in loop niter trees.
1692 2018-11-22 Richard Biener <rguenther@suse.de>
1696 * lto-streamer-out.c (lto_is_streamable): Allow CALL_EXPRs
1697 which can appear in size expressions.
1698 * tree-streamer-in.c (unpack_ts_base_value_fields): Stream
1699 CALL_EXPR_BY_DESCRIPTOR.
1700 (streamer_read_tree_bitfields): Stream CALL_EXPR_IFN.
1701 * tree-streamer-out.c (pack_ts_base_value_fields): Stream
1702 CALL_EXPR_BY_DESCRIPTOR.
1703 (streamer_write_tree_bitfields): Stream CALL_EXPR_IFN.
1707 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
1710 2018-11-22 Richard Biener <rguenther@suse.de>
1712 PR tree-optimization/88069
1713 * tree-ssa-sccvn.c (visit_phi): Tweak previous fix to not
1714 apply to default defs.
1716 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1720 * config/i386/i386.c (ix86_option_override_internal): Default
1721 ix86_stack_protector_guard to SSP_TLS only if TARGET_THREAD_SSP_OFFSET
1723 * config/i386/i386.md (stack_protect_set, stack_protect_set_<mode>,
1724 stack_protect_test, stack_protect_test_<mode>): Use empty condition
1725 instead of TARGET_SSP_TLS_GUARD.
1727 2018-11-22 Martin Liska <mliska@suse.cz>
1730 * config/i386/i386.c (ix86_option_override_internal): Error
1731 about usage -mabi=ms and -fsanitize=thread.
1733 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1736 2013-10-26 Vladimir Makarov <vmakarov@redhat.com>
1739 2013-10-25 Vladimir Makarov <vmakarov@redhat.com>
1741 * lra-spills.c (lra_final_code_change): Remove useless move insns.
1743 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1746 * function.c (expand_function_end): Do not emit extra blockage insn.
1748 2018-11-21 Lokesh Janghel <lokeshjanghel91@gmail.com>
1751 * config/i386/i386.c (function_value_ms_64): Return AX_REG instead
1752 of FIRST_SSE_REG for 4 or 8 byte modes.
1754 2018-11-21 Jan Hubicka <jh@suse.cz>
1757 * tree.c (fld_decl_context): Break out from ...
1758 (free_lang_data_in_decl): ... here; free TREE_PUBLIC, TREE_PRIVATE
1759 DECL_ARTIFICIAL of TYPE_DECL; do not free TREE_TYPE of TYPE_DECL.
1760 (fld_incomplete_type_of): Build copy of TYP_DECL.
1761 * ipa-devirt.c (free_enum_values): Rename to ...
1762 (free_odr_warning_data): ... this one; free also duplicated TYPE_DECLs
1763 and TREE_TYPEs of TYPE_DECLs.
1764 (get_odr_type): Initialize odr_vtable_hash if needed.
1766 2018-11-21 Alexandre Oliva <oliva@adacore.com>
1768 * final.c (compute_discriminator): Declare. Renamed from...
1769 (maybe_set_discriminator): ... this. Set and return a local.
1770 (override_discriminator): New.
1771 (final_scan_insn_1): Set it.
1772 (notice_source_line): Adjust. Always set discriminator.
1774 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1777 * config/aarch64/atomics.md (@aarch64_compare_and_swap<mode>): Use
1778 rIJ constraint for aarch64_plus_operand rather than rn.
1780 2018-11-21 Renlin Li <renlin.li@arm.com>
1783 * explow.h (get_dynamic_stack_size): Declare it as external.
1784 * explow.c (record_new_stack_level): Remove function static attribute.
1785 * function.c (assign_stack_local_1): Dynamically align the stack slot
1786 addr for parameter copy on the stack.
1788 2018-11-21 Richard Biener <rguenther@suse.de>
1791 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
1792 * Makefile.in (bitmap.o-warn): Remove again.
1794 2018-11-20 Jeff Law <law@redhat.com>
1796 PR tree-optimization/88069
1797 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
1798 degenerate virtual PHIs.
1800 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1803 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
1804 nonimmediate_operand with register_operand.
1805 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
1806 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
1807 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
1808 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
1809 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
1810 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
1811 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
1812 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
1813 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
1814 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
1815 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
1816 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
1817 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
1818 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
1819 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
1820 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
1821 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
1822 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
1823 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
1824 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
1825 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
1826 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
1827 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
1828 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
1829 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
1830 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
1831 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
1833 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1835 * read-rtl.c (apply_subst_iterator): Handle define_split and
1836 define_insn_and_split.
1838 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1840 PR rtl-optimization/87817
1841 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
1842 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
1843 in the pattern to avoid triggering UB when operands[2] is zero.
1844 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
1845 (*tbm_bextri_<mode>): ... this.
1847 2018-11-21 Tom de Vries <tdevries@suse.de>
1850 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
1853 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
1855 * configure.ac: Add check for Binutils to determine whether vector
1856 load/store alignments hints are being supported.
1857 * config.in: Regenerate.
1858 * configure: Regenerate.
1859 * config/s390/s390.c (print_operand): Support new output
1861 * config/s390/s390.md ("movti"): Append alignment hint output
1862 using the new output modifier 'A'.
1863 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
1864 ("*vec_ti_to_v1ti"): Likewise.
1866 2018-11-20 Martin Sebor <msebor@redhat.com>
1868 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
1870 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1873 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
1874 compare ENUM values.
1876 2018-11-20 Jakub Jelinek <jakub@redhat.com>
1878 PR tree-optimization/87895
1879 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
1880 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
1881 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
1882 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
1883 has any preds and always use EDGE_PRED (, 0).
1885 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
1888 * mode-switching.c (create_pre_exit): After reload, always split the
1889 fallthrough edge to the exit block.
1891 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1893 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
1894 * tree.c (build_array_type_1): Forward declare.
1895 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
1896 (fld_type_variant): Likewise.
1897 (fld_simplified_types): New hash.
1898 (fld_process_array_type): New function.
1899 (fld_incomplete_type_of): Handle array and enumeration types.
1900 (fld_simplified_type): Handle simplification of arrays.
1901 (free_lang_data): Allocate and free simplified types hash.
1903 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1906 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
1907 when sybtype already violates ODR.
1908 (get_odr_type): Do not ICE when insert is false and type duplicate
1909 is not registered yet.
1910 (register_odr_type): Be sure to register subtypes first.
1912 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
1914 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
1915 DImode parts of the target operand.
1917 2018-11-20 Nathan Sidwell <nathan@acm.org>
1920 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
1922 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1925 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
1927 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
1929 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1932 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
1933 * ipa.c (possible_inline_candidate_p): Break out from ..
1934 (process_references): ... here ; drop before_inlining_p;
1935 cleanup handling of alises.
1936 (walk_polymorphic_call_targets): Likewise.
1937 (symbol_table::remove_unreachable_nodes): Likewise.
1938 * passes.c (pass_data_ipa_remove_symbols): New structure.
1939 (pass_ipa_remove_symbols): New pass.
1940 (make_pass_ipa_remove_symbols): New function.
1941 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
1942 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
1944 2018-11-20 Richard Biener <rguenther@suse.de>
1946 * tree-vect-stmts.c (vectorizable_condition): Do not get
1947 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
1948 pointless vect_is_simple_use calls.
1950 2018-11-20 Richard Biener <rguenther@suse.de>
1952 PR tree-optimization/88074
1953 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
1956 2018-11-20 Richard Biener <rguenther@suse.de>
1958 PR tree-optimization/88069
1959 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
1960 virtual PHI arguments.
1962 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
1965 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
1967 * rtl.h (contains_constant_pool_address_p): New function.
1968 * rtlanal.c (contains_constant_pool_address_p): Likewise.
1970 2018-11-20 Richard Biener <rguenther@suse.de>
1973 * alias.c (component_uses_parent_alias_set_from): Remove
1974 alias-set zero and TYPE_TYPELESS_STORAGE case both already
1975 handled in other ways.
1977 2018-11-20 Richard Biener <rguenther@suse.de>
1979 PR tree-optimization/88087
1980 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
1982 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
1985 2018-11-20 Richard Biener <rguenther@suse.de>
1988 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
1990 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
1993 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
1995 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
1998 * doc/invoke.texi (Link Options): Mention shared libraries
1999 in documentation for the -l option. Simplify discussion and
2000 point to the system linker documentation for details.
2002 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
2004 PR rtl-optimization/88033
2005 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
2006 to itself. Use HARD_REGISTER_NUM_P.
2008 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2010 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
2011 of the loop if vf is non-constant.
2013 2018-11-19 David Malcolm <dmalcolm@redhat.com>
2015 PR tree-optimization/87025
2016 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
2017 immediately after creating the scope optinfo.
2018 (selftest::test_pr87025): New function.
2019 (selftest::dumpfile_c_tests): Call it.
2020 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
2021 that we're not popping the top-level records array.
2022 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
2023 not changing the kind of a "scope" optinfo.
2025 2018-11-19 David Malcolm <dmalcolm@redhat.com>
2027 PR tree-optimization/87025
2028 * dump-context.h: Include "optinfo.h".
2029 (class optrecord_json_writer): New forward decl.
2030 (dump_context::forcibly_enable_optinfo_p): Delete.
2031 (dump_context::optinfo_enabled_p): New member function.
2032 (dump_context::optimization_records_enabled_p): New member
2034 (dump_context::set_json_writer): New member function.
2035 (dump_context::emit_optinfo): New member function.
2036 (dump_context::m_forcibly_enable_optinfo): Delete.
2037 (dump_context::m_json_writer): New member data.
2038 * dumpfile.c (dump_context::set_json_writer): New member function.
2039 (dump_context::finish_any_json_writer): New member function.
2040 (dump_context::end_scope): Replace call to
2041 optimization_records_maybe_pop_dump_scope with call to
2042 m_json_writer->pop_scope.
2043 (dump_context::optinfo_enabled_p): New member function.
2044 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
2045 to dump_context::emit_optinfo.
2046 (dump_context::emit_optinfo): New member function.
2047 (temp_dump_context::temp_dump_context): Replace
2048 m_forcibly_enable_optinfo with call to set_json_writer.
2049 (temp_dump_context::~temp_dump_context): Clean up any json writer.
2050 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
2052 (the_json_writer): Delete.
2053 (optimization_records_start): Delete.
2054 (optimization_records_finish): Delete.
2055 (optimization_records_enabled_p): Delete, in favor of
2056 dump_context::optimization_records_enabled_p.
2057 (optimization_records_maybe_record_optinfo): Delete.
2058 (optimization_records_maybe_pop_dump_scope): Delete.
2059 * optinfo-emit-json.h: Include "json.h". Delete forward
2061 (optimization_records_start): Delete.
2062 (optimization_records_finish): Delete.
2063 (optimization_records_enabled_p): Delete.
2064 (optimization_records_maybe_record_optinfo): Delete.
2065 (optimization_records_maybe_pop_dump_scope): Delete.
2066 (class optrecord_json_writer): Move here from
2067 optinfo-emit-json.cc.
2068 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
2069 to optinfo::emit with call to dump_context::emit_optinfo.
2070 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
2071 (optinfo_enabled_p): Delete, in favor of
2072 dump_context::optinfo_enabled_p.
2073 (optinfo_wants_inlining_info_p): Update for conversion o
2074 optimization_records_enabled_p to a member function of
2076 * optinfo.h (optinfo_enabled_p): Delete, in favor of
2077 dump_context::optinfo_enabled_p.
2078 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
2079 * toplev.c: Include "dump-context.h".
2080 (compile_file): Replace call to optimization_records_finish with
2081 dump_context::finish_any_json_writer.
2082 (do_compile): Replace call to optimization_records_start with
2083 conditionally creating a optrecord_json_writer for the
2086 2018-11-19 Jakub Jelinek <jakub@redhat.com>
2088 PR tree-optimization/88071
2089 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
2090 of false as last argument to gsi_remove.
2091 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
2092 false as last argument to gsi_replace.
2095 * omp-expand.c: Don't include debug.h.
2096 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
2097 from innermost outer parallel, task, teams or target that has a
2098 child_fn set, or, if there is no such outer region, use
2099 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
2100 whether a suitable BLOCK is found or not.
2101 (expand_parallel_call, expand_teams_call): Don't call
2102 adjust_context_and_scope here.
2103 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
2104 (expand_omp_taskreg, expand_omp_target): Likewise. Call
2105 adjust_context_and_scope.
2106 * dwarf2out.c (dwarf2out_early_global_decl): For
2107 decl_function_context recurse instead of calling dwarf2out_decl.
2109 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
2111 PR rtl-optimization/85925
2112 * rtl.h (word_register_operation_p): New predicate.
2113 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
2114 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
2115 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
2116 (num_sign_bit_copies1): Likewise.
2118 2018-11-19 Richard Biener <rguenther@suse.de>
2121 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
2124 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
2126 * compare-elim.c (struct comparison): Add not_in_a field.
2127 (is_not): New static function.
2128 (strip_not): Likewise.
2129 (conforming_compare): Handle a NOT in the first operand.
2130 (can_eliminate_compare): Likewise.
2131 (find_comparison_dom_walker::before_dom_children): Likewise.
2132 (try_eliminate_compare): Likewise.
2133 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
2134 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
2136 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
2138 * doc/extend.texi (Common Type Attributes): Fix typos.
2140 2018-11-19 Martin Liska <mliska@suse.cz>
2142 PR gcov-profile/88045
2143 * coverage.c (coverage_begin_function): Add assert.
2145 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
2148 * doc/extend.texi (Common Function Attributes): Document that
2149 no_instrument_function applies to -p and -pg, too.
2150 * doc/invoke.texi (Instrumentation Options): Add cross-references
2151 to docs for -p, -pg, and -finstrument-functions.
2153 2018-11-18 Alan Modra <amodra@gmail.com>
2155 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
2156 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
2157 (GNU_USER_TARGET_CRTN): Define.
2158 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
2159 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
2161 2018-11-18 Alan Modra <amodra@gmail.com>
2163 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
2164 !HAVE_LD_PIE variant.
2165 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2167 2018-11-17 Nick Clifton <nickc@redhat.com>
2168 Sandra Loosemore <sandra@codesourcery.com>
2171 * doc/invoke.texi (Overall Options): Document that --help and
2172 --help=<value> options cannot be combined.
2174 2018-11-17 Jim Wilson <jimw@sifive.com>
2176 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
2177 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
2178 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
2179 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
2180 to step2 if frame_pointer_needed and step1 is 0.
2182 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
2185 * doc/cpp.texi (System Headers): Add note about implicit
2186 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
2188 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
2191 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
2192 -fno-implicit-templates.
2194 2018-11-17 Jakub Jelinek <jakub@redhat.com>
2196 PR tree-optimization/87546
2197 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
2198 min_precision variable, initially set it to orig_precision, only does
2199 something if op_type's precision is <= min_precision and update
2200 min_precision whenever calling set_op.
2202 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
2205 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
2207 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
2209 * doc/invoke.texi (Option Summary): Fix whitespace and line
2210 breaks in @gccoptlist environments.
2211 (Warning Options): Likewise.
2212 (Optimize Options): Likewise.
2213 (PowerPC SPE Options): Likewise.
2214 (RS/6000 and PowerPC Options): Likewise.
2216 2018-11-16 Jeff Law <law@redhat.com>
2218 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
2221 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
2224 * doc/invoke.texi (Optimize Options): Update options enabled by
2225 fprofile-generate, -fprofile-use, and -fauto-profile.
2227 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2230 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
2231 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
2232 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
2234 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
2235 * targhooks.c (default_print_patchable_function_entry): Assert
2236 asm_int_op does not return a NULL string.
2237 * varasm.c (integer_asm_op): Return the op for a partial int type
2238 when the requested size does not correspond to an integer type.
2239 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
2240 * doc/tm.texi: Regenerate.
2242 2018-11-12 Jason Merrill <jason@redhat.com>
2244 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
2246 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
2248 * config/rs6000/constraints.md (wF constraint): Remove power9
2249 fusion documentation. Just document wF's use for power8 fusion.
2250 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
2251 (fusion_gpr_addis): Delete power9 fusion support. Change power8
2252 fusion support to require the upper 12 bits to be all 0's or all
2254 (fusion_gpr_mem_load): Add comment.
2255 (fusion_addis_mem_combo_load): Remove power9 fusion support.
2256 (fusion_addis_mem_combo_store): Delete.
2257 (fusion_offsettable_mem_operand): Delete.
2258 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
2260 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
2261 (POWERPC_MASKS): Delete power9 fusion.
2262 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
2263 (fusion_p9_p): Delete.
2264 (expand_fusion_p9_load): Delete.
2265 (expand_fusion_p9_store): Delete.
2266 (emit_fusion_p9_load): Delete.
2267 (emit_fusion_p9_store): Delete.
2268 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
2270 (rs6000_option_override_internal): Set power8 fusion based on
2271 whether we are tuning for power8. Delete power9 fusion support.
2272 (rs6000_opt_masks): Delete -mpower9-fusion switch.
2273 (emit_fusion_load): Rename emit_fusion_load_store to
2274 emit_fusion_load, and drop fusion store support. Update callers.
2275 (emit_fusion_load_store): Likewise.
2276 (emit_fusion_gpr_load): Likewise.
2277 (fusion_p9_p): Delete.
2278 (expand_fusion_p9_load): Delete.
2279 (expand_fusion_p9_store): Delete.
2280 (emit_fusion_p9_load): Delete.
2281 (emit_fusion_p9_store): Delete.
2282 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
2283 (GPR_FUSION): Delete.
2284 (FPR_FUSION): Delete.
2285 (power9 fusion peephole2s): Delete.
2286 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
2287 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
2288 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
2289 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
2290 (fusion_p9_<mode>_constant): Delete.
2291 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
2292 power9 fusion switch.
2293 * doc/md.texi (PowerPC constraints): Update wF constraint
2294 documentation for power8 fusion only.
2296 2018-11-16 Jakub Jelinek <jakub@redhat.com>
2298 PR rtl-optimization/87475
2299 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
2300 CROSSING_JUMP_P insns.
2301 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
2303 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
2306 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
2307 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
2308 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
2310 2018-11-16 Jakub Jelinek <jakub@redhat.com>
2313 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
2314 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
2316 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
2318 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
2319 location_t parameter.
2320 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
2321 (switch_decision_tree::do_jump_if_equal): Likewise.
2322 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
2323 of switch statement to emit_case_nodes.
2324 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
2325 set it on the newly built GIMPLE comparison statement.
2326 (switch_decision_tree::do_jump_if_equal): Likewise.
2327 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
2328 in calls to do_jump_if_equal as well as recursive calls.
2330 2018-11-16 Andi Kleen <ak@linux.intel.com>
2332 * config/i386/i386.md: Allow memory operands to ptwrite.
2334 2018-11-16 Michael Matz <matz@suse.de>
2336 * system.h (PRsa): New macro.
2337 (SIZE_AMOUNT): Cast number to uint64_t.
2338 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
2339 (pool_usage::dump_footer): Likewise and also use PRIu64.
2340 * bitmap.h (bitmap_usage::dump): Likewise.
2341 * ggc-common.c (ggc_usage::dump): Likewise.
2342 * ggc-page.c (ggc_print_statistics): Likewise.
2343 * input.c (dump_line_table_statistics): Likewise.
2344 * mem-stats.h (mem_usage::dump): Likewise.
2345 (mem_usage::dump_footer): Likewise.
2346 * rtl.c (dump_rtx_statistics): Likewise.
2347 * tree-cfg.c (dump_cfg_stats): Likewise.
2348 * tree-dfa.c (dump_dfa_stats): Likewise.
2349 * tree-phinodes.c (phinodes_print_statistics): Likewise.
2350 * tree-ssanames (ssanames_print_statistics): Likewise.
2351 * vec.c (vec_usage::dump): Likewise.
2352 (vec_usage::dump_footer): Likewise.
2354 2018-11-16 Richard Biener <rguenther@suse.de>
2356 PR tree-optimization/88011
2357 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
2358 replacing set_value_range_to_undefined and
2359 set_value_range_to_varying with method calls.
2361 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
2363 * config/s390/s390.md
2364 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
2366 (*r<noxa>sbg_<mode>_sll): Likewise.
2367 (*r<noxa>sbg_<mode>_srl): Likewise.
2368 (*r<noxa>sbg_sidi_srl): New pattern.
2370 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
2372 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
2373 Default to TARGET_VXWORKS_RTP.
2374 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
2375 of TARGET_VXWORKS_RTP.
2376 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
2377 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
2378 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
2380 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
2383 * doc/extend.texi (Common Type Attributes): Make it explicit
2384 that attribute "packed" can apply to C++ classes.
2386 2018-11-15 Martin Sebor <msebor@redhat.com>
2390 * tree.c (type_argument_type): New function.
2391 * tree.h (type_argument_type): Declare it.
2392 * gcc/doc/extend.texi (alloc_align): Update and clarify.
2393 (alloc_size, nonnull, sentinel): Same.
2395 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
2396 Kwok Cheung Yeung <kcy@codesourcery.com>
2398 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
2399 int_mode_for_size fails.
2400 (vectorizable_load): Likewise.
2402 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2404 * doc/ux.texi (Group logically-related diagnostics): Move
2405 discussion of auto_diagnostic_group into this new subsection.
2406 Give an example of where this grouping is used.
2408 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2411 * Makefile.in (OBJS): Move json.o to...
2412 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
2413 * common.opt (fdiagnostics-format=): New option.
2414 (diagnostics_output_format): New enum.
2415 * diagnostic-format-json.cc: New file.
2416 * diagnostic.c (default_diagnostic_final_cb): New function, taken
2417 from start of diagnostic_finish.
2418 (diagnostic_initialize): Initialize final_cb to
2419 default_diagnostic_final_cb.
2420 (diagnostic_finish): Move "being treated as errors" messages to
2421 default_diagnostic_final_cb. Call any final_cb.
2422 (default_diagnostic_finalizer): Add diagnostic_t param.
2423 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
2424 diagnostic_finalizer callback.
2425 * diagnostic.h (enum diagnostics_output_format): New enum.
2426 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
2427 (struct diagnostic_context): Add "final_cb".
2428 (default_diagnostic_finalizer): Add diagnostic_t param.
2429 (diagnostic_output_format_init): New decl.
2430 * doc/invoke.texi (-fdiagnostics-format): New option.
2431 * dwarf2out.c (gen_producer_string): Ignore
2432 OPT_fdiagnostics_format_.
2433 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
2434 * lto-wrapper.c (append_diag_options): Ignore it.
2435 * opts.c (common_handle_option): Handle it.
2437 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2439 PR tree-optimization/88015
2440 * graphite-isl-ast-to-gimple.c
2441 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
2443 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
2445 2018-11-15 Richard Biener <rguenther@suse.de>
2448 * gimple.c (gimple_call_flags): Union flags from decl, type
2450 * trans-mem.c (is_tm_pure_call): Simplify.
2452 2018-11-15 Richard Biener <rguenther@suse.de>
2454 PR tree-optimization/88031
2455 * tree-vect-loop.c (vectorizable_reduction): Move check
2456 for multiple types earlier so we get the expected dump.
2457 Simplify calls to vectorizable_condition.
2458 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
2459 * tree-vect-stmts.c (vectorizable_condition): Instead of
2460 reduc_def and reduc_index take just a flag. Simplify
2461 code-generation now that we can rely on the defs being set up.
2462 (vectorizable_comparison): Remove unused argument.
2464 2018-11-15 Richard Biener <rguenther@suse.de>
2466 PR tree-optimization/88030
2467 * tree-complex.c (need_eh_cleanup): New global.
2468 (update_complex_assignment): Mark blocks that need EH update.
2469 (expand_complex_comparison): Likewise.
2470 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
2471 perform EH cleanup and schedule CFG cleanup if that did anything.
2473 2018-11-15 Jakub Jelinek <jakub@redhat.com>
2475 PR rtl-optimization/88018
2476 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
2477 on the presence of fallthru edge, rather than if it is a USE or not.
2479 2018-11-15 Richard Biener <rguenther@suse.de>
2482 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
2483 analyze_subscript_affine_affine properly.
2485 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
2487 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
2488 (GLIBC_DYNAMIC_LINKER): Define.
2489 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
2491 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2493 PR tree-optimization/84648
2494 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
2495 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
2496 by calling adjust_cond_for_loop_until_wrap.
2498 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
2501 * doc/extend.texi (Common Function Attributes): Clarify linker
2502 restrictions on "aligned" attribute.
2503 (Common Variable Attributes): Likewise. Mention that linker
2504 restrictions don't apply to stack-allocated variables.
2506 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
2508 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
2511 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
2515 * doc/invoke.texi (Optimize Options): Remove bad example about
2516 interaction between -flto and -O. Replace it with a note that
2517 you need to compile with -O and not just link. Copy-edit -flto
2518 discussion to reduce verbiage and improve flow.
2520 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
2523 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
2524 also suppress many optimizations. Alphabetize option lists for
2525 -O1, -O2, and -Os. Add list of options disabled with -Og, and
2526 correct documentation for those options to say that.
2527 * opts.c (default_options_table): Sort table by level and option
2528 name, to make it easier to correlate to the manual.
2530 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
2532 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
2533 printing of the names of segment registers.
2535 2018-11-14 Nathan Sidwell <nathan@acm.org>
2539 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
2542 2018-11-14 David Malcolm <dmalcolm@redhat.com>
2544 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
2546 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2548 P1236R1 - Signed integers are two's complement
2549 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
2552 * hash-map.h (hash_map::iterator::reference_pair): New class.
2553 (hash_map::iterator::operator*): Return it rather than std::pair.
2555 2018-11-14 Jeff Law <law@redhat.com>
2557 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
2558 iff the operand is a constant.
2560 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
2562 * gimple-ssa-evrp-analyze.c
2563 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
2564 ignore_equivs_equal_p to equal_p.
2565 * ipa-cp.c (meet_with_1): Use equal_p instead of
2566 ignore_equivs_equal_p.
2567 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
2568 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
2569 (value_range::operator==): Remove.
2570 (value_range::operator!=): Remove.
2571 (vrp_prop::visit_stmt): Use equal_p.
2572 * tree-vrp.h (value_range): Remove operator==, operator!=,
2573 ignore_equivs_equal_p.
2574 * vr-values.c (update_value_range): Use equal_p.
2576 2018-11-14 Michael Matz <matz@suse.de>
2579 * gimplify.c (collect_fallthrough_labels): Add new argument,
2580 return location via that, don't modify statements.
2581 (warn_implicit_fallthrough_r): Adjust call, don't use
2582 statement location directly.
2584 2018-11-14 Richard Biener <rguenther@suse.de>
2587 * tree-data-ref.c (split_constant_offset): Add wrapper
2588 allocating a cache hash-map.
2589 (split_constant_offset_1): Cache results of expanding
2590 expressions from SSA def stmts.
2592 2018-11-14 Richard Biener <rguenther@suse.de>
2595 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
2597 (lambda_vector_mult_const): Likewise.
2598 (lambda_matrix_right_hermite): Use lambda_int temporaries.
2600 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
2601 Jackson Woodruff <jackson.woodruff@arm.com>
2603 PR 71026/tree-optimization
2604 * match.pd: Simplify floating point comparisons.
2606 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2608 PR rtl-optimization/87817
2609 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
2610 last argument has low 8 bits clear, fold to 0.
2612 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
2614 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
2615 MAX_OFILE_ALIGNMENT as needed.
2617 2018-11-14 Richard Biener <rguenther@suse.de>
2619 PR tree-optimization/88019
2620 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
2621 COND_EXPR arguments to non-trapping overflow.
2623 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
2625 * config/arc/arc.h (reg_class): Reorder registers classes, remove
2626 unused register classes.
2627 (REG_CLASS_NAMES): Likewise.
2628 (REG_CLASS_CONTENTS): Likewise.
2629 (FIXED_REGISTERS): Make lp_count fixed.
2630 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
2631 (PROGRAM_COUNTER_REGNO): Remove.
2632 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
2633 register classes, use constants for register numbers, remove
2634 reg_class_contents references.
2635 (arc_process_double_reg_moves): Add asserts.
2636 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
2637 lra_in_progress predicate.
2638 (arc_init_reg_tables): Remove unused register classes.
2639 (arc_register_move_cost): Likewise.
2640 (arc_preferred_reload_class): Likewise.
2641 (hwloop_optimize): Update rtx patterns involving lp_count register.
2642 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
2643 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
2644 CHEAP_CORE_REGS. Former one will be used for LRA.
2645 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
2646 one will be used for LRA.
2647 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
2648 one will be used for LRA.
2649 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
2650 Former one will be used for LRA.
2651 ("f"): Delete constraint.
2654 ("l"): Change it from register constraint to constraint.
2655 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
2656 (movhi_insn): Likewise.
2657 (movsi_insn): Update pattern.
2660 (stack_tie): Remove 'b' constraint letter.
2662 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
2663 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
2664 (R57_REG, R59_REG, PCL_REG): Likewise.
2665 (ILINK1_REGNUM): Renamed to ILINK1_REG.
2666 (ILINK2_REGNUM): Renamed to ILINK2_REG.
2668 (SP_REGS): Likewise.
2669 (Rcw): Remove unused reg classes.
2670 * config/arc/predicates.md (dest_reg_operand): Just default on
2671 register_operand predicate.
2672 (mpy_dest_reg_operand): Likewise.
2673 (move_dest_operand): Use macros instead of constants.
2674 * config/arc/arc.opt (mlra): Switch to lra as default.
2676 2018-11-14 Richard Biener <rguenther@suse.de>
2678 PR tree-optimization/87974
2679 * tree-vect-loop.c (vectorizable_reduction): When computing
2680 the vectorized reduction PHI vector type ignore constant
2683 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2685 PR tree-optimization/87977
2686 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
2687 stmt, build a new one and replace the old one with it. Formatting fix.
2688 Call release_ssa_name (x) if !has_other_use and !delete_div.
2689 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
2690 verify lhs of stmt is still def.
2692 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2694 PR rtl-optimization/87507
2695 * lower-subreg.c (operand_for_swap_move_operator): New function.
2696 (simple_move): Strip simple operators.
2697 (find_pseudo_copy): Likewise.
2698 (resolve_operand_for_swap_move_operator): New function.
2699 (resolve_simple_move): Strip simple operators and swap operands.
2701 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2703 PR tree-optimization/87898
2704 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
2705 (ipa_simd_modify_function_body): Remove debug stmts where the first
2706 argument was changed into a non-decl.
2708 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2710 PR rtl-optimization/87899
2711 * lra-lives.c (start_living): Update white space in comment.
2712 (enum point_type): New.
2713 (sparseset_contains_pseudos_p): New function.
2714 (update_pseudo_point): Likewise.
2715 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
2716 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
2717 handling. Move early exit after adding conflicts.
2718 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
2719 if regno is already live. Remove all handling of program points.
2720 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
2721 after adding conflicts. Remove all handling of program points and
2722 ignore_reg_for_conflicts.
2723 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
2724 and do not guard call to mark_pseudo_live.
2725 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
2726 and do not guard call to mark_pseudo_dead.
2727 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
2728 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
2729 Use new function update_pseudo_point. Handle register copies by
2730 removing the source register from the live set. Handle INOUT operands.
2731 Update to the next program point using the unused_set, dead_set and
2733 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
2735 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2737 * builtins.c: Replace "source_location" with "location_t".
2738 * diagnostic-show-locus.c: Likewise.
2739 * diagnostic.c: Likewise.
2740 * dumpfile.c: Likewise.
2741 * gcc-rich-location.h: Likewise.
2742 * genmatch.c: Likewise.
2743 * gimple.h: Likewise.
2744 * gimplify.c: Likewise.
2745 * input.c: Likewise.
2746 * input.h: Likewise. Eliminate the typedef.
2747 * omp-expand.c: Likewise.
2748 * selftest.h: Likewise.
2749 * substring-locations.h (get_source_location_for_substring):
2751 (get_location_within_string): ...this.
2752 * tree-cfg.c: Replace "source_location" with "location_t".
2753 * tree-cfgcleanup.c: Likewise.
2754 * tree-diagnostic.c: Likewise.
2755 * tree-into-ssa.c: Likewise.
2756 * tree-outof-ssa.c: Likewise.
2757 * tree-parloops.c: Likewise.
2758 * tree-phinodes.c: Likewise.
2759 * tree-phinodes.h: Likewise.
2760 * tree-ssa-loop-ivopts.c: Likewise.
2761 * tree-ssa-loop-manip.c: Likewise.
2762 * tree-ssa-phiopt.c: Likewise.
2763 * tree-ssa-phiprop.c: Likewise.
2764 * tree-ssa-threadupdate.c: Likewise.
2765 * tree-ssa.c: Likewise.
2766 * tree-ssa.h: Likewise.
2767 * tree-vect-loop-manip.c: Likewise.
2769 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2771 * doc/extend.texi: Fix typo in the weakref description.
2773 2018-11-13 Richard Biener <rguenther@suse.de>
2775 PR tree-optimization/86991
2776 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
2777 group building until we have successfully detected the SLP
2779 (vect_is_simple_reduction): Remove fixup code here.
2781 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2783 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
2784 (dump_gimple_stmt): Use it.
2785 (dump_gimple_stmt_loc): Likewise.
2786 (dump_gimple_expr): Likewise.
2787 (dump_gimple_expr_loc): Likewise.
2788 (dump_generic_expr): Likewise.
2789 (dump_generic_expr_loc): Likewise.
2790 (dump_printf): Likewise.
2791 (dump_printf_loc): Likewise.
2792 (dump_dec): Likewise.
2793 (dump_dec): Likewise.
2794 (dump_hex): Likewise.
2795 (dump_symtab_node): Likewise.
2796 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2797 Guard dump call with dump_enabled_p.
2798 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
2799 * graphite-optimize-isl.c (optimize_isl): Likewise.
2800 * graphite.c (graphite_transform_loops): Likewise.
2801 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
2802 * tree-parloops.c (parallelize_loops): Likewise.
2803 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
2804 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
2805 (vect_prune_runtime_alias_test_list): Likewise.
2806 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
2807 (vect_estimate_min_profitable_iters): Likewise.
2808 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
2809 (vect_build_slp_tree_2): Likewise.
2810 (vect_supported_load_permutation_p): Likewise.
2811 (vect_slp_analyze_operations): Likewise.
2812 (vect_slp_analyze_bb_1): Likewise.
2813 (vect_slp_bb): Likewise.
2814 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
2815 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
2816 (pass_slp_vectorize::execute): Likewise.
2817 (increase_alignment): Likewise.
2819 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2822 * ipa-inline.c (report_inline_failed_reason): Guard calls to
2823 cl_target_option_print_diff and cl_optimization_print_diff with
2826 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2828 * doc/invoke.texi (-fsave-optimization-record): Note that the
2829 output is compressed.
2830 * optinfo-emit-json.cc: Include <zlib.h>.
2831 (optrecord_json_writer::write): Compress the output.
2833 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
2835 * tree-vrp.c (value_range_base::dump): Dump type.
2836 Do not use INF nomenclature for 1-bit types.
2837 (dump_value_range): Group all variants to common dumping code.
2838 (debug): New overloaded functions for value_ranges.
2839 (value_range_base::dump): Remove no argument version.
2840 (value_range::dump): Same.
2842 2018-11-13 Richard Biener <rguenther@suse.de>
2844 PR tree-optimization/87931
2845 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
2846 nested cycles we support to latch computations vectorizable_reduction
2849 2018-11-13 Martin Liska <mliska@suse.cz>
2851 PR tree-optimization/87885
2852 * cfghooks.c (account_profile_record): Rename
2854 (profile_record_check_consistency): ... this.
2855 Calculate missing num_mismatched_freq_in.
2856 (profile_record_account_profile): New function
2857 that calculates time and size of a function.
2858 * cfghooks.h (struct profile_record): Remove
2860 (struct cfg_hooks): Remove after_pass flag.
2861 (account_profile_record): Rename to ...
2862 (profile_record_check_consistency): ... this.
2863 (profile_record_account_profile): New.
2864 * cfgrtl.c (rtl_account_profile_record): Remove
2866 * passes.c (check_profile_consistency): Do only
2868 (account_profile): Calculate size and time of
2870 (pass_manager::dump_profile_report): Reformat
2872 (execute_one_ipa_transform_pass): Call
2873 consistency check before clean upand call account_profile
2874 after a clean up is done.
2875 (execute_one_pass): Call check_profile_consistency and
2876 account_profile instead of using after_pass flag..
2877 * tree-cfg.c (gimple_account_profile_record): Likewise.
2879 2018-11-13 Martin Liska <mliska@suse.cz>
2882 * config/i386/i386.c (ix86_option_override_internal): Error
2883 about usage -mabi=ms and -fsanitize={,kernel-}address.
2885 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
2887 * config/aarch64/aarch64.c
2888 (aarch64_vectorize_preferred_vector_alignment): Change return type to
2890 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
2891 alignment being a poly int.
2892 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
2893 return type to poly_uint64.
2894 * target.def (default_preferred_vector_alignment): Likewise.
2895 * targhooks.c (default_preferred_vector_alignment): Likewise.
2896 * targhooks.h (default_preferred_vector_alignment): Likewise.
2897 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
2898 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
2900 (vect_update_misalignment_for_peel): Likewise.
2901 (vect_enhance_data_refs_alignment): Likewise.
2902 (vect_find_same_alignment_drs): Likewise.
2903 (vect_duplicate_ssa_name_ptr_info): Likewise.
2904 (vect_setup_realignment): Likewise.
2905 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2907 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
2908 mask with a compile time variable vector alignment.
2909 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
2911 (vect_do_peeling): Exit early if vector alignment is not constant.
2912 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
2914 (vectorizable_store): Likewise.
2915 (vectorizable_load): Likweise.
2916 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
2918 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
2920 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2923 2018-11-13 Richard Biener <rguenther@suse.de>
2925 PR tree-optimization/87962
2926 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
2927 detect outer reduction for disqualifying in-loop uses.
2929 2018-11-13 Richard Biener <rguenther@suse.de>
2931 PR tree-optimization/87967
2932 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
2933 for constants for the scalar loop.
2935 2018-11-13 Alan Modra <amodra@gmail.com>
2937 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
2938 integer constants when -mcmodel=medium.
2940 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2942 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
2943 * config/arc/elf.h (PROFILE_HOOK): Define.
2944 * config/arc/linux.h (PROFILE_HOOK): Likewise.
2946 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2948 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
2949 jump table data in the text section.
2951 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2953 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
2954 to fit the eh_return pattern.
2955 * config/arc/arc.md (eh_return): Define.
2956 (VUNSPEC_ARC_EH_RETURN): Likewise.
2957 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
2959 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
2962 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2964 * common/config/arc/arc-common.c (arc_option_optimization_table):
2965 Millicode optimization is default on for size optimizations.
2966 * config/arc/arc-protos.h (arc_check_multi): New function.
2967 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
2968 (ENTER_LEAVE_START_REG): Define.
2969 (ENTER_LEAVE_END_REG): Likewise.
2970 (arc_override_options): Disable millicode when long calls option
2972 (arc_frame_info): Change it from int to bool.
2973 (arc_compute_frame_size): Clean up.
2974 (arc_save_restore): Remove.
2975 (frame_save_reg): New function.
2976 (frame_restore_reg): Likewise.
2977 (arc_enter_leave_p): Likewise.
2978 (arc_save_callee_saves): Likewise.
2979 (arc_restore_callee_saves): Likewise.
2980 (arc_save_callee_enter): Likewise.
2981 (arc_restore_callee_leave): Likewise.
2982 (arc_save_callee_milli): Likewise.
2983 (arc_restore_callee_milli): Likewise.
2984 (arc_expand_prologue): Reimplement to emit enter/leave
2986 (arc_expand_epilogue): Likewise.
2987 (arc_check_multi): New function.
2988 * config/arc/arc.md (push_multi_fp): New pattern.
2989 (push_multi_fp_blink): Likewise.
2990 (pop_multi_fp): Likewise.
2991 (pop_multi_fp_blink): Likewise.
2992 (pop_multi_fp_ret): Likewise.
2993 (pop_multi_fp_blink_ret): Likewise.
2994 * config/arc/arc.opt (mmillicode): Update option.
2995 (mcode-density-frame): New option.
2996 * config/arc/predicates.md (push_multi_operand): New predicate.
2997 (pop_multi_operand): Likewise.
2998 * doc/invoke.texi (ARC): Update ARC options information.
3001 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
3003 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
3004 * config/arc/arc.c (operands_ok_ldd_std): New function.
3005 (mem_ok_for_ldd_std): Likewise.
3006 (gen_operands_ldd_std): Likewise.
3007 * config/arc/arc.md: Add peephole2 rules for std/ldd.
3009 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
3011 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
3012 test on stack_usage_file.
3013 (lang_dependent_init): Do not open the .su file if generating LTO.
3015 2018-11-13 Jakub Jelinek <jakub@redhat.com>
3017 PR rtl-optimization/87918
3018 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
3019 simplify_gen_relational rather than simplify_gen_binary.
3021 2018-11-13 Richard Biener <rguenther@suse.de>
3023 * tree-ssanames.h (set_range_info): Use value_range_base.
3024 (get_range_info): Likewise.
3025 * tree-ssanames.c (set_range_info): Likewise.
3026 (get_range_info): Likewise.
3027 * tree-vrp.c (value_range_base::union_helper): Split
3028 out common parts of value_range[_base]::union_.
3029 (value_range_base::union_): Update.
3030 (value_range::union_): Likewise.
3031 (determine_value_range_1): Use value_range_base.
3032 (determine_value_range): Likewise.
3033 * tree-vrp.h (value_range_base::union_helper): Move ...
3034 (value_range::union_helper): ... from here.
3036 2018-11-13 Alan Modra <amodra@gmail.com>
3038 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
3040 (rs6000_secondary_reload_gpr): Don't call find_replacement.
3042 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
3045 * doc/invoke.texi (Optimize Options): Clarify that the
3046 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
3047 apply to data cache size.
3049 2018-11-13 Alan Modra <amodra@gmail.com>
3051 * config/rs6000/predicates.md (logical_const_operand),
3052 (logical_operand): Correct comment.
3053 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
3055 2018-11-13 Alan Modra <amodra@gmail.com>
3057 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
3059 2018-11-13 Alan Modra <amodra@gmail.com>
3061 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
3063 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
3064 covered by alternative.
3065 (movcc_internal1): Ignore h for register preference.
3066 (mov<mode>_hardfloat64): Likewise.
3067 (mov<mode>_softfloat): Ignore c, l, h for register preference.
3069 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
3071 PR preprocessor/47823
3072 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
3075 * doc/extend.texi (Pragmas): Note additional pragmas documented
3078 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3082 * tree-scalar-evolution.c (expression_expensive_p): Make
3083 BUILTIN POPCOUNT as expensive when backend does not define it.
3085 2018-11-12 Fredrik Noring <noring@nocrew.org>
3087 * config.gcc: Update with-llsc defaults for MIPS r5900.
3089 2018-11-12 Martin Liska <mliska@suse.cz>
3091 PR gcov-profile/87442
3092 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
3094 * doc/invoke.texi: Document them.
3095 * tree-profile.c (parse_profile_filter): New.
3096 (parse_profile_file_filtering): Likewise.
3097 (release_profile_file_filtering): Likewise.
3098 (include_source_file_for_profile): Likewise.
3099 (tree_profiling): Filter source files based on the
3100 newly added options.
3102 2018-11-12 Martin Liska <mliska@suse.cz>
3105 * doc/extend.texi: Add missing values for __builtin_cpu_is and
3106 __builtin_cpu_supports for x86 target.
3108 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
3111 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
3113 (Arithmetic): Likewise for "lo_sum".
3115 2018-11-12 Renlin Li <renlin.li@arm.com>
3118 * dse.c (get_stored_val): Add check for compile-time constantness
3121 2018-11-12 Sudakshina Das <sudi.das@arm.com>
3123 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
3124 (ARMv8_5a): New fgroup.
3125 (armv8.5-a): New arch.
3126 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
3128 * config/arm/arm-tables.opt: Regenerate.
3129 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
3130 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
3131 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
3132 Add matching rules for -march=armv8.5-a and extensions.
3133 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
3134 Add sb and predres to all armv8-a except armv8.5-a.
3136 2018-11-12 Richard Biener <rguenther@suse.de>
3138 * tree-vrp.h (value_range[_base]::set): Make public. Provide
3139 overload for single value.
3140 (value_range[_base]::set_nonnull): New.
3141 (value_range[_base]::set_null): Likewise.
3142 (value_range): Document bitmap copying behavior, mark
3143 copy constructor and assignment operator deleted.
3144 (value_range::move): New.
3145 (value_range::set_and_canonicalize): Default bitmap to zero.
3146 (set_value_range_to_nonnull): Remove.
3147 (set_value_range_to_null): Likewise.
3148 (set_value_range): Likewise.
3149 (set_value_range_to_value): Likewise.
3150 (extract_range_from_unary_expr): Work on value_range_base.
3151 (extract_range_from_binary_expr_1): Likewise. Rename to...
3152 (extract_range_from_binary_expr): ... this.
3153 * tree-vrp.c (value_range::update): Clear equiv bitmap
3155 (value_range::move): New, move equiv bitmap.
3156 (value_range_base::set_undefined): Avoid assignment.
3157 (value_range::set_undefined): Likewise.
3158 (value_range_base::set_varying): Likewise.
3159 (value_range::set_varying): Likewise.
3160 (set_value_range): Remove.
3161 (value_range_base::set): New overload for value.
3162 (value_range::set): Likewise.
3163 (set_value_range_to_nonnull): Remove.
3164 (value_range_base::set_nonnull): New.
3165 (value_range::set_nonnull): Likewise.
3166 (set_value_range_to_null): Remove.
3167 (value_range_base::set_null): New.
3168 (value_range::set_null): Likewise.
3169 (range_is_null): Work on value_range_base.
3170 (range_is_nonnull): Likewise.
3171 (ranges_from_anti_range): Likewise.
3172 (extract_range_into_wide_ints): Likewise.
3173 (extract_range_from_multiplicative_op): Likewise.
3174 (extract_range_from_binary_expr): Likewise. Update for API changes.
3175 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
3177 (value_range::intersect_helper): Avoid copy and assignment.
3178 (value_range::union_helper): Likewise.
3179 (determine_value_range_1): Adjust.
3180 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3181 Avoid assignment by using move.
3182 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
3183 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3185 * tree-ssanames.c (get_range_info): Likewise.
3186 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
3187 * vr-values.c (vr_values::get_value_range): Adjust.
3188 (vr_values::update_value_range): Likewise.
3189 (symbolic_range_based_on_p): Work on value_range_base.
3190 (vr_values::extract_range_from_binary_expr): Use value_range_base.
3191 (vr_values::extract_range_from_unary_expr): Likewise.
3192 (vr_values::extract_range_from_cond_expr): Avoid assignment.
3193 (vr_values::extract_range_from_comparison): Adjust.
3194 (vr_values::check_for_binary_op_overflow): Use value_range_base.
3195 (vr_values::extract_range_basic): Adjust.
3196 (vr_values::adjust_range_with_scev): Likewise.
3197 (vr_values::vrp_visit_assignment_or_call): Likewise.
3198 (vr_values::get_vr_for_comparison): Change API to avoid
3199 assignment and copy construction.
3200 (vr_values::compare_name_with_value): Adjust accordingly.
3201 (vr_values::compare_names): Likewise.
3202 (vr_values::extract_range_from_phi_node): Avoid assignment and
3203 bogus in-place modify of equiv bitmap.
3204 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
3205 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
3206 for extract_range_from_unary_expr API change.
3207 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
3209 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
3211 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
3213 2018-11-12 Richard Biener <rguenther@suse.de>
3215 * tree-vrp.h (value_range_base::symbolic_p,
3216 value_range_base::constant_p, value_range_base::zero_p,
3217 value_range_base::singleton_p): Move from value_range.
3218 (value_range::dump): Add.
3219 * gimple-ssa-evrp-analyze.c
3220 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
3221 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
3222 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3224 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
3225 (value_range::constant_p): Likewise.
3226 (value_range::singleton_p): Likewise.
3227 (value_range_base::dump): Add.
3228 (set_value_range_to_undefined): Remove.
3229 (set_value_range_to_varying): Likewise.
3230 (range_int_cst_p): Take value_range_base argument.
3231 (range_int_cst_singleton_p): Likewise.
3232 (value_range_constant_singleton): Likewise.
3233 (vrp_set_zero_nonzero_bits): Likewise.
3234 (extract_range_from_multiplicative_op): Use set_varying.
3235 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
3236 (extract_range_from_unary_expr): Likewise.
3237 (dump_value_range_base): Change to overload of dump_value_range.
3238 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
3239 (vrp_prop::visit_stmt): Likewise.
3240 (value_range::intersect_helper): Likewise.
3241 (value_range::union_helper): Likewise.
3242 (determine_value_range_1): Likewise.
3244 2018-11-12 Richard Biener <rguenther@suse.de>
3246 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
3247 (set_value_range_to_null): Likewise.
3248 * vr-values.c (vr_values::extract_range_from_comparison):
3249 Clear equiv for constant singleton ranges.
3251 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
3253 * config/i386/sse.md: Combine VFIXUPIMM* patterns
3254 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3255 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3256 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
3257 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3258 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3259 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
3261 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3264 * doc/extend.texi (Common Type Attributes): For the align type
3265 attribute, copy language about decreasing alignment from the
3266 corresponding variable attribute.
3268 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3270 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
3273 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3276 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
3277 -frtti and -fno-rtti code.
3279 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3282 * doc/extend.texi (Other Builtins): Document probability associated
3283 with __builtin_expect.
3285 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
3288 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
3289 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
3290 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
3291 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
3293 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3295 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
3297 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3299 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
3301 2018-11-11 Richard Biener <rguenther@suse.de>
3303 * tree-vrp.h (class value_range_base): New base class for
3304 value_range containing all but the m_equiv member.
3305 (dump_value_range_base): Add.
3306 (range_includes_zero_p): Work on value_range_base.
3307 * tree-vrp.c (value_range_base::set): Split out base handling
3309 (value_range::set): this.
3310 (value_range::set_equiv): New.
3311 (value_range_base::value_range_base): New constructors.
3312 (value_range_base::check): Split out base handling from...
3313 (value_range::check): this.
3314 (value_range::equal_p): Refactor in terms of
3315 ignore_equivs_equal_p which is now member of the base.
3316 (value_range_base::set_undefined): New.
3317 (value_range_base::set_varying): Likewise.
3318 (value_range_base::dump):Split out base handling from...
3319 (value_range::dump): this.
3320 (value_range_base::set_and_canonicalize): Split out base handling
3322 (value_range::set_and_canonicalize): this.
3323 (value_range_base::union_): New.
3324 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
3326 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
3327 instead of value_range everywhere.
3328 (ipcp_vr_lattice::print): Use dump_value_range_base.
3329 (ipcp_vr_lattice::meet_with): Adjust.
3330 (ipcp_vr_lattice::meet_with_1): Likewise.
3331 (ipa_vr_operation_and_type_effects): Likewise.
3332 (propagate_vr_across_jump_function): Likewise.
3333 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
3334 (ipa_get_value_range): Likewise.
3335 (ipa_set_jfunc_vr): Likewise.
3336 (ipa_compute_jump_functions_for_edge): Likewise.
3338 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
3341 * doc/invoke.texi (Optimize Options): Add @opindex entries
3342 for the positive forms of -fno-xxx and -mno-xxx options
3343 that were lacking them.
3345 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
3347 * combine.c (make_more_copies): Only make an intermediate copy if the
3348 dest of a move is a pseudo.
3350 2018-11-09 Maya Rashish <coypu@sdf.org>
3353 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
3354 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
3356 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
3360 * doc/invoke.texi (Optimize Options): Clarify default behavior
3361 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
3363 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
3365 PR tree-optimization/87940
3366 * expr.c (string_constant): Don't strip NOPS in subexpressions.
3367 Fold PLUS_EXPR correctly.
3369 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
3372 * config/s390/s390.c (s390_safe_relative_long_p): New function.
3373 (annotate_constant_pool_refs): Skip insns which support
3374 relative addressing.
3375 (annotate_constant_pool_refs_1): New helper function.
3376 (find_constant_pool_ref): Skip insns which support relative
3378 (find_constant_pool_ref_1): New helper function.
3379 (replace_constant_pool_ref): Skip insns which support
3380 relative addressing.
3381 (replace_constant_pool_ref_1): New helper function.
3382 (s390_mainpool_start): Adapt to the new signature.
3383 (s390_mainpool_finish): Likewise.
3384 (s390_chunkify_start): Likewise.
3385 (s390_chunkify_finish): Likewise.
3386 (pass_s390_early_mach::execute): Likewise.
3387 (s390_prologue_plus_offset): Likewise.
3388 (s390_emit_prologue): Likewise.
3389 (s390_emit_epilogue): Likewise.
3391 2018-11-09 Jakub Jelinek <jakub@redhat.com>
3393 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
3394 but unsupported lastprivate with conditional modifier.
3396 2018-11-09 Jeff Law <law@redhat.com>
3398 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
3399 unused argument better. Add gcc_unreachable to silence warning.
3401 2018-11-09 Martin Sebor <msebor@redhat.com>
3404 * attribs.c (has_attribute): New helper function.
3405 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
3406 * attribs.h (decls_mismatched_attributes): Declare.
3407 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
3408 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
3409 * common.opt (-Wattribute-alias): Take an argument.
3410 (-Wno-attribute-alias): New option.
3411 * doc/extend.texi (Common Function Attributes): Document copy.
3412 (Common Variable Attributes): Same.
3413 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
3414 (-Wattribute-alias): Document new option argument.
3416 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
3418 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
3419 matching criteria. Remove unused array initializer.
3421 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
3422 Jinsong Ji <jji@us.ibm.com>
3424 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
3425 constraints by introducing a new temporary.
3426 (_mm_cvtss_si64): Likewise.
3428 2018-11-09 Martin Liska <mliska@suse.cz>
3430 * common.opt: Add -fipa-stack-alignment flag.
3431 * doc/invoke.texi: Document it.
3432 * final.c (rest_of_clean_state): Guard stack
3433 shrinking with flag.
3435 2018-11-09 Martin Liska <mliska@suse.cz>
3437 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
3439 (ipa_discover_variable_flags): ... this.
3440 * common.opt: Come up with new flag -fipa-reference-addressable.
3441 * doc/invoke.texi: Document it.
3442 * ipa-reference.c (propagate): Call the renamed fn.
3443 * ipa-visibility.c (whole_program_function_and_variable_visibility):
3445 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
3447 (ipa_discover_variable_flags): ... this. Discover
3448 non-addressable variables only with the newly added flag.
3449 * opts.c: Enable the newly added flag with -O1 and higher
3452 2018-11-09 David Malcolm <dmalcolm@redhat.com>
3454 * json.cc (selftest::test_writing_literals): Fix comment.
3456 2018-11-09 Martin Liska <mliska@suse.cz>
3458 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
3459 string to a stack buffer.
3460 (aarch64_parse_cpu): Likewise.
3461 (aarch64_parse_tune): Likewise.
3463 2018-11-09 Richard Biener <rguenther@suse.de>
3465 PR tree-optimization/87953
3466 * tree-vect-loop.c (vectorizable_reduction): For analysis
3467 always pass ops[0] to vectorizable_condition.
3469 2018-11-09 Stafford Horne <shorne@gmail.com>
3470 Richard Henderson <rth@twiddle.net>
3471 Joel Sherrill <joel@rtems.org>
3473 * common/config/or1k/or1k-common.c: New file.
3474 * config/or1k/*: New.
3475 * config.gcc (or1k*-*-*): New.
3476 * configure.ac (or1k*-*-*): New test for openrisc tls.
3477 * configure: Regenerated.
3478 * doc/install.texi: Document OpenRISC triplets.
3479 * doc/invoke.texi: Document OpenRISC arguments.
3480 * doc/md.texi: Document OpenRISC.
3482 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
3484 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
3485 (arm7tdmi-s): Delete CPU.
3486 (arm710t): Add aliases for arm720t and arm740t.
3487 (arm720t, arm740t): Delete CPUs.
3488 (arm920t): Add aliases for arm920, arm922t and arm940t.
3489 (arm920, arm922t, arm940t): Delete CPUs.
3490 (arm10tdmi): Add alias for arm1020t.
3491 (arm1020t): Delete CPU.
3492 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
3493 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
3494 (arm10e): Add aliases for arm1020e and arm1022e.
3495 (arm1020e, arm1022e): Delete CPU.
3496 * config/arm/arm.md (generic_sched): Remove entries that are now
3498 (generic_vfp): Likewise.
3499 * config/arm/arm1020e.md: Simplify tuning selection based on alias
3501 * config/arm/arm-tune.md: Regenerated.
3502 * config/arm/arm-tables.opt: Regenerated.
3504 2018-11-09 Richard Biener <rguenther@suse.de>
3506 PR tree-optimization/87621
3507 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
3508 op with only phi inputs.
3509 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
3510 (ch_base::copy_headers): Run CSE on copied loop headers.
3511 (pass_ch_vect::process_loop_p): Simplify.
3513 2018-11-09 Alexandre Oliva <oliva@adacore.com>
3515 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
3516 for non-w64 x86_64 biarch.
3518 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
3520 PR rtl-optimization/86438
3521 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
3522 of in_b for the compare if in_b is SET_DEST.
3525 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
3526 non-toplevel UNSPEC.
3528 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3530 * tree-vrp.c (value_range::check): Do not access internals
3532 (value_range::singleton_p): Same.
3533 (value_range::type): Same.
3534 (vrp_finalize): Use value_range API.
3536 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3538 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
3540 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3542 * vr-values.c (vr_values::get_value_range): Use value_range API
3543 instead of piecing together ranges.
3544 (vr_values::update_value_range): Same.
3546 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3548 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
3549 of performing ad-hoc calculations.
3550 * tree-ssanames.c (set_range_info): New overloaded function
3551 accepting value_range &.
3552 (get_range_info): Same.
3553 * tree-ssanames.h (set_range_info_raw): Remove.
3554 (set_range_info): New prototype.
3555 (get_range_info): Same.
3556 * tree-vrp.h (value_range::null_p): Rename to zero_p.
3557 * tree-vrp.c (value_range::null_p): Same.
3559 2018-11-09 Jan Hubicka <jh@suse.cz>
3561 * tree.c (fld_type_variant_equal_p): Test user align flag.
3562 (flt_type_variant): Copy user align flag.
3563 (fld_incomplete_type_of): Clear it.
3565 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3567 * config/arm/neon.md (div<mode>3): New pattern.
3569 2018-11-08 Andi Kleen <ak@linux.intel.com>
3571 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
3572 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
3573 (ix86_handle_option): Handle OPT_mptwrite.
3574 * config/i386/cpuid.h (bit_PTWRITE): Add.
3575 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
3576 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
3577 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
3578 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
3579 (ix86_option_override_internal): Handle PTA_PTWRITE.
3580 (ix86_valid_target_attribute_inner_p): Define ptwrite.
3581 (def_builtin2): Force UINT64 to be 64bit only.
3582 * config/i386/i386.h (TARGET_PTWRITE): Add.
3583 (TARGET_PTWRITE_P): Add.
3585 * config/i386/i386.md: Define ptwrite.
3586 * config/i386/i386.opt: Add -mptwrite.
3587 * config/i386/immintrin.h (_ptwrite64): Add.
3589 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
3590 * doc/invoke.texi: Document -mptwrite.
3592 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
3594 PR rtl-optimization/87600
3595 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
3596 * lra-constraints.c (process_alt_operands): Skip illegal hard
3597 register usage. Prefer reloading non hard register operands.
3599 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
3602 * doc/invoke.texi (Optimize Options): Clarify default behavior
3603 for -fno-sched-interblock and -fno-sched-spec.
3605 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
3607 * collect2.c (linker_select): Add USE_LLD_LD.
3608 (ld_suffixes): Add ld.lld.
3609 (main): Handle -fuse-ld=lld.
3610 * common.opt (-fuse-ld=lld): New option.
3611 * doc/invoke.texi (-fuse-ld=lld): Document.
3612 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
3614 2018-11-08 Paul Koning <ni1d@arrl.net>
3616 * config/pdp11/constraints.md: Add "Z" series constraints for use
3617 with pre-dec and post-inc addressing.
3618 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
3619 (pdp11_expand_operands): Add int argument (word count).
3620 (pdp11_sp_frame_offset): Delete.
3621 (pdp11_cmp_length): New function.
3622 (pushpop_regeq): New function.
3623 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
3625 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
3627 (pdp11_initial_elimination_offset): Ditto.
3628 (pdp11_expand_operands): Add word count argument. Bugfixes.
3629 (output_move_multiple): Change how pointer adjustment is done.
3630 (pdp11_gen_int_label): Correct format.
3631 (output_ascii): Ditto.
3632 (pdp11_asm_output_var): Add code for DEC assembler case.
3633 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
3635 (legitimate_const_double_p): Ditto.
3636 (pdp11_register_move_cost): Adjust for new register classes.
3637 (pdp11_regno_reg_class): Ditto.
3638 (expand_block_move): Delete.
3639 (pushpop_regeq): New function.
3640 (pdp11_legitimate_address_p): Bugfix in check for constant
3642 (pdp11_sp_frame_offset): Delete.
3643 (pdp11_reg_save_size): New helper function for new frame layout.
3644 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
3645 (pdp11_expand_shift): Bugfix in check for constant shift count.
3646 (pdp11_shift_length): Ditto.
3647 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
3648 (pdp11_cmp_length): New function.
3649 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
3650 some compile options.
3651 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
3652 (CALL_USED_REGISTERS): Ditto.
3653 (ELIMINABLE_REGS): Ditto.
3654 (REGISTER_NAMES): Ditto.
3655 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
3657 (REG_CLASS_NAMES): Ditto.
3658 (REG_CLASS_CONTENTS): Ditto. Also remove
3659 HARD_FRAME_POINTER_REGNUM.
3660 (CPU_REG_CLASS): New macro.
3661 (CLASS_MAX_NREGS): Adjust for new register classes.
3662 (FUNCTION_PROFILER): Make no-op.
3663 (may_call_alloca): Remove unused declaration.
3664 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
3665 (ASM_OUTPUT_SKIP): Fix format.
3666 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
3667 (HARD_FRAME_POINTER_REGNUM): Remove.
3669 (*rts): Rename. Remove epilogue related checks.
3670 (cmpsi, cmpdi): New insn.
3671 (cbranch<mode>4): Change to apply to SI and DI modes as well.
3672 (mov<mode>): Change constraints to enforce that push/pop
3673 destination cannot use the same register as source.
3674 (*mov<mode><cc_cc>): Ditto.
3675 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
3676 at assembly output rather than as RTL expander.
3677 (zero_extendqihi2): Bugfix in check for same registers.
3678 (adddi3_nocc): Bugfix in check for constant operand.
3679 (addsi3_nocc): Ditto.
3680 (subdi3_nocc): Ditto.
3681 (subsi3_nocc): Ditto.
3682 (negdi2_nocc): Copy input to pdp11_expand_operands.
3683 (negsi2_nocc): Ditto.
3684 (bswap2_nocc): Ditto.
3685 * config/pdp11/pdp11.opt (mlra): Fix documentation.
3686 * config/pdp11/t-pdp11: Use -Os.
3688 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
3690 * config/arm/parsecpu.awk (/alias/): New parsing rule.
3691 (/begin cpu/): Check that the cpu name hasn't been previously defined.
3692 (gen_comm_data): Print out CPU alias tables.
3693 (check_cpu): Match aliases when checking the CPU name.
3694 * config/arm/arm-protos.h (cpu_alias): New structure.
3695 (cpu_option): Add entry for aliases.
3696 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
3697 strongarm1100 and strongarm1110.
3698 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
3699 (config/arm/arm-generic.md): Remove redundant references to
3700 strongarm110, strongarm1100 and strongarm1110.
3701 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
3702 Scan aliases for additional hints.
3703 (arm_parse_cpu_option_name): Also match a cpu name against the list
3705 * config/arm/arm-tables.opt: Regenerated.
3706 * config/arm/arm-tune.md: Regenerated.
3708 2018-11-08 Jakub Jelinek <jakub@redhat.com>
3710 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
3711 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
3712 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3713 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
3714 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3715 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
3716 * gengtype.c (open_base_files): Add omp-general.h.
3717 * gimple.c (gimple_build_omp_critical):
3718 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
3719 gimple_omp_taskgroup_set_clauses.
3720 (gimple_build_omp_atomic_load): Add mo argument, call
3721 gimple_omp_atomic_set_memory_order.
3722 (gimple_build_omp_atomic_store): Likewise.
3723 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
3724 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
3726 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
3727 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
3728 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
3729 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
3730 different value for GF_OMP_ATOMIC_NEED_VALUE.
3731 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
3733 (struct gimple_statement_omp_single_layout): And remove here.
3734 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
3735 than gimple_statement_omp_single_layout.
3736 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
3738 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
3739 (gimple_omp_subcode): Formatting fix.
3740 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
3741 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
3742 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
3743 gimple_omp_teams_host, gimple_omp_teams_set_host,
3744 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
3745 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
3746 gimple_omp_taskgroup_set_clauses): New inline functions.
3747 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
3748 (gimple_build_omp_atomic_store): Likewise.
3749 (gimple_omp_atomic_seq_cst_p): Remove.
3750 (gimple_omp_atomic_memory_order): New function.
3751 (gimple_omp_atomic_set_seq_cst): Remove.
3752 (gimple_omp_atomic_set_memory_order): New function.
3753 (gimple_build_omp_taskgroup): Add clauses argument.
3754 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
3755 (dump_gimple_omp_task): Print taskwait with depend clauses.
3756 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
3757 dump_omp_atomic_memory_order.
3758 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
3759 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
3760 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
3761 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
3762 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
3763 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
3764 ORT_UNTIED_TASKLOOP enumerators.
3765 (enum gimplify_defaultmap_kind): New.
3766 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
3767 target_map_pointers_as_0len_arrays members, add defaultmap.
3768 (new_omp_context): Initialize defaultmap member.
3769 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3770 (maybe_fold_stmt): Don't fold even in host teams regions.
3771 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
3772 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
3773 ctx->omp_firstprivatize_variable.
3774 (omp_add_variable): Don't add private/firstprivate for VLAs in
3776 (omp_default_clause): Print "taskloop" rather than "task" if
3778 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3779 Handle new defaultmap clause kinds.
3780 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
3781 iterator to be lastprivate or private. Fix up diagnostics if linear
3782 is used on collapse>1 simd iterator.
3783 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3784 (gimplify_omp_depend): New function.
3785 (gimplify_scan_omp_clauses): Add shared clause on parallel for
3786 combined parallel master taskloop{, simd} if taskloop has
3787 firstprivate, lastprivate or reduction clause. Handle
3788 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
3789 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
3790 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
3791 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
3792 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
3794 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
3795 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
3796 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
3797 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3798 (gimplify_omp_task): Handle taskwait with depend clauses.
3799 (gimplify_omp_for): Add shared clause on parallel for combined
3800 parallel master taskloop{, simd} if taskloop has firstprivate,
3801 lastprivate or reduction clause. Use ORT_TASKLOOP or
3802 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
3803 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
3804 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
3805 __for_end and __for_range temporaries on OMP_PARALLEL for
3806 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
3807 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
3808 sandwiched in between two taskloops.
3809 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
3810 instead of ctx->omp_firstprivatize_variable.
3811 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
3812 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
3813 host teams, use gimplify_and_return_first etc. for body like
3814 for target or target data constructs, and at the end call
3815 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
3816 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
3817 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
3818 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
3819 gimple_omp_atomic_set_seq_cst calls.
3820 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
3821 case, handle taskgroup clauses.
3822 * lto-streamer-out.c (hash_tree): Handle
3823 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3824 * Makefile.in (GTFILES): Add omp-general.h.
3825 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
3826 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
3827 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
3828 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
3829 BUILT_IN_GOMP_LOOP_DOACROSS_START,
3830 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
3831 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3832 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
3833 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
3834 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
3835 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
3836 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
3837 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3838 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
3839 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
3840 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
3841 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
3842 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
3843 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
3844 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
3845 * omp-expand.c (workshare_safe_to_combine_p): Return false for
3846 non-worksharing loops.
3847 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
3848 (determine_parallel_type): Don't combine parallel with worksharing
3849 which has _reductemp_ clause.
3850 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
3851 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
3852 if there is nonmonotonic modifier or if there is no modifier and no
3853 ordered clause. For dynamic and guided schedule without monotonic
3854 and nonmonotonic modifier, default to nonmonotonic.
3855 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
3856 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
3858 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
3859 there are any reduction clauses.
3860 (expand_taskwait_call): New function.
3861 (expand_teams_call): New function.
3862 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
3863 expand_teams_call for it. Formatting fix. Handle taskwait with
3865 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
3866 of worksharing loops with task reductions.
3867 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
3868 expansion of worksharing loops with task reductions.
3869 (expand_omp_sections): Handle expansion of sections with task
3871 (expand_omp_synch): For host teams call expand_omp_taskreg.
3872 (omp_memory_order_to_memmodel): New function.
3873 (expand_omp_atomic_load, expand_omp_atomic_store,
3874 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
3875 instead of gimple_omp_atomic_seq_cst_p.
3876 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
3877 depend clauses as a standalone directive.
3878 * omp-general.c (enum omp_requires): New variable.
3879 (omp_extract_for_data): Initialize have_reductemp member. Allow
3880 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
3881 GT_EXPR loops depending on incr sign. Formatting fixes.
3882 * omp-general.h (struct omp_for_data): Add have_reductemp member.
3883 (enum omp_requires): New enum.
3884 (omp_requires_mask): Declare.
3885 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
3887 * omp-low.c (struct omp_context): Add task_reductions and
3888 task_reduction_map fields.
3889 (is_host_teams_ctx): New function.
3890 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
3891 (use_pointer_for_field): Use is_global_var instead of
3892 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
3894 (build_outer_var_ref): Ignore taskgroup outer contexts.
3895 (delete_omp_context): Release task_reductions and task_reduction_map.
3896 (scan_sharing_clauses): Don't add any fields for reduction clause on
3897 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
3898 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
3899 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
3900 Handle OMP_CLAUSE_NONTEMPORAL.
3901 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
3903 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
3904 clauses with task modifier.
3905 (scan_omp_task): Handle taskwait with depend clauses.
3906 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
3907 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
3908 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
3909 (scan_omp_for): Fix comment formatting.
3910 (scan_omp_teams): Handle host teams constructs.
3911 (check_omp_nesting_restrictions): Allow teams with no outer
3912 OpenMP context. Adjust diagnostics for teams strictly nested into
3913 some explicit OpenMP construct other than target. Allow OpenMP atomics
3914 inside of simd regions.
3915 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
3916 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
3917 taskreg_nesting_level while scanning host teams construct.
3918 (task_reduction_read): New function.
3919 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
3920 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
3921 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
3922 second argument create_tmp_var if it is NULL. Don't ignore shared
3923 clauses in is_host_teams_ctx contexts. Handle
3924 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
3926 (lower_reduction_clauses): Ignore reduction clauses with task
3927 modifier. Remove second argument create_tmp_var if it is NULL.
3928 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
3929 (lower_send_clauses): Ignore reduction clauses with task modifier.
3930 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
3931 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
3932 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
3933 rely that it is the last stmt in body so far. Ignore outer taskgroup
3935 (omp_task_reductions_find_first, omp_task_reduction_iterate,
3936 lower_omp_task_reductions): New functions.
3937 (lower_omp_sections): Handle reduction clauses with taskgroup
3938 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
3939 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
3940 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
3942 (lower_omp_taskgroup): Handle taskgroup reductions.
3943 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
3944 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
3945 (lower_depend_clauses): If there are any
3946 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3947 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
3948 seen, assume lowering is done already and return early. Set kind
3949 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
3950 (lower_omp_taskreg): Handle reduction clauses with task modifier on
3951 parallel construct. Handle reduction clause on taskloop construct.
3952 Handle taskwait with depend clauses.
3953 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
3954 for host teams constructs.
3955 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
3956 nontemporal and _reductemp_ clause entries.
3957 (omp_clause_code_name): Likewise.
3958 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
3959 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
3960 * tree-core.h (enum omp_clause_code): Add
3961 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
3962 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
3963 (struct tree_base): Add omp_atomic_memory_order field into union.
3964 Remove OMP_ATOMIC_SEQ_CST comment.
3965 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3966 and OMP_CLAUSE_DEPEND_DEPOBJ.
3967 (struct tree_omp_clause): Add subcode.defaultmap_kind.
3968 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
3969 OpenMP constructs with body and clauses operands.
3970 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
3971 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
3972 (OMP_TASKGROUP_CLAUSES): Define.
3973 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
3974 OMP_CLAUSE__LOOPTEMP_.
3975 (OMP_ATOMIC_SEQ_CST): Remove.
3976 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
3977 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
3978 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
3979 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
3980 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
3981 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
3982 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
3983 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
3984 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
3985 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
3987 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
3988 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
3989 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
3990 (convert_local_omp_clauses): Likewise. Remove useless test.
3991 * tree-parloops.c (create_call_for_reduction_1): Pass
3992 OMP_MEMORY_ORDER_RELAXED as new argument to
3993 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
3994 * tree-pretty-print.c (dump_omp_iterators): New function.
3995 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
3996 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
3997 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
3998 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
3999 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
4000 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
4001 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
4002 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
4003 (dump_omp_atomic_memory_order): New function.
4004 (dump_generic_node): Use it. Print taskgroup clauses. Print
4005 taskwait with depend clauses.
4006 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
4007 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
4008 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
4009 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
4010 write_ts_omp_clause_tree_pointers): Likewise.
4012 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4015 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
4016 "-missed", "-note", and "-all" sub-options.
4017 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
4019 (can_early_inline_edge_p): Likewise.
4020 (want_early_inline_function_p): Likewise.
4021 (want_inline_self_recursive_call_p): Likewise.
4022 (recursive_inlining): Likewise.
4023 (inline_small_functions): Likewise.
4024 (flatten_function): Likewise.
4025 (ipa_inline): Likewise.
4026 (inline_always_inline_functions): Likewise.
4027 (early_inline_small_functions): Likewise.
4028 (early_inliner): Likewise.
4029 * tree-inline.c (expand_call_inline): Likewise.
4031 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4033 * pretty-print.c (pp_format): Handle %f.
4034 (selftest::test_pp_format): Add test of %f.
4035 * pretty-print.h (pp_double): New macro.
4037 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4039 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
4040 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
4041 (dump_pretty_printer::decode_format): Implement "%C" for
4043 (selftest::test_capture_of_dump_calls): Rename "where" to
4044 "stmt_loc". Convert test_decl to a function decl and set its
4045 location. Add a symbol_table_test RAII instance and a
4046 cgraph_node, using it to test "%C" and dump_symtab_node.
4048 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
4051 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
4053 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4055 * cgraph.c: Include "selftest.h".
4056 (saved_symtab): New variable.
4057 (selftest::symbol_table_test::symbol_table_test): New ctor.
4058 (selftest::symbol_table_test::~symbol_table_test): New dtor.
4059 (selftest::test_symbol_table_test): New test.
4060 (selftest::cgraph_c_tests): New.
4061 * cgraph.h (saved_symtab): New decl.
4062 (selftest::symbol_table_test): New class.
4063 * selftest-run-tests.c (selftest::run_tests): Call
4064 selftest::cgraph_c_tests.
4065 * selftest.h (selftest::cgraph_c_tests): New decl.
4067 2018-11-08 Richard Biener <rguenther@suse.de>
4069 * tree-data-ref.h (lambda_int): New typedef.
4070 (lambda_vector_gcd): Adjust.
4071 (lambda_vector_new): Likewise.
4072 (lambda_matrix_new): Likewise.
4073 * tree-data-ref.c (print_lambda_vector): Adjust.
4075 2018-11-08 Richard Biener <rguenther@suse.de>
4077 PR tree-optimization/87929
4078 * tree-complex.c (expand_complex_comparison): Clean EH.
4080 2018-11-08 Martin Liska <mliska@suse.cz>
4082 * doc/extend.texi: Reword.
4083 * predict.c (expr_expected_value_1): Likewise.
4085 2018-11-08 Richard Biener <rguenther@suse.de>
4087 PR tree-optimization/87913
4088 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
4089 of extreme values to ordered comparisons.
4091 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4094 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
4096 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4098 * doc/invoke.texi: Remove leading dash from @opindex entries
4099 throughout the file.
4101 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4104 * doc/invoke.texi (Option Summary): Add -e and --entry.
4105 (Link Options): Likewise.
4107 2018-11-07 Nathan Sidwell <nathan@acm.org>
4110 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
4111 --disable-checking bootstrap.
4113 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
4115 * configure: Regenerated.
4117 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4120 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
4121 to the mode of the widest field iff the widest field has mode class
4122 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
4125 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
4127 * common.opt: Add -Wattribute-warning.
4128 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
4129 * expr.c (expand_expr_real_1): Add new attribute to warning_at
4130 call to allow user configure behavior of "warning" attribute.
4132 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4134 * target.def: Put @: after every vs., e.g., and i.e. where it is
4135 followed by whitespace.
4136 * doc/extend.texi: Ditto.
4137 * doc/fragments.texi: Ditto.
4138 * doc/gimple.texi: Ditto.
4139 * doc/implement-c.texi: Ditto.
4140 * doc/install.texi: Ditto.
4141 * doc/invoke.texi: Ditto.
4142 * doc/md.texi: Ditto.
4143 * doc/plugins.texi: Ditto.
4144 * doc/rtl.texi: Ditto.
4145 * doc/sourcebuild.texi: Ditto.
4146 * doc/tm.texi.in: Ditto.
4147 * doc/ux.texi: Ditto.
4148 * doc/tm.texi: Regenerate.
4150 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4152 * config/arm/arm-cpus.in (ares): New entry.
4153 * config/arm/arm-tables.opt: Regenerate.
4154 * config/arm/arm-tune.md: Likewise.
4155 * doc/invoke.texi (ARM Options): Document ares.
4157 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4159 * config/aarch64/aarch64-cores.def (ares): Define.
4160 * config/aarch64/aarch64-tune.md: Regenerate.
4161 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
4163 2018-11-07 Jan Hubicka <jh@suse.cz>
4165 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
4166 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
4167 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
4170 2018-11-07 Richard Biener <rguenther@suse.de>
4172 PR tree-optimization/87914
4173 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
4175 (vectorizable_reduction): Handle shifts and rotates by dispatching
4176 to vectorizable_shift.
4177 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
4178 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
4180 (vectorizable_shift): Export and handle being called as
4182 (vect_analyze_stmt): Call vectorizable_shift after
4183 vectorizable_reduction.
4184 * tree-vectorizer.h (vectorizable_shift): Declare.
4186 2018-11-07 Jan Hubicka <jh@suse.cz>
4188 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
4189 than const decls in TREE_VALUE of enum.
4190 (dump_type_inheritance_graph): Improve duplicate dumping.
4191 (free_enum_values): New.
4192 (build_type_inheritance_graph): Use it.
4193 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
4194 which are not main variants or not ODR types.
4195 (verify_type_variant): Expect variants to have no TYPE_VALUES.
4197 2018-11-07 Richard Biener <rguenther@suse.de>
4199 * ipa-inline.c (want_inline_small_function_p): Compute
4200 big_speedup_p lazily and last.
4202 2018-11-07 Jan Hubicka <jh@suse.cz>
4204 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
4205 building incomplete variant of complete type.
4206 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
4207 variant of complete type.
4209 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4211 * config/mips/mips.c: Fix typo in documentation of
4212 mips_loongson_ext2_prefetch_cookie.
4213 (mips_option_override): fix brain twister logical.
4214 * config/mips/mips.h: Fix typo in documentation of
4215 ISA_HAS_CTZ_CTO and define pattern.
4216 * config/mips/mips.md (prefetch): Hoist EXT2 above
4218 (prefetch_indexed): Hoist EXT2 above the EXT block.
4220 2018-11-07 Jan Hubicka <jh@suse.cz>
4222 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
4223 return and parameter types of function and method types.
4224 (free_lang_data_in_cgraph): Update.
4226 2018-11-07 Martin Liska <mliska@suse.cz>
4228 PR rtl-optimization/87868
4229 * postreload-gcse.c (eliminate_partially_redundant_load): Set
4230 threshold to max_count if we would overflow.
4231 * profile-count.h: Make max_count a public constant.
4233 2018-11-07 Martin Liska <mliska@suse.cz>
4235 * mem-stats.h: Fix GNU coding style.
4237 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4239 * config/mips/gs264e.md: New.
4240 * config/mips/mips-cpus.def: Define gs264e.
4241 * config/mips/mips-tables.opt: Regenerate.
4242 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
4244 (mips_issue_rate): Add support for gs264e.
4245 (mips_multipass_dfa_lookahead): Likewise.
4246 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
4247 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
4248 (MIPS_ASE_MSA_SPEC): New.
4249 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
4250 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
4251 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
4252 * config/mips/mips.md: Include gs264e.md.
4253 (processor): Add gs264e.
4254 * config/mips/mips.opt (MSA): Use Mask instead of Var.
4255 * doc/invoke.texi: Add gs264e to supported architectures.
4257 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4259 * config/mips/gs464e.md: New.
4260 * config/mips/mips-cpus.def: Define gs464e.
4261 * config/mips/mips-tables.opt: Regenerate.
4262 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
4264 (mips_issue_rate): Add support for gs464e.
4265 (mips_multipass_dfa_lookahead): Likewise.
4266 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
4267 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
4268 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
4269 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
4270 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
4271 * config/mips/mips.md: Include gs464e.md.
4272 (processor): Add gs464e.
4273 * doc/invoke.texi: Add gs464e to supported architectures.
4275 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4277 * config/mips/loongson3a.md: Rename to ...
4278 * config/mips/gs464.md: ... here.
4279 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
4280 as an alias of gs464 processor.
4281 * config/mips/mips-tables.opt: Regenerate.
4282 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
4283 instead of PROCESSOR_LOONGSON_3A.
4284 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
4286 (mips_option_override): Enable MMI and EXT for gs464.
4287 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
4288 Rename TUNE_LOONGSON_3A to TUNE_GS464.
4289 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
4290 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
4291 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
4293 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
4294 (processor): Add gs464;
4295 * doc/invoke.texi: Add gs464 to supported architectures.
4297 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4299 * config/mips/mips-protos.h
4300 (mips_loongson_ext2_prefetch_cookie): New prototype.
4301 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
4302 (mips_option_override): Enable TARGET_LOONGSON_EXT when
4303 TARGET_LOONGSON_EXT2 is true.
4304 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
4305 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
4306 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
4307 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
4308 TARGET_LOONGSON_EXT2.
4309 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
4310 (define_insn "ctz<mode>2"): New insn pattern.
4311 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
4312 (define_insn "prefetch_indexed_<mode>"): Include
4313 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
4314 * config/mips/mips.opt (-mloongson-ext2): Add option.
4315 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
4317 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4319 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
4320 __mips_loongson_ext.
4321 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
4322 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
4324 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
4325 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
4326 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
4327 instead of TARGET_LOONGSON_3A.
4328 * config/mips/mips.opt (-mloongson-ext): Add option.
4329 * gcc/doc/invoke.texi (-mloongson-ext): Document.
4331 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4333 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
4334 * config/mips/loongson.md: Move to ...
4335 * config/mips/loongson-mmi.md: here; Adjustment.
4336 * config/mips/loongson.h: Move to ...
4337 State as deprecated. Include loongson-mmiintrin.h for back
4338 compatibility and warning.
4339 * config/mips/loongson-mmiintrin.h: ... here.
4340 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
4341 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
4342 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
4343 (mips_option_override): Make sure MMI use hard float;
4344 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
4345 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
4346 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
4347 TARGET_LOONGSON_VECTORS.
4348 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
4349 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
4350 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
4351 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
4353 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
4354 TARGET_LOONGSON_VECTORS.
4355 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
4356 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
4357 (Loongson MMI patterns): Include loongson-mmi.md instead of
4359 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
4360 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
4362 2018-11-07 Richard Biener <rguenther@suse.de>
4365 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
4366 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
4368 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
4370 PR rtl-optimization/87874
4371 * lra.c (lra_substitute_pseudo): Do not create a subreg for
4374 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
4376 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
4377 if not in indexed or indirect form.
4378 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
4379 (bswap<mode>2_store): Ditto.
4381 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
4383 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
4386 2018-11-06 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/86850
4389 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
4390 instead of src.m_vec.
4392 2018-11-06 Jan Hubicka <jh@suse.cz>
4394 * tree.c (fld_simplified_type_name): Break out form ...
4395 (free_lang_data_in_type): ... here.
4396 (fld_type_variant_equal_p): Use it.
4398 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4400 * config/default-d.c: Include memmodel.h.
4402 * config/sol2-d.c: New file.
4403 * config/t-sol2 (sol2-d.o): New rule.
4404 * config.gcc <*-*-solaris2*>: Set d_target_objs,
4405 target_has_targetdm.
4407 2018-11-06 Jan Hubicka <jh@suse.cz>
4409 * tree.c (fld_type_variant): Also copy alignment; be sure that
4410 new variant is equal.
4412 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
4415 * config/s390/s390.md: Add relative_long attribute.
4417 2018-11-06 Jan Hubicka <jh@suse.cz>
4419 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
4420 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
4421 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
4422 stream TYPE_NEEDS_CONSTRUCTING.
4423 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
4424 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
4426 2018-11-06 Richard Biener <rguenther@suse.de>
4428 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
4430 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
4432 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4434 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
4435 REG_CLASS_CONTENTS[GEN_REGS].
4436 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
4438 2018-11-06 Jan Hubicka <jh@suse.cz>
4440 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
4442 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4445 * config/sol2.h (ASAN_CC1_SPEC): Define.
4446 (LD_WHOLE_ARCHIVE_OPTION): Define.
4447 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
4448 (ASAN_REJECT_SPEC): Provide default.
4449 (LIBASAN_EARLY_SPEC): Define.
4450 (LIBTSAN_EARLY_SPEC): Define.
4451 (LIBLSAN_EARLY_SPEC): Define.
4452 * config/i386/sol2.h (CC1_SPEC): Redefine.
4453 (ASAN_REJECT_SPEC): Define.
4455 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
4456 (TARGET_ASAN_SHADOW_OFFSET): Define.
4457 (sparc_asan_shadow_offset): New function.
4458 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
4459 (ASAN_REJECT_SPEC): Define.
4461 2018-11-06 Jan Hubicka <jh@suse.cz>
4463 * tree.c (fld_type_variant): Copy canonical type.
4464 (fld_incomplete_type_of): Check that canonical types looks sane;
4465 copy canonical type.
4466 (verify_type): Accept when incomplete type has complete canonical type.
4468 2018-11-06 Jan Hubicka <jh@suse.cz>
4470 * tree.c (free_lang_data): Reset overwite_assembler_name,
4471 print_xnode, print_decl, print_type and print_identifier of
4474 2018-11-06 Richard Biener <rguenther@suse.de>
4476 PR tree-optimization/87889
4477 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
4478 Do nothing if old and new arg are the same
4480 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
4483 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
4484 attributes for operands 3 and 4.
4486 2018-11-06 Richard Biener <rguenther@suse.de>
4489 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
4490 matching bitfield insertion.
4492 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
4494 * auto-inc-dec.c: Include valtrack.h. Improve comments.
4495 (reg_next_debug_use): New.
4496 (attempt_change): Propagate adjusted expression into affected
4498 (merge_in_block): Track uses in debug insns.
4499 (pass_inc_dec::execute): Allocate and release
4502 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
4504 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
4505 (_mm512_fixupimm_round_pd): Update parameters and builtin.
4506 (_mm512_maskz_fixupimm_round_pd): Ditto.
4507 (_mm512_fixupimm_round_ps): Ditto.
4508 (_mm512_maskz_fixupimm_round_ps): Ditto.
4509 (_mm_fixupimm_round_sd): Ditto.
4510 (_mm_maskz_fixupimm_round_sd): Ditto.
4511 (_mm_fixupimm_round_ss): Ditto.
4512 (_mm_maskz_fixupimm_round_ss): Ditto.
4513 (_mm512_fixupimm_pd): Ditto.
4514 (_mm512_maskz_fixupimm_pd): Ditto.
4515 (_mm512_fixupimm_ps): Ditto.
4516 (_mm512_maskz_fixupimm_ps): Ditto.
4517 (_mm_fixupimm_sd): Ditto.
4518 (_mm_maskz_fixupimm_sd): Ditto.
4519 (_mm_fixupimm_ss): Ditto.
4520 (_mm_maskz_fixupimm_ss): Ditto.
4521 (_mm512_mask_fixupimm_round_pd): Update builtin.
4522 (_mm512_mask_fixupimm_round_ps): Ditto.
4523 (_mm_mask_fixupimm_round_sd): Ditto.
4524 (_mm_mask_fixupimm_round_ss): Ditto.
4525 (_mm512_mask_fixupimm_pd): Ditto.
4526 (_mm512_mask_fixupimm_ps): Ditto.
4527 (_mm_mask_fixupimm_sd): Ditto.
4528 (_mm_mask_fixupimm_ss): Ditto.
4529 * config/i386/avx512vlintrin.h:
4530 (_mm256_fixupimm_pd): Update parameters and builtin.
4531 (_mm256_maskz_fixupimm_pd): Ditto.
4532 (_mm256_fixupimm_ps): Ditto.
4533 (_mm256_maskz_fixupimm_ps): Ditto.
4534 (_mm_fixupimm_pd): Ditto.
4535 (_mm_maskz_fixupimm_pd): Ditto.
4536 (_mm_fixupimm_ps): Ditto.
4537 (_mm_maskz_fixupimm_ps): Ditto.
4538 (_mm256_mask_fixupimm_pd): Update builtin.
4539 (_mm256_mask_fixupimm_ps): Ditto.
4540 (_mm_mask_fixupimm_pd): Ditto.
4541 (_mm_mask_fixupimm_ps): Ditto.
4542 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
4543 * config/i386/i386-builtin.def: Update builtin definitions.
4544 * config/i386/i386.c: Handle new builtin types and remove useless ones.
4545 * config/i386/sse.md: Update VFIXUPIMM* patterns.
4546 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
4547 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
4548 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
4549 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
4550 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
4551 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
4552 * config/i386/subst.md:
4553 (round_saeonly_sd_mask_operand4): Add new subst_attr.
4554 (round_saeonly_sd_mask_op4): Ditto.
4555 (round_saeonly_expand_operand5): Ditto.
4556 (round_saeonly_expand): Update.
4558 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
4560 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
4562 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
4564 PR rtl-optimization/87871
4565 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
4567 2018-11-05 Paul Koning <ni1d@arrl.net>
4569 * doc/sourcebuild.texi (target attributes): Document new "inf"
4570 effective target keyword.
4572 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4574 * config/s390/s390.c (s390_register_move_cost): Increase costs for
4575 moves involving the CC reg.
4577 2018-11-05 Richard Biener <rguenther@suse.de>
4579 PR tree-optimization/87873
4580 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
4582 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
4583 * tree-vect-loop.c (vect_transform_loop): When splitting the
4584 loop exit also create forwarder PHIs for constants.
4585 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
4586 Handle constant to_arg, add extra checking we match up the correct
4589 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4591 * config/s390/s390.md: QImode and HImode for load on condition.
4593 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4595 * config/s390/predicates.md: Fix typo.
4596 * config/s390/s390.md: Allow immediates for load on condition.
4598 2018-11-05 Martin Liska <mliska@suse.cz>
4600 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
4601 * bitmap.h (struct bitmap_usage): Likewise.
4602 * ggc-common.c (SCALE): Remove.
4604 (struct ggc_usage): Use SIZE_AMOUNT. And update
4606 * ggc-page.c (SCALE): Remove.
4607 (STAT_LABEL): Remove.
4608 (ggc_print_statistics): Use SIZE_AMOUNT.
4609 * gimple.h (SCALE): Remove.
4611 * input.c (ONE_K): Remove.
4614 (STAT_LABEL): Likewise.
4615 (FORMAT_AMOUNT): Likewise.
4616 (dump_line_table_statistics): Use SIZE_AMOUNT.
4617 * mem-stats.h (struct mem_usage): Likewise.
4618 * rtl.c (dump_rtx_statistics): Likewise.
4619 (rtx_alloc_counts): Change type to size_t.
4620 (rtx_alloc_sizes): Likewise.
4621 (rtx_count_cmp): New.
4622 (dump_rtx_statistics): Sort first based on counts.
4623 * tree.c (tree_nodes_cmp): New.
4624 (tree_codes_cmp): New.
4625 (dump_tree_statistics): Sort first based on counts.
4626 * system.h (ONE_K): New.
4628 (SIZE_SCALE): Likewise.
4629 (SIZE_LABEL): Likewise.
4630 (SIZE_AMOUNT): Likewise.
4631 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
4632 * tree-dfa.c (dump_dfa_stats): Likewise.
4633 * tree-phinodes.c (phinodes_print_statistics): Likewise.
4634 * tree-ssanames.c (ssanames_print_statistics): Likewise.
4635 * tree.c (dump_tree_statistics): Likewise.
4636 * vec.c (struct vec_usage): Likewise.
4637 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
4638 trigger a -Werror=format-overflow with
4639 --enable-gather-detailed-stats.
4641 2018-11-05 Martin Liska <mliska@suse.cz>
4643 * mem-stats.h (mem_alloc_description::release_instance_overhead):
4645 * vec.c (struct vec_usage): Register m_element_size.
4646 (vec_prefix::register_overhead): New arguments: elements and
4648 (vec_prefix::release_overhead): Subtract elements.
4649 * vec.h (struct vec_prefix): Change signature.
4650 (va_heap::reserve): Pass proper arguments.
4651 (va_heap::release): Likewise.
4653 2018-11-05 Martin Liska <mliska@suse.cz>
4655 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
4659 2018-11-05 Richard Biener <rguenther@suse.de>
4661 * tree-scalar-evolution.h (final_value_replacement_loop): Update
4663 * tree-scalar-evolution.c (final_value_replacement_loop): Return
4664 whether anything was done.
4665 (scev_const_prop): Remove constant propagation part, fold
4667 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
4668 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
4671 2018-11-05 Jakub Jelinek <jakub@redhat.com>
4673 PR tree-optimization/87859
4674 * gimple-ssa-store-merging.c (struct merged_store_group): Add
4675 only_constants and first_nonmergeable_order members.
4676 (merged_store_group::merged_store_group): Initialize them.
4677 (merged_store_group::do_merge): Clear only_constants member if
4678 adding something other than INTEGER_CST store.
4679 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
4680 stores with order >= first_nonmergeable_order. Use
4681 merged_store->only_constants instead of always recomputing it.
4682 Set merged_store->first_nonmergeable_order if we've skipped any
4683 stores. Attempt to merge overlapping INTEGER_CST stores that
4684 we would otherwise skip.
4687 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
4689 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
4692 * config/i386/emmintrin.h (__v16qs): New to cope with option
4694 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
4695 (_mm_cmplt_epi8): Likewise.
4696 (_mm_cmpgt_epi8): Likewise.
4698 2018-11-05 Richard Biener <rguenther@suse.de>
4700 PR rtl-optimization/87852
4701 * fwprop.c (use_killed_between): Only consider single-defs of the
4702 use whose definition statement dominates the use.
4704 2018-11-05 Martin Liska <mliska@suse.cz>
4707 * doc/invoke.texi: Remove options that are
4708 not disabled with -Os.
4710 2018-11-05 Martin Liska <mliska@suse.cz>
4713 * doc/extend.texi: Update constrain about the last argument
4714 of __builtin_expect_with_probability.
4716 2018-11-05 Martin Liska <mliska@suse.cz>
4719 * predict.c (expr_expected_value_1): Verify
4720 that last argument is a real constants and emit
4723 2018-11-05 Martin Liska <mliska@suse.cz>
4725 PR gcov-profile/77698
4726 * ipa-profile.c (ipa_profile): Adjust hotness threshold
4729 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4731 PR tree-optimization/86572
4732 * builtins.c (c_strlen): Handle negative offsets in a safe way.
4734 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4736 PR tree-optimization/87672
4737 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
4738 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
4740 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
4743 * cfgexpand.c (pass_expand::execute): Move the call to
4744 finish_eh_generation in front of the call to expand_stack_alignment.
4746 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
4748 * common/config/i386/i386-common.c (processor_alias_table): Add
4750 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
4751 (case ${target}): Add znver2.
4752 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
4753 -march=native recognize znver2 processors.
4754 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
4755 * config/i386/i386.c (m_znver2): New definition.
4756 (m_ZNVER): New definition.
4757 (m_AMD_MULTIPLE): Includes m_znver2.
4758 (processor_cost_table): Add znver2 entry.
4759 (processor_target_table): Add znver2 entry.
4760 (get_builtin_code_for_version): Set priority for
4762 (processor_model): Add M_AMDFAM17H_ZNVER2.
4763 (arch_names_table): Ditto.
4764 (ix86_reassociation_width): Include znver2.
4765 * config/i386/i386.h (TARGET_znver2): New definition.
4766 (struct ix86_size_cost): Add TARGET_ZNVER2.
4767 (enum processor_type): Add PROCESSOR_ZNVER2.
4768 * config/i386/i386.md (define_attr "cpu"): Add znver2.
4769 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
4770 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
4771 (ix86_adjust_cost): Add znver2.
4772 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
4773 * gcc/doc/extend.texi: Add details about znver2.
4774 * gcc/doc/invoke.texi: Add details about znver2.
4776 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
4780 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
4783 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4785 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
4786 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
4788 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
4790 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
4791 shrink-wrapping when -mtrack-speculation.
4793 2018-11-02 Richard Biener <rguenther@suse.de>
4795 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
4796 (pop_cost_one_pair): Do not free pair.
4797 (pop_best_coalesce): Likewise.
4798 (create_coalesce_list): Initialize obstack.
4799 (delete_coalesce_list): Free obstack.
4800 (find_coalesce_pair): Obstack-allocate coalesce pairs.
4801 (add_cost_one_coalesce): Likewise.
4802 (struct live_track): Remove bitmap pointer indirections.
4803 (new_live_track): Adjust.
4804 (delete_live_track): Likewise.
4805 (live_track_remove_partition): Likewise.
4806 (live_track_add_partition): Likewise.
4807 (live_track_live_p): Likewise.
4808 (live_track_process_def): Likewise.
4809 (live_track_clear_base_vars): Likewise.
4811 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4813 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
4814 * configure: Regenerate.
4815 * config.in: Regenerate.
4816 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
4817 HAVE_AS_REGISTER_PSEUDO_OP guard.
4818 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
4819 !HAVE_AS_REGISTER_PSEUDO_OP support.
4820 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
4823 2018-11-02 Richard Biener <rguenther@suse.de>
4825 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
4828 2018-11-02 Richard Biener <rguenther@suse.de>
4830 PR tree-optimization/87776
4831 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
4832 executable when iterating but running into rpo-vn-max-loop-depth
4833 and not eliding the iteration.
4835 2018-11-30 Jan Hubicka <jh@suse.cz>
4837 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
4840 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
4842 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
4844 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
4845 Combine with rs6000_address_for_fpconvert.
4846 (rs6000_address_for_fpconvert): Combine with
4847 rs6000_force_indexed_or_indirect_mem.
4848 (rs6000_expand_vector_init): Change function call from
4849 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4850 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
4851 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4852 (floatsi<mode>2_lfiwax_mem): Ditto.
4853 (floatunssi<mode>2_lfiwzx): Ditto.
4854 (floatunssi<mode>2_lfiwzx_mem): Ditto.
4855 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
4856 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
4857 (fix_trunc<mode>si2_stfiwx): Ditto.
4858 (fixuns_trunc<mode>si2_stfiwx): Ditto.
4859 (float_<mode>si2_hw): Ditto.
4860 (floatuns_<mode>si2_hw): Ditto.
4861 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
4862 (vsx_splat_<mode>): Ditto.
4864 2018-11-01 Joseph Myers <joseph@codesourcery.com>
4866 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
4867 at end of assembler input text.
4868 * configure: Regenerate.
4870 2018-11-01 Jakub Jelinek <jakub@redhat.com>
4872 PR tree-optimization/87826
4873 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
4874 negative or larger or equal to type's precision.
4876 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
4878 * opts.c (default_options_table): Do not enable
4879 OPT_fdelayed_branch at -Og.
4880 * doc/invoke.texi (-fdelayed-branch): Document it.
4882 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4884 * optabs-libfuncs.c (build_libfunc_function_visibility):
4885 New, split out from...
4886 (build_libfunc_function): ... here.
4887 (init_one_libfunc_visibility): New, split out from ...
4888 (init_one_libfunc): ... here.
4890 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
4891 scratch register need not be early-clobber. Document the reason
4892 why we cannot use ST<OP>.
4894 2018-10-31 Joseph Myers <joseph@codesourcery.com>
4897 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
4898 line for second argument of AC_DEFINE_UNQUOTED.
4899 * doc/install.texi (Tools/packages necessary for modifying GCC):
4900 Update to autoconf 2.69 and automake 1.15.1.
4901 * aclocal.m4, config.in, configure: Regenerate.
4903 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
4905 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
4908 2018-10-31 Martin Liska <mliska@suse.cz>
4911 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
4912 Add new argument invalid_extension.
4913 (aarch64_get_all_extension_candidates): New function.
4914 (aarch64_rewrite_selected_cpu): Add NULL to function call.
4915 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
4917 (aarch64_get_all_extension_candidates): New function.
4918 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
4919 argument invalid_extension.
4920 (aarch64_parse_cpu): Likewise.
4921 (aarch64_print_hint_for_extensions): New function.
4922 (aarch64_validate_mcpu): Provide hint about invalid extension.
4923 (aarch64_validate_march): Likewise.
4924 (aarch64_handle_attr_arch): Pass new argument.
4925 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
4926 (aarch64_handle_attr_isa_flags): Likewise.
4928 2018-10-31 Richard Biener <rguenther@suse.de>
4932 * tree-outof-ssa.c (insert_backedge_copies): Restrict
4933 copy generation to useful cases. Place the copy before
4934 the definition of the backedge value when possible.
4936 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4938 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
4939 * config/arc/arc.c (arc_active_insn): New function.
4940 (check_store_cacheline_hazard): Likewise.
4941 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
4942 (arc_override_options): Disable delay slot scheduler for older
4944 (arc_store_addr_hazard_p): New implementation, old one renamed to
4946 (arc_store_addr_hazard_internal_p): Renamed.
4947 (arc_reorg): Don't combine into brcc instructions which are part
4948 of hardware hazard solution.
4949 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
4950 (tune_arc700): Likewise.
4951 * config/arc/arc.opt (arc7xx): New tune value.
4952 * config/arc/arc700.md: Improve A7 scheduler.
4954 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4956 * config/arc/arc.c (arc_override_options): Remove
4957 TARGET_COMPACT_CASESI.
4958 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
4959 (CASE_VECTOR_MODE): Likewise.
4960 (CASE_VECTOR_PC_RELATIVE): Likewise.
4961 (CASE_VECTOR_SHORTEN_MODE): Likewise.
4962 (CASE_VECTOR_SHORTEN_MODE1): Delete.
4963 (ADDR_VEC_ALIGN): Update.
4964 (ASM_OUTPUT_CASE_LABEL): Undefine.
4965 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
4966 (TARGET_BI_BIH): Define.
4967 (DEFAULT_BRANCH_INDEX): Likewise.
4968 * config/arc/arc.md (casesi): Rework to accept BI/BIH
4969 instructions, remove compact_casesi use case.
4970 (casesi_compact_jump): Remove.
4971 (casesi_dispatch): New pattern.
4972 * config/arc/arc.opt: Add mbranch-index option. Deprecate
4973 compact_casesi option.
4974 * doc/invoke.texi: Document mbranch-index option.
4976 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4978 * config/arc/arc.c (arc_get_tp): Remove function.
4979 (arc_emit_call_tls_get_addr): Likewise.
4980 (arc_call_tls_get_addr): New function.
4981 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
4982 * config/arc/arc.md (tls_load_tp_soft): Remove.
4983 (tls_gd_get_addr): Likewise.
4985 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4987 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
4988 (mulsi3_600_lib): Remove pattern.
4989 (umulsi3_highpart_600_lib_le): Likewise.
4990 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
4991 (umulsidi3): Remove call to umulsidi3_600_lib.
4992 (umulsidi3_600_lib): Remove pattern.
4993 (peephole2): Remove peephole using the above deprecated patterns.
4995 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
4998 * config/arm/arm.c (arm_option_check_internal): Disable the combined
4999 use of -mslow-flash-data and -mword-relocations.
5000 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
5001 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
5003 * doc/invoke.texi (-mword-relocations): Mention conflict with
5005 (-mslow-flash-data): Reciprocally.
5007 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
5009 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
5010 16-byte modes held in GP registers to use an even regno.
5012 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
5013 (aarch64_atomic_ldop_supported_p): Remove.
5014 (aarch64_gen_atomic_ldop): Remove.
5015 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
5016 Fully expand LSE operations here.
5017 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
5018 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
5019 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
5020 and use ATOMIC_LDOP instead; use register_operand for the input;
5021 drop the split and emit insns directly.
5022 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
5023 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
5024 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
5026 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
5027 (aarch64_gen_atomic_ldop): Don't call it.
5028 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
5029 Use aarch64_reg_or_zero.
5030 (aarch64_atomic_exchange<ALLI>): Likewise.
5031 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
5032 operand 0; use aarch64_reg_or_zero for input; merge ...
5033 (@aarch64_atomic_swp<ALLI>): ... this and remove.
5035 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
5036 (aarch64_split_compare_and_swap): Use it.
5037 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
5038 test oldval against the proper predicate.
5039 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
5040 Use nonmemory_operand for expected.
5041 (cas_short_expected_pred): New.
5042 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
5043 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
5044 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
5045 (aarch64_plushi_operand): New.
5047 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
5048 Force oldval into the rval register for TARGET_LSE; emit the compare
5049 during initial expansion so that it may be deleted if unused.
5050 (aarch64_gen_atomic_cas): Remove.
5051 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
5052 Change =&r to +r for operand 0; use match_dup for operand 2;
5053 remove is_weak and mod_f operands as unused. Drop the split
5055 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
5056 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
5057 (@aarch64_atomic_cas<GPI>): Similarly.
5059 2018-10-31 Richard Biener <rguenther@suse.de>
5061 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
5064 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5066 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
5067 * config/aarch64/aarch64.md: Include saphira.md
5068 * config/aarch64/saphira.md: New file for pipeline description.
5070 2018-10-30 Martin Sebor <msebor@redhat.com>
5073 * gimple-ssa-sprintf.c (format_directive): Use %G to include
5075 (sprintf_dom_walker::compute_format_length):
5076 Avoid setting POSUNDER4K here.
5077 (get_destination_size): Handle null argument values.
5078 (get_user_idx_format): New function.
5079 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
5080 functions, including user-defined with attribute format printf.
5081 Use %G to include inlining context.
5082 Set POSUNDER4K here.
5084 2018-10-30 Jan Hubicka <jh@suse.cz>
5086 * params.def (lto-partitions): Bump from 32 to 128.
5088 2018-10-30 Jan Hubicka <jh@suse.cz>
5091 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
5093 (free_lang_data_in_type): Forward declare.
5094 (fld_type_variant_equal_p): New function.
5095 (fld_type_variant): New function
5096 (fld_incomplete_types): New hash.
5097 (fld_incomplete_type_of): New function
5098 (fld_simplfied-type): New function.
5099 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
5100 (free_lang_data): Allocate and free fld_incomplete_type; update call
5101 of free_lang_data_in_decl.
5103 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
5105 * gcov.c (output_lines): Remove duplicate line.
5107 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5109 * config/rs6000/rs6000.md (bswapdi2): Force address into register
5110 if not in indexed or indirect form.
5111 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
5112 (bswapdi2_store): Ditto.
5113 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
5115 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
5116 Prototype for helper function.
5118 2018-10-30 Martin Sebor <msebor@redhat.com>
5120 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
5121 (target, pragma GCC optimize, pragma GCC target): Ditto.
5123 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
5125 * doc/extend.texi: Fix prototype and description of
5126 __builtin_expect_with_probability.
5128 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
5130 * cgraph.h (clone_function_name_1): Replaced by new
5131 clone_function_name_numbered that takes name as string; for
5132 privatize_symbol_name_1 use only.
5133 (clone_function_name): Renamed to
5134 clone_function_name_numbered to be explicit about numbering.
5135 (clone_function_name): New two-argument function that does
5136 not number its output.
5137 (clone_function_name): New three-argument function that
5138 takes a number to append to its output.
5139 * cgraphclones.c (duplicate_thunk_for_node):
5140 (clone_function_name_1): Renamed.
5141 (clone_function_name_numbered): Two new functions.
5142 (clone_function_name): Improved documentation.
5143 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
5144 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
5145 * final.c (final_scan_insn_1): Use the new clone_function_name
5147 * multiple_target.c (create_dispatcher_calls): Ditto.
5148 (create_target_clone): Ditto.
5149 * omp-expand.c (grid_expand_target_grid_body): Ditto.
5150 * omp-low.c (create_omp_child_function_name): Ditto.
5151 * omp-simd-clone.c (simd_clone_create): Ditto.
5152 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
5153 new clone_function_name without numbering.
5155 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
5157 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
5158 Assert that the allocation size is not zero.
5160 2018-10-30 Richard Biener <rguenther@suse.de>
5162 PR tree-optimization/87800
5163 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
5164 non-induction or reduction PHIs.
5166 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
5168 * config/aarch64/falkor-tag-collision-avoidance.c
5169 (execute_tag_collision_avoidance): Call df_note_add_problem.
5171 2018-10-30 Martin Liska <mliska@suse.cz>
5173 * doc/extend.texi: Fix typo in documentation
5174 of __builtin_expect_with_probability.
5176 2018-10-29 David Malcolm <dmalcolm@redhat.com>
5179 * input.c (get_substring_ranges_for_loc): Detect if
5180 linemap_resolve_location gives us a NULL map, and reject
5183 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
5185 * config.gcc (xstormy16-*-elf): Set tm_d_file.
5187 2018-10-29 David Malcolm <dmalcolm@redhat.com>
5188 Martin Sebor <msebor@redhat.com>
5189 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
5191 * doc/ux.texi (Quoting): New subsection, adapted from material at
5192 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
5193 MartinSebor and ManuelLopezIbanez.
5194 (Fix-it hints): Note that fix-it hints shouldn't be marked for
5197 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
5200 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
5203 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
5205 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
5207 2018-10-29 Paul Koning <ni1d@arrl.net>
5209 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
5211 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
5213 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
5214 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
5215 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
5216 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
5217 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
5218 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
5219 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
5220 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
5221 Change 'vector' to '__vector'.
5222 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
5223 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
5224 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
5225 _mm_avg_pu16): Likewise. And, whitespace corrections.
5227 2018-10-29 Richard Biener <rguenther@suse.de>
5229 PR tree-optimization/87785
5230 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
5233 2018-10-29 Olivier Hainque <hainque@adacore.com>
5235 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
5237 2018-10-29 Olivier Hainque <hainque@adacore.com>
5239 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
5240 ports configuration macro, defaults to "gnu".
5241 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
5242 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
5244 2018-10-29 Olivier Hainque <hainque@adacore.com>
5246 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
5248 2018-10-29 Richard Biener <rguenther@suse.de>
5250 PR tree-optimization/87790
5251 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
5252 (vect_make_slp_decision): Adjust.
5253 (vect_slp_analyze_bb_1): Likewise.
5254 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
5257 2018-10-29 Richard Biener <rguenther@suse.de>
5259 PR tree-optimization/87785
5260 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
5262 (vect_build_slp_tree): Likewise.
5263 (vect_gather_slp_loads): New function.
5264 (vect_analyze_slp_instance): Gather loads separately from the
5267 2018-10-29 Martin Liska <mliska@suse.cz>
5269 * Makefile.in: Make dependency to json.o.
5270 * doc/gcov.texi: Document new JSON format, remove
5271 old intermediate format documentation.
5272 * gcov.c (struct function_info): Come up with m_name and
5274 (function_info::function_info): Initialize it.
5275 (function_info::~function_info): Release it.
5276 (main): Rename flag_intermediate_format to flag_json_format.
5277 (print_usage): Describe --json-format.
5278 (process_args): Set flag_json_format.
5279 (output_intermediate_line): Remove.
5280 (output_intermediate_json_line): Likewise.
5281 (get_gcov_intermediate_filename): Return new extension
5283 (output_intermediate_file): Implement JSON emission.
5284 (output_json_intermediate_file): Implement JSON emission.
5285 (generate_results): Use ::get_name for function name.
5286 Handle JSON output file.
5287 (read_graph_file): Use ::get_name instead of cplus_demangle.
5288 (read_count_file): Likewise.
5289 (solve_flow_graph): Likewise.
5290 (add_line_counts): Likewise.
5291 (accumulate_line_counts): Use new flag_json_format.
5292 (output_function_details): Use ::get_name instead of cplus_demangle.
5293 (output_lines): Likewise.
5294 * json.cc (test_writing_literals): Add new tests.
5295 * json.h (class literal): Add new boolean constructor.
5297 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
5299 PR rtl-optimization/87701
5300 PR rtl-optimization/87780
5301 * combine.c (make_more_copies): Rewrite.
5303 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
5305 * doc/generic.texi (ABSU_EXPR): Document.
5306 * match.pd (absu(x)*absu(x) -> x*x): Handle.
5307 (absu(absu(X)) -> absu(X)): Likewise.
5308 (absu(-X) -> absu(X)): Likewise.
5309 (absu(X) where X is nonnegative -> X): Likewise.
5311 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
5313 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
5314 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
5315 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
5316 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
5317 (s-tm-texi): Also check timestamp on d-target.def.
5318 (generated_files): Add TM_D_H and d-target-hooks-def.h.
5319 (build/genhooks.o): Also depend on D_TARGET_DEF.
5320 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
5322 * config/aarch64/aarch64-d.c: New file.
5323 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
5325 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
5327 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
5328 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
5329 * config/arm/arm-d.c: New file.
5330 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
5331 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
5332 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5333 * config/arm/t-arm (arm-d.o): New rule.
5334 * config/default-d.c: New file.
5335 * config/glibc-d.c: New file.
5336 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5337 * config/i386/i386-d.c: New file.
5338 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
5339 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
5340 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5341 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
5342 * config/i386/t-i386 (i386-d.o): New rule.
5343 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5344 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5345 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
5346 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5347 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5348 * config/mips/mips-d.c: New file.
5349 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
5350 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
5351 * config/mips/t-mips (mips-d.o): New rule.
5352 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5353 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5354 * config/powerpcspe/powerpcspe-d.c: New file.
5355 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
5357 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
5358 Support GNU D by using 0 as the language type.
5359 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
5360 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
5361 * config/riscv/riscv-d.c: New file.
5362 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
5364 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
5365 * config/riscv/t-riscv (riscv-d.o): New rule.
5366 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5367 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5368 * config/rs6000/rs6000-d.c: New file.
5369 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
5371 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
5372 Support GNU D by using 0 as the language type.
5373 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
5374 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
5375 * config/s390/s390-d.c: New file.
5376 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
5377 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
5378 * config/s390/t-s390 (s390-d.o): New rule.
5379 * config/sparc/sparc-d.c: New file.
5380 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
5382 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
5383 * config/sparc/t-sparc (sparc-d.o): New rule.
5384 * config/t-glibc (glibc-d.o): New rule.
5385 * configure: Regenerated.
5386 * configure.ac (tm_d_file): New variable.
5387 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
5388 * doc/contrib.texi (Contributors): Add self for the D frontend.
5389 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
5390 * doc/install.texi (Configuration): Mention libphobos as an option for
5391 --enable-shared. Mention d as an option for --enable-languages.
5392 (Testing): Mention check-d as a target.
5393 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
5394 name suffixes. Mention d as a -x option.
5395 * doc/sourcebuild.texi (Top Level): Mention libphobos.
5396 * doc/standards.texi (Standards): Add section on D language.
5397 * doc/tm.texi: Regenerated.
5398 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
5399 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
5400 * dwarf2out.c (is_dlang): New function.
5401 (gen_compile_unit_die): Use DW_LANG_D for D.
5402 (declare_in_namespace): Return module die for D, instead of adding
5403 extra declarations into the namespace.
5404 (gen_namespace_die): Generate DW_TAG_module for D.
5405 (gen_decl_die): Handle CONST_DECLSs for D.
5406 (dwarf2out_decl): Likewise.
5407 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
5408 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
5409 kinds of aggregates.
5410 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
5411 * genhooks.c: Include d/d-target.def.
5413 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
5416 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
5417 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
5418 (STACK_DYNAMIC_OFFSET): Likewise.
5420 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
5423 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
5426 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
5428 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
5429 return type and other typos.
5431 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
5433 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
5434 a shorter sequence with fewer branches.
5435 (emit_final_str_compare_gpr): Ditto.
5437 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5439 * config/rs6000/tmmintrin.h: New file.
5440 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
5442 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5444 * config/rs6000/mmintrin.h: Enable 32bit compilation.
5445 * config/rs6000/xmmintrin.h: Likewise.
5447 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5449 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
5451 2018-10-26 Richard Biener <rguenther@suse.de>
5453 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
5455 (vect_mark_slp_stmts_relevant): Likewise.
5456 (vect_detect_hybrid_slp_stmts): Likewise.
5457 (vect_bb_slp_scalar_cost): Likewise.
5458 (vect_remove_slp_scalar_calls): Likewise.
5460 2018-10-26 Jan Hubicka <jh@suse.cz>
5462 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
5463 (warn_types_mismatch): Fix walk of DECL_NAME.
5464 (odr_types_equivalent_p): Fix overactive assert.
5466 2018-10-26 Richard Biener <rguenther@suse.de>
5468 PR tree-optimization/87105
5469 * tree-vectorizer.h (_slp_tree::refcnt): New member.
5470 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
5472 (vect_create_new_slp_node): Initialize refcnt to one.
5474 (scalar_stmts_set_t, bst_fail): Remove.
5475 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
5476 (vect_build_slp_tree): Add bst_map argument and lookup
5477 already created SLP nodes.
5478 (vect_print_slp_tree): Handle a SLP graph, print SLP node
5480 (vect_slp_rearrange_stmts): Handle a SLP graph.
5481 (vect_analyze_slp_instance): Adjust and free SLP nodes from
5482 the CSE map. Fix indenting.
5483 (vect_schedule_slp_instance): Add short-cut.
5485 2018-10-26 Martin Liska <mliska@suse.cz>
5488 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
5489 addr_expr and not with pointers.
5491 2018-10-26 Jan Hubicka <jh@suse.cz>
5493 * tree.c (free_lang_data_in_type): Only check main variants.
5494 * ipa-devirt.c (warn_odr): Make static.
5495 (types_same_for_odr): Drop strict variant.
5496 (types_odr_comparable): Likewise.
5497 (odr_or_derived_type_p): Look for main variants.
5498 (odr_name_hasher::equal): Cleanup comment.
5499 (odr_subtypes_equivalent): Add warn and warned arguments; check main
5501 (type_variants_equivalent_p): break out from ...
5502 (odr_types_equivalent): ... here; go for main variants where needed.
5503 (warn_odr): ... here; turn static.
5504 (warn_types_mismatch): Compare mangled names of main variants.
5505 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
5506 (type_with_linkage_p): Sanity check that we look at main variant.
5507 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
5508 * tree.h (types_same_for_odr): Drop strict argument.
5510 2018-10-26 Richard Biener <rguenther@suse.de>
5512 PR tree-optimization/87746
5513 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
5514 Simplify and fix WRT strided store groups with size not
5515 equal to step in element count.
5516 (vect_analyze_group_access_1): Dump the whole group.
5518 2018-10-25 Carl Love <cel@us.ibm.com>
5520 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
5521 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
5522 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
5523 precicion entry for each overloaded builtin.
5524 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
5525 VSCEDPUO): Rename overloaded name.
5526 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
5527 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
5528 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
5529 define_expand for xscmpexqp instruction.
5530 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
5532 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5533 Jinsong Ji <jji@us.ibm.com>
5535 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
5536 function with vec_sl.
5537 (_mm_slli_epi32): Likewise.
5538 (_mm_slli_epi64): Likewise.
5539 (_mm_srai_epi16): Replace deprecated function with vec_sra.
5540 (_mm_srai_epi32): Likewise.
5541 (_mm_srli_epi16): Replace deprecated function with vec_sr.
5542 (_mm_srli_epi32): Likewise.
5543 (_mm_srli_epi64): Likewise.
5544 (_mm_sll_epi16): Replace deprecated function with vec_sl.
5545 (_mm_sll_epi32): Likewise.
5546 (_mm_sll_epi64): Likewise.
5547 (_mm_sra_epi16): Replace deprecated function with vec_sra.
5548 (_mm_sra_epi32): Likewise.
5549 (_mm_srl_epi16): Replace deprecated function with vec_sr.
5550 (_mm_srl_epi32): Likewise.
5551 (_mm_srl_epi64): Likewise.
5553 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5554 Jinsong Ji <jji@us.ibm.com>
5556 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
5557 comparison operators with vec_cmp* for compatibility due to
5558 unfortunate history; clean up formatting and use types more
5560 (_mm_sll_epi32): Likewise.
5561 (_mm_sll_epi64): Likewise.
5562 (_mm_srl_epi16): Likewise.
5563 (_mm_srl_epi32): Likewise.
5564 (_mm_srl_epi64): Likewise.
5566 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5567 Jinsong Ji <jji@us.ibm.com>
5569 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
5570 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
5571 __vector __bool int. Use vec_cmpgt in preference to deprecated
5572 function vec_vcmpgtfp.
5573 (_mm_max_ps): Likewise.
5575 2018-10-25 Jeff Law <law@redhat.com>
5577 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
5580 2018-10-25 Martin Sebor <msebor@redhat.com>
5582 * doc/extend.texi (aligned): Expand attribute description.
5583 (Alignment): Rename section. Discuss function arguments.
5585 2018-10-25 Jan Hubicka <jh@suse.cz>
5587 * ipa-devirt.c (main_odr_variant): Remove.
5588 (hash_odr_name, types_same_for_odr, types_odr_comparable,
5589 odr_name_hasher::equal, odr_subtypes_equivalent_p):
5590 Drop use of main_odr_variant.
5591 (add_type_duplicate): Silence confused warnings on integer types.
5592 (get_odr_type): Always look for main variant.
5593 (register_odr_type): Simplify.
5595 2018-10-25 Richard Biener <rguenther@suse.de>
5597 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
5598 Initialize ng to silence error with release checking bootstrap.
5600 2018-10-25 Richard Biener <rguenther@suse.de>
5602 * tree-if-conv.c: Include tree-ssa-sccvn.h.
5603 (tree_if_conversion): Run CSE on the if-converted loop body.
5605 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
5607 * config/s390/constraints.md (ZL): New constraint.
5608 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
5610 * config/s390/s390.md (movdi_larl): Remove.
5611 (movdi_64): Add the LARL alternative.
5613 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
5616 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
5617 (RTX_CODE_HWINT_P): New macro.
5618 (rtx_code_size): Use RTX_CODE_HWINT_P ().
5620 2018-10-25 Jan Hubicka <jh@suse.cz>
5622 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
5625 2018-10-25 Richard Biener <rguenther@suse.de>
5627 PR tree-optimization/87665
5628 PR tree-optimization/87745
5629 * tree-vectorizer.h (get_earlier_stmt): Remove.
5630 (get_later_stmt): Pick up UID from the original non-pattern stmt.
5632 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
5634 * options.texi (Deprecated): Move list to Var section.
5636 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
5637 Jinsong Ji <jji@us.ibm.com>
5639 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
5640 __vector long to __vector long long.
5641 (_mm_cvtpd_ps): Likewise.
5642 (_mm_cvttpd_epi32): Likewise.
5643 (_mm_cvtpi32_pd): Likewise.
5644 (_mm_unpackhi_epi64): Likewise.
5645 (_mm_unpacklo_epi64): Likewise.
5647 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
5649 PR rtl-optimization/87720
5650 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
5652 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
5654 * gimple-ssa-isolate-paths.c
5655 (find_implicit_erroneous_behavior): Do not change code if the
5656 pass is running for warnings only.
5657 (find_explicit_erroneous_behavior): Likewise.
5659 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
5661 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
5662 Define as rs6000_mangle_decl_assembler_name.
5663 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
5664 long double to IEEE long double, switch the names of the long
5665 double built-in functions to be <func>f128 instead of <func>l.
5667 2018-10-24 Martin Sebor <msebor@redhat.com>
5669 * doc/extend.texi (nonnull): List no-argument form. Reference
5670 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
5672 2018-10-24 Richard Biener <rguenther@suse.de>
5674 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
5676 2018-10-24 Martin Liska <mliska@suse.cz>
5678 PR tree-optimization/84436
5679 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
5681 (switch_conversion::contains_linear_function_p): New.
5682 (switch_conversion::build_one_array): Support linear
5683 transformation on input.
5684 * tree-switch-conversion.h (struct switch_conversion): Add
5685 contains_linear_function_p declaration.
5687 2018-10-24 Richard Biener <rguenther@suse.de>
5689 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
5690 if its argument is CONSTANT_CLASS_P.
5692 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5694 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
5695 it is wrong for forward declarations.
5697 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
5699 * config/s390/s390.c (s390_check_qrst_address): Add the missing
5700 SYMBOL_REF_P () check.
5702 2018-10-24 Richard Biener <rguenther@suse.de>
5704 PR tree-optimization/87105
5705 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
5706 dump classification.
5707 (vect_analyze_data_ref_accesses): Handle duplicate loads and
5708 stores by splitting the affected group after the fact.
5709 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
5710 fail the SLP build because of size constraints.
5712 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5714 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
5715 * configure: Regenerate.
5716 * config.in: Regenerate.
5717 * varasm.c (mergeable_string_section): Use readonly_data_section
5718 if linker doesn't support SHF_MERGE with alignment > 8.
5719 (mergeable_constant_section): Likewise.
5721 2018-10-24 Richard Biener <rguenther@suse.de>
5723 PR tree-optimization/84013
5724 * tree-ssa-structalias.c (struct msdi_data): New struct for
5725 marshalling data to walk_stmt_load_store_ops.
5726 (maybe_set_dependence_info): Refactor as callback for
5727 walk_stmt_load_store_ops.
5728 (compute_dependence_clique): Set restrict info on all stmt kinds.
5730 2018-10-24 Martin Liska <mliska@suse.cz>
5732 * cgraph.c (cgraph_node::dump):
5733 Remove reduntant dumps and make tp_first_run dump more compact.
5735 2018-10-24 Richard Biener <rguenther@suse.de>
5737 PR tree-optimization/87665
5738 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
5741 2018-10-12 Jeff Law <law@redhat.com>
5743 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
5746 2018-10-23 Richard Biener <rguenther@suse.de>
5748 * tree-vrp.c (add_assert_info): Guard dump_printf with
5750 * gimple-ssa-evrp-analyze.c
5751 (evrp_range_analyzer::record_ranges_from_incoming_edge):
5752 Use value_range::ignore_equivs_equal_p.
5754 2018-10-23 Richard Biener <rguenther@suse.de>
5756 PR tree-optimization/87105
5757 PR tree-optimization/87608
5758 * passes.def (pass_all_early_optimizations): Add early phi-opt
5760 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
5761 addition to debug stmts.
5762 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
5763 and abs replacement early.
5764 * tree-cfg.c (gimple_empty_block_p): Likewise.
5766 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
5769 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
5770 specified to configure.
5771 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
5773 2018-10-23 Richard Biener <rguenther@suse.de>
5775 PR tree-optimization/87700
5776 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
5778 2018-10-23 Jakub Jelinek <jakub@redhat.com>
5781 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
5782 second argument from __mmask16 to __mmask8.
5783 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
5784 _mm_mask_packs_epi32): Likewise.
5785 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
5787 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
5789 2018-10-23 Richard Biener <rguenther@suse.de>
5791 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
5793 2018-10-23 Richard Biener <rguenther@suse.de>
5795 PR tree-optimization/86144
5796 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
5797 over simd attribute.
5799 2018-10-23 Richard Biener <rguenther@suse.de>
5801 PR tree-optimization/87693
5802 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
5803 the case we do not find the taken edge.
5805 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
5806 Jinsong Ji <jji@us.ibm.com>
5808 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
5809 (_mm_store_pd): Use unaligned vector type for pointer cast.
5810 (_mm_maskmoveu_si128): Likewise.
5811 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
5812 (_mm_store_ps): Use unaligned vector type for pointer cast.
5814 2018-10-22 Paul Koning <ni1d@arrl.net>
5816 * symtab.c (symtab_node::increase_alignment): Correct max
5819 2018-10-22 Yury Gribov <tetra2005@gmail.com>
5821 PR tree-optimization/87633
5822 * match.pd: Do not generate unordered integer comparisons.
5824 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5826 PR rtl-optimization/87600
5827 * combine.c: Add include of expr.h.
5828 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
5829 register to a pseudo.
5830 (make_more_copies): New function, add a copy to a new pseudo after
5831 the moves from hard registers into pseudos.
5832 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
5833 later. Call make_more_copies.
5835 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
5837 * lra-constraints.c (process_alt_operands): New local array,
5838 matching_early_clobber. Check matching_early_clobber before
5839 decrementing reject, and set matching_early_clobber after.
5841 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5844 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
5845 call output_addr_const and hope for the best.
5847 2018-10-22 Richard Biener <rguenther@suse.de>
5849 * gimple-ssa-evrp-analyze.c
5850 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
5851 smarter about what ranges to use.
5852 * tree-vrp.c (add_assert_info): Dump here.
5853 (register_edge_assert_for_2): Instead of here at multiple but
5856 * gcc.dg/tree-ssa/evrp12.c: New testcase.
5857 * gcc.dg/predict-6.c: Adjust.
5858 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
5859 * gcc.dg/tree-ssa/vrp02.c: Likewise.
5860 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
5862 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
5863 Richard Biener <rguenther@suse.de>
5865 * bitmap.h: Update data structure documentation, including a
5866 description of bitmap views as either linked-lists or splay trees.
5867 (struct bitmap_element_def): Update comments for splay tree bitmaps.
5868 (struct bitmap_head_def): Likewise.
5869 (bitmap_list_view, bitmap_tree_view): New prototypes.
5870 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
5872 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
5873 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
5874 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
5875 released bitmap element here.
5876 (bitmap_element_free): Remove.
5877 (bitmap_elt_clear_from): Work on splay tree bitmaps.
5878 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
5879 this function similar ones such that linked-list bitmap implementation
5880 functions are grouped.
5881 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
5882 and moved for grouping.
5883 (bitmap_list_insert_element_after): Renamed from
5884 bitmap_elt_insert_after, and moved for grouping.
5885 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
5886 (bitmap_tree_link_left, bitmap_tree_link_right,
5887 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
5888 bitmap_tree_link_element, bitmap_tree_unlink_element,
5889 bitmap_tree_find_element): New functions for splay-tree bitmap
5891 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
5892 Renamed and moved, see above entries.
5893 (bitmap_tree_listify_from): New function to convert part of a splay
5894 tree bitmap to a linked-list bitmap.
5895 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
5896 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
5897 (bitmap_find_bit): Remove.
5898 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
5899 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
5900 Handle splay tree bitmaps.
5901 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
5902 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
5903 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
5904 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
5905 bitmap_intersect_compl_p, bitmap_ior_and_compl,
5906 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
5907 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
5908 corresponding changes to use linked-list specific bitmap_element
5909 manipulation functions as applicable for efficiency.
5910 (bitmap_tree_to_vec): New function.
5911 (debug_bitmap_elt_file): New function split out from ...
5912 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
5913 (bitmap_print): Likewise.
5915 PR tree-optimization/63155
5916 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
5918 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
5921 2018-10-22 Martin Liska <mliska@suse.cz>
5923 PR tree-optimization/87686
5925 2018-08-29 Martin Liska <mliska@suse.cz>
5927 * tree-switch-conversion.c (switch_conversion::expand):
5928 Strenghten assumption about gswitch statements.
5930 2018-10-22 Martin Liska <mliska@suse.cz>
5932 * ipa-icf.c (sem_item::compare_attributes): Remove.
5933 (sem_item::compare_referenced_symbol_properties): Use
5934 attribute_list_equal instead.
5935 (sem_function::equals_wpa): Likewise.
5936 * ipa-icf.h: Remove compare_attributes.
5938 2018-10-22 Richard Biener <rguenther@suse.de>
5941 * mem-stats.h (mem_usage::operator==): Fix pasto.
5943 2018-10-22 Richard Biener <rguenther@suse.de>
5945 PR tree-optimization/87640
5946 * tree-vrp.c (set_value_range_with_overflow): Decompose
5948 (extract_range_from_binary_expr_1): Adjust.
5950 2018-10-22 Martin Jambor <mjambor@suse.cz>
5952 * tree-eh.h (stmt_could_throw_p): Add function parameter.
5953 (stmt_can_throw_external): Likewise.
5954 (stmt_can_throw_internal): Likewise.
5955 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
5956 (lower_eh_constructs_2): Likewise.
5957 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
5958 (stmt_can_throw_external): Likewise.
5959 (stmt_can_throw_internal): Likewise.
5960 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
5961 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
5962 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
5963 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
5964 (pass_lower_eh_dispatch::execute): Pass cfun to
5965 stmt_can_throw_external.
5966 (cleanup_empty_eh): Likewise.
5967 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
5968 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
5969 stmt_can_throw_external instead of pushing it to cfun.
5970 (symbol_table::create_edge): Likewise.
5971 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
5972 stmt_can_throw_internal.
5973 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
5974 to stmt_could_throw_p.
5975 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
5976 stmt_can_throw_internal.
5977 (pass_store_merging::execute): Likewise.
5978 * gimple-ssa-strength-reduction.c
5979 (find_candidates_dom_walker::before_dom_children): Pass cfun to
5981 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
5982 stmt_can_throw_internal.
5983 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
5984 to stmt_can_throw_external.
5985 (check_stmt): Pass cfun to stmt_could_throw_p.
5986 (check_stmt): Pass cfun to stmt_can_throw_external.
5987 (pass_nothrow::execute): Likewise.
5988 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
5989 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
5990 stmt_can_throw_internal.
5991 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
5992 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
5993 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
5994 * tree-complex.c (expand_complex_libcall): Pass cfun to
5995 stmt_could_throw_p and to stmt_can_throw_internal.
5996 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
5997 * tree-inline.c (copy_edges_for_bb): Likewise.
5998 (maybe_move_debug_stmts_to_successors): Likewise.
5999 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
6001 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
6002 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
6003 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
6004 stmt_can_throw_internal.
6005 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
6006 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
6008 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
6009 stmt_can_throw_internal.
6010 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
6011 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
6013 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
6014 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
6015 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
6016 (convert_mult_to_fma_1): Likewise.
6017 (convert_to_divmod): Likewise.
6018 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
6019 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
6020 * tree-ssa-propagate.c
6021 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
6022 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
6023 (maybe_optimize_range_tests): Likewise.
6024 (linearize_expr_tree): Likewise.
6025 (reassociate_bb): Likewise.
6026 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
6027 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
6028 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
6029 (handle_char_store): Likewise.
6030 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
6031 stmt_can_throw_internal.
6032 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
6034 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
6035 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
6036 (vectorizable_simd_clone_call): Likewise.
6037 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
6038 (gimple_stringop_fixed_value): Likewise.
6040 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
6042 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
6043 literal pool references.
6044 (s390_check_qrst_address): Adapt to the new behavior of
6045 s390_loadrelative_operand_p ().
6047 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6050 * config/i386/sse.md (*andnot<mode>3_bcst): New.
6052 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6055 * config/i386/sse.md (*<code><mode>3_bcst): New.
6057 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6060 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
6061 V4DI, V16SI and V8DI.
6062 (*sub<mode>3<mask_name>_bcst): New.
6063 (*add<mode>3<mask_name>_bcst): Likewise.
6065 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
6066 Jinsong Ji <jji@us.ibm.com>
6068 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
6069 __m64 with __vector unsigned long long for compatibility.
6070 (_mm_movemask_epi8): Likewise.
6071 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
6072 (_mm_cvttps_pi32): Likewise.
6073 (_mm_cvtpi32_ps): Likewise.
6074 (_mm_cvtps_pi16): Likewise.
6075 (_mm_loadh_pi): Likewise.
6076 (_mm_storeh_pi): Likewise.
6077 (_mm_movehl_ps): Likewise.
6078 (_mm_movelh_ps): Likewise.
6079 (_mm_loadl_pi): Likewise.
6080 (_mm_storel_pi): Likewise.
6081 (_mm_movemask_ps): Likewise.
6082 (_mm_shuffle_pi16): Likewise.
6084 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6087 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
6088 __builtin_ia32_vfnmsubpd512_mask.
6089 (_mm512_mask_fnmsub_round_pd): Likewise.
6090 (_mm512_fnmsub_pd): Likewise.
6091 (_mm512_mask_fnmsub_pd): Likewise.
6092 (_mm512_maskz_fnmsub_round_pd): Use
6093 __builtin_ia32_vfnmsubpd512_maskz.
6094 (_mm512_maskz_fnmsub_pd): Likewise.
6095 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
6096 (_mm512_mask_fnmsub_round_ps): Likewise.
6097 (_mm512_fnmsub_ps): Likewise.
6098 (_mm512_mask_fnmsub_ps): Likewise.
6099 (_mm512_maskz_fnmsub_round_ps): Use
6100 __builtin_ia32_vfnmsubps512_maskz.
6101 (_mm512_maskz_fnmsub_ps): Likewise.
6102 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
6103 __builtin_ia32_vfnmsubpd256_mask.
6104 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
6105 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
6106 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
6107 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
6108 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
6109 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
6110 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
6111 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
6112 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
6113 __builtin_ia32_vfnmsubpd.
6114 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
6115 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
6116 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
6117 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
6118 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
6119 * config/i386/i386-builtin.def: Add
6120 __builtin_ia32_vfnmsubpd256_mask,
6121 __builtin_ia32_vfnmsubpd256_maskz,
6122 __builtin_ia32_vfnmsubpd128_mask,
6123 __builtin_ia32_vfnmsubpd128_maskz,
6124 __builtin_ia32_vfnmsubps256_mask,
6125 __builtin_ia32_vfnmsubps256_maskz,
6126 __builtin_ia32_vfnmsubps128_mask,
6127 __builtin_ia32_vfnmsubps128_maskz,
6128 __builtin_ia32_vfnmsubpd512_mask,
6129 __builtin_ia32_vfnmsubpd512_maskz,
6130 __builtin_ia32_vfnmsubps512_mask,
6131 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
6132 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
6133 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
6134 __builtin_ia32_vfnmsubpd256.
6135 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
6136 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
6137 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
6139 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
6141 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
6143 (fmai_vmfnmsub_<mode><round_name>): Likewise.
6145 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6148 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
6149 __builtin_ia32_vfnmaddpd512_mask.
6150 (_mm512_mask_fnmadd_round_pd): Likewise.
6151 (_mm512_fnmadd_pd): Likewise.
6152 (_mm512_mask_fnmadd_pd): Likewise.
6153 (_mm512_maskz_fnmadd_round_pd): Use
6154 __builtin_ia32_vfnmaddpd512_maskz.
6155 (_mm512_maskz_fnmadd_pd): Likewise.
6156 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
6157 (_mm512_mask_fnmadd_round_ps): Likewise.
6158 (_mm512_fnmadd_ps): Likewise.
6159 (_mm512_mask_fnmadd_ps): Likewise.
6160 (_mm512_maskz_fnmadd_round_ps): Use
6161 __builtin_ia32_vfnmaddps512_maskz.
6162 (_mm512_maskz_fnmadd_ps): Likewise.
6163 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
6164 __builtin_ia32_vfnmaddpd256_mask.
6165 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
6166 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
6167 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
6168 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
6169 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
6170 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
6171 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
6172 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
6173 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
6174 __builtin_ia32_vfnmaddpd.
6175 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
6176 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
6177 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
6178 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
6179 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
6180 * config/i386/i386-builtin.def: Add
6181 __builtin_ia32_vfnmaddpd256_mask,
6182 __builtin_ia32_vfnmaddpd256_maskz,
6183 __builtin_ia32_vfnmaddpd128_mask,
6184 __builtin_ia32_vfnmaddpd128_maskz,
6185 __builtin_ia32_vfnmaddps256_mask,
6186 __builtin_ia32_vfnmaddps256_maskz,
6187 __builtin_ia32_vfnmaddps128_mask,
6188 __builtin_ia32_vfnmaddps128_maskz,
6189 __builtin_ia32_vfnmaddpd512_mask,
6190 __builtin_ia32_vfnmaddpd512_maskz,
6191 __builtin_ia32_vfnmaddps512_mask,
6192 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
6193 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
6194 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
6195 __builtin_ia32_vfnmaddpd256.
6196 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
6197 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
6198 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
6200 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
6202 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
6204 (fmai_vmfnmadd_<mode><round_name>): Likewise.
6206 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6209 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
6210 __builtin_ia32_vfmsubpd512_mask.
6211 (_mm512_mask_fmsub_round_pd): Likewise.
6212 (_mm512_fmsub_pd): Likewise.
6213 (_mm512_mask_fmsub_pd): Likewise.
6214 (_mm512_maskz_fmsub_round_pd): Use
6215 __builtin_ia32_vfmsubpd512_maskz.
6216 (_mm512_maskz_fmsub_pd): Likewise.
6217 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
6218 (_mm512_mask_fmsub_round_ps): Likewise.
6219 (_mm512_fmsub_ps): Likewise.
6220 (_mm512_mask_fmsub_ps): Likewise.
6221 (_mm512_maskz_fmsub_round_ps): Use
6222 __builtin_ia32_vfmsubps512_maskz.
6223 (_mm512_maskz_fmsub_ps): Likewise.
6224 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
6225 __builtin_ia32_vfmsubpd256_mask.
6226 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
6227 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
6228 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
6229 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
6230 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
6231 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
6232 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
6233 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
6234 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
6235 __builtin_ia32_vfmsubpd.
6236 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
6237 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
6238 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
6239 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
6240 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
6241 * config/i386/i386-builtin.def: Add
6242 __builtin_ia32_vfmsubpd256_mask,
6243 __builtin_ia32_vfmsubpd256_maskz,
6244 __builtin_ia32_vfmsubpd128_mask,
6245 __builtin_ia32_vfmsubpd128_maskz,
6246 __builtin_ia32_vfmsubps256_mask,
6247 __builtin_ia32_vfmsubps256_maskz,
6248 __builtin_ia32_vfmsubps128_mask,
6249 __builtin_ia32_vfmsubps128_maskz,
6250 __builtin_ia32_vfmsubpd512_mask,
6251 __builtin_ia32_vfmsubpd512_maskz,
6252 __builtin_ia32_vfmsubps512_mask,
6253 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
6254 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
6255 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
6256 __builtin_ia32_vfmsubpd256.
6257 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
6258 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
6259 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
6261 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
6263 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
6265 (fmai_vmfmsub_<mode><round_name>): Likewise.
6267 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6269 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
6270 Remove plus. Renamed to ...
6271 (*sub<mode>3<mask_name>_bcst): This.
6272 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
6273 (*add<mode>3<mask_name>_bcst): This.
6275 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6278 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
6280 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6283 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
6284 (_mm_or_epi32): Likewise.
6285 (_mm256_xor_epi32): Likewise.
6286 (_mm_xor_epi32): Likewise.
6287 (_mm256_or_epi64): Likewise.
6288 (_mm_or_epi64): Likewise.
6289 (_mm256_xor_epi64): Likewise.
6290 (_mm_xor_epi64): Likewise.
6292 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
6295 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
6297 2018-10-20 Jakub Jelinek <jakub@redhat.com>
6300 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
6302 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
6304 * doc/ux.texi: Move @section directly after @node.
6306 2018-10-19 Jakub Jelinek <jakub@redhat.com>
6310 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
6311 depend closely nested inside of loop with ordered clause with
6314 2018-10-19 David Malcolm <dmalcolm@redhat.com>
6316 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
6317 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
6318 * doc/ux.texi: New file.
6320 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6322 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
6323 be the first CR field allocated.
6325 2018-10-19 Richard Biener <rguenther@suse.de>
6328 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
6329 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
6331 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
6334 * config/i386/sse.md
6335 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
6336 (*add<mode>3<mask_name>_bcst_2): Likewise.
6338 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
6340 * config/i386/sse.md
6341 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
6342 Replace nonimmediate_operand with register_operand.
6343 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
6345 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
6348 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
6350 PR rtl-optimization/87596
6351 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
6352 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
6353 for instructions in FROM..TO range.
6355 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
6357 * cfgexpand.c (expand_one_var): Use specific wording in error message
6358 for non-local frame variables.
6359 * stor-layout.c (layout_decl): Do not issue a warning for them.
6361 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
6363 * haifa-sched.c (priority): Add force_recompute parameter.
6364 (apply_replacement): Call priority () with force_recompute = true.
6365 (restore_pattern): Likewise.
6367 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6369 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
6370 HOST_BITS_PER_WIDE_INT.
6371 (test_vector_ops_duplicate): Likewise.
6373 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6376 * config/i386/sse.md (VF_AVX512): New.
6377 (avx512bcst): Likewise.
6378 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
6380 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
6382 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
6385 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
6387 * doc/invoke.texi (-dumpversion): Improve grammar.
6388 (-dumpfullversion): Make more consistent with -dumpversion.
6390 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
6392 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
6393 Set JUMP_LABEL to the jump insn.
6394 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
6395 Predict emitted jump and add label to jump insn.
6397 2018-10-18 David Malcolm <dmalcolm@redhat.com>
6399 PR tree-optimization/87562
6400 * input.c (get_substring_ranges_for_loc): Use
6401 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
6402 getting the linemap for the endpoint. Verify that it's either
6403 in the same linemap as the start point's spelling location, or
6404 at least in the same file.
6406 2018-10-18 Richard Biener <rguenther@suse.de>
6408 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
6409 feed width-specific load/store costs through ix86_vec_cost.
6410 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
6411 (k8_cost): Likewise.
6412 (bdver_cost): Likewise.
6413 (znver1_cost): Likewise.
6414 (btver1_cost): Likewise.
6415 (btver2_cost): Likewise.
6417 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6419 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
6420 to simplify subreg of vec_merge.
6422 2018-10-18 Richard Biener <rguenther@suse.de>
6424 * config/i386/i386.c: Fix costing of vector FMA.
6426 2018-10-18 Richard Biener <rguenther@suse.de>
6428 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
6430 (ix86_builtin_vectorization_cost): For vec_construct properly
6431 cost insertion into SSE regs.
6432 (...): Adjust calls to ix86_vec_cost.
6434 2018-10-18 Richard Biener <rguenther@suse.de>
6438 2018-02-07 Richard Biener <rguenther@suse.de>
6440 PR tree-optimization/84204
6441 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
6444 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6447 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
6449 (test_vector_ops_duplicate): Add test for a scalar subreg of a
6450 VEC_MERGE of a VEC_DUPLICATE.
6452 2018-10-17 Joseph Myers <joseph@codesourcery.com>
6454 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
6455 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
6456 * doc/standards.texi (C Language): Document C2X.
6457 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
6458 (rl78_option_override): Handle "GNU C2X" language name.
6460 2018-10-17 Joseph Myers <joseph@codesourcery.com>
6462 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
6463 Document C17 as published in 2018.
6465 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
6468 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
6469 bail out if both sides do not have the same storage order.
6471 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6473 * bitmap.c (bitmap_head::dump): New.
6474 * bitmap.h (bitmap_head): Add dump().
6475 * gimple-ssa-evrp-analyze.c
6476 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
6477 (evrp_range_analyzer::set_ssa_range_info): Same.
6478 (evrp_range_analyzer::record_ranges_from_phis): Same.
6479 (evrp_range_analyzer::record_ranges_from_stmt): Same.
6480 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
6481 * gimple-ssa-sprintf.c (get_int_range): Same.
6482 (format_integer): Same.
6483 (sprintf_dom_walker::handle_gimple_call): Same.
6484 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
6485 (ipcp_vr_lattice::top_p): Same.
6486 (ipcp_vr_lattice::bottom_p): Same.
6487 (ipcp_vr_lattice::set_to_bottom): Same.
6488 (ipa_vr_operation_and_type_effects): Same.
6489 (propagate_vr_across_jump_function): Same.
6490 (ipcp_store_vr_results): Same.
6491 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
6492 (ipa_print_node_jump_functions_for_edge): Same.
6493 (ipa_get_value_range): Same.
6494 (ipa_compute_jump_functions_for_edge): Same.
6495 (ipa_write_jump_function): Same.
6496 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
6497 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
6499 * vr-values.c (set_value_range_to_nonnegative): Same.
6500 (set_value_range_to_truthvalue): Same.
6501 (vr_values::get_value_range): Same.
6502 (vr_values::set_defs_to_varying): Same.
6503 (vr_values::update_value_range): Same.
6504 (symbolic_range_based_on_p): Same.
6505 (vr_values::op_with_boolean_value_range_p): Same.
6506 (vr_values::extract_range_for_var_from_comparison_expr): Same.
6507 (vr_values::extract_range_from_ssa_name): Same.
6508 (vr_values::extract_range_from_binary_expr): Same.
6509 (vr_values::extract_range_from_unary_expr): Same.
6510 (vr_values::extract_range_from_cond_expr): Same.
6511 (vr_values::extract_range_from_comparison): Same.
6512 (vr_values::check_for_binary_op_overflow): Same.
6513 (vr_values::extract_range_basic): Same.
6514 (vr_values::extract_range_from_assignment): Same.
6515 (compare_ranges): Same.
6516 (compare_range_with_value): Same.
6517 (vr_values::adjust_range_with_scev): Same.
6518 (vrp_valueize): Same.
6519 (vrp_valueize_1): Same.
6520 (vr_values::get_vr_for_comparison): Same.
6521 (vr_values::compare_name_with_value): Same.
6522 (vr_values::compare_names): Same.
6523 (vr_values::vrp_evaluate_conditional): Same.
6524 (find_case_label_ranges): Same.
6525 (vr_values::vrp_visit_switch_stmt): Same.
6526 (vr_values::extract_range_from_phi_node): Same.
6527 (vr_values::simplify_div_or_mod_using_ranges): Same.
6528 (vr_values::simplify_bit_ops_using_ranges): Same.
6529 (test_for_singularity): Same.
6530 (range_fits_type_p): Same.
6531 (vr_values::simplify_cond_using_ranges_1): Same.
6532 (vr_values::simplify_switch_using_ranges): Same.
6533 (vr_values::simplify_float_conversion_using_ranges): Same.
6534 (vr_values::two_valued_val_range_p): Same.
6535 (vr_values::add_equivalence): Move to value_range::equiv_add.
6536 * vr-values.h (vr_values::add_equivalence): Remove.
6537 (VR_INITIALIZER): Remove.
6538 * tree-vrp.c (value_range::set): New.
6539 (value_range::equiv_add): New.
6540 (value_range::value_range): New.
6541 (value_range::deep_copy): New.
6542 (value_range::check): New.
6543 (value_range::equal_p): New.
6544 (value_range::ignore_equivs_equal_p): New.
6545 (value_range::operator==): New.
6546 (value_range::operator!=): New.
6547 (value_range::symbolic_p): New.
6548 (value_range::numeric_p): New.
6549 (value_range::set_undefined): New.
6550 (value_range::set_varying): New.
6551 (value_range::may_contain_p): New.
6552 (value_range::equiv_clear): New.
6553 (value_range::singleton_p): New.
6554 (value_range::intersect): New.
6555 (value_range::dump): New.
6556 (value_range::set_and_canonicalize): New.
6557 (set_value_range): Adjust for value_range API.
6558 (set_value_range_to_undefined): Same.
6559 (set_value_range_to_varying): Same.
6560 (set_and_canonicalize_value_range): Same.
6561 (set_value_range_to_nonnull): Same.
6562 (set_value_range_to_null): Same.
6563 (range_is_null): Same.
6564 (range_is_nonnull): Same.
6565 (range_int_cst_p): Same.
6566 (range_int_cst_singleton_p): Same.
6567 (symbolic_range_p): Same.
6568 (range_includes_zero_p): Same.
6569 (value_range_constant_singleton): Same.
6570 (vrp_set_zero_nonzero_bits): Same.
6571 (ranges_from_anti_range): Same.
6572 (extract_range_into_wide_ints): Same.
6573 (extract_range_from_multiplicative_op): Same.
6574 (set_value_range_with_overflow): Same.
6575 (extract_range_from_binary_expr_1): Same.
6576 (extract_range_from_unary_expr): Same.
6577 (dump_value_range): Same.
6578 (debug_value_range): Same.
6579 (vrp_prop::check_array_ref): Same.
6580 (vrp_prop::check_mem_ref): Same.
6581 (vrp_prop::vrp_initialize): Same.
6582 (vrp_prop::visit_stmt): Same.
6583 (intersect_ranges): Same.
6584 (vrp_prop::visit_phi): Same.
6585 (vrp_prop::vrp_finalize): Same.
6586 (determine_value_range_1): Same.
6587 (determine_value_range): Same.
6588 (vrp_intersect_ranges_1): Rename to...
6589 (vrp_intersect_1): this.
6590 (vrp_intersect_ranges): Rename to...
6591 (value_range::intersect_helper): ...this.
6592 (vrp_meet_1): Rename to...
6593 (value_range::union_helper): ...this.
6594 (vrp_meet): Rename to...
6595 (value_range::union_): ...this.
6596 (copy_value_range): Remove.
6597 * tree-vrp.h (struct value_range): Rewrite into a proper class.
6598 (value_range::vrtype): New.
6599 (value_range::type): New.
6600 (value_range::equiv): New.
6601 (value_range::min): New.
6602 (value_range::max): New.
6603 (value_range::varying_p): New.
6604 (value_range::undefined_p): New.
6605 (value_range::null_p): New.
6606 (value_range::equiv_add): New.
6607 (copy_value_range): Remove.
6609 2018-10-17 David Malcolm <dmalcolm@redhat.com>
6611 * Makefile.in (SELFTEST_TARGETS): New.
6612 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
6613 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
6614 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
6616 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
6617 (selftest-c++-gdb, selftest-c++-valgrind): Move to
6619 * configure: Regenerate.
6620 * configure.ac (selftest_languages): New.
6622 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6624 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
6625 overflow wraps argument.
6626 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
6627 wide_int_range_multiplicative_op.
6628 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
6629 overflow wraps argument.
6630 (wide_int_range_multiplicative_op): Same.
6631 (wide_int_range_lshift): Same.
6632 (wide_int_range_div): Same.
6633 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
6634 (wide_int_range_lshift): Same.
6635 (wide_int_range_div): Same.
6637 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6639 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
6640 use sign as argument.
6641 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
6642 wide_int_range_shift_undefined_p.
6644 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6646 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
6648 (@despeculate_copy<ALLI_TI:mode>): ... This.
6649 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
6652 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6654 * config.gcc: Obsolete *-*-solaris2.10*.
6655 * doc/install.texi (Specific, *-*-solaris2*): Document it.
6657 2018-10-12 Jeff Law <law@redhat.com>
6659 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
6660 reg + sym +- const_int addressing modes.
6662 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6664 * common.opt (fdiagnostics-minimum-margin-width=): New option.
6665 * diagnostic-show-locus.c (layout::layout): Apply the minimum
6667 (layout::start_annotation_line): Only print up to 3 of the
6668 margin character, to avoid touching the left-hand side.
6669 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
6670 minimum margin width, as set by test_diagnostic_context's ctor.
6671 (selftest::test_fixit_insert_containing_newline): Likewise.
6672 (selftest::test_fixit_insert_containing_newline_2): Likewise.
6673 (selftest::test_line_numbers_multiline_range): Clear
6674 dc.min_margin_width.
6675 * diagnostic.c (diagnostic_initialize): Initialize
6677 * diagnostic.h (struct diagnostic_context): Add field
6679 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
6680 * opts.c (common_handle_option): Handle
6681 OPT_fdiagnostics_minimum_margin_width_.
6682 * selftest-diagnostic.c
6683 (selftest::test_diagnostic_context::test_diagnostic_context):
6684 Initialize min_margin_width to 6.
6685 * toplev.c (general_init): Initialize global_dc->min_margin_width.
6687 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6689 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6690 Fix usage of "error_at_rich_loc" in the comment.
6692 2018-10-15 Renlin Li <renlin.li@arm.com>
6695 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
6696 if-conversioned loop when it contains ifn with types not
6697 supported by backend.
6698 * internal-fn.c (expand_direct_optab_fn): Add an assert.
6699 (direct_internal_fn_supported_p): New helper function.
6700 * internal-fn.h (direct_internal_fn_supported_p): Declare.
6702 2018-10-15 Jakub Jelinek <jakub@redhat.com>
6705 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6706 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6707 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6709 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
6711 PR tree-optimization/87022
6712 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
6713 bits in dist vector rather than the first one.
6715 2018-10-15 Richard Biener <rguenther@suse.de>
6718 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
6719 (visit_loadstore): When a used restrict tag escaped verify that
6720 the points-to solution of "other" pointers do not include
6722 (compute_dependence_clique): If a used restrict tag escaped
6723 communicated that down to visit_loadstore.
6725 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
6727 * config/s390/s390.c (s390_expand_vec_init): Force vector element
6728 into reg if it isn't a general operand.
6730 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6733 * config/i386/sse.md (*vec_dupv2di): Add register source to
6736 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6739 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6740 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
6741 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6742 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6744 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
6746 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
6747 (notice_args_size): Set it in the current trace if no insn that can
6748 throw internally has been seen yet.
6749 (connect_traces): When connecting args_size between traces, allow the
6750 incoming values not to match if there is an insn setting it before the
6751 first insn that can throw internally; in that case, force the creation
6752 of a CFI note on this latter insn.
6754 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
6756 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
6757 the base class, because using the injected-class-name was not clearly
6758 specified until DR 176.
6760 2018-10-12 Paul Koning <ni1d@arrl.net>
6762 * config/pdp11/pdp11.md (doloop_end): New expander.
6763 (doloop_end_insn): renamed from "doloop_end".
6764 (addqi3): New pattern.
6765 (subqi3): New pattern.
6766 * config/pdp11/predicates.md (incdec_operand): New predicate.
6768 2018-10-12 Yury Gribov <tetra2005@gmail.com>
6771 * real.c (format_helper::can_represent_integral_type_p): New function
6772 * real.h (format_helper::can_represent_integral_type_p): Ditto.
6773 * match.pd: New pattern.
6775 2018-10-12 Alexandre Oliva <oliva@adacore.com>
6777 * configure.ac: Introduce --enable-large-address-aware
6778 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
6779 * doc/install.texi: Document it.
6780 * configure, config.in: Rebuilt.
6781 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
6782 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
6783 (LINK_SPEC): Insert it.
6784 * config/i386/mingw-w64.h: Likewise.
6786 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
6788 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
6790 PR rtl-optimization/87600
6791 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
6793 2018-10-12 Paul Koning <ni1d@arrl.net>
6795 * doc/md.texi (doloop_end): Document that the pattern code may
6796 need to check operand mode.
6798 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
6800 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
6801 to zero-extend between int and floating-point registers.
6802 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
6803 ldp into floating-point registers. Add type and arch attributes.
6804 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
6805 Use f_loads for type attribute.
6807 2018-10-11 Martin Sebor <msebor@redhat.com>
6809 * doc/extend.texi (attribute packed): Correct typos.
6811 2018-10-11 Martin Sebor <msebor@redhat.com>
6813 * doc/extend.texi (attribute flatten): Mention interaction with
6816 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
6819 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
6820 Set new_decl virtual flag to zero.
6822 2018-10-11 Martin Sebor <msebor@redhat.com>
6825 * doc/extend.texi (attribute format_arg): Discuss using multiple
6826 attributes on a single function.
6828 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
6830 PR tree-optimization/86829
6831 * match.pd (sin (atan (x))): New simplification rules.
6832 (cos (atan (x))): Likewise.
6833 * real.c (build_sinatan_real): New function.
6834 * real.h (build_sinatan_real): Prototype.
6836 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
6838 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
6840 (fold_mergeeo_helper): New helper function.
6841 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
6842 intrinsics. Correct some whitespace indentation issues.
6844 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
6847 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
6848 Use HOST_WIDE_INT_1U for shift.
6850 2018-10-11 Doug Rupp <rupp@adacore.com>
6851 Olivier Hainque <hainque@adacore.com>
6853 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
6854 Pass --relax to the linker for RTPs.
6855 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
6857 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
6858 Jan Hubicka <jh@suse.cz>
6859 Martin Jambor <mjambor@suse.cz>
6861 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
6862 the same elements are repeated rather than printing all of them.
6863 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
6864 "repeated" elements.
6865 * read-rtl-function.c (test_loading_repeat): New function.
6866 (read_rtl_function_c_tests): Call test_loading_repeat.
6867 * rtl-tests.c (test_dumping_repeat): New function.
6868 (rtl_tests_c_tests): Call test_dumping_repeat.
6870 2018-10-11 Richard Biener <rguenther@suse.de>
6872 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
6873 bdver?_cost): Unify to ...
6874 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
6875 * config/i386/i386.c (processor_cost_table): Adjust.
6877 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
6880 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
6881 the thunk when expanding to GIMPLE.
6883 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
6886 * varasm.c (mergeable_string_section): Don't try to move zero-length
6887 strings to the merge section.
6889 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
6892 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
6894 2018-10-10 Jakub Jelinek <jakub@redhat.com>
6897 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
6898 to special_args set.
6900 2018-10-10 Richard Biener <rguenther@suse.de>
6902 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
6903 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
6904 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
6905 and recursing and pattern terminating the recursion on SSE
6906 vector width using ix86_expand_reduc.
6907 (reduc_sminmax_scal_<mode>): Split into part reducing to half
6908 width and recursing and SSE2 vector variant doing the final
6909 reduction with ix86_expand_reduc.
6910 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
6911 with terminating the recursion at AVX level, splitting that
6914 2018-10-09 David Malcolm <dmalcolm@redhat.com>
6916 * genmatch.c (error_cb): Rename to...
6917 (diagnostic_cb): ...this, converting int params to enums.
6918 (fatal_at): Update for renaming.
6919 (warning_at): Likewise.
6921 * input.c (selftest::ebcdic_execution_charset::apply):
6922 Update for renaming of...
6923 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
6925 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
6926 converting level and reason to enums.
6927 (class selftest::lexer_error_sink): Rename to...
6928 (class selftest::lexer_test_options): ...this, renaming field
6929 "m_errors" to "m_diagnostics".
6930 (selftest::lexer_test_options::apply): Update for renaming of...
6931 (selftest::lexer_test_options::on_error): ...this, renaming to...
6932 (selftest::lexer_test_options::on_diagnostic): ...this
6933 converting level and reason to enums.
6934 (selftest::test_lexer_string_locations_raw_string_unterminated):
6935 Update for renamings.
6936 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
6939 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
6941 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
6942 * config/rs6000/pmmintrin.h: New file.
6944 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
6946 PR tree-optimization/86659
6947 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
6949 2018-10-09 Richard Biener <rguenther@suse.de>
6951 PR tree-optimization/63155
6952 * tree-ssa-structalias.c: Include tree-ssa.h.
6953 (get_constraint_for_ssa_var): For undefs return nothing_id.
6954 (find_func_aliases): Cleanup PHI handling.
6956 2018-10-09 Richard Biener <rguenther@suse.de>
6958 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
6961 2018-10-09 Martin Liska <mliska@suse.cz>
6963 * asan.c (asan_emit_stack_protection): If a stack variable
6964 is located in a same file as current function, then emit
6965 line info into variable definition string.
6967 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6969 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
6972 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6974 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
6977 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6979 PR tree-optimization/86659
6980 * gimple-match.h (struct gimple_match_op): Add reverse field.
6981 (gimple_match_op::set_op): New overloaded method.
6982 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
6983 the REF_REVERSE_STORAGE_ORDER flag on the value.
6984 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
6985 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
6987 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
6990 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
6991 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
6993 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
6996 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
6997 Defined with __builtin_ia32_vfmaddsubpd512_mask.
6999 2018-10-08 Richard Biener <rguenther@suse.de>
7001 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
7002 cost the same as AVX128 ones.
7004 2018-10-08 Paul Koning <ni1d@arrl.net>
7006 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
7007 (expand_block_move): New function.
7008 * config/pdp11/pdp11.c (output_block_move): Remove.
7009 (expand_block_move): New function.
7010 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
7011 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
7012 (*movmemhi1): Remove.
7014 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
7016 * config/s390/2827.md: Increase latencies for some FP instructions.
7018 2018-10-08 Richard Biener <rguenther@suse.de>
7020 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
7022 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
7023 * tree-vectorizer.h (dump_stmt_cost): Adjust.
7024 (add_stmt_cost): Dump return value of the hook.
7026 2018-10-08 Richard Biener <rguenther@suse.de>
7028 PR tree-optimization/63155
7029 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
7030 (ssa_propagation_engine::ssa_propagate): Remove redundant
7031 bitmap bit clearing.
7033 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
7035 PR rtl-optimization/86939
7036 PR rtl-optimization/87479
7037 * ira.h (non_conflicting_reg_copy_p): New prototype.
7038 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
7039 (make_hard_regno_dead): Don't add conflicts for register
7040 ignore_reg_for_conflicts.
7041 (make_object_dead): Likewise.
7042 (non_conflicting_reg_copy_p): New function.
7043 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
7044 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
7045 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
7046 (make_hard_regno_dead): Don't add conflicts for register
7047 ignore_reg_for_conflicts. Remove special conflict handling of
7048 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
7049 check_pic_pseudo_p and update callers.
7050 (mark_pseudo_dead): Don't add conflicts for register
7051 ignore_reg_for_conflicts.
7052 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
7054 2018-10-05 Andrew Waterman <andrew@sifive.com>
7055 Jim Wilson <jimw@sifive.com>
7057 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
7058 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
7059 new pattern using HONOR_SNANS that emits one extra instruction.
7061 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7063 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
7064 patterns): Merge SI and DI patterns to a GPR pattern.
7065 (unnamed define_insn and define_split for record form of that): Merge
7066 to a single define_insn_and_split pattern.
7068 2018-10-05 David Malcolm <dmalcolm@redhat.com>
7071 * input.c (expand_location_to_spelling_point): Add param "aspect"
7072 and use rather than hardcoding LOCATION_ASPECT_CARET.
7073 (get_substring_ranges_for_loc): Handle the case of a single token
7074 within a macro expansion.
7075 * input.h (expand_location_to_spelling_point): Add "aspect" param,
7076 defaulting to LOCATION_ASPECT_CARET.
7078 2018-10-05 Paul Koning <ni1d@arrl.net>
7080 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
7081 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
7082 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
7083 (pdp11_guard_type): New function.
7085 2018-10-05 Paul Koning <ni1d@arrl.net>
7087 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
7088 * config/pdp11/pdp11.opt (mfloat32): Remove.
7090 * doc/invoke.texi (pdp11 -mfloat32): Remove:
7091 (pdp11 -mfloat64): Remove.
7093 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7095 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
7096 (*cmp<mode>_cc_i387): Ditto.
7097 (*cmpu<mode>_cc_i387): Ditto.
7098 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7099 * config/i386/i386.c (ix86_expand_fp_compare): Remove
7101 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
7102 Emit x86_sahf_1 pattern.
7103 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
7104 (ix86_expand_carry_flag_compare): Ditto.
7106 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7108 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
7109 to reg_or_0_operand. Add "C" constraint.
7110 (*cmpxf_cc_i387): Ditto.
7111 (*cmp<mode>_i387): Change operand 2 predicate
7112 to nonimm_or_0_operand. Add "C" constraint.
7113 (*cmp<mode>_cc_i387): Ditto.
7114 (*cmp<mode>_0_i387): Remove insn pattern.
7115 (*cmp<mode>_0_cc_i387): Ditto.
7117 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7119 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
7120 * config/i386/predicates.md (nonimm_or_0_operand): Rename
7121 from vector_move_operand. Update all uses.
7123 2018-10-05 Martin Sebor <msebor@redhat.com>
7125 PR tree-optimization/87490
7126 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
7129 2018-10-05 Richard Biener <rguenther@suse.de>
7131 PR tree-optimization/63155
7132 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
7133 vertical space in dumpfiles.
7134 * tree-ssa-propagate.h
7135 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
7136 * tree-ssa-propagate.c (cfg_blocks_back): New global.
7137 (ssa_edge_worklist_back): Likewise.
7138 (curr_order): Likewise.
7139 (cfg_blocks_get): Remove abstraction.
7140 (cfg_blocks_add): Likewise.
7141 (cfg_blocks_empty_p): Likewise.
7142 (add_ssa_edge): Add to current or next worklist based on
7144 (add_control_edge): Likewise.
7145 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
7147 (ssa_propagation_engine::ssa_propagate): ... here. Unify
7148 iteration from CFG and SSA edge worklist so we process
7149 everything in RPO order, prioritizing forward progress
7151 (ssa_prop_init): Allocate new worklists, do not dump
7153 (ssa_prop_fini): Free new worklists.
7155 2018-10-05 Richard Biener <rguenther@suse.de>
7157 * tree-core.h (tree_block::abstract_flag): Remove.
7158 (tree_block::block_num): Make full 32bits.
7159 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
7160 * tree.h (BLOCK_ABSTRACT): Remove.
7161 * dwarf2out.c (gen_lexical_block_die): Remove dead code
7162 resulting from BLOCK_ABSTRACT being always false.
7163 (gen_inlined_subroutine_die): Likewise.
7164 (gen_block_die): Likewise.
7165 * tree.c (block_ultimate_origin): Likewise.
7166 * tree-pretty-print.c (dump_block_node): Remove code dealing
7167 with BLOCK_ABSTRACT.
7168 * tree-ssa-live.c (dump_scope_block): Likewise.
7169 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
7170 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
7172 2018-10-05 Richard Biener <rguenther@suse.de>
7174 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
7175 is asked for initialize mode to the component mode of the
7178 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
7181 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
7182 assembler for -mavx.
7183 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
7185 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7188 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
7190 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
7191 to be DImode. When using mffscrn, force the operand to a register.
7193 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
7195 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
7196 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
7197 X87MODEF mode iterator.
7198 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
7199 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
7200 X87MODEF mode iterator.
7202 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
7204 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
7205 -Wno-prio-ctor-dtor.
7207 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7209 * Makefile.in (OBJS): Add opt-problem.o.
7210 * dump-context.h: Include "selftest.h.
7211 (selftest::temp_dump_context): New forward decl.
7212 (class dump_context): Make friend of class
7213 selftest::temp_dump_context.
7214 (dump_context::dump_loc_immediate): New decl.
7215 (class dump_pretty_printer): Move here from dumpfile.c.
7216 (class temp_dump_context): Move to namespace selftest.
7217 (temp_dump_context::temp_dump_context): Add param
7218 "forcibly_enable_dumping".
7219 (selftest::verify_dumped_text):
7220 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
7221 (selftest::verify_item):
7222 (ASSERT_IS_TEXT): Move here from dumpfile.c.
7223 (ASSERT_IS_TREE): Likewise.
7224 (ASSERT_IS_GIMPLE): Likewise.
7225 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
7227 (dump_context::dump_loc_immediate): ...this new function.
7228 (class dump_pretty_printer): Move to dump-context.h.
7229 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
7230 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
7231 (temp_dump_context::temp_dump_context): Move to "selftest"
7232 namespace. Add param "forcibly_enable_dumping", and use it to
7233 conditionalize the use of m_pp;
7234 (selftest::verify_dumped_text): Make non-static.
7235 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
7236 (selftest::verify_item): Make non-static.
7237 (ASSERT_IS_TEXT): Move to dump-context.h.
7238 (ASSERT_IS_TREE): Likewise.
7239 (ASSERT_IS_GIMPLE): Likewise.
7240 (selftest::test_capture_of_dump_calls): Pass "true" for new
7241 param of temp_dump_context.
7242 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
7243 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
7245 * opt-problem.cc: New file.
7246 * opt-problem.h: New file.
7247 * optinfo-emit-json.cc
7248 (selftest::test_building_json_from_dump_calls): Pass "true" for
7249 new param of temp_dump_context.
7250 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
7251 (optinfo::emit_for_opt_problem): New function.
7252 (optinfo::emit): Clarity which emit_item is used.
7253 * optinfo.h (optinfo::get_dump_location): New accessor.
7254 (optinfo::emit_for_opt_problem): New decl.
7255 (optinfo::emit): Make const.
7256 * selftest-run-tests.c (selftest::run_tests): Call
7257 selftest::opt_problem_cc_tests.
7258 * selftest.h (selftest::opt_problem_cc_tests): New decl.
7259 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
7260 bool to opt_result, converting fprintf messages to
7261 opt_result::failure_at calls. Add "stmt" param for use by the
7263 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
7264 (runtime_alias_check_p): Convert return type from bool to
7265 opt_result, converting dump_printf calls to
7266 opt_result::failure_at, using the statement DDR_A for their
7268 (find_data_references_in_stmt): Convert return type from bool to
7269 opt_result, converting "return false" to opt_result::failure_at
7271 * tree-data-ref.h: Include "opt-problem.h".
7272 (dr_analyze_innermost): Convert return type from bool to opt_result,
7273 and add a const gimple * param.
7274 (find_data_references_in_stmt): Convert return type from bool to
7276 (runtime_alias_check_p): Likewise.
7277 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
7278 dr_analyze_innermost.
7279 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
7280 Convert return type from bool to opt_result, adding a message for
7281 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
7282 (vect_analyze_data_ref_dependence): Convert return type from bool
7283 to opt_result. Change sense of return type from "false"
7284 effectively meaning "no problems" to "false" meaning a problem,
7285 so that "return false" becomes "return opt_result::success".
7286 Convert "return true" calls to opt_result::failure_at, using
7287 the location of statement A rather than vect_location.
7288 (vect_analyze_data_ref_dependences): Convert return type from bool
7290 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
7291 calls to opt_result::failure_at, using the stmt location rather
7293 (vect_verify_datarefs_alignment): Convert return type from bool
7295 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
7296 into multiple more-tightly-scoped copies.
7297 (vect_analyze_data_refs_alignment): Convert return type from bool
7299 (vect_analyze_data_ref_accesses): Likewise, converting a
7300 "return false" to a "return opt_result::failure_at", adding a
7302 (vect_prune_runtime_alias_test_list): Convert return type from
7303 bool to opt_result, converting dump_printf_loc to
7304 opt_result::failure_at. Add a %G to show the pertinent statement,
7305 and use the stmt's location rather than vect_location.
7306 (vect_find_stmt_data_reference): Convert return type from
7307 bool to opt_result, converting dump_printf_loc to
7308 opt_result::failure_at, using stmt's location.
7309 (vect_analyze_data_refs): Convert return type from bool to
7310 opt_result. Convert "return false" to "return
7311 opt_result::failure_at", adding messages as needed.
7312 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
7313 type from bool to opt_result.
7314 (vect_determine_vf_for_stmt): Likewise.
7315 (vect_determine_vectorization_factor): Likewise, converting
7316 dump_printf_loc to opt_result::failure_at, using location of phi
7317 rather than vect_location.
7318 (vect_analyze_loop_form_1): Convert return type from bool to
7319 opt_result, converting dump_printf_loc calls, retaining the use of
7321 (vect_analyze_loop_form): Convert return type from loop_vec_info
7322 to opt_loop_vec_info.
7323 (vect_analyze_loop_operations): Convert return type from bool to
7324 opt_result, converting dump_printf_loc calls, using the location
7325 of phi/stmt rather than vect_location where available. Convert
7326 various "return false" to "return opt_result::failure_at" with
7327 "unsupported phi" messages.
7328 (vect_get_datarefs_in_loop): Convert return type from bool to
7329 opt_result. Add a message for the
7330 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
7331 (vect_analyze_loop_2): Convert return type from bool to
7332 opt_result. Ensure "ok" is set to a opt_result::failure_at before
7333 each "goto again;", adding new messages where needed.
7334 Add "unsupported grouped {store|load}" messages.
7335 (vect_analyze_loop): Convert return type from loop_vec_info to
7337 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
7339 * tree-vect-stmts.c (process_use): Likewise, converting
7340 dump_printf_loc call and using stmt location, rather than
7342 (vect_mark_stmts_to_be_vectorized): Likeise.
7343 (vect_analyze_stmt): Likewise, adding a %G.
7344 (vect_get_vector_types_for_stmt): Convert return type from bool to
7345 opt_result, converting dump_printf_loc calls and using stmt
7346 location, rather than vect_location.
7347 (vect_get_mask_type_for_stmt): Convert return type from tree to
7348 opt_tree, converting dump_printf_loc calls and using stmt location.
7349 * tree-vectorizer.c: Include "opt-problem.h.
7350 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
7351 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
7352 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
7353 enabled, use it to report at the top level "couldn't vectorize
7354 loop" followed by the problem.
7355 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
7356 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
7358 (vect_analyze_stmt): Likewise.
7359 (vect_get_vector_types_for_stmt): Likewise.
7360 (tree vect_get_mask_type_for_stmt): Likewise.
7361 (vect_analyze_data_ref_dependences): Likewise.
7362 (vect_enhance_data_refs_alignment): Likewise.
7363 (vect_analyze_data_refs_alignment): Likewise.
7364 (vect_verify_datarefs_alignment): Likewise.
7365 (vect_analyze_data_ref_accesses): Likewise.
7366 (vect_prune_runtime_alias_test_list): Likewise.
7367 (vect_find_stmt_data_reference): Likewise.
7368 (vect_analyze_data_refs): Likewise.
7369 (vect_analyze_loop): Convert return type from loop_vec_info to
7371 (vect_analyze_loop_form): Likewise.
7372 (vect_analyze_slp): Convert return type from bool to opt_result.
7374 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7376 * doc/invoke.texi (-fopt-info): Document new "internals"
7378 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
7379 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
7381 (optinfo_verbosity_options): Add "internals".
7382 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
7383 (dump_context::apply_dump_filter_p): New member function.
7384 (dump_context::dump_loc): Use apply_dump_filter_p rather than
7385 explicitly masking the dump_kind.
7386 (dump_context::begin_scope): Increment the scope depth first. Use
7387 apply_dump_filter_p rather than explicitly masking the dump_kind.
7388 (dump_context::emit_item): Use apply_dump_filter_p rather than
7389 explicitly masking the dump_kind.
7390 (dump_dec): Likewise.
7391 (dump_hex): Likewise.
7392 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
7393 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
7394 (opt_info_switch_p): Update handling of default
7395 MSG_OPTIMIZED_LOCATIONS to cope with default of
7396 MSG_PRIORITY_USER_FACING.
7397 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
7398 masking the dump_kind.
7399 (selftest::test_capture_of_dump_calls): Update test_dump_context
7400 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
7401 than MSG_ALL. Generalize scope test to be run at all four
7402 combinations of with/without MSG_PRIORITY_USER_FACING and
7403 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
7404 for each of the two values.
7405 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
7406 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
7407 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
7408 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
7409 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
7410 with MSG_PRIORITY_*.
7411 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
7412 dump messages as MSG_PRIORITY_USER_FACING.
7413 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
7414 about the interaction with MSG_PRIORITY_*.
7416 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7418 * varasm.c (output_constant): Add new parameter merge_strings.
7419 Make strings properly zero terminated in merge string sections.
7420 (mergeable_string_section): Don't fail if the last char is non-zero.
7421 (assemble_variable_contents): Handle merge string sections.
7422 (assemble_variable): Likewise.
7423 (assemble_constant_contents): Likewise.
7424 (output_constant_def_contents): Likewise.
7425 (output_constructor_array_range,
7426 output_constructor_regular_field): Adjust call to output_constant.
7427 (output_object_block): Adjust call to assemble_constant_contents
7428 and assemble_variable_contents.
7430 2018-10-04 Martin Liska <mliska@suse.cz>
7433 * cgraphunit.c (process_function_and_variable_attributes):
7434 Warn about a function with alias attribute and a body.
7436 2018-10-04 Martin Liska <mliska@suse.cz>
7439 * multiple_target.c (redirect_to_specific_clone): New function.
7440 (ipa_target_clone): Use it.
7441 * tree-inline.c: Fix comment.
7443 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7445 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
7447 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
7448 (gcc::dump_manager::register_pass): New member function, adapted
7449 from loop body in gcc::pass_manager::register_pass, adding a
7450 call to update_dfi_for_opt_info.
7451 (gcc::dump_manager::opt_info_enable_passes): Store the
7452 -fopt-info options into the new fields. Move the loop
7454 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
7456 * dumpfile.h (struct opt_pass): New forward decl.
7457 (gcc::dump_manager::register_pass): New decl.
7458 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
7459 (class gcc::dump_manager): Add fields "m_optgroup_flags",
7460 "m_optinfo_flags", and "m_optinfo_filename".
7461 * passes.c (gcc::pass_manager::register_pass): Move all of the
7462 dump-handling code to gcc::dump_manager::register_pass.
7464 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
7466 PR rtl-optimization/87466
7467 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
7468 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
7469 * doc/tm.texi: Regenerate.
7470 * ira-lives.c (process_bb_node_lives): Use the new target hook.
7471 * lra-lives.c (process_bb_lives): Likewise.
7472 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
7475 2018-10-04 Tamar Christina <tamar.christina@arm.com>
7477 * params.c (add_params): Fix initialization.
7479 2018-10-04 Martin Liska <mliska@suse.cz>
7481 PR gcov-profile/84107
7482 * tree-profile.c (init_ic_make_global_vars):
7483 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
7484 Come up with new ic_tuple* variables. Emit
7485 __gcov_indirect_call{,_topn} variables.
7486 (gimple_gen_ic_profiler): Access the variable
7488 (gimple_gen_ic_func_profiler): Access
7489 __gcov_indirect_call.callee field.
7490 (gimple_init_gcov_profiler): Use ptr_type_node.
7491 * value-prof.c (gimple_ic): Use ptr_type_node.
7493 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7495 PR tree-optimization/85787
7496 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
7497 malloc_candidate_p into this function and add support for
7498 detecting multiple phis.
7499 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
7501 2018-10-04 Martin Liska <mliska@suse.cz>
7504 * ipa-inline.c (inline_to_all_callers_1):
7505 Call ultimate_alias_target for node being inlined.
7507 2018-10-03 Jeff Law <law@redhat.com>
7509 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
7511 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
7512 * tree.h (get_typenode_from_name): Prototype.
7514 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
7516 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7517 Change operand 2 predicate to nonimmediate_operand.
7518 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7520 2018-10-03 Martin Sebor <msebor@redhat.com>
7521 Jeff Law <law@redhat.com>
7523 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
7525 (get_string_length): Detect unterminated arrays.
7526 (format_string): Same.
7527 (format_directive): Warn about unterminated arrays.
7528 (handle_gimple_call): Mark statements with no_warning as needed.
7530 2018-10-03 Jim Wilson <jimw@sifive.com>
7532 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
7533 also define __riscv_abi_rve. Delete trailing white space.
7535 2018-10-03 Paul Koning <ni1d@arrl.net>
7537 Enable LRA register allocator for PDP11.
7538 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
7541 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
7542 * config/pdp11/pdp11.opt (-mlra): New option.
7543 * doc/invoke.texi (PDP-11 Options): Document -mlra.
7545 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
7547 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
7548 (*<absneg:code>extend<mode>xf2): Ditto.
7550 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
7552 PR tree-optimization/87415
7553 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
7556 2018-10-02 Jeff Law <law@redhat.com>
7558 * gimple-fold.c (get_range_strlen): Only set *nonstr when
7559 an unterminated string is discovered. Bubble up range
7560 even for unterminated strings.
7561 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
7562 indicates the string was not terminated via NONSTR.
7564 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
7566 * tree-vrp.c (extract_range_from_unary_expr): Special case all
7567 pointer conversions.
7568 Do not do anything special for anti-ranges.
7570 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
7572 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
7573 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
7575 2018-10-03 Martin Liska <mliska@suse.cz>
7577 PR gcov-profile/86109
7578 * coverage.c (coverage_begin_function): Do not
7579 mark lambdas as artificial.
7580 * tree-core.h (struct GTY): Remove tm_clone_flag
7581 and introduce new lambda_function.
7582 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
7584 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7587 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
7588 P8_VECTOR and VSX are enabled.
7590 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
7592 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7593 0x3907 as CPU model number.
7595 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
7597 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
7598 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
7599 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
7600 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
7601 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
7602 * config/s390/s390.md: Likewise. Rename also the cpu attribute
7603 value from arch12 to z14.
7605 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
7607 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
7609 (isinf<mode>2): Ditto.
7611 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
7613 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
7614 before emitting fxam. Perform calculations in XFmode.
7616 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
7618 * match.pd (((X /[ex] A) +- B) * A): New transformation.
7620 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
7623 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
7624 * tree.c (signed_or_unsigned_type_for): Handle complex.
7626 2018-10-02 Jeff Law <law@redhat.com>
7628 * gimple-fold.c (get_range_strlen): Remove dead code.
7630 2018-10-02 Martin Sebor <msebor@redhat.com>
7631 Jeff Law <law@redhat.com>
7633 * builtins.c (unterminated_array): Add new arguments.
7634 If argument is not terminated, bubble up size and exact
7636 (expand_builtin_strnlen): Detect, avoid expanding
7637 and diagnose unterminated arrays.
7638 (c_strlen): Fill in offset of start of unterminated strings.
7639 * builtins.h (unterminated_array): Update prototype.
7641 2018-10-02 Richard Biener <rguenther@suse.de>
7643 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
7644 of haddv4df, first reduce to SSE width and exploit the fact
7645 that we only need element zero with the reduction result.
7646 (reduc_plus_scal_v2df): Likewise.
7648 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7650 * dojump.h (do_jump): Delete.
7651 (do_jump_1): Likewise.
7652 (split_comparison): Move around.
7653 * dojump.c (do_jump): Make static.
7654 (do_jump_1): Likewise.
7655 (jumpifnot): Move around.
7656 (jumpifnot_1): Likewise.
7658 (jumpif_1): Likewise.
7659 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
7661 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7663 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
7664 insns in the delay slot and add_insn_after for the jump insn.
7666 2018-10-02 Richard Biener <rguenther@suse.de>
7668 * tree-inline.c (expand_call_inline): Use the location of
7669 the callee declaration for the inline-entry marker.
7670 * final.c (notice_source_line): Remove special-casing of
7671 NOTE_INSN_INLINE_ENTRY.
7673 2018-10-01 Carl Love <cel@us.ibm.com>
7676 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
7677 (__builtin_mtfsb0): New.
7678 (__builtin_mtfsb1): New.
7679 ( __builtin_set_fpscr_rn): New.
7680 (__builtin_set_fpscr_drn): New.
7681 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
7682 (rs6000_expand_set_fpscr_rn_builtin): Add.
7683 (rs6000_expand_set_fpscr_drn_builtin): Add.
7684 (rs6000_expand_builtin): Add case statement entries for
7685 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
7686 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
7687 RS6000_BUILTIN_MFFSL.
7688 (rs6000_init_builtins): Add ftype initialization and def_builtin
7689 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
7690 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
7691 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
7692 rs6000_mffscdrn): Add define_insn.
7693 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
7694 * doc/extend.texi: Add documentation for the builtins.
7696 2018-10-01 Richard Biener <rguenther@suse.de>
7698 PR tree-optimization/87465
7699 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
7700 causing branch miscounts.
7702 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7704 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
7705 aarch64_option_default_param): New.
7706 (params.h): Include.
7707 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
7708 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
7709 stack-clash protection validation code.
7711 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7713 * params.c (validate_param): New.
7714 (add_params): Use it.
7715 (set_param_value): Refactor param validation into validate_param.
7716 (diagnostic.h): Include.
7717 * diagnostic.h (diagnostic_ready_p): New.
7719 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7721 * params.c (set_param_value):
7722 Add index of parameter being validated.
7723 * common/common-target.def (option_validate_param): New.
7724 * common/common-targhooks.h (default_option_validate_param): New.
7725 * common/common-targhooks.c (default_option_validate_param): New.
7726 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
7727 * doc/tm.texi: Regenerate.
7729 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7732 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7733 Add validation for stack-clash parameters and set defaults.
7735 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7738 * configure.ac: Add stack-clash-protection-guard-size.
7739 * doc/install.texi: Document it.
7740 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
7741 * params.def: Update comment for guard-size.
7742 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
7743 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
7744 * configure: Regenerate.
7746 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7749 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
7750 STACK_DYNAMIC_OFFSET): New.
7751 * config/aarch64/aarch64.c (aarch64_layout_frame):
7752 Update outgoing args size.
7753 (aarch64_stack_clash_protection_alloca_probe_range,
7754 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7756 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7759 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
7761 * target.def (stack_clash_protection_alloca_probe_range): New.
7762 (stack_clash_protection_final_dynamic_probe): Remove.
7763 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
7764 (default_stack_clash_protection_final_dynamic_probe): Remove.
7765 * targhooks.c: Likewise.
7766 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7767 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
7768 * doc/tm.texi: Regenerate.
7770 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7773 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
7774 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
7775 aarch64_clamp_to_uimm12_shift): New.
7776 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
7777 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
7779 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7782 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
7784 2018-10-01 Jeff Law <law@redhat.com>
7785 Richard Sandiford <richard.sandiford@linaro.org>
7786 Tamar Christina <tamar.christina@arm.com>
7789 * config/aarch64/aarch64.md
7790 (probe_stack_range): Add k (SP) constraint.
7791 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
7792 STACK_CLASH_MAX_UNROLL_PAGES): New.
7793 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
7794 stack probes for stack clash.
7795 (aarch64_allocate_and_probe_stack_space): New.
7796 (aarch64_expand_prologue): Use it.
7797 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
7798 (aarch64_sub_sp): Add emit_move_imm optional param.
7800 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
7802 PR tree-optimization/87261
7803 * match.pd: Remove trailing whitespace.
7804 Add (x & y) | ~(x | y) -> ~(x ^ y),
7805 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
7807 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7809 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
7811 (ashlsi3_insn): Update instruction constraints.
7812 (ashrsi3_insn): Likewise.
7813 (rotrsi3): Likewise.
7814 (add_shift): Likewise.
7815 * config/arc/constraints.md (Csz): New 32 bit constraint. It
7816 avoids placing in the limm field small constants which, otherwise,
7817 could end into a small instruction.
7819 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7821 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
7822 destination register is not odd-even.
7823 (umaddsidi4_split): Likewise.
7825 2018-10-01 Richard Biener <rguenther@suse.de>
7827 * tree-inline.c (expand_call_inline): Store origin of fn
7828 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
7829 * tree.c (block_ultimate_origin): Simplify and do some
7832 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
7834 * config/i386/mmx.md (EMMS): New int iterator.
7835 (emms): New int attribute.
7836 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
7837 EMMS int iterator. Explicitly declare clobbers.
7838 (mmx_emms): Remove expander.
7840 * config/i386/predicates.md (emms_operation): Remove predicate.
7841 (vzeroall_pattern): New predicate.
7842 (vzeroupper_pattern): Rename from vzeroupper_operation.
7843 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
7844 vzeroupper_pattern and vzeroall_pattern predicates.
7846 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
7848 PR rtl-optimization/86939
7849 * ira-lives.c (make_hard_regno_born): Rename from this...
7850 (make_hard_regno_live): ... to this. Remove update to conflict
7851 information. Update function comment.
7852 (make_hard_regno_dead): Add conflict information update. Update
7854 (make_object_born): Rename from this...
7855 (make_object_live): ... to this. Remove update to conflict information.
7856 Update function comment.
7857 (make_object_dead): Add conflict information update. Update function
7859 (mark_pseudo_regno_live): Call make_object_live.
7860 (mark_pseudo_regno_subword_live): Likewise.
7861 (mark_hard_reg_dead): Update function comment.
7862 (mark_hard_reg_live): Call make_hard_regno_live.
7863 (process_bb_node_lives): Likewise.
7864 * lra-lives.c (make_hard_regno_born): Rename from this...
7865 (make_hard_regno_live): ... to this. Remove update to conflict
7866 information. Remove now uneeded check_pic_pseudo_p argument.
7867 Update function comment.
7868 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
7869 to conflict information. Update function comment.
7870 (mark_pseudo_live): Remove update to conflict information. Update
7872 (mark_pseudo_dead): Add conflict information update.
7873 (mark_regno_live): Call make_hard_regno_live.
7874 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
7875 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
7877 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
7880 * config/i386/i386.c (construct_container): Use TImode for
7881 BLKmode values in 2 integer registers.
7883 2018-09-29 Jeff Law <law@redhat.com>
7885 * builtins.c (unterminated_array): Pass in c_strlen_data * to
7886 c_strlen rather than just a tree *.
7887 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
7888 Update recursive calls appropriately. If caller did not provide a
7889 suitable data pointer, create a local one. When a non-terminated
7890 string is discovered, bubble up information about the string via the
7891 c_strlen_data object.
7892 * builtins.h (c_strlen): Update prototype.
7893 (c_strlen_data): New structure.
7894 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
7895 For a type 2 call, if c_strlen indicates a non-terminated string
7896 use the length of the non-terminated string.
7897 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
7899 2018-09-29 Jakub Jelinek <jakub@redhat.com>
7902 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
7903 __m512d type for __A argument rather than __m512.
7905 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
7907 * match.pd (simple_comparison): Don't optimize if either operand is
7908 a function pointer when target needs function pointer canonicalization.
7910 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7912 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
7913 power5 .. power9 to remove indirection.
7914 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
7915 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
7916 ASM_CPU_476_SPEC): Delete.
7917 (ASM_CPU_SPEC): Adjust.
7918 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
7919 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
7921 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7923 * config.in: Delete HAVE_AS_DCI.
7924 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
7925 * config/rs6000/rs6000.h: Ditto.
7926 * configure.ac: Delete HAVE_AS_DCI.
7927 * configure: Regenerate.
7929 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7931 * config.in (HAVE_AS_LWSYNC): Delete.
7932 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7933 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
7935 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7936 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
7938 * configure.ac: Delete HAVE_AS_LWSYNC.
7939 * configure: Regenerate.
7941 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7942 Pierre-Marie de Rodat <derodat@adacore.com>
7944 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
7945 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
7946 (cgraph_node::create_thunk): Add indirect_offset parameter.
7947 (thunk_adjust): Likewise.
7948 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
7949 and initialize the corresponding field with it.
7950 (cgraph_node::dump): Dump indirect_offset field.
7951 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
7952 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
7953 (thunk_adjust): Add indirect_offset parameter and deal with it.
7954 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
7955 pass it to thunk_adjust. Do not call the target hook if it's non-zero
7956 or if the thunk is external or local. Fix formatting. Do not chain
7957 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
7958 if any, in the GIMPLE representation.
7959 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
7960 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
7961 (input_node): Read indirect_offset field.
7962 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
7963 call to thunk_adjust.
7964 * tree-nested.c (struct nesting_info): Add thunk_p field.
7965 (create_nesting_tree): Set it.
7966 (convert_all_function_calls): Copy static chain from targets to thunks.
7967 (finalize_nesting_tree_1): Return early for thunks.
7968 (unnest_nesting_tree_1): Do not finalize thunks.
7969 (gimplify_all_functions): Do not gimplify thunks.
7971 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7973 * opt-suggestions.c (option_proposer::build_option_suggestions):
7974 Release "option_values".
7976 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7978 * coverage.c (get_coverage_counts): Convert problem-reporting dump
7979 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
7980 * dumpfile.c (kind_as_string): New function.
7981 (dump_loc): Rather than a hardcoded prefix of "note: ", use
7982 kind_as_string to vary the prefix based on dump_kind.
7983 (selftest::test_capture_of_dump_calls): Update for above.
7985 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7987 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
7988 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
7990 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7992 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
7993 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
7994 INVALID_REGNUM instead of FPSR_REG.
7995 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
7996 * config/i386/i386.md: Update comment of FP compares.
7997 (fldenv): Do not clobber FPSR_REG.
7999 2018-09-28 Richard Biener <rguenther@suse.de>
8001 * tree.h (BLOCK_ORIGIN): New.
8002 * omp-expand.c (grid_expand_target_grid_body): Assign
8003 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
8004 * tree-inline.c (remap_block): Likewise.
8005 * auto-profile.c (get_function_decl_from_block): Simplify
8006 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
8007 * langhooks.c (lhd_print_error_function): Likewise.
8008 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
8010 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
8011 * tree.c (block_nonartificial_location): Likewise.
8012 (block_ultimate_origin): Likewise.
8013 * tree-pretty-print.c (percent_K_format): Likewise. Remove
8014 no longer needed LTO case.
8016 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
8017 Jan Hubicka <jh@suse.cz>
8018 Martin Jambor <mjambor@suse.cz>
8020 * simplify-rtx.c (simplify_merge_mask): New function.
8021 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
8022 same masks are used in op1 or op2.
8023 (test_vec_merge): New function.
8024 (test_vector_ops): Call test_vec_merge.
8026 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
8028 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
8029 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
8030 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
8031 (cypress_costs): Set it.
8032 (supersparc_costs): Likewise.
8033 (hypersparc_costs): Likewise.
8034 (leon_cost): Likewise.
8035 (leon3_costs): Likewise.
8036 (sparclet_costs): Likewise.
8037 (ultrasparc_costs): Likewise.
8038 (ultrasparc_costs): Likewise.
8039 (niagara_costs): Likewise.
8040 (niagara2_costs): Likewise.
8041 (niagara3_costs): Likewise.
8042 (niagara4_costs): Likewise.
8043 (niagara7_costs): Likewise.
8044 (m8_costs): Likewise.
8045 (TARGET_CAN_FOLLOW_JUMP): Define.
8046 (pass_work_around_errata::gate): Minor tweak.
8047 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
8048 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
8049 Automaitcally clear MASK_FSMULD mask for V7 processors.
8050 (sparc_can_follow_jump): New static function.
8051 (output_ubranch): Deal with CROSSING_JUMP_P.
8052 (sparc_use_sched_lookahead): Rewrite using switch statement.
8053 (sparc_issue_rate): Reorder.
8054 (sparc_branch_cost): New function.
8056 2018-09-27 Martin Sebor <msebor@redhat.com>
8058 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
8059 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
8060 (int_fits_type_p): Same.
8062 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
8064 * config/i386/i386.md (FPCR_REG): Remove.
8065 (UNSPEC_FLDCW): Remove.
8066 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
8067 (x86_fldcw_1): Remove insn pattern.
8068 (fnstenv): Do not clobber FPCR_REG.
8070 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
8071 (CALL_USED_REGISTERS): Ditto.
8072 (REG_ALLOC_ORDER): Ditto.
8073 (REG_CLASS_CONTENTS): Ditto.
8074 (HI_REGISTER_NAMES): Ditto.
8075 (ADDITIONAL_REGISTER_NAMES): Use defines instead
8076 of numerical constants.
8077 * config/i386/i386.c (regclass_map): Remove fpsr register.
8078 (dbx_register_map): Ditto.
8079 (dbx64_register_map): Ditto.
8080 (svr4_dbx_register_map): Ditto.
8081 (print_reg): Do not handle FPCR_REG.
8083 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
8086 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
8087 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
8088 Delete, always treat as true.
8089 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
8090 Ditto. Simplify remaining code.
8091 * config/powerpcspe/powerpcspe.h: Ditto.
8092 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
8093 Simplify remaining code.
8094 (rs6000_expand_builtin): Ditto.
8095 * config/rs6000/rs6000.h: Ditto.
8096 * configure.ac: Ditto.
8097 * configure: Regenerate.
8099 2018-09-27 Martin Liska <mliska@suse.cz>
8101 * coverage.c (get_coverage_counts): Revert the formatting
8102 of missing profile opt info.
8104 2018-09-27 Richard Biener <rguenther@suse.de>
8108 * dwarf2out.c (set_block_origin_self): Do not mark outermost
8109 block as we do not output that.
8110 (gen_inlined_subroutine_die): Elide the originally outermost
8111 block, matching what we do for concrete instances.
8112 (decls_for_scope): Add parameter specifying whether to recurse
8115 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
8116 Tom de Vries <tom@codesourcery.com>
8120 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
8121 STORE_FLAG_VALUE == 1.
8123 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
8125 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
8126 constant definitions.
8127 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
8128 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
8129 ("speculation_barrier"): New expander definition.
8131 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
8133 PR gcov-profile/86957
8134 * common.opt: New warning option -Wmissing-profile.
8135 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
8136 * doc/invoke.texi: Document -Wmissing-profile.
8138 2018-09-26 Jim Wilson <jimw@sifive.com>
8140 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
8141 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
8143 2018-09-26 Martin Sebor <msebor@redhat.com>
8145 * tree.c (zerop): Change return type to bool.
8146 (integer_zerop, integer_onep, integer_each_onep): Same.
8147 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
8148 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
8149 (real_onep, real_minus_onep, chain_index): Same.
8150 (print_type_hash_statistics, type_list_equal): Same.
8151 * tree.h (zerop): Same.
8152 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
8153 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
8154 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
8155 (real_onep, real_minus_onep, chain_index): Same.
8156 (print_type_hash_statistics, type_list_equal): Same.
8158 2018-09-26 Jim Wilson <jimw@sifive.com>
8160 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
8162 2018-09-26 Jakub Jelinek <jakub@redhat.com>
8165 * config/i386/i386.c: Include debug.h and dwarf2out.h.
8166 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
8169 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
8171 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
8173 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
8175 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
8176 and FP_SECOND_SSE_REGS.
8177 (REG_CLASS_NAMES): Ditto.
8178 (REG_CLASS_CONTENTS): Ditto.
8179 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
8180 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
8181 (ix86_preferred_output_reload_class): Ditto.
8182 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
8183 clobber constraint to "=&f".
8184 (fix_truncdi_i387): Ditto.
8185 (lrintxfdi2): Ditto.
8186 (fistdi2_<rounding>): Ditto.
8187 (fpremxf4_i387): Change "=u" constraint to "=f".
8188 (fprem1xf4_i387): Ditto.
8190 (fptanxf4_i387): Ditto.
8191 (fxtractxf3_i387): Ditto.
8192 (fscalexf4_i387): Ditto.
8193 (atan2xf3): Change "u" constraint to "f".
8194 (fyl2xxf3_i387): Ditto.
8195 (fyl2xp1xf3_i387): Ditto.
8197 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
8200 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
8201 for removed I387_MASK_PM entity.
8204 2018-09-26 Jeff Law <law@redhat.com>
8206 2018-09-26 Alexey Neyman <stilor@att.net>
8208 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
8209 headers are no longer pulled in by <isl/val.h>.
8211 2018-09-26 Richard Biener <rguenther@suse.de>
8214 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
8215 or concrete instance DIE to the tree. Create abstract origin
8216 attributes also for concrete instances.
8218 2018-09-26 Alexey Neyman <stilor@att.net>
8220 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
8221 headers are no longer pulled in by <isl/val.h>.
8223 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
8225 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
8226 Use new helper functions.
8227 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
8228 Use new helper functions.
8229 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
8230 aarch_mm_needs_release): New declarations.
8231 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
8232 aarch_mm_needs_release): New.
8234 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
8236 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
8237 (arm32_output_mi_thunk): Deal with long calls.
8239 2018-09-26 Richard Biener <rguenther@suse.de>
8243 * tree-inline.c (expand_call_inline): When the location
8244 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
8245 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
8246 the inserted BLOCK to make inlined_function_outer_scope_p
8248 * dwarf2out.c (add_call_src_coords_attributes): Do not add
8249 coords for reserved locations.
8251 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8253 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
8254 (*call_indirect_nonlocal_sysv<mode>): Ditto.
8255 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
8256 (*sibcall_nonlocal_sysv<mode>): Ditto.
8257 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
8258 (<bd>_<mode>): Ditto.
8259 (<bd>tf_<mode>): Ditto.
8261 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8263 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
8264 control string as a list of templates instead of as C code.
8265 (*altivec_movti): Ditto.
8266 * config/rs6000/darwin.md (movdf_low_di): Ditto.
8268 2018-09-25 Jim Wilson <jimw@sifive.com>
8270 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
8271 when target symbol is weak.
8273 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
8276 * builtins.c (unterminated_array): Simplify.
8277 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
8278 where pointer arithmetic is safe.
8280 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8283 * config/rs6000/altivec.md (altivec_vspltb): Use
8284 const_0_to_15_operand instead of u5bit_cint_operand.
8285 (*altivec_vspltb_internal): Ditto.
8286 (altivec_vspltb_direct): Ditto.
8287 (altivec_vsplth): Use const_0_to_7_operand instead of
8289 (*altivec_vsplth_internal): Ditto.
8290 (altivec_vsplth_direct): Ditto.
8291 (altivec_vspltw): Use const_0_to_3_operand instead of
8293 (*altivec_vspltw_internal): Ditto.
8294 (altivec_vspltw_direct): Ditto.
8295 (altivec_vspltsf): Ditto.
8296 (*altivec_vspltsf_internal): Ditto.
8297 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
8298 various splats with the proper size immediate. Reorder the various
8299 cases by ascending size of immediate, and put all such together.
8301 2018-09-25 Richard Biener <rguenther@suse.de>
8304 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
8306 (maybe_create_die_with_external_ref): Do not create
8307 DW_TAG_imported_unit here.
8308 (add_abstract_origin_attribute): Handle external BLOCK refs.
8309 (dwarf2out_abstract_function): Simplify LTO case.
8310 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
8311 rather than using maybe_create_die_with_external_ref.
8313 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
8316 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
8317 (frndintxf2_mask_pm_i387): Ditto.
8318 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
8319 Enable for !flag_trapping_math.
8320 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
8321 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
8322 Change operand 1 predicate to nonimmediate_operand.
8323 (attr "i387_cw"): Remove mask_pm.
8324 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
8325 (enum ix86_entity): Remove I387_MASK_PM.
8326 * config/i386/i386.c (ix86_i387_mode_needed): Do not
8327 handle I387_MASK_PM.
8328 (ix86_mode_needed): Ditto.
8329 (ix86_mode_after): Ditto.
8330 (ix86_mode_entry): Ditto.
8331 (ix86_mode_exit): Ditto.
8332 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
8334 2018-09-25 Jakub Jelinek <jakub@redhat.com>
8336 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
8337 to_update_switch_stmts to vNULL instead of calling create on them
8340 2018-09-25 Richard Biener <rguenther@suse.de>
8342 PR tree-optimization/87402
8343 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
8344 (visit_phi): Re-instantiate handling of supposed to be VARYING
8345 but non-VARYING backedge value.
8347 2018-09-25 Richard Biener <rguenther@suse.de>
8350 * dwarf2out.c (struct sym_off_pair): New.
8351 (external_die_map): New global.
8352 (lookup_decl_die): When in LTO create DIEs lazily from the
8354 (lookup_block_die): New function, create DIEs lazily in LTO.
8355 (equate_block_to_die): New function.
8356 (dwarf2out_die_ref_for_decl): During WPA get the association
8357 from the external DIE map.
8358 (dwarf2out_register_external_die): Record mapping into the
8360 (maybe_create_die_with_external_ref): New function split out from
8361 DIE generation part of old dwarf2out_register_external_die.
8362 (add_abstract_origin_attribute): Do not return the DIE. When
8363 in LTO reference externals directly.
8364 (dwarf2out_abstract_function): When in LTO ignore calls for
8365 decls with external DIEs (already present abstract instances).
8366 (gen_call_site_die): Adjust.
8367 (add_high_low_attributes): Likewise.
8368 (gen_lexical_block_die): Likewise.
8369 (gen_inlined_subroutine_die): Likewie.
8370 (gen_block_die): Likewise.
8371 (dwarf2out_inline_entry): Likewise.
8372 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
8375 2018-09-25 Martin Liska <mliska@suse.cz>
8377 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
8378 integers and not by a float value.
8380 2018-09-25 Martin Liska <mliska@suse.cz>
8383 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
8385 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
8386 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
8387 * godump.c (go_format_type): Remove extra parenthesis.
8389 2018-09-25 Martin Liska <mliska@suse.cz>
8391 * alias.c (set_dest_equal_p): Remove unused function.
8392 * config/i386/i386.c (def_builtin_pure2): Likewise.
8393 * diagnostic-show-locus.c (class layout): Remove
8395 (layout::layout): Likewise here.
8396 * dump-context.h (class temp_dump_context): Likewise.
8397 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
8398 (add_AT_loclistsptr): Likewise.
8399 (add_AT_offset): Likewise.
8400 (get_AT_hi_pc): Likewise.
8401 (is_comdat_die): Likewise.
8402 (type_is_enum): Likewise.
8403 (ceiling): Likewise.
8404 (add_AT_vms_delta): Likewise.
8405 (is_class_die): Likewise.
8406 * edit-context.c (class line_event): Remove unused field.
8407 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
8409 * ipa-cp.c (ipa_get_vr_lat): Likewise.
8410 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
8411 (ok_for_base_p_nonstrict): Likewise.
8412 * tree-chrec.c (is_not_constant_evolution): Likewise.
8413 (chrec_fold_poly_cst): Likewise.
8414 * tree-if-conv.c (has_pred_critical_p): Likewise.
8415 * tree-ssa-coalesce.c (print_exprs): Likewise.
8416 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
8417 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
8418 * tree-vrp.c (value_ranges_intersect_p): Likewise.
8419 (value_range_nonnegative_p): Likewise.
8421 2018-09-25 Martin Liska <mliska@suse.cz>
8423 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
8424 Do not handle "GNU Pascal".
8425 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
8427 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
8428 from documentation. Likewise.
8429 * dbxout.c (dbxout_range_type): Likewise.
8430 * doc/cpp.texi: Likewise.
8431 * doc/extend.texi: Likewise.
8432 * doc/frontends.texi: Likewise.
8433 * doc/invoke.texi: Remove Pascal entry.
8434 * tree.def (CLEANUP_POINT_EXPR): Likewise.
8435 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
8437 2018-09-25 Martin Liska <mliska@suse.cz>
8440 * doc/invoke.texi: Document all parameters and remove default
8443 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
8446 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
8447 contains HOST_WIDE_INTs when computing its size.
8449 2018-09-24 Jim Wilson <jimw@sifive.com>
8452 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
8454 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
8456 2018-09-24 Andrew Pinski <apinski@marvell.com>
8458 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
8459 access prev before checking it for NULLness in the
8460 AARCH64_FUSE_CMP_BRANCH case.
8462 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
8465 * config/i386/i386.c (rest_of_insert_endbranch): Set
8466 endbr_queued_at_entrance to true and don't insert ENDBR if
8467 x86_function_profiler will be called.
8468 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
8470 * config/i386/i386.h (machine_function): Add
8471 endbr_queued_at_entrance.
8473 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
8475 * genattrtab.c (mk_attr_alt): Use alternative_mask.
8476 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
8478 (check_attr_test): Use alternative_mask.
8479 (get_attr_value): Likewise.
8480 (compute_alternative_mask): Use alternative_mask and XWINT.
8481 (make_alternative_compare): Use alternative_mask.
8482 (attr_alt_subset_p): Use XWINT.
8483 (attr_alt_subset_of_compl_p): Likewise.
8484 (attr_alt_intersection): Use alternative_mask and XWINT.
8485 (attr_alt_union): Likewise.
8486 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
8487 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
8488 (simplify_test_exp): Use alternative_mask and XWINT.
8489 (write_test_expr): Use alternative_mask and XWINT, adjust bit
8490 number calculation to support 64 bits. Generate code that
8491 checks 64-bit masks.
8492 (main): Use alternative_mask.
8493 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
8495 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
8498 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
8499 RETURN+USE when returning via %r14.
8501 2018-09-24 Martin Liska <mliska@suse.cz>
8503 * gcov.c (output_lines): Print colorization legend
8504 for both flag_use_colors and flag_use_hotness_colors.
8507 2018-09-24 Martin Liska <mliska@suse.cz>
8509 * coverage.c (get_coverage_counts): Use warning_at
8510 with current_function_decl location. Use %qD in warning
8513 2018-09-24 Martin Liska <mliska@suse.cz>
8515 * memory-block.h (memory_block_pool::release): Annotate with
8516 valgrind that the memory is not accessible.
8518 2018-09-24 Martin Liska <mliska@suse.cz>
8521 * asan.c: Make asan_handled_variables extern.
8523 * cfgexpand.c (expand_stack_vars): Make sure
8524 a representative is unpoison if another
8525 variable in the partition is handled by
8526 use-after-scope sanitization.
8528 2018-09-24 Richard Biener <rguenther@suse.de>
8530 PR tree-optimization/63155
8531 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
8532 the worklist when the edge of the respective argument isn't
8535 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8537 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
8538 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
8539 (MASK_CLASS_P): Update for rename.
8540 (MAYBE_MASK_CLASS_P): Ditto.
8541 (REG_CLASS_NAMES): Update.
8542 (REG_CLASS_CONTENT): Update.
8543 * config/i386/i386.c (regclass_map): Update for MASK_REG
8544 and ALL_MASK_REGS rename.
8545 * config/i386/constraints.md (Yk): Update for rename.
8548 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8550 * config/i386/i386.h (enum reg_class): Remove
8551 EVEX_SSE_REGS and MOD4_SSE_REGS.
8552 (REG_CLASS_NAMES): Update.
8553 (REG_CLASS_CONTENT): Update.
8554 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
8555 registers as ALL_SSE_REGS.
8556 (ix86_additional_allocno_class_p): Remove.
8557 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
8558 (ix86_register_priority): Lower priority of EVEX SSE registers.
8559 Use IN_RANGE macro where appropriate.
8560 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
8561 AVX-5124VNNIW checks.
8562 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
8563 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
8564 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
8565 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
8566 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
8567 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
8568 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
8569 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
8570 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
8571 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
8572 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
8573 * config/i386/constraints.md (Yh): Remove.
8575 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8577 * config/i386/i386.c (regclass_map): Declare integer REX registers
8580 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
8582 * doc/service.texi (Service): Switch the fsf.org link to https.
8584 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
8587 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8588 Define to speculation_safe_value_not_needed.
8590 2018-09-21 Florian Weimer <fweimer@redhat.com>
8593 * doc/extend.texi (Common Function Attributes): Mention that
8594 noreturn suppresses tail call optimization.
8596 2018-09-21 Jeff Law <law@redhat.com>
8598 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
8599 vr_values::cleanup_edges_and_switches.
8600 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
8602 (identify_jump_threads): Remove EDGE_IGNORE handling.
8603 (execute_vrp): Move handling of to_remove_edges and
8604 to_update_switch_stmts into vr_values class member functions.
8605 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
8606 (to_update_switch_stmts): Likewise.
8607 * vr-values.c: Include cfghooks.h.
8608 (vr_values::vr_values): Initialize to_remove_edges and
8609 to_update_switch_stmts.
8610 (vr_values::~vr_values): Verify to_remove_edges and
8611 to_update_switch_stmts are empty.
8612 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
8613 (vr_values::cleanup_edges_and_switches): New member function.
8614 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
8615 function. Add new data members.
8617 2018-09-21 David Malcolm <dmalcolm@redhat.com>
8619 PR tree-optimization/87309
8620 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
8621 calls with pflags and alt_flags.
8622 (selftest::test_capture_of_dump_calls): Add test of interaction of
8623 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
8625 2018-09-21 Olivier Hainque <hainque@adacore.com>
8627 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
8629 2018-09-21 Olivier Hainque <hainque@adacore.com>
8631 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
8633 2018-09-21 Olivier Hainque <hainque@adacore.com>
8635 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
8636 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
8637 (VXWORKS_LIBS_RTP): Minor reordering.
8639 2018-09-21 Olivier Hainque <hainque@adacore.com>
8641 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
8642 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
8644 2018-09-21 Olivier Hainque <hainque@adacore.com>
8646 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
8647 (PTRDIFF_TYPE): Likewise.
8649 2018-09-21 Olivier Hainque <hainque@adacore.com>
8651 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
8652 triplet, similar to support for VxWorks7.
8653 * config/vxworks-dummy.h: Provide a default definition
8654 of TARGET_VXWORKS64 to 0.
8656 2018-09-21 Olivier Hainque <hainque@adacore.com>
8658 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
8659 * config/vxworks-dummy.h: here.
8661 2018-09-21 Olivier Hainque <hainque@adacore.com>
8663 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
8665 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
8666 Bo Zhou <zbo.zhou@hisilicon.com>
8668 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
8669 * config/aarch64/aarch64-tune.md: Regenerated.
8670 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
8671 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
8672 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
8674 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
8675 Julian Brown <julian@codesourcery.com>
8677 * builtins.c (get_builtin_sync_mem): Handle address spaces.
8679 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8681 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
8682 if the call takes a static chain.
8684 2018-09-21 Martin Liska <mliska@suse.cz>
8686 * auto-profile.c (autofdo_source_profile::read): Do not
8688 (read_profile): Do not add working sets.
8689 (read_autofdo_file): Remove sum_all.
8690 (afdo_callsite_hot_enough_for_early_inline): Remove const
8692 * coverage.c (struct counts_entry): Remove gcov_summary.
8693 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
8694 do not support GCOV_TAG_PROGRAM_SUMMARY.
8695 (get_coverage_counts): Remove summary and expected
8697 * coverage.h (get_coverage_counts): Likewise.
8698 * doc/gcov-dump.texi: Remove -w option.
8699 * gcov-dump.c (dump_working_sets): Remove.
8700 (main): Do not support '-w' option.
8701 (print_usage): Likewise.
8702 (tag_summary): Likewise.
8703 * gcov-io.c (gcov_write_summary): Do not dump
8705 (gcov_read_summary): Likewise.
8706 (gcov_histo_index): Remove.
8707 (gcov_histogram_merge): Likewise.
8708 (compute_working_sets): Likewise.
8709 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
8711 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
8712 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
8713 (GCOV_HISTOGRAM_SIZE): Remove.
8714 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
8715 (struct gcov_summary): Simplify rapidly just
8716 to runs and sum_max fields.
8717 (gcov_histo_index): Remove.
8718 (NUM_GCOV_WORKING_SETS): Likewise.
8719 (compute_working_sets): Likewise.
8720 * gcov-tool.c (print_overlap_usage_message): Remove
8721 trailing empty line.
8722 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
8723 (output_lines): Remove program related line.
8724 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
8725 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
8727 (input_profile_summary): Do not read it.
8728 (merge_profile_summaries): And do not merge it.
8729 (input_symtab): Do not call removed function.
8730 * modulo-sched.c (sms_schedule): Do not print sum_max.
8731 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
8732 removed when histogram method was invented.
8733 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
8735 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
8737 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
8738 and dump selected value.
8739 * profile.c (add_working_set): Remove.
8740 (get_working_sets): Likewise.
8741 (find_working_set): Likewise.
8742 (get_exec_counts): Do not work with working sets.
8743 (read_profile_edge_counts): Do not inform as sum_max is removed.
8744 (compute_branch_probabilities): Likewise.
8745 (compute_value_histograms): Remove argument for call of
8746 get_coverage_counts.
8747 * profile.h: Do not make gcov_summary const.
8749 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
8751 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
8753 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8755 PR tree-optimization/86990
8756 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
8757 Check that the entire merged store group is made of constants only for
8760 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
8762 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
8763 (VTABLE_VERIFICATION_SPEC): Likewise.
8764 (SANITIZER_EARLY_SPEC): Likewise.
8765 (SANITIZER_SPEC): Likewise.
8766 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
8767 * doc/invoke.texi (Link Options): Document -r.
8769 2018-09-20 Richard Biener <rguenther@suse.de>
8772 * gimplify.c (gimplify_expr): Retain alignment of
8773 addressable lvalue in dereference.
8775 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
8778 * configure.ac: Check for .loc is_stmt support.
8779 * configure, config.in: Rebuilt.
8780 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
8783 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8785 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
8786 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
8789 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8791 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
8793 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
8796 2018-09-20 Marek Polacek <polacek@redhat.com>
8798 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
8800 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8802 PR tree-optimization/87288
8803 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
8804 into account when determining PEELING_FOR_NITERS.
8806 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8808 PR tree-optimization/86877
8809 * tree-vect-loop.c (vect_analyze_loop_2): Call
8810 vect_verify_datarefs_alignment.
8812 2018-09-19 Marek Polacek <polacek@redhat.com>
8814 * doc/invoke.texi: Document -Wclass-conversion.
8816 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
8818 * config/pa/pa.c (pa_adjust_priority): Delete.
8819 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
8821 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
8822 (atomic_storehi): Likewise.
8823 (atomic_storesi): Likewise.
8824 (atomic_loaddi): Restore compare and swap exchange loop code.
8826 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8828 PR rtl-optimization/86902
8829 * combine.c (try_combine): When changing the CC mode used, don't change
8830 an unrelated mode in other_insn to that new CC mode.
8832 2018-09-19 David Malcolm <dmalcolm@redhat.com>
8834 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
8835 with %T in place of calls to dump_generic_expr.
8836 (prune_runtime_alias_test_list): Likewise.
8837 (create_runtime_alias_checks): Likewise.
8838 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
8839 (vect_analyze_data_ref_dependence): Likewise.
8840 (vect_slp_analyze_data_ref_dependence): Likewise.
8841 (vect_record_base_alignment): Likewise. Use %G in place of call
8842 to dump_gimple_stmt.
8843 (vect_compute_data_ref_alignment): Likewise.
8844 (verify_data_ref_alignment): Likewise.
8845 (vect_find_same_alignment_drs): Likewise.
8846 (vect_analyze_group_access_1): Likewise.
8847 (vect_analyze_data_ref_accesses): Likewise.
8848 (dependence_distance_ge_vf): Likewise.
8849 (dump_lower_bound): Likewise.
8850 (vect_prune_runtime_alias_test_list): Likewise.
8851 (vect_find_stmt_data_reference): Likewise.
8852 (vect_analyze_data_refs): Likewise.
8853 (vect_create_addr_base_for_vector_ref): Likewise.
8854 (vect_create_data_ref_ptr): Likewise.
8855 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
8856 (vect_can_advance_ivs_p): Likewise.
8857 (vect_update_ivs_after_vectorizer): Likewise.
8858 (vect_gen_prolog_loop_niters): Likewise.
8859 (vect_prepare_for_masked_peels): Likewise.
8860 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8861 (vect_determine_vectorization_factor): Likewise.
8862 (vect_is_simple_iv_evolution): Likewise.
8863 (vect_analyze_scalar_cycles_1): Likewise.
8864 (vect_analyze_loop_operations): Likewise.
8865 (report_vect_op): Likewise.
8866 (vect_is_slp_reduction): Likewise.
8867 (check_reduction_path): Likewise.
8868 (vect_is_simple_reduction): Likewise.
8869 (vect_create_epilog_for_reduction): Likewise.
8870 (vect_finalize_reduction:): Likewise.
8871 (vectorizable_induction): Likewise.
8872 (vect_transform_loop_stmt): Likewise.
8873 (vect_transform_loop): Likewise.
8874 (optimize_mask_stores): Likewise.
8875 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
8876 (vect_split_statement): Likewise.
8877 (vect_recog_over_widening_pattern): Likewise.
8878 (vect_recog_average_pattern): Likewise.
8879 (vect_determine_min_output_precision_1): Likewise.
8880 (vect_determine_precisions_from_range): Likewise.
8881 (vect_determine_precisions_from_users): Likewise.
8882 (vect_mark_pattern_stmts): Likewise.
8883 (vect_pattern_recog_1): Likewise.
8884 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8885 (vect_record_max_nunits): Likewise.
8886 (vect_build_slp_tree_1): Likewise.
8887 (vect_build_slp_tree_2): Likewise.
8888 (vect_print_slp_tree): Likewise.
8889 (vect_analyze_slp_instance): Likewise.
8890 (vect_detect_hybrid_slp_stmts): Likewise.
8891 (vect_detect_hybrid_slp_1): Likewise.
8892 (vect_slp_analyze_operations): Likewise.
8893 (vect_slp_analyze_bb_1): Likewise.
8894 (vect_transform_slp_perm_load): Likewise.
8895 (vect_schedule_slp_instance): Likewise.
8896 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8897 (vect_mark_stmts_to_be_vectorized): Likewise.
8898 (vect_init_vector_1): Likewise.
8899 (vect_get_vec_def_for_operand): Likewise.
8900 (vect_finish_stmt_generation_1): Likewise.
8901 (vect_check_load_store_mask): Likewise.
8902 (vectorizable_call): Likewise.
8903 (vectorizable_conversion): Likewise.
8904 (vectorizable_operation): Likewise.
8905 (vectorizable_load): Likewise.
8906 (vect_analyze_stmt): Likewise.
8907 (vect_is_simple_use): Likewise.
8908 (vect_get_vector_types_for_stmt): Likewise.
8909 (vect_get_mask_type_for_stmt): Likewise.
8910 * tree-vectorizer.c (increase_alignment): Likewise.
8912 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
8914 * doc/rtl.texi: Adjust vec_select description.
8915 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
8916 non-constant selectors.
8918 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
8920 * config/aarch64/aarch64-protos.h
8921 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
8922 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
8923 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
8924 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
8925 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
8926 (AARCH64_FL_PROFILE): Move index so flags are ordered.
8927 (AARCH64_ISA_RCPC8_4): New flag.
8928 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
8929 to aarch64_offset_9bit_signed_unscaled_p.
8930 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
8932 * config/aarch64/constraints.md (Ust): New constraint.
8933 * config/aarch64/predicates.md.
8934 (aarch64_9bit_offset_memory_operand): New predicate.
8935 (aarch64_rcpc_memory_operand): New predicate.
8937 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
8939 PR rtl-optimization/87361
8940 * rtlanal.c (nonzero_bits1): Revert accidental change.
8942 2018-09-19 Richard Biener <rguenther@suse.de>
8944 PR tree-optimization/87349
8945 PR tree-optimization/87342
8946 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
8948 2018-09-18 Marek Polacek <polacek@redhat.com>
8950 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
8951 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
8953 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8955 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
8957 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8959 PR rtl-optimization/86882
8960 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
8962 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
8964 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
8965 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
8967 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
8970 * doc/invoke.texi (Link Options): Fix formatting and grammar.
8972 2018-09-18 Richard Biener <rguenther@suse.de>
8975 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
8976 (compute_samebase_partition_bases): Likewise.
8977 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
8978 (gimple_can_coalesce_p): Simplify.
8980 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
8982 Handle a library implementation of ffs calling __builtin_ffs.
8983 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
8984 (mmix_init_libfuncs): New function: make __builtin_ffs expand
8987 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8989 * diagnostic-show-locus.c (class layout_range): Add field
8991 (layout_range::layout_range): Add "original_idx" param and use it
8992 to initialize new field.
8993 (make_range): Use 0 for original_idx.
8994 (layout::layout): Pass in index to calls to
8995 maybe_add_location_range.
8996 (layout::maybe_add_location_range): Add param "original_idx" and
8997 pass it on to layout_range.
8998 (layout::print_any_labels): Pass on range->m_original_idx to
9000 (gcc_rich_location::add_location_if_nearby): Use 0 for
9002 * gcc-rich-location.h (text_range_label::get_text): Update for new
9004 (range_label_for_type_mismatch::get_text): Likewise.
9006 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
9008 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
9010 2018-09-17 David Malcolm <dmalcolm@redhat.com>
9012 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
9013 format_string_diagnostic_t.
9014 (fmtwarn_n): Likewise.
9015 * substring-locations.c
9016 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
9017 (format_warning_n_va): Convert to...
9018 (format_string_diagnostic_t::emit_warning_n_va): ...this.
9019 (format_warning_va): Convert to...
9020 (format_string_diagnostic_t::emit_warning_va): ...this.
9021 (format_warning_at_substring): Convert to...
9022 (format_string_diagnostic_t::emit_warning): ...this.
9023 (format_warning_at_substring_n): Convert to...
9024 (format_string_diagnostic_t::emit_warning_n): ...this.
9025 * substring-locations.h (class format_string_diagnostic_t): New
9027 (format_warning_va): Convert to
9028 format_string_diagnostic_t::emit_warning_va.
9029 (format_warning_n_va): Convert to
9030 format_string_diagnostic_t::emit_warning_n_va.
9031 (format_warning_at_substring): Convert to
9032 format_string_diagnostic_t::emit_warning.
9033 (format_warning_at_substring_n): Convert to
9034 format_string_diagnostic_t::emit_warning_n.
9036 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
9037 Bernd Schmidt <bernds_cb1@t-online.de>
9039 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
9042 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
9044 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
9045 operand 0 predicate to nonimmediate operand.
9046 (rint<mode>2_frndint): Remove insn pattern.
9047 (rint<mode>2): Change operand 1 predicate to general_operand.
9048 Extend operand 1 to XFmode and generate rintxf2 insn.
9049 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
9050 Do not use X87MODEF mode macro.
9051 (frndintxf2_<rounding>_i387): Rename from
9052 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
9053 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
9054 to XFmode and generate significandxf3 insn.
9056 2018-09-17 Richard Biener <rguenther@suse.de>
9058 PR tree-optimization/87328
9059 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
9060 visiting unexecutable backedges when not iterating.
9061 (do_rpo_vn): Mark all edges not executable even when not
9064 2018-09-17 Martin Jambor <mjambor@suse.cz>
9067 * doc/invoke.texi (Warning Options): Likewise.
9069 2018-09-17 Richard Biener <rguenther@suse.de>
9071 PR tree-optimization/87301
9072 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
9073 clean EH info from leftover copy assignments.
9075 2018-09-17 Martin Liska <mliska@suse.cz>
9077 PR gcov-profile/85871
9078 * gcov.c (output_intermediate_file): Fix out of bounds
9081 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
9083 * config/arc/arc.c: Object attributes for core4 not reflected
9085 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
9088 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
9090 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
9092 2018-09-17 Martin Liska <mliska@suse.cz>
9094 * doc/gcov.texi: Document new option --use-hotness-colors.
9095 * gcov.c (struct source_info): Declare new field.
9096 (source_info::source_info): Set default for maximum_count.
9097 (print_usage): Add new -q option.
9098 (process_args): Process it.
9099 (accumulate_line_info): Save src->maximum_count.
9100 (output_line_beginning): Make color line number if
9101 flag_use_hotness_colors is set.
9102 (output_line_details): Pass default argument value.
9103 (output_lines): Pass src->maximum_count.
9105 2018-09-17 Martin Liska <mliska@suse.cz>
9107 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
9108 Use processor_names table.
9109 * config/i386/i386.c (ix86_default_align): Use
9110 processor_cost_table for alignment values.
9111 (ix86_option_override_internal): Use processor_names.
9112 (ix86_function_specific_print): Likewise.
9113 * config/i386/i386.h (struct processor_costs):
9114 Add alignment values.
9115 (struct ptt): Remove and replace with const char *.
9116 * config/i386/x86-tune-costs.h (struct processor_costs):
9117 Declare default alignments for all costs.
9119 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
9121 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
9122 symbolics or VR_VARYING ranges for ABS_EXPR.
9123 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
9124 when range will wrap.
9126 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
9129 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
9130 before and after a JUMP_TABLE_DATA.
9132 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
9135 * dojump.c (do_compare_and_jump): Canonicalize function pointers
9136 when one operand is a function pointer. Use POINTER_TYPE_P and
9137 FUNC_OR_METHOD_TYPE_P.
9138 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
9139 * fold-const.c (build_range_check): Likewise.
9140 * match.pd (simple_comparison): Likewise.
9142 2018-09-14 David Malcolm <dmalcolm@redhat.com>
9145 * spellcheck.c (get_edit_distance_cutoff): New function.
9146 (selftest::test_edit_distance_unit_test_oneway): Rename to...
9147 (selftest::test_get_edit_distance_one_way): ...this.
9148 (selftest::test_get_edit_distance_unit): Rename to...
9149 (selftest::test_get_edit_distance_both_ways): ...this.
9150 (selftest::test_edit_distances): Move tests to this new function,
9151 and test some more pairs of strings. Update for above renaming.
9152 (selftest::get_old_cutoff): New function.
9153 (selftest::test_get_edit_distance_cutoff): New function.
9154 (selftest::assert_suggested_for): New function.
9155 (ASSERT_SUGGESTED_FOR): New macro.
9156 (selftest::assert_not_suggested_for): New function.
9157 (ASSERT_NOT_SUGGESTED_FOR): New macro.
9158 (selftest::test_suggestions): New function.
9159 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
9160 tests to selftest::test_edit_distances and call it. Add calls to
9161 selftest::test_get_edit_distance_cutoff and
9162 selftest::test_suggestions.
9163 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
9164 (best_match::consider): Replace hard-coded cutoff calculation with
9166 (best_match::get_cutoff): New declaration.
9167 (best_match::get_best_meaningful_candidate): Likewise.
9169 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
9171 * builtins.c (fold_builtin_strlen): Remove TODO comment.
9173 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
9176 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9178 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9179 terminated string literal.
9181 2018-09-14 Martin Sebor <msebor@redhat.com>
9183 * builtins.c (unterminated_array): Handle ARRAY_REF.
9184 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
9185 * builtins.h (unterminated_array): Declare extern.
9186 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
9188 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
9191 2018-09-14 Martin Sebor <msebor@redhat.com>
9192 Jeff Law <law@redhat.com>
9194 * builtins.c (unterminated_array): New.
9195 (expand_builtin_strcpy): Adjust.
9196 (expand_builtin_strcpy_args): Detect unterminated arrays.
9197 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
9198 unterminated arrays.
9199 * gimple-fold.h (get_maxval_strlen): Add argument.
9200 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
9202 * gimple-fold.c (get_range_strlen): Add argument.
9203 (get_maxval_strlen): Adjust.
9204 * gimple-fold.h (get_range_strlen): Add argument.
9206 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
9208 * config/i386/movdirintrin.h: Fix copyright year.
9210 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
9212 * reg-stack.c: Include regs.h.
9213 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
9214 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
9215 FIRST_STACK_REG, not DFmode.
9216 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
9217 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
9218 (change stack): Default register mode to the reg_raw_mode of
9219 FIRST_STACK_REG, not DFmode.
9220 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
9221 (*swapxf): Rename from swapxf.
9223 2018-09-14 Carl Love <cel@us.ibm.com>
9225 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
9226 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
9228 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
9231 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
9234 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
9237 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
9239 2018-09-14 Jason Merrill <jason@redhat.com>
9241 Fix --enable-gather-detailed-mem-stats.
9242 * hash-table.c (hash_table_usage): Change from variable to function.
9243 * hash-table.h: Adjust.
9244 * Makefile.in: Add missing dependencies on hash-table.h.
9246 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9248 PR tree-optimization/87259
9250 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
9251 execute_cse_reciprocals_1 has tried transforming.
9253 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
9255 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
9256 VR_VARYING for PLUS/MINUS_EXPR.
9258 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
9260 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
9263 2018-09-14 Richard Biener <rguenther@suse.de>
9266 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
9267 bits for the merged partition.
9269 2018-09-13 Martin Sebor <msebor@redhat.com>
9270 Bernd Edlinger <bernd.edlinger@hotmail.de>
9272 * builtins.h (c_srlen): Add argument.
9273 * builtins.c (warn_string_no_nul): New function.
9274 (c_strlen): Add argument and use it. Update recursive calls.
9275 Pass DECL argument to string_constant to get info on non
9276 terminated strings. Update *NONSTR as needed.
9277 (fold_builtin_strlen): Add argument to calls to c_strlen.
9278 Warn for unterminated arrays.
9279 (warn_string_no_null): Add prototype.
9280 * expr.c (string_constant): Update arguments. Update recursive
9281 calls appropriately. Detect missing NUL terminator and outermost
9282 declaration its missing in.
9283 Improve checks for arrays with nonzero lower bound or elements
9284 that are not a single byte. Simplify offset computation.
9285 Simplify checks for non-NUL terminated strings.
9286 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
9287 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
9289 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
9291 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
9293 * fold-const.c (c_getstr): Fix function comment. Remove unused third
9294 argument. Fix range checks.
9295 * fold-const.h (c_getstr): Adjust protoype.
9296 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
9297 string is constant but contains no NUL byte.
9299 * expr.c (string_constant): Adjust function comment.
9300 Remove bogus check for zero termination.
9302 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
9304 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
9305 (get_constant_size): Don't make STRING_CSTs larger than they are.
9306 (check_string_literal): New check function for STRING_CSTs.
9307 (output_constant): Use it.
9309 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
9312 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
9314 2018-09-13 Richard Biener <rguenther@suse.de>
9316 PR tree-optimization/87263
9317 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
9318 (struct unwind_state): Add max_rpo field.
9319 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
9320 Compute max_rpo, the max RPO number a block can be backwards reached
9321 from. Re-write non-iterating mode to a RPO ordered worklist approach,
9322 separating it from the iterating mode.
9324 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
9326 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
9327 (rfs_decision): New scheduling decision.
9329 2018-09-13 Richard Biener <rguenther@suse.de>
9332 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
9333 (vn_nary_op_insert_pieces_predicated): Do not write useless
9334 valid_dominated_by_p entry outside of the allocated storage.
9336 2018-09-13 Omar Sandoval <osandov@osandov.com>
9337 Tom de Vries <tdevries@suse.de>
9340 * dwarf2out.c (is_c): New function.
9341 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
9343 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
9346 * config/aarch64/aarch64.md (*aarch64_bfxil):
9348 * config/aarch64/constraints.md (Ulc): Define.
9349 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
9351 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
9354 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
9356 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9357 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
9358 aarch64_layout_frame call.
9359 (aarch64_expand_epilogue): Likewise.
9360 (aarch64_initial_elimination_offset): Likewise.
9361 (aarch64_get_separate_components): Likewise.
9362 (aarch64_use_return_insn_p): Likewise.
9363 (aarch64_layout_frame): Remove unneeded check.
9365 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9367 * configure.ac: Only append
9368 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
9369 gcc_config_arguments if it was never reconfigured or last reconfigure
9370 was with different arguments.
9371 * configure: Regenerated.
9373 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9374 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9377 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
9378 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
9380 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9382 PR tree-optimization/87287
9383 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
9384 X % C == 0 to X % (unsigned) C == 0 optimization to ...
9385 * match.pd (X % C == 0): ... here. New optimization.
9387 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9390 * expr.h (maybe_optimize_mod_cmp): Declare.
9391 * expr.c (mod_inv): New function.
9392 (maybe_optimize_mod_cmp): New function.
9393 (do_store_flag): Use it.
9394 * cfgexpand.c (expand_gimple_cond): Likewise.
9396 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
9397 Julian Brown <julian@codesourcery.com>
9400 * gimplify.c (gimplify_scan_omp_clauses): Set
9401 target_firstprivatize_array_bases in OpenACC parallel and kernels
9402 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
9403 OpenACC data regions.
9405 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
9407 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
9408 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
9410 2018-09-12 Richard Biener <rguenther@suse.de>
9412 PR tree-optimization/87280
9413 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
9414 edge but unreachable target.
9415 (do_rpo_vn): For conservatively handling a PHI only mark
9416 the backedge executable but not the block reachable.
9418 2018-09-12 Richard Biener <rguenther@suse.de>
9420 PR tree-optimization/87266
9421 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
9424 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
9426 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
9428 ("trunc<BFP:mode><DFP_ALL:mode>2")
9429 ("trunc<DFP_ALL:mode><BFP:mode>2")
9430 ("extend<BFP:mode><DFP_ALL:mode>2")
9431 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
9432 according to the target operand type.
9434 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9435 Andreas Krebbel <krebbel@linux.ibm.com>
9437 PR tree-optimization/86844
9438 * gimple-ssa-store-merging.c
9439 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
9440 there are any overlapping stores in between them, make sure they are
9441 also coalesced or we give up completely.
9443 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9446 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
9447 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
9449 2018-09-12 Tom de Vries <tdevries@suse.de>
9451 * common.opt (gdescribe-dies): Add option.
9452 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
9453 attribute for artifical and nameless decls.
9454 (dwarf2out_register_external_die): Add description attribute to
9455 external reference die.
9456 (add_desc_attribute): New functions.
9457 (gen_subprogram_die): Add description attribute to
9458 DW_TAG_call_site_parameter.
9459 * tree-pretty-print.c (print_generic_expr_to_str): New function.
9460 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
9461 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
9463 (@item -gdescribe-dies): Add.
9465 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
9467 * tree-vrp.c (vrp_shift_undefined_p): Remove.
9468 (extract_range_from_binary_expr_1: Call
9469 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
9470 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
9473 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
9475 * gimple-ssa-warn-alloca.c
9476 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
9477 field for ALLOCA_BOUND_*_LARGE.
9479 2018-09-11 Nathan Sidwell <nathan@acm.org>
9481 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
9483 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9485 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
9486 for clobbers. Remove obsolete comment.
9488 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9490 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
9491 mpxchk, mpxld and mpxst types.
9492 (define_attr length_immediate): Remove all processing of mpx types.
9493 (define_attr prefix_0f): Ditto.
9494 (define_attr memory): Ditto.
9496 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9498 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
9499 (log<mode>2): Change operand 1 predicate to general_operand.
9500 Extend operand 1 to XFmode and generate logxf3 insn.
9501 (log10<mode>2): Change operand 1 predicate to general_operand.
9502 Extend operand 1 to XFmode and generate log10xf3 insn.
9503 (log2<mode>2): Change operand 1 predicate to general_operand.
9504 Extend operand 1 to XFmode and generate log2xf3 insn.
9505 (fyl2xp1_extend<mode>xf3_i387): Remove.
9506 (log1p<mode>2): Change operand 1 predicate to general_operand.
9507 Extend operand 1 to XFmode and generate log1pxf3 insn.
9508 (fxtract_extend<mode>xf3_i387): Remove.
9509 (logb<mode>2): Change operand 1 predicate to general_operand.
9510 Extend operand 1 to XFmode and generate logbxf3 insn.
9511 (ilogb<mode>2): Change operand 1 predicate to general_operand.
9512 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
9513 (significand<mode>2): Change operand 1 predicate to general_operand.
9514 Extend operand 1 to XFmode and generate significandxf3 insn.
9516 2018-09-11 Nathan Sidwell <nathan@acm.org>
9518 * gcc.c (perror_with_name, pfatal_with_name): Delete.
9519 (load_specs): Use fatal_error.
9520 (DELETE_IF_ORDINARY, process_command): Use error.
9521 (execute, run_attempt): Use fatal_error.
9523 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
9525 * diagnostic-core.h (sorry_at): New prototype.
9526 * diagnostic.c (sorry_at): New function.
9528 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
9530 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
9531 by zero as VR_UNDEFINED.
9533 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
9535 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
9536 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
9537 (<sincos>mode2): New expander.
9538 (sincos_extend<mode>xf3_i387): Remove insn pattern.
9539 (sincos -> sin, cos splitters): Remove splitter patterns.
9540 (sincos<mode>3): Change operand 2 predicate to general_operand.
9541 Extend operand 2 to XFmode and generate sincosxf3 insn.
9542 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
9543 Change operand 3 predicate to const1_operand.
9544 (fptan_extend<mode>xf4_i387): Remove insn pattern.
9545 (tanxf2): Update operands in the call to fptanxf4_i387.
9546 (tan<mode>2): Change operand 1 predicate to general_operand.
9547 Extend operand 1 to XFmode and generate tanxf3 insn.
9548 (atan2xf3): Rename from *fpatanxf3_i387.
9549 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
9550 (atan2xf3): Remove expander.
9551 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
9552 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
9553 (atan<mode>2): Change operand 1 predicate to general_operand.
9554 Extend operand 1 to XFmode and generate atanxf3 insn.
9556 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
9558 * config/i386/i386.md (x87/SSE constant load splitter): Use
9559 memory_operand instead of nonimmediate_operand for input operand
9562 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
9564 * config/i386/i386.md (float partial SSE register stall splitter): Move
9565 splitter near its instruction pattern.
9566 (float_extend partial SSE register stall splitter): Ditto.
9567 (float_truncate partial SSE register stall splitter): Ditto.
9569 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
9572 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9573 to speculation_safe_value_not_needed.
9576 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
9577 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
9578 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
9579 leaf_function_p, instead use has_hard_reg_initial_val.
9581 2018-09-09 Nathan Sidwell <nathan@acm.org>
9583 * gcc.h (pfatal_with_name): Don't declare here.
9584 * gcc.c (pfatal_with_name): Make static.
9586 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
9588 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
9591 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
9593 PR rtl-optimization/85458
9594 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
9595 priority hook to reduce the priority of EXPR.
9597 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
9599 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
9600 DImode for x87 on 32bit targets. Conditionally disable x87 modes
9601 with X87_ENABLE_FLOAT. Remove preparation code.
9602 (*float<SWI48:mode><MODEF:mode>2): Rename from
9603 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
9604 math using "enabled" attribute.
9605 (*floatdi<MODEF:mode>2_i387): Rename from
9606 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
9607 enable for 32bit targets only.
9608 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
9610 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
9611 as operand 1 predicate. Rewrite as define_insn_and_split.
9612 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
9614 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
9616 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
9617 to fallthru to FLOAT case.
9619 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9622 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
9623 around folding of vec_sl to handle out of range shift values.
9625 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
9627 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
9628 Update callers to gen_fix_trunc<mode>_i387_fisttp
9629 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
9630 nonimmediate_operand.
9631 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
9632 and corresponding splitters.
9633 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
9634 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
9635 (fix_truncdi_i387_with_temp): Remove insn pattern
9636 and corresponding splitters.
9637 (fix_trunc<mode>_i387): Change operand 0 predicate to
9638 nonimmediate_operand.
9639 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
9640 and corresponding splitters.
9641 (*fistdi2_1): Remove.
9643 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
9644 (lrintxfdi2): Remove expander. Reimplement as define_insn.
9645 (*fist<mode>2_1): Remove.
9646 (fist<mode>2): Ditto.
9647 (fist<mode>2_with_temp): Remove insn pattern and corresponding
9649 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
9650 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
9651 (fistdi2_<rounding>): Change operand 0 predicate to
9652 nonimmediate_operand.
9653 (fistdi2_<rounding>_with_temp): Remove insn pattern
9654 and corresponding splitters.
9655 (fist<mode>2_<rounding>): Change operand 0 predicate to
9656 nonimmediate_operand.
9657 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
9658 and corresponding splitters.
9660 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
9662 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9664 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
9667 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9669 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9670 early gimple folding of vec_splat().
9671 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
9672 * gimple-fold.h: Add an extern define for tree_vec_extract().
9674 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9676 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
9677 wrappers around TREE_TYPE comparisons.
9679 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9682 * config/s390/predicates.md: Add nonsym_memory_operand.
9683 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
9684 contains a SYMBOL_REF, load it into an intermediate pseudo.
9685 (s390_emit_compare_and_swap): Legitimize operand.
9686 * config/s390/s390.md: Use the new nonsym_memory_operand
9687 with UNSPECV_CAS patterns.
9689 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9692 * config/s390/s390-passes.def: New file.
9693 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
9695 (make_pass_s390_early_mach): Add declaration.
9696 * config/s390/s390.c (make_pass_s390_early_mach):
9697 (s390_option_override): Remove dynamic registration.
9698 * config/s390/t-s390: Add s390-passes.def.
9700 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9702 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9703 Remove UNSPEC_LTREL_BASE check.
9704 (annotate_constant_pool_refs): Likewise.
9705 (find_constant_pool_ref): Likewise.
9706 (find_ltrel_base): Removed.
9707 (replace_ltrel_base): Removed.
9708 (s390_mainpool_finish): Remove replace_ltrel_base call.
9709 (s390_chunkify_start): Remove pending LTREL_BASE logic.
9710 (s390_chunkify_finish): Remove replace_ltrel_base call.
9711 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
9713 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
9716 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9717 to speculation_safe_value_not_needed.
9719 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
9720 Bernd Schmidt <bernds_cb1@t-online.de>
9722 * config/nvptx/nvptx-opts.h: New file.
9723 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
9724 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
9726 (TARGET_SM35): New macro.
9727 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
9729 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
9731 (misa=): New option.
9732 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
9734 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
9736 * config/i386/i386.md (truncdfsf2): Remove expander.
9737 (truncdfsf2_with_temp): Ditto.
9738 (truncxf<mode>2): Ditto.
9739 (*truncdfsf_fast_mixed): Remove insn pattern.
9740 (*truncdfsf_fast_i387): Ditto.
9741 (*truncdfsf_mixed): Ditto.
9742 (*truncdfsf_i387): Ditto.
9743 (*truncdfsf2_i387_1): Ditto.
9744 (*truncxfsf2_mixed): Ditto.
9745 (*truncxfdf2_mixed): Ditto.
9746 (*truncxf<mode>2_i387_noop): Ditto. Update callers
9747 to call gen_truncxf<mode>2 instead.
9748 (*truncxf<mode>2_i387): Remove.
9749 (reg->reg splitters): Remove splitter pattern.
9750 (reg->mem splitters): Ditto.
9752 (truncdfsf2): New insn pattern.
9753 (truncxf<mode>2): Ditto.
9755 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9757 * tree-ssa-math-opts.c (is_mult_by): New function.
9758 (is_square_of): Use the above.
9759 (optimize_recip_sqrt): New function.
9760 (pass_cse_reciprocals::execute): Use the above.
9762 2018-09-05 Richard Biener <rguenther@suse.de>
9765 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
9766 to zero-init the emplaced vec.
9768 2018-09-05 Martin Liska <mliska@suse.cz>
9770 PR tree-optimization/87205
9771 * tree-switch-conversion.c (pass_lower_switch::execute):
9772 Group cases for switch statements.
9774 2018-09-05 Richard Biener <rguenther@suse.de>
9776 PR tree-optimization/87217
9777 * tree-ssa-sccvn.c (vuse_valueize): New.
9778 (vn_reference_lookup_pieces): Use it.
9779 (vn_reference_lookup): Likewise.
9781 2018-09-05 Nathan Sidwell <nathan@acm.org>
9784 * stor-layout.c (place_field): Scan forwards to check last
9785 bitfield when ms_bitfield_placement is in effect.
9787 2018-09-05 Richard Biener <rguenther@suse.de>
9790 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
9793 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
9794 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
9796 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
9797 * config.gcc (extra_objs): Build it.
9798 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
9800 * config/aarch64/aarch64-passes.def
9801 (pass_tag_collision_avoidance): New pass.
9802 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
9803 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
9804 (aarch64_classify_address): Remove static qualifier.
9805 (aarch64_address_info, aarch64_address_type): Move to...
9806 * config/aarch64/aarch64-protos.h: ... here.
9807 (make_pass_tag_collision_avoidance): New function.
9808 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
9811 2018-09-05 Martin Liska <mliska@suse.cz>
9813 * doc/gcov.texi: Update documentation of humar
9815 * gcov.c (format_count): Print one decimal place, it provides
9816 more fine number of situations like '1G' vs. '1.4G'.
9818 2018-09-05 Martin Liska <mliska@suse.cz>
9821 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
9822 * optc-gen.awk: Allow 'Var' for Deprecated options in order
9823 to generate a MASK value.
9825 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
9828 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
9829 if frame pointer isn't used.
9830 (compute_frame_pointer_to_fb_displacement): Likewise.
9832 2018-09-04 Jakub Jelinek <jakub@redhat.com>
9835 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
9836 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
9837 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
9838 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
9839 and OPTION_MASK_ISA_XSAVEC_UNSET.
9841 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
9843 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
9844 XOR operations in NAND case.
9846 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9848 * wide-int-range.cc (wide_int_range_convert): New.
9849 * wide-int-range.h (wide_int_range_convert): New.
9850 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
9851 code into wide_int_range_convert.
9852 (extract_range_into_wide_ints): Do not munge anti range constants
9853 into the entire domain. Just return the range back.
9855 2018-09-04 Martin Liska <mliska@suse.cz>
9857 * genmatch.c (output_line_directive): Add new argument
9859 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
9861 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9863 * doc/invoke.texi (Option Summary): Add whitespace.
9865 * doc/invoke.texi (Option Summary): Add -Waligned-new.
9867 2018-09-04 Richard Biener <rguenther@suse.de>
9869 PR tree-optimization/87211
9870 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
9871 backedge value we're supposed to treat as VARYING also number
9872 the PHI to VARYING in case it got a different value-number already.
9874 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9876 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
9877 (extract_range_from_binary_expr_1): Do not call
9878 vrp_can_optimize_bit_op.
9879 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
9881 (wide_int_range_get_mask_and_bounds): New.
9882 (wide_int_range_optimize_bit_op): New.
9883 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
9884 (wide_int_range_bit_and): Same.
9885 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
9886 (wide_int_range_optimize_bit_op): New.
9887 (wide_int_range_get_mask_and_bounds): New.
9889 2018-09-04 Richard Biener <rguenther@suse.de>
9891 PR tree-optimization/87176
9892 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
9893 variable. When value-numbering a virtual PHI node make sure
9894 to not value-number to the backedge value.
9896 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9898 * doc/extend.texi (Long Long, Hex Floats): Document support for
9899 long long and hex floats in more recent versions of ISO C++.
9901 2018-09-03 Richard Biener <rguenther@suse.de>
9903 PR tree-optimization/87177
9904 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
9907 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9909 * bb-reorder.c (edge_order): Convert to C-qsort-style
9910 tri-state comparator.
9911 (reorder_basic_blocks_simple): Change std::stable_sort to
9914 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9916 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
9917 tri-state comparator.
9918 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
9920 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9922 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
9923 (mergesort): ... here as maximum count for using netsort.
9924 (gcc_qsort): Set nlim to 3 if stable sort is requested.
9925 (gcc_stablesort): New.
9926 * system.h (gcc_stablesort): Declare.
9928 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9930 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
9931 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
9932 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
9934 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
9936 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
9937 lxsdx and stxsdx alternatives.
9938 (*mov<mode>_hardfloat64): Ditto.
9939 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
9941 2018-09-03 Richard Biener <rguenther@suse.de>
9943 PR tree-optimization/87200
9944 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
9947 2018-09-03 Martin Liska <mliska@suse.cz>
9949 PR tree-optimization/87201
9950 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9951 Fix parenthesis in an expression.
9953 2018-09-03 Richard Biener <rguenther@suse.de>
9955 PR tree-optimization/87197
9956 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
9957 visited. CSE the VN_INFO hashtable lookup.
9959 PR tree-optimization/87169
9960 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
9961 iterating make sure there's no extra backedges from irreducible
9962 regions feeding the header. Mark the destination block
9965 2018-09-03 Martin Liska <mliska@suse.cz>
9968 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
9969 * common/common-targhooks.c (default_get_valid_option_values):
9971 * common/common-targhooks.h (default_get_valid_option_values):
9973 * common/config/i386/i386-common.c: Move processor_target_table
9975 (ix86_get_valid_option_values): New function.
9976 (TARGET_GET_VALID_OPTION_VALUES): New macro.
9977 * config/i386/i386.c (struct ptt): Move to i386-common.c.
9978 (PTA_*): Move all defined masks into i386-common.c.
9979 (ix86_function_specific_restore): Use new processor_cost_table.
9980 * config/i386/i386.h (struct ptt): Moved from i386.c.
9981 (struct pta): Likewise.
9982 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
9983 * doc/tm.texi.in: Likewise.
9984 * opt-suggestions.c (option_proposer::suggest_option):
9985 Pass prefix to build_option_suggestions.
9986 (option_proposer::get_completions): Likewise.
9987 (option_proposer::build_option_suggestions): Use the new target
9989 * opts.c (struct option_help_tuple): New struct.
9990 (print_filtered_help): Use the new target hook.
9992 2018-09-03 Martin Liska <mliska@suse.cz>
9995 * predict.c (set_even_probabilities): Add likely_edges
9996 argument and handle cases where we have precisely one
9998 (combine_predictions_for_bb): Catch also likely_edges.
9999 (tree_predict_by_opcode): Handle gswitch statements.
10000 * tree-cfg.h (find_case_label_for_value): New declaration.
10001 (find_taken_edge_switch_expr): Likewise.
10002 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
10003 Find pivot in decision tree based on probabily, not by number of
10006 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
10008 * doc/standards.texi (Standards): Update Objective-C reference.
10010 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
10012 * doc/install.texi (Prerequisites): Update link for MPC.
10014 2018-09-01 Michael Matz <matz@suse.de>
10016 PR tree-optimization/87074
10017 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
10018 PHIs for outer-loop uses.
10020 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
10022 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
10023 * doc/invoke.texi (C Dialect Options): Ditto.
10025 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
10027 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
10029 2018-08-31 Richard Biener <rguenther@suse.de>
10031 PR tree-optimization/87168
10032 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
10033 (rpo_elim::eliminate_avail): When OP was not visited it must
10036 2018-08-31 David Malcolm <dmalcolm@redhat.com>
10038 * tree-vrp.c (copy_value_range): Convert param "from" from
10039 "value_range *" to "const value_range *".
10040 (range_is_null): Likewise for param "vr".
10041 (range_int_cst_p): Likewise.
10042 (range_int_cst_singleton_p): Likewise.
10043 (symbolic_range_p): Likewise.
10044 (value_ranges_intersect_p): Likewise for both params.
10045 (value_range_nonnegative_p): Likewise for param "vr".
10046 (value_range_constant_singleton): Likewise.
10047 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
10048 (extract_range_into_wide_ints): Likewise for param "vr".
10049 (extract_range_from_multiplicative_op): Likewise for params "vr0"
10051 (vrp_can_optimize_bit_op): Likewise.
10052 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
10054 (extract_range_from_unary_expr): Likewise.
10055 (debug_value_range): Likewise for param "vr".
10056 (value_range::dump): Add "const" qualifier.
10057 (vrp_prop::check_array_ref): Convert local "vr" from
10058 "value_range *" to "const value_range *".
10059 (vrp_prop::check_mem_ref): Likewise.
10060 (vrp_prop::visit_stmt): Likewise for local "old_vr".
10061 (vrp_intersect_ranges_1): Likewise for param "vr_1".
10062 (vrp_intersect_ranges): Likewise.
10063 (simplify_stmt_for_jump_threading): Likewise for local "vr".
10064 (vrp_prop::vrp_finalize): Likewise.
10065 * tree-vrp.h (value_range::dump): Add "const" qualifier.
10066 (vrp_intersect_ranges): Add "const" qualifier to params as above.
10067 (extract_range_from_unary_expr): Likewise.
10068 (value_range_constant_singleton): Likewise.
10069 (symbolic_range_p): Likewise.
10070 (copy_value_range): Likewise.
10071 (extract_range_from_binary_expr_1): Likewise.
10072 (range_int_cst_p): Likewise.
10073 (vrp_set_zero_nonzero_bits): Likewise.
10074 (range_int_cst_singleton_p): Likewise.
10076 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
10078 * config/aarch64/arm_neon.h (vabsd_s64): New.
10079 (vnegd_s64): Likewise.
10081 2018-08-31 Martin Jambor <mjambor@suse.cz>
10083 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
10085 2018-08-31 Martin Liska <mliska@suse.cz>
10087 * ipa-icf.c (sem_item::add_type): Use
10088 sem_item::m_type_hash_cache.
10089 * ipa-icf.h: Move the cache from sem_item_optimizer
10092 2018-08-31 Nathan Sidwell <nathan@acm.org>
10094 * doc/extend.texi (Backwards Compatibility): Remove implicit
10095 extern C leeway of () being (...).
10097 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10099 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
10101 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
10105 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
10107 2018-08-31 Jakub Jelinek <jakub@redhat.com>
10109 PR middle-end/87138
10110 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
10111 gen_int_mode. Formatting fixes.
10113 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
10115 * target.def (custom_function_descriptors): Improve documentation.
10116 * doc/tm.texi.in (Trampolines): Expand discussion of function
10117 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
10118 beginning of the section.
10119 * doc/tm.texi: Regenerated.
10121 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
10123 * cfg.h (class auto_edge_flag): Spell out the template-id of the
10124 base class in the initializer list. This is a workaround for
10125 building with older GCC.
10126 (class auto_bb_flag): Likewise.
10128 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
10130 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
10131 (altivec_vcmpequ<VI_char>_p): Remove star.
10132 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
10134 (expand_strncmp_vec_sequence): New function.
10135 (emit_final_str_compare_vec): New function.
10136 (expand_strn_compare): Add support for vector strncmp.
10137 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
10138 length specification to bytes.
10139 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
10140 (vcmpnezb_p): New pattern.
10141 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
10142 for option -mstring-compare-inline-limit.
10144 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
10146 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
10147 (PTA_SKYLAKE): Add PTA_AES.
10148 (PTA_GOLDMONT): Likewise.
10150 2018-08-29 Jan Hubicka <jh@suse.cz>
10153 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
10154 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
10156 2018-08-29 Jan Hubicka <jh@suse.cz>
10158 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
10160 (hash_tree): Do not visit TYPE_STUB_DECL.
10161 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
10162 stream TYPE_STUB_DECL.
10163 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
10164 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
10165 after free_lang_data.
10166 (type_in_anonymous_namespace_p): Likewise.
10168 2018-08-29 Jan Hubicka <jh@suse.cz>
10170 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
10171 comment that it has to be even number.
10172 (class sreal): Change m_sig type to int32_t.
10173 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
10174 int64_t for temporary calculations.
10175 (sreal_verify_basics): Drop one bit from minimum and maximum.
10177 2018-08-30 Richard Biener <rguenther@suse.de>
10179 PR tree-optimization/87147
10180 * tree-ssa-sccvn.c (SSA_VISITED): New function.
10181 (visit_phi): When the degenerate result is from the backedge and
10182 we didn't visit its definition yet drop to VARYING.
10183 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
10185 2018-08-29 Jan Hubicka <jh@suse.cz>
10187 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
10189 (hash_tree): Likewise.
10191 2018-08-29 Jan Hubicka <jh@suse.cz>
10193 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
10194 and TYPE_NEXT_REF_TO.
10196 2018-08-29 Jan Hubicka <jh@suse.cz>
10198 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
10199 comment that it has to be even number.
10200 (class sreal): Change m_sig type to int32_t.
10201 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
10202 int64_t for temporary calculations.
10203 (sreal_verify_basics): Drop one bit from minimum and maximum.
10205 2018-08-30 Tamar Christina <tamar.christina@arm.com>
10207 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
10209 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
10211 PR middle-end/86995
10212 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
10213 if to_add is negative.
10215 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
10217 PR middle-end/87053
10218 * builtins.c (c_strlen): Improve range checks.
10220 2018-08-29 Martin Sebor <msebor@redhat.com>
10221 Jeff Law <law@redhat.com>
10223 PR tree-optimization/86714
10224 PR tree-optimization/86711
10225 * builtins.c (c_strlen): Add arguments to call to string_constant.
10226 * expr.c (string_constant): Add argument. Detect missing nul
10227 terminator and outermost declaration it's missing in.
10228 * expr.h (string_constant): Add argument.
10229 * fold-const.c (read_from_constant_string): Add arguments to call to
10231 (c_getstr): Likewise.
10232 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
10233 to string_constant.
10234 * tree-ssa-strlen.c (get_stridx): Likewise.
10236 2018-08-29 Jan Hubicka <jh@suse.cz>
10238 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
10239 Do not stream DECL_VINDEX.
10240 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
10241 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
10242 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
10244 2018-08-29 Richard Biener <rguenther@suse.de>
10246 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
10247 virtual operands that are not default defs to honor region
10249 (rpo_vn_valueize): Remove ineffective code here.
10251 2018-08-29 Richard Biener <rguenther@suse.de>
10253 PR tree-optimization/87132
10254 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
10255 when skipping defs reachable over backedges.
10257 2018-08-29 Richard Biener <rguenther@suse.de>
10259 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
10260 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
10261 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
10262 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
10263 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
10264 (vn_reference_lookup_3): Remove use of const_parms.
10265 (free_rpo_vn): Do not free const_parms.
10266 (do_rpo_vn): Do not call init_const_parms.
10267 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
10268 SSA_NAME_POINTS_TO_READONLY_MEMORY.
10269 (call_may_clobber_ref_p_1): Likewise.
10271 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
10274 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
10276 (-ftree-scev-cprop): Document.
10278 2018-08-29 Jan Hubicka <jh@suse.cz>
10280 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
10282 (sreal constructor): Update.
10283 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
10284 sreal:operator/): Update.
10286 2018-08-29 Martin Liska <mliska@suse.cz>
10288 * tree-switch-conversion.c (switch_conversion::expand):
10289 Strenghten assumption about gswitch statements.
10291 2018-08-29 Richard Biener <rguenther@suse.de>
10293 PR tree-optimization/87117
10294 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
10295 re-value-number released SSA VDEFs.
10297 2018-08-29 Richard Biener <rguenther@suse.de>
10299 PR tree-optimization/87126
10300 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
10302 2018-08-28 Jim Wilson <jimw@sifive.com>
10304 * config/riscv/pic.md: Rewrite.
10305 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
10307 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
10308 (SOFTF, default_load, softload, softstore): New.
10310 2018-08-28 Jeff Law <law@redhat.com>
10312 * fold-const.c (fold_binary_loc): Remove recently added assert.
10314 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
10316 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
10317 to OP parmeter of generated function.
10319 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
10321 PR tree-optimization/87009
10322 * match.pd: Add boolean optimizations.
10324 2018-08-28 Martin Sebor <msebor@redhat.com>
10326 PR middle-end/86631
10327 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
10328 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
10329 (pass_walloca::gate): Use it.
10330 (alloca_call_type): Same.
10331 (pass_walloca::execute): Same.
10332 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
10334 2018-08-28 David Malcolm <dmalcolm@redhat.com>
10336 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
10337 GCC_VERSION for usage of "__gcc_dump_printf__" format from
10338 >= 3005 to >= 9000.
10340 2018-08-28 Richard Biener <rguenther@suse.de>
10342 PR tree-optimization/87124
10343 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
10344 constants before looking up avail.
10346 2018-08-28 Jakub Jelinek <jakub@redhat.com>
10348 PR middle-end/87099
10349 * calls.c (maybe_warn_nonstring_arg): Punt early if
10350 warn_stringop_overflow is zero. Don't call get_range_strlen
10351 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
10352 Swap comparison operands to have constants on rhs. Only use
10353 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
10354 increment lenrng[0].
10356 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
10358 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
10359 use of tree_to_shwi. Remove duplicated test for the size being
10360 a whole number of bytes.
10362 2018-08-28 Richard Biener <rguenther@suse.de>
10364 PR tree-optimization/87117
10365 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
10366 Handle removed stmt without LHS (GIMPLE_NOP).
10368 2018-08-28 Richard Biener <rguenther@suse.de>
10370 PR tree-optimization/87117
10371 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
10372 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
10374 2018-08-28 Richard Biener <rguenther@suse.de>
10376 PR tree-optimization/87117
10377 * tree-ssa-pre.c (compute_avail): Do not make expressions
10378 with predicated values available.
10379 (get_expr_value_id): Assert we do not run into predicated value
10382 2018-08-28 Richard Biener <rguenther@suse.de>
10384 PR tree-optimization/87117
10385 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
10386 get virtual operands.
10387 (get_expr_operands): Handle STRING_CST like other decls.
10389 2018-08-28 Martin Liska <mliska@suse.cz>
10391 * tree.h: Update documentation of fndecl_built_in_p
10395 2018-08-27 Jeff Law <law@redhat.com>
10396 PR tree-optimization/87110
10397 * tree-ssa-dse.c (compute_trims): Handle non-constant
10400 2018-08-27 Martin Sebor <msebor@redhat.com>
10402 PR tree-optimization/86914
10403 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
10405 2018-08-27 Martin Sebor <msebor@redhat.com>
10407 PR tree-optimization/87112
10408 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
10409 the type of the bound argument.
10411 2018-08-27 Jeff Law <law@redhat.com>
10413 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
10414 type does not have a TYPE_SIZE_UNIT.
10416 2018-08-27 Steve Ellcey <sellcey@cavium.com>
10418 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
10419 with include of backend.h.
10421 2018-08-27 Richard Biener <rguenther@suse.de>
10423 PR tree-optimization/86927
10424 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
10425 use const cond reduction code.
10427 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
10429 PR tree-optimization/85758
10430 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
10432 2018-08-27 David Malcolm <dmalcolm@redhat.com>
10435 * diagnostic-show-locus.c (class layout_range): Update for
10436 conversion of show_caret_p to a tri-state.
10437 (layout_range::layout_range): Likewise.
10438 (make_range): Likewise.
10439 (layout::maybe_add_location_range): Likewise.
10440 (layout::should_print_annotation_line_p): Don't show annotation
10441 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
10442 (layout::get_state_at_point): Update for conversion of
10443 show_caret_p to a tri-state. Bail out early for
10444 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
10445 underlining or source colorization.
10446 (gcc_rich_location::add_location_if_nearby): Update for conversion
10447 of show_caret_p to a tri-state.
10448 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
10449 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
10451 (selftest::test_one_liner_labels): Likewise.
10452 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
10453 conversion of show_caret_p to a tri-state.
10454 * pretty-print.c (text_info::set_location): Likewise.
10455 * pretty-print.h (text_info::set_location): Likewise.
10456 * substring-locations.c (format_warning_n_va): Likewise.
10457 * tree-diagnostic.c (default_tree_printer): Likewise.
10458 * tree-pretty-print.c (newline_and_indent): Likewise.
10460 2018-08-27 David Malcolm <dmalcolm@redhat.com>
10463 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
10464 line above for line-insertion fix-it hints.
10465 (selftest::test_fixit_insert_containing_newline): Update the
10466 expected results, and add a test with line-numbering enabled.
10468 2018-08-27 Martin Liska <mliska@suse.cz>
10471 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
10472 params with DECL_HAS_VALUE_EXPR_P.
10474 2018-08-27 Martin Liska <mliska@suse.cz>
10476 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
10477 selected expansion strategy.
10479 2018-08-27 Martin Liska <mliska@suse.cz>
10481 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
10482 * builtins.c (is_builtin_fn): Likewise.
10483 * attribs.c (diag_attr_exclusions): Use new function
10484 fndecl_built_in_p and remove check for FUNCTION_DECL if
10486 (builtin_mathfn_code): Likewise.
10487 (fold_builtin_expect): Likewise.
10488 (fold_call_expr): Likewise.
10489 (fold_builtin_call_array): Likewise.
10490 (fold_call_stmt): Likewise.
10491 (set_builtin_user_assembler_name): Likewise.
10492 (is_simple_builtin): Likewise.
10493 * calls.c (gimple_alloca_call_p): Likewise.
10494 (maybe_warn_nonstring_arg): Likewise.
10495 * cfgexpand.c (expand_call_stmt): Likewise.
10496 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
10497 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
10498 (cgraph_node::verify_node): Likewise.
10499 * cgraphclones.c (build_function_decl_skip_args): Likewise.
10500 (cgraph_node::create_clone): Likewise.
10501 * config/arm/arm.c (arm_insert_attributes): Likewise.
10502 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
10503 * dse.c (scan_insn): Likewise.
10504 * expr.c (expand_expr_real_1): Likewise.
10505 * fold-const.c (operand_equal_p): Likewise.
10506 (fold_binary_loc): Likewise.
10507 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10508 * gimple-low.c (lower_stmt): Likewise.
10509 * gimple-pretty-print.c (dump_gimple_call): Likewise.
10510 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
10512 * gimple.c (gimple_build_call_from_tree): Likewise.
10513 (gimple_call_builtin_p): Likewise.
10514 (gimple_call_combined_fn): Likewise.
10515 * gimplify.c (gimplify_call_expr): Likewise.
10516 (gimple_boolify): Likewise.
10517 (gimplify_modify_expr): Likewise.
10518 (gimplify_addr_expr): Likewise.
10519 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
10520 * ipa-cp.c (determine_versionability): Likewise.
10521 * ipa-fnsummary.c (compute_fn_summary): Likewise.
10522 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
10523 * ipa-split.c (visit_bb): Likewise.
10524 (split_function): Likewise.
10525 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10526 * lto-cgraph.c (input_node): Likewise.
10527 * lto-streamer-out.c (write_symbol): Likewise.
10528 * omp-low.c (setjmp_or_longjmp_p): Likewise.
10529 (lower_omp_1): Likewise.
10530 * predict.c (strip_predict_hints): Likewise.
10531 * print-tree.c (print_node): Likewise.
10532 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
10533 * trans-mem.c (is_tm_irrevocable): Likewise.
10534 (is_tm_load): Likewise.
10535 (is_tm_simple_load): Likewise.
10536 (is_tm_store): Likewise.
10537 (is_tm_simple_store): Likewise.
10538 (is_tm_abort): Likewise.
10539 (tm_region_init_1): Likewise.
10540 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
10541 * tree-cfg.c (verify_gimple_call): Likewise.
10542 (move_stmt_r): Likewise.
10543 (stmt_can_terminate_bb_p): Likewise.
10544 * tree-eh.c (lower_eh_constructs_2): Likewise.
10545 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
10546 * tree-inline.c (remap_gimple_stmt): Likewise.
10547 (copy_bb): Likewise.
10548 (estimate_num_insns): Likewise.
10549 (fold_marked_statements): Likewise.
10550 * tree-sra.c (scan_function): Likewise.
10551 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
10552 (optimize_stack_restore): Likewise.
10553 (pass_fold_builtins::execute): Likewise.
10554 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10555 (mark_all_reaching_defs_necessary_1): Likewise.
10556 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
10557 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
10558 (pass_forwprop::execute): Likewise.
10559 * tree-ssa-loop-im.c (stmt_cost): Likewise.
10560 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10561 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
10562 * tree-ssa-strlen.c (get_string_length): Likewise.
10563 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
10564 (find_func_aliases_for_call): Likewise.
10565 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
10566 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
10567 * tree-tailcall.c (find_tail_calls): Likewise.
10568 * tree.c (need_assembler_name_p): Likewise.
10569 (free_lang_data_in_decl): Likewise.
10570 (get_call_combined_fn): Likewise.
10571 * ubsan.c (is_ubsan_builtin_p): Likewise.
10572 * varasm.c (incorporeal_function_p): Likewise.
10573 * tree.h (DECL_BUILT_IN): Remove and replace with
10575 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
10576 (fndecl_built_in_p): New.
10578 2018-08-27 Martin Liska <mliska@suse.cz>
10580 PR tree-optimization/86847
10581 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
10582 Dump also subtree probability.
10583 (switch_decision_tree::do_jump_if_equal): New function.
10584 (switch_decision_tree::emit_case_nodes): Handle special
10585 situations in balanced tree that can be emitted much simpler.
10586 Fix calculation of probabilities that happen in tree expansion.
10587 * tree-switch-conversion.h (struct cluster): Add
10589 (struct simple_cluster): Likewise.
10590 (struct case_tree_node): Add new function has_child.
10591 (do_jump_if_equal): New.
10593 2018-08-27 Martin Liska <mliska@suse.cz>
10595 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
10596 Add new argument to bit_test_cluster constructor.
10597 (bit_test_cluster::emit): Set bits really number of values
10599 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
10600 probability argument.
10601 * tree-switch-conversion.h (struct bit_test_cluster):
10602 Add m_handles_entire_switch.
10604 2018-08-27 Martin Liska <mliska@suse.cz>
10606 PR tree-optimization/86702
10607 * tree-switch-conversion.c (jump_table_cluster::emit):
10608 Make probabilities even for values in jump table
10609 according to number of cases handled.
10610 (switch_decision_tree::compute_cases_per_edge): Pass
10611 argument to reset_out_edges_aux function.
10612 (switch_decision_tree::analyze_switch_statement): Likewise.
10613 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
10616 2018-08-27 Martin Liska <mliska@suse.cz>
10618 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
10619 cfun argument explicitly.
10620 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
10621 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
10622 function gimple_switch_default_bb.
10623 (convert_switch_statements):
10625 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
10626 * stmt.c (label_to_block_fn): Use label_to_block and pass
10627 cfun argument explicitly and use gimple_switch_label_bb.
10628 (expand_case): Likewise.
10629 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
10630 cfun argument explicitly. Likewise.
10631 (make_edges_bb): Likewise.
10632 (make_cond_expr_edges): Likewise.
10633 (get_cases_for_edge): Likewise.
10634 (make_gimple_switch_edges): Likewise.
10635 (label_to_block_fn): Likewise.
10636 (label_to_block): Likewise.
10637 (make_goto_expr_edges): Likewise.
10638 (make_gimple_asm_edges): Likewise.
10639 (main_block_label): Likewise.
10640 (group_case_labels_stmt): Likewise.
10641 (find_taken_edge_computed_goto): Likewise.
10642 (find_taken_edge_switch_expr): Likewise.
10643 (gimple_verify_flow_info): Likewise.
10644 (gimple_redirect_edge_and_branch): Likewise.
10645 (gimple_switch_label_bb): New function.
10646 (gimple_switch_default_bb): Likewise.
10647 (gimple_switch_edge): Likewise.
10648 (gimple_switch_default_edge): Likewise.
10649 * tree-cfg.h (label_to_block_fn): Remove and replace ...
10650 (label_to_block): ... with this.
10651 (gimple_switch_label_bb): New.
10652 (gimple_switch_default_bb): Likewise.
10653 (gimple_switch_edge): Likewise.
10654 (gimple_switch_default_edge): Likewise.
10655 * tree-cfgcleanup.c (convert_single_case_switch): Use
10656 new gimple functions and pass new argument to label_to_block.
10657 (cleanup_control_flow_bb):
10658 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
10659 cfun argument explicitly.
10660 (make_eh_edges): Likewise.
10661 (redirect_eh_dispatch_edge): Likewise.
10662 (lower_resx): Likewise.
10663 (lower_eh_dispatch): Likewise.
10664 (maybe_remove_unreachable_handlers): Likewise.
10665 (unsplit_eh): Likewise.
10666 (cleanup_empty_eh): Likewise.
10667 (verify_eh_edges): Likewise.
10668 (verify_eh_dispatch_edge): Likewise.
10669 * tree-ssa-dom.c (record_edge_info): Likewise.
10670 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
10671 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
10672 (thread_through_normal_block): Likewise.
10673 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
10674 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
10675 * tree-switch-conversion.c (switch_conversion::collect): Use new
10677 (switch_conversion::check_final_bb): Likewise.
10678 (switch_conversion::gather_default_values): Pass new argument
10680 (switch_conversion::build_constructors): Likewise.
10681 (switch_decision_tree::compute_cases_per_edge): Use new
10682 gimple_switch_edge function.
10683 (switch_decision_tree::analyze_switch_statement): Pass new argument
10685 (switch_decision_tree::try_switch_expansion): Use
10686 gimple_switch_default_edge.
10687 * tree-vrp.c (find_switch_asserts): Pass new argument
10689 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
10690 (vr_values::simplify_switch_using_ranges): Likewise.
10692 2018-08-27 Richard Biener <rguenther@suse.de>
10694 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
10695 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
10697 * tree-ssa-sccvn.h (struct vn_pval): New structure.
10698 (struct vn_nary_op_s): Add unwind_to member. Add
10699 predicated_values flag and put result into a union together
10700 with a linked list of vn_pval.
10701 (struct vn_ssa_aux): Add name member to make maintaining
10702 a map of SSA name to vn_ssa_aux possible. Remove no longer
10703 needed info, dfsnum, low, visited, on_sccstack, use_processed
10704 and range_info_anti_range_p members.
10705 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
10706 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
10708 (vn_valueize): New global.
10709 (vn_context_bb): Likewise.
10710 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
10711 VN_INFO_PTR_INFO): Remove.
10712 * tree-ssa-sccvn.c: ... (rewrite)
10713 (pass_fre::execute): For -O2+ initialize loops and run
10714 RPO VN in optimistic mode (iterating). For -O1 and -Og
10715 run RPO VN in non-optimistic mode.
10716 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
10717 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
10718 * doc/invoke.texi (sccvn-max-scc-size): Remove.
10719 (rpo-vn-max-loop-depth): Document.
10720 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
10721 when valuezing the VUSE signals we walked out of the region.
10722 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
10723 (phi_translate): Set VN context block to use for availability
10725 (compute_avail): Likewise.
10726 (pre_valueize): New function.
10727 (pass_pre::execute): Adjust to the RPO VN API.
10729 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
10730 (propagate_constants_for_unrolling): Remove.
10731 (tree_unroll_loops_completely): Perform value-numbering
10732 on the unrolled bodies loop parent.
10734 2018-08-27 Richard Biener <rguenther@suse.de>
10736 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
10737 for partial antic compute.
10739 2018-08-27 Jakub Jelinek <jakub@redhat.com>
10741 PR rtl-optimization/87065
10742 * combine.c (simplify_if_then_else): Formatting fix.
10743 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
10745 (known_cond): Don't return const_true_rtx for vector modes. Use
10746 CONST0_RTX instead of const0_rtx. Formatting fixes.
10748 2018-08-27 Martin Liska <mliska@suse.cz>
10750 PR gcov-profile/87069
10751 * gcov.c (process_file): Record files already processed
10752 and warn about a file being processed multiple times.
10754 2018-08-27 Martin Liska <mliska@suse.cz>
10757 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10758 Set default values for x_aarch64_*_string strings.
10759 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
10760 prefix. For -mabi do not print '=ABI' in help and use
10761 <option_value> format for -msve-vector-bits and -moverride
10764 2018-08-26 Jeff Law <law@redhat.com>
10766 * config/mips/frame-header-opt.c: Include "backend.h" rather than
10769 2018-08-26 Marek Polacek <polacek@redhat.com>
10771 PR c++/87029, Implement -Wredundant-move.
10772 * doc/invoke.texi: Document -Wredundant-move.
10774 2018-08-25 Martin Sebor <msebor@redhat.com>
10776 PR tree-optimization/87059
10777 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
10778 to the same type as the other.
10779 * fold-const.c (fold_binary_loc): Assert expectation.
10781 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
10783 * config/darwin.c (machopic_legitimize_pic_address): Clean up
10784 extraneous parentheses, dead code section and formatting.
10786 2018-08-24 David Malcolm <dmalcolm@redhat.com>
10789 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
10790 wide enough for jumps in the line-numbering to be visible.
10791 (layout::print_gap_in_line_numbering): New member function.
10792 (layout::calculate_line_spans): When using line numbering, merge
10793 line spans that are only 1 line apart.
10794 (diagnostic_show_locus): When printing line numbers, show gaps in
10795 line numbering directly, rather than printing headers.
10796 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
10797 line-numbering with multiple line spans.
10798 (selftest::test_fixit_insert_containing_newline_2): Add test of
10799 line-numbering, in which the spans are close enough to be merged.
10801 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10803 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
10804 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
10805 * tree-vrp.c (range_is_nonnull): Remove.
10806 (range_includes_zero_p): Accept value_range instead of min/max.
10807 (extract_range_from_binary_expr_1): Do not early bail on
10809 Use range_includes_zero_p instead of range_is_nonnull.
10810 (extract_range_from_unary_expr): Use range_includes_zero_p instead
10811 of range_is_nonnull.
10812 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
10813 special case VR_ANTI_RANGE.
10814 (vrp_finalize): Same.
10815 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
10816 instead of min/max.
10817 (range_is_nonnull): Remove.
10818 * vr-values.c (vrp_stmt_computes_nonzero): Use
10819 range_includes_zero_p instead of range_is_nonnull.
10820 (extract_range_basic): Pass value_range to range_includes_zero_p
10821 instead of range_is_nonnull.
10823 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
10825 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
10826 * emit-rtl.h (rtl_data): Remove return_bnd.
10827 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
10828 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
10829 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
10830 (POINTER_BOUNDS_MODE): Remove definition.
10831 (make_pointer_bounds_mode): Remove.
10832 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
10833 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
10834 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
10835 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
10836 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
10837 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
10838 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
10840 * config/i386/i386-modes.def (BND32, BND64): Remove.
10841 * config/i386/i386.c (dbx_register_map): Remove bound registers.
10842 (dbx64_register_map): Ditto.
10843 (svr4_dbx_register_map): Ditto.
10844 (indirect_thunk_bnd_needed): Remove.
10845 (indirect_thunks_bnd_used): Ditto.
10846 (indirect_return_bnd_needed): Ditto.
10847 (indirect_return_via_cx_bnd): Ditto.
10848 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
10849 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
10850 (output_indirect_thunk): Ditto. Remove need_prefix argument.
10851 (output_indirect_thunk_function): Remove handling of
10852 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
10853 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
10854 (ix86_save_reg): Remove handling of crtl->return_bnd.
10855 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
10856 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
10857 and UNSPEC_BNDLX_ADDR.
10858 (ix86_output_indirect_branch_via_reg): Remove handling of
10859 indirect_thunk_prefix_bnd.
10860 (ix86_output_indirect_branch_via_push): Ditto.
10861 (ix86_output_function_return): Ditto.
10862 (ix86_output_indirect_function_return): Ditto.
10863 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
10864 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
10865 (CALL_USED_REGISTERS): Ditto.
10866 (REG_ALLOC_ORDER): Update for removal of bound registers.
10867 (HI_REGISTER_NAMES): Ditto.
10868 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
10869 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
10870 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
10871 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
10872 (FIRST_PSEUDO_REG): Update.
10873 (BND): Remove mode iterator.
10874 * config/i386/predicates.md (bnd_mem_operator): Remove.
10876 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10878 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
10881 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10883 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
10884 the case in which the permute needs only a single element and
10885 repeats for every vector of the result. Extend that case to
10886 handle variable-length vectors.
10887 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
10889 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
10892 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
10893 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
10895 2018-08-24 Richard Biener <rguenther@suse.de>
10897 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
10898 bb_flags_allocated members.
10899 (auto_flag): New RAII class for allocating flags.
10900 (auto_edge_flag): New RAII class for allocating edge flags.
10901 (auto_bb_flag): New RAII class for allocating bb flags.
10902 * cfgloop.c (verify_loop_structure): Allocate temporary edge
10904 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
10905 in favor of temporarily allocated BB flag.
10906 * hsa-brig.c: Re-order includes.
10907 * hsa-dump.c: Likewise.
10908 * hsa-regalloc.c: Likewise.
10909 * print-rtl.c: Likewise.
10910 * profile-count.c: Likewise.
10912 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
10915 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
10918 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10921 * wide-int-range.cc (wide_int_range_div): Do not ignore result
10922 from wide_int_range_multiplicative_op.
10924 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10926 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
10929 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
10931 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
10934 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
10936 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
10937 full GENERIC expression used for replacement.
10939 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
10941 * tree-vrp.c (abs_extent_range): Remove.
10942 (extract_range_into_wide_ints): Pass wide ints by reference.
10943 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
10944 Pass wide ints by reference in all calls to
10945 extract_range_into_wide_ints.
10946 * wide-int-range.cc (wide_int_range_div): New.
10947 * wide-int-range.h (wide_int_range_div): New.
10948 (wide_int_range_includes_zero_p): New.
10949 (wide_int_range_zero_p): New.
10951 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
10953 * config/aarch64/aarch64.md (arches): New enum.
10954 (arch): New enum attr.
10955 (arch_enabled): New attr.
10956 (enabled): Now uses arch_enabled only.
10957 (simd, sve, fp16): Removed attribute.
10958 (fp): Attr now defined in terms of 'arch'.
10959 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
10960 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
10961 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
10962 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
10963 attributes into 'arch'.
10964 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
10965 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
10966 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
10967 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
10968 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
10969 'simd' attribute into 'arch'.
10970 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
10971 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
10972 Convert use of 'fp' attribute to 'arch'.
10973 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
10974 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
10976 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
10977 (different modes) Merge 'fp' and 'simd' into 'arch'.
10978 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
10979 'simd' into 'arch'.
10981 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
10983 PR rtl-optimization/87026
10984 * expmed.c (canonicalize_comparison): If we can no longer create
10985 pseudoregisters, don't.
10987 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
10990 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
10992 * config/arm/arm.c (speculation_barrier_libfunc): New static
10994 (arm_init_libfuncs): Initialize it.
10995 (arm_emit_speculation_barrier): New function.
10996 * config/arm/arm.md (speculation_barrier): Call
10997 arm_emit_speculation_barrier for architectures that do not have
10999 (speculation_barrier_insn): Only match on Armv7 or later.
11001 2018-08-23 Richard Biener <rguenther@suse.de>
11003 PR middle-end/87024
11004 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
11007 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
11009 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
11010 of single-vector TBLs.
11011 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
11012 one input is given.
11014 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
11017 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
11018 aarch64_evpc_tbl guard.
11020 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
11022 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
11025 2018-08-22 Martin Sebor <msebor@redhat.com>
11027 PR middle-end/87052
11028 * tree-pretty-print.c (pretty_print_string): Add argument.
11029 (dump_generic_node): Call to pretty_print_string with string size.
11031 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
11033 PR rtl-optimization/86771
11034 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
11035 of two SETs into those two SETs, one to be placed at i2, if that SETs
11036 destination is modified between i2 and i3.
11038 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11040 PR tree-optimization/86725
11041 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
11043 (vect_analyze_scalar_cycles_1): Check it.
11045 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11047 PR tree-optimization/86725
11048 * tree-vect-loop.c (vect_is_simple_reduction): When treating
11049 an outer loop phi as a double reduction, make sure that the
11050 single user of the phi result is an inner loop phi.
11052 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11054 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
11055 grouped stores with gaps to a strided group.
11057 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11059 * tree-vect-stmts.c (get_group_load_store_type)
11060 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
11061 first statement in a group.
11063 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11065 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
11066 the sequence used in gcc/gcc.c.
11068 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11071 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
11074 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11076 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
11077 Darwin10-specific unwinder-shim.
11078 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
11079 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
11080 New to cater for Darwin10 Rosetta.
11082 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11084 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
11085 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
11087 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11092 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
11093 (dwarf2out_switch_text_section): Generate a local label for the
11094 second function sub-section and apply it as the second FDE start
11096 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
11097 second sub-section start.
11099 2018-08-22 Richard Biener <rguenther@suse.de>
11101 PR tree-optimization/86988
11102 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
11104 2018-08-22 Richard Biener <rguenther@suse.de>
11106 PR tree-optimization/86945
11107 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
11109 2018-08-22 Alexandre Oliva <oliva@adacore.com>
11111 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
11112 a comment about how uses of r2 for .sdata2 come about.
11114 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
11116 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
11118 2018-08-21 Marek Polacek <polacek@redhat.com>
11120 PR c++/86981, Implement -Wpessimizing-move.
11121 * doc/invoke.texi: Document -Wpessimizing-move.
11123 2018-08-21 Jan Hubicka <jh@suse.cz>
11125 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
11126 * tree.h (is_redundant_typedef): Remove.
11127 * dwarf2out.c (is_redundant_typedef): Turn into static function.
11129 2018-08-21 Jan Hubicka <jh@suse.cz>
11131 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
11134 2018-08-21 Tamar Christina <tamar.christina@arm.com>
11136 * expmed.c (extract_low_bits): Reject invalid subregs early.
11138 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
11140 PR middle-end/86121
11141 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
11144 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11146 * config/vxworks.h: Guard vxworks_asm_out_constructor and
11147 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
11148 * config/vxworks.c: Likewise.
11150 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11152 * config/vxworks.c: Set targetm.have_ctors_dtors
11153 if HAVE_INITFINI_ARRAY_SUPPORT.
11154 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
11155 if HAVE_INITFINI_ARRAY_SUPPORT.
11157 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11159 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
11160 default search path for VxWorks < 7.
11162 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
11164 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
11165 (get_format_string): Refer to c_getstr.
11167 2018-08-21 Tom de Vries <tdevries@suse.de>
11169 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
11170 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
11171 (debuginfo_early_stop): Declare.
11172 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
11173 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
11174 (debuginfo_early_stop): New function.
11175 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
11176 and debuginfo_early_stop.
11177 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
11178 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
11179 (general_init): Call debuginfo_early_init.
11180 (finalize): Call debuginfo_fini.
11181 (do_compile): Call debuginfo_init.
11182 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
11183 -fdump-early-debug.
11184 (@item -fdump-debug, @item -fdump-earlydebug): Add.
11186 2018-08-21 Tom de Vries <tdevries@suse.de>
11188 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
11189 flag_dump_noaddr and flag_dump_unnumbered.
11191 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
11193 * wide-int-range.cc (wide_int_range_abs): New.
11194 (wide_int_range_order_set): Rename from wide_int_range_min_max.
11195 * wide-int-range.h (wide_int_range_abs): New.
11196 (wide_int_range_min_max): New.
11197 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
11198 case to call wide_int_range_abs.
11199 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
11200 (extract_range_from_abs_expr): Delete.
11202 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
11205 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
11206 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
11209 2018-08-20 Nathan Sidwell <nathan@acm.org>
11210 Jeff Law <law@redhat.com>
11212 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
11213 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
11215 2018-08-20 David Malcolm <dmalcolm@redhat.com>
11218 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
11219 (decl_attributes): Likewise.
11220 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
11222 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11223 * diagnostic-core.h (class auto_diagnostic_group): New class.
11224 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
11225 (diagnostic_report_diagnostic): Handle the first diagnostics within
11227 (emit_diagnostic): Add auto_diagnostic_group instance.
11228 (inform): Likewise.
11229 (inform_n): Likewise.
11230 (warning): Likewise.
11231 (warning_at): Likewise.
11232 (warning_n): Likewise.
11233 (pedwarn): Likewise.
11234 (permerror): Likewise.
11236 (error_n): Likewise.
11237 (error_at): Likewise.
11239 (fatal_error): Likewise.
11240 (internal_error): Likewise.
11241 (internal_error_no_backtrace): Likewise.
11242 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
11243 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
11244 * diagnostic.h (struct diagnostic_context): Add fields
11245 "diagnostic_group_nesting_depth",
11246 "diagnostic_group_emission_count", "begin_group_cb",
11248 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
11249 Add auto_diagnostic_group instance(s).
11250 (find_explicit_erroneous_behavior): Likewise.
11251 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
11252 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
11253 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
11254 (gimplify_va_arg_expr): Likewise.
11255 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
11256 (HSA_SORRY_AT): Likewise.
11257 * ipa-devirt.c (compare_virtual_tables): Likewise.
11258 (warn_odr): Likewise.
11259 * multiple_target.c (expand_target_clones): Likewise.
11260 * opts-common.c (cmdline_handle_error): Likewise.
11261 * reginfo.c (globalize_reg): Likewise.
11262 * substring-locations.c (format_warning_n_va): Likewise.
11263 * tree-inline.c (expand_call_inline): Likewise.
11264 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
11265 * tree-ssa-loop-niter.c
11266 (do_warn_aggressive_loop_optimizations): Likewise.
11267 * tree-ssa-uninit.c (warn_uninit): Likewise.
11268 * tree.c (warn_deprecated_use): Likewise.
11270 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
11273 * config/i386/i386.md (eh_return): Always update EH return
11274 address in word_mode.
11276 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
11278 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
11279 TARGET_SPLIT_COMPLEX_ARG is defined.
11281 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11283 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
11285 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11288 * expr.c (expand_assignment): Assert that bitpos is positive.
11289 (store_field): Likewise
11290 (expand_expr_real_1): Make sure that bitpos is positive.
11291 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
11294 2018-08-20 Nathan Sidwell <nathan@acm.org>
11296 * Makefile.in (CPP_ID_DATA_H): Delete.
11297 (CPP_INTERNAL_H): Don't add it.
11298 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
11299 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
11301 2018-08-20 Richard Biener <rguenther@suse.de>
11303 PR tree-optimization/78655
11304 * tree-vrp.c (extract_range_from_binary_expr_1): Make
11305 pointer + offset nonnull if either operand is nonnull work.
11307 2018-08-20 Tom de Vries <tdevries@suse.de>
11309 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
11310 unless the referenced die describes the added property using
11311 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
11312 Otherwise, add a DW_AT_location to the referenced die.
11314 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
11317 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
11318 register_operand when calling ix86_set_reg_reg_cost.
11319 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
11320 Set *total to 0 for operands that satisfy x86_64_immediate_operand
11321 predicate and to 1 otherwise.
11323 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
11325 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
11326 emit a diagnostic that it is not supported and reset the option.
11327 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
11328 supported and consume the option. (ASM_FINAL_SPEC): New.
11330 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
11332 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
11335 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
11337 C-SKY port: Documentation
11339 * doc/extend.texi (C-SKY Function Attributes): New section.
11340 * doc/invoke.texi (Option Summary): Add C-SKY options.
11341 (C-SKY Options): New section.
11342 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
11344 2018-08-17 Jojo <jijie_rong@c-sky.com>
11345 Huibin Wang <huibin_wang@c-sky.com>
11346 Sandra Loosemore <sandra@codesourcery.com>
11347 Chung-Lin Tang <cltang@codesourcery.com>
11349 C-SKY port: Backend implementation
11351 * config/csky/*: New.
11352 * common/config/csky/*: New.
11354 2018-08-17 Jojo <jijie_rong@c-sky.com>
11355 Huibin Wang <huibin_wang@c-sky.com>
11356 Sandra Loosemore <sandra@codesourcery.com>
11357 Chung-Lin Tang <cltang@codesourcery.com>
11358 Andrew Jenner <andrew@codesourcery.com>
11360 C-SKY port: Configury
11362 * config.gcc (csky-*-*): New.
11363 * configure.ac: Add csky to targets for dwarf2 debug_line support.
11364 * configure: Regenerated.
11366 2018-08-17 David Malcolm <dmalcolm@redhat.com>
11368 * dump-context.h: Include "dumpfile.h".
11369 (dump_context::dump_printf_va): Convert final param from va_list
11370 to va_list *. Convert from ATTRIBUTE_PRINTF to
11371 ATTRIBUTE_GCC_DUMP_PRINTF.
11372 (dump_context::dump_printf_loc_va): Likewise.
11373 * dumpfile.c: Include "stringpool.h".
11374 (make_item_for_dump_printf_va): Delete.
11375 (make_item_for_dump_printf): Delete.
11376 (class dump_pretty_printer): New class.
11377 (dump_pretty_printer::dump_pretty_printer): New ctor.
11378 (dump_pretty_printer::emit_items): New member function.
11379 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
11381 (dump_pretty_printer::emit_item): New member function.
11382 (dump_pretty_printer::stash_item): New member function.
11383 (dump_pretty_printer::format_decoder_cb): New member function.
11384 (dump_pretty_printer::decode_format): New member function.
11385 (dump_context::dump_printf_va): Reimplement in terms of
11386 dump_pretty_printer.
11387 (dump_context::dump_printf_loc_va): Convert final param from va_list
11389 (dump_context::begin_scope): Reimplement call to
11390 make_item_for_dump_printf.
11391 (dump_printf): Update for change to dump_printf_va.
11392 (dump_printf_loc): Likewise.
11393 (selftest::test_capture_of_dump_calls): Convert "stmt" from
11394 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
11395 with %T, %E, and %G.
11396 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
11397 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
11398 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
11399 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
11400 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
11401 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
11402 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
11403 within a dump_printf_loc call to "%wu".
11404 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
11405 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
11406 missing space after "=".
11407 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
11408 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
11409 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
11410 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
11411 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
11412 duplicate "vectorized" from message.
11414 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
11416 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
11417 polyNxK_t element's TYPE_STRING_FLAG.
11419 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
11421 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
11422 (they were unnamed before). Fix comments.
11424 2018-08-17 Nathan Sidwell <nathan@acm.org>
11426 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
11428 2018-08-17 Richard Biener <rguenther@suse.de>
11430 PR tree-optimization/86841
11431 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
11433 2018-08-17 Martin Liska <mliska@suse.cz>
11435 * common.opt: Remove Warn, Init and Report for options with
11436 Ignore/Deprecated flag. Warning is done automatically for
11438 * config/i386/i386.opt: Likewise.
11439 * config/ia64/ia64.opt: Likewise.
11440 * config/rs6000/rs6000.opt: Likewise.
11441 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
11442 Remove usage of flag_check_pointer_bounds.
11443 * lto-wrapper.c (merge_and_complain): Do not handle
11444 OPT_fcheck_pointer_bounds.
11445 (append_compiler_options): Likewise.
11446 * opt-functions.awk: Do not handle Deprecated.
11447 * optc-gen.awk: Check that Var, Report and Init are not
11448 used for an option with Ignore/Deprecated flag.
11449 * opts-common.c (decode_cmdline_option): Do not report
11451 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
11453 * opts.h (struct cl_option): Remove cl_deprecated flag.
11454 (CL_ERR_DEPRECATED): Remove error enum value.
11456 2018-08-17 Richard Biener <rguenther@suse.de>
11458 PR middle-end/86505
11459 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
11460 across a va-arg-pack using call adjust its return value accordingly.
11462 2018-08-16 Martin Sebor <msebor@redhat.com>
11464 PR tree-optimization/86853
11465 * gimple-ssa-sprintf.c (struct format_result): Rename member.
11466 (struct fmtresult): Add member and initialize it in ctors.
11467 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
11468 (format_string): Handle %S the same as %ls. Set MAYFAIL.
11469 (format_directive): Set POSUNDER4K when MAYFAIL is set.
11470 (parse_directive): Handle %C same as %c.
11471 (sprintf_dom_walker::compute_format_length): Adjust.
11472 (is_call_safe): Adjust.
11474 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
11476 * builtins.c (c_strlen): Add new parameter eltsize. Use it
11477 for determining how to count the elements.
11478 * builtins.h (c_strlen): Adjust prototype.
11479 * expr.c (string_constant): Add new parameter mem_size.
11480 Set *mem_size appropriately.
11481 * expr.h (string_constant): Adjust protoype.
11482 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
11483 * gimple-fold.h (get_range_strlen): Adjust prototype.
11484 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
11485 (format_string): Call get_string_length with eltsize.
11487 2018-08-16 David Malcolm <dmalcolm@redhat.com>
11489 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
11490 to emit the span, rather than setting it as the prefix.
11492 2018-08-16 David Malcolm <dmalcolm@redhat.com>
11494 * diagnostic-show-locus.c (layout::start_annotation_line): Add
11495 "margin_char" parameter, defaulting to space. Use it in place
11496 of pp_space for the initial part of the margin.
11497 (layout::print_leading_fixits): Use '+' when filling the margin
11498 of line-insertion fix-it hints.
11500 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11502 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
11505 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11507 * config/rs6000/altivec.md: Don't set length attribute to the default
11509 * config/rs6000/darwin.md: Ditto.
11510 * config/rs6000/dfp.md: Ditto.
11511 * config/rs6000/htm.md: Ditto.
11512 * config/rs6000/rs6000.md: Ditto.
11513 * config/rs6000/sync.md: Ditto.
11514 * config/rs6000/vsx.md: Ditto.
11516 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11518 * config/rs6000/altivec.md: Don't set length attribute to the default
11519 value, for branch instructions.
11520 * config/rs6000/darwin.md: Ditto.
11521 * config/rs6000/rs6000.md: Ditto.
11523 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11525 * config/rs6000/rs6000.md (length): Always define as const_int 4.
11526 (unnamed conditional branch define_insn): Set length to 4 or 8
11527 depending on offset.
11528 (<bd>_<mode>): Similar, for alternative 0.
11529 (<bd>tf_<mode>): Ditto.
11531 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11533 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
11535 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
11537 * doc/rtl.texi: Replace old RTX class names with new names.
11540 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
11542 * expmed.h (canonicalize_comparison): New declaration.
11543 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
11544 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
11545 * optabs.c (prepare_cmp_insn): Likewise.
11546 * rtl.h (unsigned_condition_p): New function which checks if a
11547 comparison operator is unsigned.
11549 2018-08-16 Nathan Sidwell <nathan@acm.org>
11551 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
11552 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
11554 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11557 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
11558 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
11559 (mov<mov>): ..this and enable unconditionally.
11561 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11563 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
11565 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
11567 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
11568 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
11569 "Common" with "Target".
11571 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
11573 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
11574 * doc/invoke.texi (mmitigate-rop): Remove.
11575 * config/i386/i386.c: Do not include "regrename.h".
11576 (ix86_rop_should_change_byte_p, reg_encoded_number)
11577 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
11579 (ix86_reorg): Remove call to ix86_mitigate_rop.
11580 * config/i386/i386.md (attr "modrm_class"): Remove.
11581 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
11582 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
11583 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
11585 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
11587 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
11588 allow folding of mergeh() and mergel() for the float and double types.
11589 (fold_mergehl_helper): Rework to handle building a permute tree
11592 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
11594 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
11597 2018-08-15 David Malcolm <dmalcolm@redhat.com>
11599 * common.opt (fdiagnostics-show-labels): New option.
11600 * diagnostic-show-locus.c (class layout_range): Add field
11602 (class layout): Add field "m_show_labels_p".
11603 (layout_range::layout_range): Add param "label" and use it to
11604 initialize m_label.
11605 (make_range): Pass in NULL for new "label" param of layout_range's
11607 (layout::layout): Initialize m_show_labels_p.
11608 (layout::maybe_add_location_range): Pass in loc_range->m_label
11609 when constructing layout_range instances.
11610 (struct line_label): New struct.
11611 (layout::print_any_labels): New member function.
11612 (layout::print_line): Call it if label-printing is enabled.
11613 (selftest::test_one_liner_labels): New test.
11614 (selftest::test_diagnostic_show_locus_one_liner): Call it.
11615 * diagnostic.c (diagnostic_initialize): Initialize
11616 context->show_labels_p.
11617 * diagnostic.h (struct diagnostic_context): Add field
11619 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11620 -fno-diagnostics-show-labels.
11621 * dwarf2out.c (gen_producer_string): Add
11622 OPT_fdiagnostics_show_labels to the ignored options.
11623 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
11625 (gcc_rich_location::maybe_add_expr): Likewise.
11626 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
11627 label" param, defaulting to NULL.
11628 (gcc_rich_location::add_expr): Add "label" param.
11629 (gcc_rich_location::maybe_add_expr): Likewise.
11630 (class text_range_label): New class.
11631 (class range_label_for_type_mismatch): New class.
11632 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
11633 of format_warning_va.
11634 (fmtwarn_n): Likewise for new params of format_warning_n_va.
11635 * lto-wrapper.c (merge_and_complain): Add
11636 OPT_fdiagnostics_show_labels to the "pick one setting" options.
11637 (append_compiler_options): Likewise to the dropped options.
11638 (append_diag_options): Likewise to the passed-on options.
11639 * opts.c (common_handle_option): Handle the new option.
11640 * selftest-diagnostic.c
11641 (test_diagnostic_context::test_diagnostic_context): Enable
11643 * substring-locations.c: Include "gcc-rich-location.h".
11644 (format_warning_n_va): Add "fmt_label" and "param_label" params
11645 and use them as appropriate.
11646 (format_warning_va): Add "fmt_label" and "param_label" params,
11647 passing them on to format_warning_n_va.
11648 (format_warning_at_substring): Likewise.
11649 (format_warning_at_substring_n): Likewise.
11650 * substring-locations.h (format_warning_va): Add "fmt_label" and
11651 "param_label" params.
11652 (format_warning_n_va): Likewise.
11653 (format_warning_at_substring): Likewise.
11654 (format_warning_at_substring_n): Likewise.
11655 * toplev.c (general_init): Initialize global_dc->show_labels_p.
11657 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
11660 * builtins.c (expand_builtin_memcmp): Do not expand the call
11661 when overflow is detected.
11663 2018-08-15 Martin Sebor <msebor@redhat.com>
11665 PR tree-optimization/71625
11666 * config/aarch64/aarch64-builtins.c
11667 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
11669 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
11671 * config/s390/s390.c (s390_reorg): Remove loop.
11673 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11676 (darwin_function_switched_text_sections): Delete.
11677 * gcc/config/darwin.h
11678 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
11680 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11683 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
11684 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
11685 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
11687 2018-08-15 Martin Liska <mliska@suse.cz>
11689 PR tree-optimization/86925
11690 * predict.c (expr_expected_value_1): When taking
11691 later predictor, assign also probability.
11692 Use fold_build2_initializer_loc in order to fold
11693 the expression in -frounding-math.
11695 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
11697 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
11699 (expand_vec_perm_1): Try the new method.
11701 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
11704 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
11705 Check whether lra_live_max_point is 0 before dividing.
11707 2018-08-14 Martin Sebor <msebor@redhat.com>
11709 PR tree-optimization/86650
11710 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
11711 (vrp_prop::check_mem_ref): Same.
11713 2018-08-13 Liu Hao <lh_mouse@126.com>
11715 * pretty-print.c (eat_esc_sequence): Swap the foreground and
11716 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
11717 and clear it thereafter, as it only works for DBCS.
11719 2018-08-13 Liu Hao <lh_mouse@126.com>
11721 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
11722 handle returned by _get_osf_handle().
11724 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11726 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
11727 for folding vec_perm.
11729 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11731 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
11732 Add support for gimple-folding of vec_pack() and vec_unpack()
11735 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11737 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
11738 vec_xst variants to the list.
11739 (rs6000_gimple_fold_builtin): Add support for folding unaligned
11740 vector loads and stores.
11742 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
11744 * config.gcc (rs6000-ibm-aix4.x): Delete.
11745 (rs6000-ibm-aix5.1): Delete.
11746 (rs6000-ibm-aix5.2): Delete.
11747 (rs6000-ibm-aix5.3): Delete.
11748 * config/rs6000/aix43.h: Delete.
11749 * config/rs6000/aix51.h: Delete.
11750 * config/rs6000/aix52.h: Delete.
11751 * config/rs6000/t-aix43: Delete.
11753 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
11755 * config/s390/s390.c (s390_decompose_constant_pool_ref):
11757 (s390_decompose_address): Factor out constant pool ref
11760 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11762 * config/nds32/nds32-predicates.c
11763 (nds32_can_use_bclr_p): Change return type as bool.
11764 (nds32_can_use_bset_p): Ditto.
11765 (nds32_can_use_btgl_p): Ditto.
11766 (nds32_can_use_bitci_p): Ditto.
11767 * config/nds32/nds32-protos.h
11768 (nds32_can_use_bclr_p): Change declaration.
11769 (nds32_can_use_bset_p): Ditto.
11770 (nds32_can_use_btgl_p): Ditto.
11771 (nds32_can_use_bitci_p): Ditto.
11773 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11775 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
11776 Support -msched-prolog-epilog option.
11777 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
11779 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11781 * common/config/nds32/nds32-common.c
11782 (nds32_option_optimization_table): Enalbe -malways-align.
11784 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11786 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
11788 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
11789 OPT_misr_secure_ case.
11790 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
11791 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
11792 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
11794 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
11795 (nds32_isr_info): New field security_level.
11796 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
11797 * config/nds32/nds32.md (return_internal): Consider critical attribute.
11798 * config/nds32/nds32.opt (misr-secure): New option.
11799 * config/nds32/nds32_init.inc: New file.
11800 * config/nds32/nds32_isr.h: New file.
11802 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
11804 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
11805 Update comment for atomic instructions.
11806 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
11807 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
11809 (atomic_loaddi): Revise fence expansion to only emit fence prior to
11810 load for __ATOMIC_SEQ_CST model.
11811 (atomic_loaddi_1): Remove float register target.
11812 (atomic_storedi): Handle CONST_INT values.
11813 (atomic_storedi_1): Remove float register source. Add special case
11815 (memory_barrier): New expander and insn.
11817 2018-08-11 Jakub Jelinek <jakub@redhat.com>
11819 PR tree-optimization/86835
11820 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
11821 new_stmt after def_gsi, make sure to insert new_square_stmt after
11822 that stmt, not 2 stmts before it.
11824 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
11827 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
11830 2018-08-10 Martin Liska <mliska@suse.cz>
11833 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
11835 * builtins.c (expand_builtin_expect_with_probability):
11837 (expand_builtin_expect_with_probability): New function.
11838 (build_builtin_expect_predicate): Add new argumnet probability
11839 for BUILT_IN_EXPECT_WITH_PROBABILITY.
11840 (fold_builtin_expect):
11843 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
11844 * builtins.h (fold_builtin_expect): Set new argument.
11845 * doc/extend.texi: Document __builtin_expect_with_probability.
11846 * doc/invoke.texi: Likewise.
11847 * gimple-fold.c (gimple_fold_call): Pass new argument.
11848 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
11849 also BUILT_IN_EXPECT_WITH_PROBABILITY.
11850 * predict.c (get_predictor_value): New function.
11851 (expr_expected_value): Add new argument probability. Assume
11852 that predictor and probability are always non-null.
11853 (expr_expected_value_1): Likewise. For __builtin_expect and
11854 __builtin_expect_with_probability set probability. Handle
11855 combination in binary expressions.
11856 (tree_predict_by_opcode): Simplify code by simply calling
11857 get_predictor_value.
11858 (pass_strip_predict_hints::execute): Add handling of
11859 BUILT_IN_EXPECT_WITH_PROBABILITY.
11860 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
11862 * tree.h (DECL_BUILT_IN_P): New function.
11864 2018-08-10 Martin Liska <mliska@suse.cz>
11866 PR tree-optimization/85799
11867 * passes.def: Add argument for pass_strip_predict_hints.
11868 * predict.c (class pass_strip_predict_hints): Add new argument
11870 (strip_predictor_early): New function.
11871 (pass_strip_predict_hints::execute): Call the function to
11873 (strip_predict_hints): New function.
11874 * predict.def: Fix comment.
11876 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
11878 * Makefile.in: Clarify which tm.texi to copy over to assert the
11879 right to grant a GFDL license for all.
11881 2018-08-09 Jeff Law <law@redhat.com>
11883 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
11886 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
11888 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
11891 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11893 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
11894 reductions for variable-length vectors.
11896 2018-08-09 David Malcolm <dmalcolm@redhat.com>
11899 * common.opt (fdiagnostics-show-line-numbers): New option.
11900 * diagnostic-show-locus.c (class layout): Add fields
11901 "m_show_line_numbers_p" and "m_linenum_width";
11902 (num_digits): New function.
11903 (test_num_digits): New function.
11904 (layout::layout): Initialize new fields. Update m_x_offset
11905 logic to handle any left margin.
11906 (layout::print_source_line): Print line number when requested.
11907 (layout::start_annotation_line): New member function.
11908 (layout::print_annotation_line): Call it.
11909 (layout::print_leading_fixits): Likewise.
11910 (layout::print_trailing_fixits): Likewise. Update calls to
11911 move_to_column for new parameter.
11912 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
11913 it to potentially call start_annotation_line.
11914 (layout::show_ruler): Call start_annotation_line.
11915 (selftest::test_line_numbers_multiline_range): New selftest.
11916 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
11917 and selftest::test_line_numbers_multiline_range.
11918 * diagnostic.c (diagnostic_initialize): Initialize
11919 show_line_numbers_p.
11920 * diagnostic.h (struct diagnostic_context): Add field
11921 "show_line_numbers_p".
11922 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11923 -fno-diagnostics-show-line-numbers.
11924 * dwarf2out.c (gen_producer_string): Add
11925 OPT_fdiagnostics_show_line_numbers to the ignored options.
11926 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
11927 one setting" options.
11928 (append_compiler_options): Likewise to the dropped options.
11929 (append_diag_options): Likewise to the passed-on options.
11930 * opts.c (common_handle_option): Handle the new option.
11931 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
11933 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
11935 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
11936 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
11937 third argument of type "const signed char" to descriptions of
11938 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
11939 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
11940 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
11941 __builtin_bcdsub_ov functions.
11943 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11945 PR tree-optimization/86858
11946 * tree-vect-loop.c (vect_is_simple_reduction): Restore
11947 flow_bb_inside_loop_p calls.
11949 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11951 PR tree-optimization/86871
11952 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
11953 instead of gimple_assign_lhs.
11955 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
11958 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
11959 register constraint to operand 0.
11960 (add<mode>3_carryinC): Likewise.
11961 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
11963 2018-08-09 Martin Liska <mliska@suse.cz>
11966 * common.opt: Remove extra line.
11968 2018-08-09 Martin Liska <mliska@suse.cz>
11970 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
11971 at the end of a line, make first letter capital and end up
11972 a sentence with a dot.
11973 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
11974 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
11975 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
11976 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
11977 (PARAM_MAX_ISL_OPERATIONS): Likewise.
11978 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
11979 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
11980 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
11981 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
11982 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
11983 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
11984 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
11985 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
11986 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
11987 (PARAM_TREE_REASSOC_WIDTH): Likewise.
11988 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
11989 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
11990 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
11992 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
11995 * config/s390/s390.c (s390_option_override_internal): Reduce the
11996 stack-clash-protection-probe-interval param if it would be too big
11999 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
12002 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
12003 * config/m68k/m68k.c (handle_move_double): Don't call
12004 m68k_final_prescan_insn.
12005 (m68k_adjust_decorated_operand): Renamed from
12006 m68k_final_prescan_insn, remove first and third operand and
12008 (print_operand): Call it.
12009 (print_operand_address): Call it.
12011 2018-08-08 Nathan Sidwell <nathan@acm.org>
12013 * diagnostic.c (diagnostic_report_current_module): Use
12014 linemap_included_from & linemap_included_from_linemap.
12016 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
12018 * config/aarch64/aarch64-cores.def: Add phecda core.
12019 * config/aarch64/aarch64-tune.md: Regenerate.
12020 * doc/invoke.texi: Add phecda core.
12022 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
12025 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
12027 * config/s390/s390.md ("movti"): Add more alternatives for
12028 constant to GPR copies.
12030 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
12032 * config/s390/s390.c: Fix whitespace damage throughout the file.
12033 * config/s390/s390.h: Likewise.
12034 * config/s390/tpf.h: Likewise.
12036 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
12038 * config/s390/s390.c (s390_loadrelative_operand_p):
12039 Remove TARGET_CPU_ZARCH usages.
12040 (s390_rtx_costs): Likewise.
12041 (s390_legitimate_constant_p): Likewise.
12042 (s390_cannot_force_const_mem): Likewise.
12043 (legitimate_reload_constant_p): Likewise.
12044 (s390_preferred_reload_class): Likewise.
12045 (legitimize_pic_address): Likewise.
12046 (legitimize_tls_address): Likewise.
12047 (s390_split_branches): Removed.
12048 (s390_add_execute): Removed.
12049 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
12050 (s390_mainpool_start): Likewise.
12051 (s390_mainpool_finish): Likewise.
12052 (s390_mainpool_cancel): Removed.
12053 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
12054 (s390_chunkify_cancel): Likewise.
12055 (s390_return_addr_rtx): Likewise.
12056 (s390_register_info): Remove split_branches_pending_p uages.
12057 (s390_optimize_register_info): Likewise.
12058 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
12059 split_branches_pending_p usages.
12060 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
12061 (s390_load_got): Likewise.
12062 (s390_expand_split_stack_prologue): Likewise.
12063 (output_asm_nops): Likewise.
12064 (s390_function_profiler): Likewise.
12065 (s390_emit_call): Likewise.
12066 (s390_conditional_register_usage): Likewise.
12067 (s390_optimize_prologue): Likewise.
12068 (s390_reorg): Remove TARGET_CPU_ZARCH and
12069 split_branches_pending_p usages.
12070 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
12072 (s390_output_indirect_thunk_function): Likewise.
12073 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
12074 (TARGET_CPU_ZARCH_P): Removed.
12075 (struct machine_function): Remove split_branches_pending_p.
12076 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
12078 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
12080 * common/config/s390/s390-common.c (processor_flags_table):
12082 * config.gcc: Remove with_arch/with_tune support.
12083 * config/s390/2064.md: Remove cpu attribute comparisons.
12084 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12086 * config/s390/linux.h (ASM_SPEC):
12087 Remove -march support.
12088 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
12089 Use a table to get an arch level.
12090 * config/s390/s390-opts.h (enum processor_type):
12091 Remove enum values.
12092 * config/s390/s390.c
12093 (processor_table): Remove entries, add arch_level values.
12094 (s390_issue_rate): Remove cases.
12095 (s390_option_override): Adjust
12096 s390_option_override_internal() call.
12097 (s390_option_override_internal): Remove deprecation warning.
12098 (s390_valid_target_attribute_tree): Adjust
12099 s390_option_override_internal() call.
12100 * config/s390/s390.h (struct s390_processor):
12101 Share with s390-c.c, add arch_level field.
12102 * config/s390/s390.md:
12103 Remove occurrences in cpu attribute.
12104 * config/s390/s390.opt: Remove -march/-mtune support.
12105 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
12106 * doc/invoke.texi: Remove deprecation warning.
12108 2018-08-08 Luis Machado <luis.machado@linaro.org>
12110 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
12112 (qdf24xx_tunings): Set vector cost structure to
12113 qdf24xx_vector_cost.
12115 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
12116 <register_sextend>: Set to 3.
12118 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
12121 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
12122 * config/aarch64/aarch64-simd.md
12123 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
12124 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
12125 (aarch64_frecpx<mode>): ...this new pattern.
12126 * config/aarch64/aarch64-simd-builtins.def: Remove comment
12127 about aarch64_frecp<FRECP:frecp_suffix><mode>.
12129 2018-08-07 Martin Liska <mliska@suse.cz>
12131 PR middle-end/83023
12132 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
12133 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
12134 * predict.def (PRED_MALLOC_NONNULL): New predictor.
12135 * doc/extend.texi: Document that malloc attribute adds
12138 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
12141 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12142 Define to speculation_safe_value_not_needed.
12144 2018-08-06 Jeff Law <law@redhat.com>
12146 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
12147 the vr_values instance to cprop_into_stmt.
12148 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
12149 (cprop_operand): Also query EVRP to determine if OP is a constant.
12151 2018-08-06 Nathan Sidwell <nathan@acm.org>
12153 * diagnostic.c (diagnostic_report_current_module): Reroll
12154 included-at loop. Translate text.
12156 2018-08-06 David Malcolm <dmalcolm@redhat.com>
12158 * function-tests.c (selftest::test_expansion_to_rtl): Call
12159 free_after_compilation.
12161 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12163 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
12165 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
12167 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
12168 loops with memory block operations from getting unrolled.
12170 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
12173 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12174 Define to speculation_safe_value_not_needed.
12176 2018-08-06 Jeff Law <law@redhat.com>
12178 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
12181 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12184 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
12185 with all enabled __intN types.
12187 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
12189 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12191 * alias.c (record_set): Check for clobber high.
12192 * cfgexpand.c (expand_gimple_stmt): Likewise.
12193 * combine-stack-adj.c (single_set_for_csa): Likewise.
12194 * combine.c (find_single_use_1): Likewise.
12195 (set_nonzero_bits_and_sign_copies): Likewise.
12196 (get_combine_src_dest): Likewise.
12197 (is_parallel_of_n_reg_sets): Likewise.
12198 (try_combine): Likewise.
12199 (record_dead_and_set_regs_1): Likewise.
12200 (reg_dead_at_p_1): Likewise.
12201 (reg_dead_at_p): Likewise.
12202 * dce.c (deletable_insn_p): Likewise.
12203 (mark_nonreg_stores_1): Likewise.
12204 (mark_nonreg_stores_2): Likewise.
12205 * df-scan.c (df_find_hard_reg_defs): Likewise.
12206 (df_uses_record): Likewise.
12207 (df_get_call_refs): Likewise.
12208 * dwarf2out.c (mem_loc_descriptor): Likewise.
12209 * haifa-sched.c (haifa_classify_rtx): Likewise.
12210 * ira-build.c (create_insn_allocnos): Likewise.
12211 * ira-costs.c (scan_one_insn): Likewise.
12212 * ira.c (equiv_init_movable_p): Likewise.
12213 (rtx_moveable_p): Likewise.
12214 (interesting_dest_for_shprep): Likewise.
12215 * jump.c (mark_jump_label_1): Likewise.
12216 * postreload-gcse.c (record_opr_changes): Likewise.
12217 * postreload.c (reload_cse_simplify): Likewise.
12218 (struct reg_use): Add source expr.
12219 (reload_combine): Check for clobber high.
12220 (reload_combine_note_use): Likewise.
12221 (reload_cse_move2add): Likewise.
12222 (move2add_note_store): Likewise.
12223 * print-rtl.c (print_pattern): Likewise.
12224 * recog.c (decode_asm_operands): Likewise.
12225 (store_data_bypass_p): Likewise.
12226 (if_test_bypass_p): Likewise.
12227 * regcprop.c (kill_clobbered_value): Likewise.
12228 (kill_set_value): Likewise.
12229 * reginfo.c (reg_scan_mark_refs): Likewise.
12230 * reload1.c (maybe_fix_stack_asms): Likewise.
12231 (eliminate_regs_1): Likewise.
12232 (elimination_effects): Likewise.
12233 (mark_not_eliminable): Likewise.
12234 (scan_paradoxical_subregs): Likewise.
12235 (forget_old_reloads_1): Likewise.
12236 * reorg.c (find_end_label): Likewise.
12237 (try_merge_delay_insns): Likewise.
12238 (redundant_insn): Likewise.
12239 (own_thread_p): Likewise.
12240 (fill_simple_delay_slots): Likewise.
12241 (fill_slots_from_thread): Likewise.
12242 (dbr_schedule): Likewise.
12243 * resource.c (update_live_status): Likewise.
12244 (mark_referenced_resources): Likewise.
12245 (mark_set_resources): Likewise.
12246 * rtl.c (copy_rtx): Likewise.
12247 * rtlanal.c (reg_referenced_p): Likewise.
12248 (single_set_2): Likewise.
12249 (noop_move_p): Likewise.
12250 (note_stores): Likewise.
12251 * sched-deps.c (sched_analyze_reg): Likewise.
12252 (sched_analyze_insn): Likewise.
12254 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12256 * cse.c (invalidate_reg): New function extracted from...
12257 (invalidate): ...here.
12258 (canonicalize_insn): Check for clobber high.
12259 (invalidate_from_clobbers): invalidate clobber highs.
12260 (invalidate_from_sets_and_clobbers): Likewise.
12261 (count_reg_usage): Check for clobber high.
12262 (insn_live_p): Likewise.
12263 * cselib.c (cselib_expand_value_rtx_1):Likewise.
12264 (cselib_invalidate_regno): Check for clobber in setter.
12265 (cselib_invalidate_rtx): Pass through setter.
12266 (cselib_invalidate_rtx_note_stores):
12267 (cselib_process_insn): Check for clobber high.
12268 * cselib.h (cselib_invalidate_rtx): Add operand.
12270 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12272 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
12273 (mark_not_eliminable): Likewise.
12274 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
12275 * lra-lives.c (process_bb_lives): Check for clobber high.
12276 * lra.c (new_insn_reg): Remember clobber highs.
12277 (collect_non_operand_hard_regs): Check for clobber high.
12278 (lra_set_insn_recog_data): Likewise.
12279 (add_regs_to_insn_regno_info): Likewise.
12280 (lra_update_insn_regno_info): Likewise.
12282 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12284 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
12285 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
12287 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12289 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
12290 (copy_insn_1): Likewise.
12291 (gen_hard_reg_clobber_high): New gen function.
12292 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
12293 * genemit.c (gen_exp): Likewise.
12294 (gen_emit_seq): Pass through info.
12295 (gen_insn): Check for CLOBBER_HIGH.
12296 (gen_expand): Pass through info.
12297 (gen_split): Likewise.
12298 (output_add_clobbers): Likewise.
12299 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
12300 (remove_clobbers): Likewise.
12301 * rtl.h (gen_hard_reg_clobber_high): New declaration.
12303 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12305 * doc/rtl.texi (clobber_high): Add.
12306 (parallel): Add in clobber high
12307 * rtl.c (rtl_check_failed_code3): Add function.
12308 * rtl.def (CLOBBER_HIGH): Add expression.
12309 * rtl.h (RTL_CHECKC3): Add macro.
12310 (rtl_check_failed_code3): Add declaration.
12311 (XC3EXP): Add macro.
12313 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
12316 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
12317 cfun->machine->max_used_stack_alignment if needed.
12319 2018-08-04 Martin Sebor <msebor@redhat.com>
12321 PR tree-optimization/86571
12322 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
12325 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
12328 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12331 2018-08-03 Jeff Law <law@redhat.com>
12334 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12335 Define to speculation_safe_value_not_needed.
12337 2018-08-03 David Malcolm <dmalcolm@redhat.com>
12339 * doc/gcov.texi (-x): Remove duplicate "to".
12340 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
12341 (-Wif-not-aligned): Remove duplicate "is".
12342 (-flto): Remove duplicate "the".
12343 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
12344 duplicate "v5.00.b".
12345 (MSP430 Options): Remove duplicate "and" from the description
12346 of "-mgprel-sec=regexp".
12347 (x86 Options): Remove duplicate copies of "vmldLog102" and
12348 vmlsLog104 from description of "-mveclibabi=type".
12350 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
12352 * internal-fn.h (first_commutative_argument): Declare.
12353 * internal-fn.c (first_commutative_argument): New function.
12354 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
12355 restrictions for pattern statements. Use first_commutative_argument
12356 to look for commutative operands in calls to internal functions.
12358 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
12360 * Makefile.in (wide-int-range.o): New.
12361 * tree-vrp.c: Move all the wide_int_* functions to...
12362 * wide-int-range.cc: ...here.
12363 * tree-vrp.h: Move all the wide_int_* prototypes to...
12364 * wide-int-range.h: ...here.
12366 2018-08-03 Tom de Vries <tdevries@suse.de>
12368 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
12370 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
12371 * except.c (output_function_exception_table): Do early exit if
12372 targetm_common.except_unwind_info (&global_options) == UI_NONE.
12374 2018-08-03 Martin Liska <mliska@suse.cz>
12376 * predict.c (dump_prediction): Change to 2 digits
12379 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
12381 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
12383 (falkor_am_1_gtov_gtov): ... a new insn reservation.
12385 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
12387 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
12388 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
12389 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
12390 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
12391 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
12393 2018-08-02 David Malcolm <dmalcolm@redhat.com>
12395 * diagnostic-show-locus.c (diagnostic_show_locus): Use
12396 pp_take_prefix when saving the existing prefix.
12397 * diagnostic.c (diagnostic_append_note): Likewise.
12398 * langhooks.c (lhd_print_error_function): Likewise.
12399 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
12400 param's type. Free the existing prefix.
12401 (pp_take_prefix): New function.
12402 (pretty_printer::pretty_printer): Drop the prefix parameter.
12403 Rename the length parameter to match the comment.
12404 (pretty_printer::~pretty_printer): Free the prefix.
12405 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
12407 (struct pretty_printer): Drop the "const" from "prefix" field's
12408 type and clarify memory management.
12409 (pp_set_prefix): Drop the "const" from the 2nd param.
12410 (pp_take_prefix): New decl.
12412 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
12414 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
12415 for word_mode_ok here instead of passing as argument.
12416 (expand_block_compare): Change select_block_compare_mode() call.
12417 (expand_strncmp_gpr_sequence): New function.
12418 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
12420 2018-08-02 Jeff Law <law@redhat.com>
12423 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12424 Define to speculation_safe_value_not_needed.
12427 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12428 Define to speculation_safe_value_not_needed.
12430 2018-08-02 Tom de Vries <tdevries@suse.de>
12433 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
12434 function. Return UI_TARGET unconditionally.
12435 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
12436 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
12438 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12440 * genemit.c (print_overload_test): Fix typo.
12442 2018-08-02 Richard Biener <rguenther@suse.de>
12444 PR tree-optimization/86816
12445 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
12446 which checks for value availability before querying it.
12447 (gvn_uses_equal): Use it.
12448 (same_succ_hash): Likewise.
12449 (gimple_equal_p): Likewise.
12451 2018-08-02 Nick Clifton <nickc@redhat.com>
12454 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12455 Define to speculation_safe_value_not_needed.
12458 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12459 Define to speculation_safe_value_not_needed.
12462 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12463 Define to speculation_safe_value_not_needed.
12466 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12467 Define to speculation_safe_value_not_needed.
12470 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12471 Define to speculation_safe_value_not_needed.
12474 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12475 Define to speculation_safe_value_not_needed.
12478 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12479 Define to speculation_safe_value_not_needed.
12482 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12483 Define to speculation_safe_value_not_needed.
12486 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12487 speculation_safe_value_not_needed.
12490 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
12491 to speculation_safe_value_not_needed.
12493 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12495 * doc/md.texi: Expand the documentation of instruction names
12496 to mention port-local uses. Document '@' in pattern names.
12497 * read-md.h (overloaded_instance, overloaded_name): New structs.
12498 (mapping): Declare.
12499 (md_reader::handle_overloaded_name): New member function.
12500 (md_reader::get_overloads): Likewise.
12501 (md_reader::m_first_overload): New member variable.
12502 (md_reader::m_next_overload_ptr): Likewise.
12503 (md_reader::m_overloads_htab): Likewise.
12504 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
12505 m_next_overload_ptr and m_overloads_htab.
12506 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
12507 (get_mode_token, get_code_token, get_int_token): New functions.
12508 (map_attr_string): Add an optional argument that passes back
12509 the associated iterator.
12510 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
12511 (md_reader::handle_overloaded_name, add_overload_instance): New
12513 (apply_iterators): Handle '@' names. Report an error if '@'
12514 is used without iterators.
12515 (initialize_iterators): Initialize the new iterator_group fields.
12516 * genopinit.c (handle_overloaded_code_for)
12517 (handle_overloaded_gen): New functions.
12518 (main): Use them to print declarations of maybe_code_for_* and
12519 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
12520 * genemit.c (print_overload_arguments, print_overload_test)
12521 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
12522 (main): Use it to print definitions of maybe_code_for_* and
12523 maybe_gen_* functions.
12524 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
12525 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
12526 instead of explicit mode checks.
12527 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
12528 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
12529 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
12530 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
12531 (aarch64_expand_compare_and_swap): Likewise
12532 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
12533 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
12534 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
12535 (aarch64_constant_pool_reload_icode): Delete.
12536 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
12537 instead of aarch64_constant_pool_reload_icode. Use
12538 code_for_aarch64_reload_mov instead of explicit mode checks.
12539 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
12540 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
12541 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
12542 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
12543 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
12544 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
12545 (aarch64_atomic_load_op_code): Delete.
12546 (aarch64_emit_atomic_load_op): Likewise.
12547 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
12548 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
12549 instead of aarch64_emit_atomic_load_op.
12550 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
12551 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
12552 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
12553 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
12554 character before the pattern name.
12555 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
12556 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
12557 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
12558 (aarch64_frecps<mode>): Likewise.
12559 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
12560 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
12561 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
12562 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
12563 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
12565 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12567 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
12568 Allow HFmode constants if TARGET_FP_F16INST.
12570 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
12573 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
12574 No longer check last store for clobber of address register.
12576 2018-08-02 Martin Liska <mliska@suse.cz>
12578 PR gcov-profile/86817
12579 * gcov.c (process_all_functions): New function.
12581 (process_file): Move functions processing to
12582 process_all_functions.
12584 2018-08-02 David Malcolm <dmalcolm@redhat.com>
12586 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
12587 "const" to the "gimple *" and "rtx_insn *" parameters.
12588 * dumpfile.h (dump_user_location_t::dump_user_location_t):
12590 (dump_location_t::dump_location_t): Likewise.
12592 2018-08-01 Martin Sebor <msebor@redhat.com>
12594 PR tree-optimization/86650
12595 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
12596 rather than a "gcall *". Directly pass the data of interest
12597 to percent_K_format, rather than building a temporary CALL_EXPR
12599 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
12600 (gimple_fold_builtin_strncat): Adjust.
12601 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
12602 gcall* argument with gimple*.
12603 * gimple-ssa-warn-restrict.c (check_call): Same.
12604 (wrestrict_dom_walker::before_dom_children): Same.
12605 (builtin_access::builtin_access): Same.
12606 (check_bounds_or_overlap): Same
12607 (maybe_diag_overlap): Same.
12608 (maybe_diag_offset_bounds): Same.
12609 * tree-diagnostic.c (default_tree_printer): Move usage of
12610 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
12612 * tree-pretty-print.c (percent_K_format): Add argument.
12613 * tree-pretty-print.h: Add argument.
12614 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
12615 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
12616 (maybe_diag_stxncpy_trunc): Same.
12617 (handle_builtin_stxncpy): Same.
12618 (handle_builtin_strcat): Same.
12620 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12622 * match.pd: Optimise pointer range checks.
12624 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12626 PR tree-optimization/86758
12627 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
12628 to remove pattern statements.
12630 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12632 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
12633 result of dfs_enumerate_from when constructing stmt_vec_infos,
12634 instead of additionally calling get_loop_body.
12636 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12638 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
12640 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
12641 When creating an iv, assert that the step is not known to be zero.
12642 (vect_setup_realignment): Update call accordingly.
12643 * tree-vect-stmts.c (vectorizable_store): Likewise.
12644 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
12646 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12648 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
12649 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
12650 (vectorizable_reduction): Likewise.
12651 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12652 (vect_detect_hybrid_slp_stmts): Likewise.
12653 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12655 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
12657 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
12658 (wide_int_set_zero_nonzero_bits): ...this.
12659 (zero_nonzero_bits_from_vr): Rename to...
12660 (vrp_set_zero_nonzero_bits): ...this.
12661 (extract_range_from_multiplicative_op_1): Abstract wide int
12663 (wide_int_range_multiplicative_op): ...here.
12664 (extract_range_from_binary_expr_1): Extract wide int binary
12665 operations into their own functions.
12666 (wide_int_range_lshift): New.
12667 (wide_int_range_can_optimize_bit_op): New.
12668 (wide_int_range_shift_undefined_p): New.
12669 (wide_int_range_bit_xor): New.
12670 (wide_int_range_bit_ior): New.
12671 (wide_int_range_bit_and): New.
12672 (wide_int_range_trunc_mod): New.
12673 (extract_range_into_wide_ints): New.
12674 (vrp_shift_undefined_p): New.
12675 (extract_range_from_multiplicative_op): New.
12676 (vrp_can_optimize_bit_op): New.
12677 * tree-vrp.h (value_range::dump): New.
12678 (wide_int_range_multiplicative_op): New.
12679 (wide_int_range_lshift):New.
12680 (wide_int_range_shift_undefined_p): New.
12681 (wide_int_range_bit_xor): New.
12682 (wide_int_range_bit_ior): New.
12683 (wide_int_range_bit_and): New.
12684 (wide_int_range_trunc_mod): New.
12685 (zero_nonzero_bits_from_bounds): Rename to...
12686 (wide_int_set_zero_nonzero_bits): ...this.
12687 (zero_nonzero_bits_from_vr): Rename to...
12688 (vrp_set_zero_nonzero_bits): ...this.
12689 (range_easy_mask_min_max): Rename to...
12690 (wide_int_range_can_optimize_bit_op): this.
12691 * vr-values.c (simplify_bit_ops_using_ranges): Rename
12692 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
12694 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12696 * tree-vectorizer.h (vect_orig_stmt): New function.
12697 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
12698 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
12699 (vect_create_epilog_for_reduction): Likewise.
12700 (vectorizable_live_operation): Likewise.
12701 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12702 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
12703 * tree-vect-stmts.c (vectorizable_call): Likewise.
12704 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
12706 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12708 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
12710 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
12711 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
12712 (vect_transform_loop): Likewise.
12713 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12715 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12717 * tree-vectorizer.h (vect_schedule_slp): Return void.
12718 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12719 (vect_schedule_slp): Likewise.
12721 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12723 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
12725 (vect_transform_loop): Update calls accordingly. Schedule SLP
12726 instances before the main loop, if any exist.
12728 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12730 PR tree-optimization/86749
12731 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
12732 If the lhs is used in a COND_EXPR, check that it is being used
12733 as the "then" or "else" value.
12735 2018-08-01 Tom de Vries <tdevries@suse.de>
12738 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12739 speculation_safe_value_not_needed.
12741 2018-08-01 Richard Biener <rguenther@suse.de>
12743 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
12744 as base and offset.
12746 2018-08-01 Martin Liska <mliska@suse.cz>
12748 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
12749 format how successful transformation is dumped.
12750 (gimple_mod_pow2_value_transform): Likewise.
12751 (gimple_mod_subtract_transform): Likewise.
12752 (gimple_stringops_transform): Likewise.
12754 2018-08-01 Martin Liska <mliska@suse.cz>
12756 PR value-prof/35543
12757 * value-prof.c (interesting_stringop_to_profile_p):
12758 Simplify the code and add BUILT_IN_MEMMOVE.
12759 (gimple_stringops_transform): Likewise.
12761 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
12763 * config/aarch64/aarch64-simd.md
12764 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
12765 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
12766 use GPI iterator instead of SI mode.
12768 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12770 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
12771 rs6000_speculation_barrier.
12772 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
12773 new barrier pattern name.
12775 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12777 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
12778 (speculation_barrier): New insn.
12780 2018-08-01 Richard Biener <rguenther@suse.de>
12783 * graphite.h: Include isl/id.h and isl/space.h to allow build
12786 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
12789 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
12790 mode for COFF targets.
12791 * defaults.h (TARGET_COFF): Define.
12792 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
12793 TARGET_COFF): Define.
12794 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
12795 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
12796 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
12798 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12800 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
12801 * gimple-streamer-out.c (output_bb): Save it.
12802 * lto-streamer-in.c (input_struct_function_base): Restore
12803 instance discriminator if available. Create map on demand.
12804 * lto-streamer-out.c (output_struct_function_base): Save it if
12806 * final.c (decl_to_instance_map): Document LTO strategy.
12808 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12809 Olivier Hainque <hainque@adacore.com>
12811 * debug.h (decl_to_instance_map_t): New type.
12812 (decl_to_instance_map): Declare.
12813 (maybe_create_decl_to_instance_map): New inline function.
12814 * final.c (bb_discriminator, last_bb_discriminator): New statics,
12815 to track basic block discriminators.
12816 (final_start_function_1): Initialize them.
12817 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
12819 (decl_to_instance_map): New variable.
12820 (map_decl_to_instance, maybe_set_discriminator): New functions.
12821 (notice_source_line): Set discriminator.
12823 2018-07-31 Ian Lance Taylor <iant@golang.org>
12825 * targhooks.c (default_have_speculation_safe_value): Add
12828 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12830 * dump-context.h: Include "pretty-print.h".
12831 (dump_context::refresh_dumps_are_enabled): New decl.
12832 (dump_context::emit_item): New decl.
12833 (class dump_context): Add fields "m_test_pp" and
12835 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12836 (temp_dump_context::get_dumped_text): New decl.
12837 (class temp_dump_context): Add field "m_pp".
12838 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
12839 (dump_context::refresh_dumps_are_enabled): ...and add a test for
12841 (set_dump_file): Update for above change.
12842 (set_alt_dump_file): Likewise.
12843 (dump_loc): New overload, taking a pretty_printer *.
12844 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
12845 to any test pretty-printer.
12846 (make_item_for_dump_gimple_stmt): New function, adapted from
12847 optinfo::add_gimple_stmt.
12848 (dump_context::dump_gimple_stmt): Call it, and use the result,
12849 eliminating the direct usage of dump_file and alt_dump_file in
12850 favor of indirectly using them via emit_item.
12851 (make_item_for_dump_gimple_expr): New function, adapted from
12852 optinfo::add_gimple_expr.
12853 (dump_context::dump_gimple_expr): Call it, and use the result,
12854 eliminating the direct usage of dump_file and alt_dump_file in
12855 favor of indirectly using them via emit_item.
12856 (make_item_for_dump_generic_expr): New function, adapted from
12858 (dump_context::dump_generic_expr): Call it, and use the result,
12859 eliminating the direct usage of dump_file and alt_dump_file in
12860 favor of indirectly using them via emit_item.
12861 (make_item_for_dump_printf_va): New function, adapted from
12862 optinfo::add_printf_va.
12863 (make_item_for_dump_printf): New function.
12864 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
12865 and use the result, eliminating the direct usage of dump_file and
12866 alt_dump_file in favor of indirectly using them via emit_item.
12867 (make_item_for_dump_dec): New function.
12868 (dump_context::dump_dec): Call it, and use the result,
12869 eliminating the direct usage of dump_file and alt_dump_file in
12870 favor of indirectly using them via emit_item.
12871 (make_item_for_dump_symtab_node): New function, adapted from
12872 optinfo::add_symtab_node.
12873 (dump_context::dump_symtab_node): Call it, and use the result,
12874 eliminating the direct usage of dump_file and alt_dump_file in
12875 favor of indirectly using them via emit_item.
12876 (dump_context::begin_scope): Reimplement, avoiding direct usage
12877 of dump_file and alt_dump_file in favor of indirectly using them
12879 (dump_context::emit_item): New member function.
12880 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12881 Set up test pretty-printer on the underlying context. Call
12882 refresh_dumps_are_enabled.
12883 (temp_dump_context::~temp_dump_context): Call
12884 refresh_dumps_are_enabled.
12885 (temp_dump_context::get_dumped_text): New member function.
12886 (selftest::verify_dumped_text): New function.
12887 (ASSERT_DUMPED_TEXT_EQ): New macro.
12888 (selftest::test_capture_of_dump_calls): Run all tests twice, with
12889 and then without optinfo enabled. Add uses of
12890 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
12891 * dumpfile.h: Update comment for the dump_* API.
12892 * optinfo-emit-json.cc
12893 (selftest::test_building_json_from_dump_calls): Update for new
12894 param for temp_dump_context ctor.
12895 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
12896 and "m_owned" field.
12897 (optinfo_item::~optinfo_item): Likewise.
12898 (optinfo::add_item): New member function.
12899 (optinfo::emit): Update comment.
12900 (optinfo::add_string): Delete.
12901 (optinfo::add_printf): Delete.
12902 (optinfo::add_printf_va): Delete.
12903 (optinfo::add_gimple_stmt): Delete.
12904 (optinfo::add_gimple_expr): Delete.
12905 (optinfo::add_tree): Delete.
12906 (optinfo::add_symtab_node): Delete.
12907 (optinfo::add_dec): Delete.
12908 * optinfo.h (class dump_context): New forward decl.
12909 (optinfo::add_item): New decl.
12910 (optinfo::add_string): Delete.
12911 (optinfo::add_printf): Delete.
12912 (optinfo::add_printf_va): Delete.
12913 (optinfo::add_gimple_stmt): Delete.
12914 (optinfo::add_gimple_expr): Delete.
12915 (optinfo::add_tree): Delete.
12916 (optinfo::add_symtab_node): Delete.
12917 (optinfo::add_dec): Delete.
12918 (optinfo::add_poly_int): Delete.
12919 (optinfo_item::optinfo_item): Remove "owned" param.
12920 (class optinfo_item): Remove field "m_owned".
12922 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12924 PR middle-end/86705
12925 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
12926 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
12927 requested variable alignment.
12928 (expand_one_ssa_partition): Likewise.
12929 (expand_one_var): Likewise.
12931 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12933 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
12934 to speculation_safe_value_not_needed.
12936 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12938 * targhooks.h (speculation_safe_value_not_needed): New prototype.
12939 * targhooks.c (speculation_safe_value_not_needed): New function.
12940 * target.def (have_speculation_safe_value): Update documentation.
12941 * doc/tm.texi: Regenerated.
12943 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12945 * config/aarch64/iterators.md (ALLI_TI): New iterator.
12946 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
12948 (despeculate_copy<ALLI:mode>_insn): New insn.
12949 (despeculate_copyti_insn): New insn.
12950 (despeculate_simple<ALLI:mode>): New insn
12951 (despeculate_simpleti): New insn.
12952 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
12954 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
12955 aarch64_speculation_safe_value.
12956 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
12958 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12960 * config/aarch64/aarch64-speculation.cc: New file.
12961 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
12962 before pass_reorder_blocks.
12963 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
12965 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
12966 X14 and X15 when tracking speculation.
12967 * config/aarch64/aarch64.md (register name constants): Add
12968 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
12969 (unspec): Add UNSPEC_SPECULATION_TRACKER.
12970 (speculation_barrier): New insn attribute.
12971 (cmp<mode>): Allow SP in comparisons.
12972 (speculation_tracker): New insn.
12973 (speculation_barrier): Add speculation_barrier attribute.
12974 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
12975 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
12976 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
12978 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12980 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
12981 aarch64_track_speculation is true.
12982 (tb<optab><mode>1): Likewise.
12983 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
12984 generate CB[N]Z when tracking speculation.
12985 (aarch64_split_compare_and_swap): Likewise.
12986 (aarch64_split_atomic_op): Likewise.
12988 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12990 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
12992 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12994 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
12995 (speculation_barrier): New insn.
12997 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12999 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
13000 * config/arm/arm.md (speculation_barrier): New expand.
13001 (speculation_barrier_insn): New pattern.
13003 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
13005 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
13006 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
13007 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
13008 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
13010 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
13011 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
13012 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
13013 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
13014 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
13015 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
13016 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
13017 * builtins.c (expand_speculation_safe_value): New function.
13018 (expand_builtin): Call it.
13019 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
13020 * doc/extend.texi: Document __builtin_speculation_safe_value.
13021 * doc/md.texi: Document "speculation_barrier" pattern.
13022 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
13023 TARGET_HAVE_SPECULATION_SAFE_VALUE.
13024 * doc/tm.texi: Regenerated.
13025 * target.def (have_speculation_safe_value, speculation_safe_value): New
13027 * targhooks.c (default_have_speculation_safe_value): New function.
13028 (default_speculation_safe_value): New function.
13029 * targhooks.h (default_have_speculation_safe_value): Add prototype.
13030 (default_speculation_safe_value): Add prototype.
13032 2018-07-31 David Malcolm <dmalcolm@redhat.com>
13034 * dump-context.h (dump_context::dump_loc): New decl.
13035 * dumpfile.c (dump_context::dump_loc): New member function.
13036 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
13037 and dump_gimple_stmt.
13038 (dump_context::dump_gimple_expr_loc): Likewise, using
13040 (dump_context::dump_generic_expr_loc): Likewise, using
13042 (dump_context::dump_printf_loc_va): Likewise, using
13044 (dump_context::begin_scope): Explicitly using the global function
13045 "dump_loc", rather than the member function.
13047 2018-07-31 Martin Sebor <msebor@redhat.com>
13049 PR tree-optimization/86741
13050 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
13052 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
13054 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
13055 depend on whether prefetch instructions will be emitted or not.
13056 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
13057 will be emitted or not.
13058 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
13059 (TARGET_SETMEM_PFD): New macros.
13061 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13063 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
13064 (NULL_STMT_VEC_INFO): Delete.
13065 (stmt_vec_info::operator*): Likewise.
13066 (stmt_vec_info::operator gimple *): Likewise.
13067 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
13068 of NULL_STMT_VEC_INFO.
13069 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13070 (vect_reassociating_reduction_p): Likewise.
13071 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13072 (vectorizable_store): Likewise.
13073 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
13074 (vec_info::free_stmt_vec_infos): Likewise.
13076 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13078 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
13079 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
13081 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13083 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
13084 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
13085 (vec_info::free_stmt_vec_info): New private member functions.
13086 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
13087 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
13088 * tree-parloops.c (gather_scalar_reductions): Remove calls to
13089 set_stmt_vec_info_vec and free_stmt_vec_infos.
13090 * tree-vect-loop.c (_loop_vec_info): Remove call to
13091 set_stmt_vec_info_vec.
13092 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
13093 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
13094 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
13095 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
13096 (vec_info::free_stmt_vec_info): ...these new functions. Remove
13097 assignments in {vec_info::,}new_stmt_vec_info that are redundant
13098 with the clearing in the xcalloc.
13099 (stmt_vec_info_vec): Delete.
13100 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
13101 (vectorize_loops): Likewise.
13102 (vec_info::~vec_info): Remove argument from call to
13103 free_stmt_vec_infos.
13104 (vec_info::add_stmt): Remove vinfo argument from call to
13107 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13109 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
13110 rather than a gimple stmt.
13111 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
13112 information for pattern statements when passed the original
13113 statement; instead wait to be passed the pattern statement itself.
13114 Don't call set_vinfo_for_stmt here.
13115 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
13116 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
13117 stmt_vec_infos here.
13118 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
13119 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
13120 stmt_vec_infos entry.
13122 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13124 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
13125 * tree-vectorizer.c (vec_info::replace_stmt): New function.
13126 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
13127 * tree-vect-stmts.c (vectorizable_call): Likewise.
13128 (vectorizable_simd_clone_call): Likewise.
13130 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13132 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
13133 * tree-vectorizer.c (vec_info::remove_stmt): New function.
13134 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
13135 * tree-vect-loop.c (vect_transform_loop): Likewise.
13136 * tree-vect-slp.c (vect_schedule_slp): Likewise.
13137 * tree-vect-stmts.c (vect_remove_stores): Likewise.
13139 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13141 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
13142 (vect_dr_stmt): Delete.
13143 * tree-vectorizer.c (vec_info::lookup_dr): New function.
13144 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
13146 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
13147 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
13148 (vect_verify_datarefs_alignment, vect_peeling_supportable)
13149 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
13150 (vect_analyze_data_refs): Likewise.
13151 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
13153 (vect_find_same_alignment_drs): Likewise.
13154 (vect_slp_analyze_node_dependences): Update calls accordingly.
13155 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
13156 instead of DR_VECT_AUX.
13157 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
13158 of a vector data references. Use vec_info::lookup_dr instead of
13160 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
13161 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
13162 instead of DR_VECT_AUX.
13164 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13166 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
13168 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
13170 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13171 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13172 (vect_gen_prolog_loop_niters): Likewise.
13174 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13176 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
13177 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
13178 (vect_known_alignment_in_bytes, vect_dr_behavior)
13179 (vect_get_scalar_dr_size): Take references as dr_vec_infos
13180 instead of data_references. Update calls to other routines for
13181 which the same change has been made.
13182 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
13183 dr_vec_infos instead of stmt_vec_infos.
13184 (vect_analyze_data_ref_dependence): Update call accordingly.
13185 (vect_slp_analyze_data_ref_dependence)
13186 (vect_record_base_alignments): Use DR_VECT_AUX.
13187 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
13188 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
13189 (vector_alignment_reachable_p, vect_get_data_access_cost)
13190 (vect_peeling_supportable, vect_analyze_group_access_1)
13191 (vect_analyze_group_access, vect_analyze_data_ref_access)
13192 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
13193 (vect_compile_time_alias, vect_small_gap_p)
13194 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
13195 (vect_supportable_dr_alignment): Take references as dr_vec_infos
13196 instead of data_references. Update calls to other routines for
13197 which the same change has been made.
13198 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
13199 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
13200 (vect_slp_analyze_and_verify_node_alignment)
13201 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
13202 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
13203 (vect_setup_realignment): Use dr_vec_infos. Update calls after
13205 (_vect_peel_info::dr): Replace with...
13206 (_vect_peel_info::dr_info): ...this new field.
13207 (vect_peeling_hash_get_most_frequent)
13208 (vect_peeling_hash_choose_best_peeling): Update accordingly.
13209 (vect_peeling_hash_get_lowest_cost):
13210 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
13211 routines for which the same change has been made.
13212 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
13214 * tree-vect-loop-manip.c (get_misalign_in_elems)
13215 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
13217 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13218 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
13219 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
13220 (get_group_load_store_type, get_negative_load_store_type)
13221 (vect_get_data_ptr_increment, vectorizable_store)
13222 (vectorizable_load): Likewise.
13223 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
13224 Update calls to other routines for which the same change has been made.
13226 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13228 * tree-vectorizer.h (vec_info::move_dr): New member function.
13229 (dataref_aux): Rename to...
13230 (dr_vec_info): ...this and add "dr" and "stmt" fields.
13231 (_stmt_vec_info::dr_aux): Update accordingly.
13232 (_stmt_vec_info::data_ref_info): Delete.
13233 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
13234 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
13235 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
13236 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
13238 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
13239 (STMT_VINFO_DR_INFO): New macro.
13240 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
13241 (set_dr_misalignment): Update after rename of dataref_aux.
13242 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
13243 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
13244 initialization of STMT_VINFO_DATA_REF.
13245 * tree-vectorizer.c (vec_info::move_dr): New function.
13246 * tree-vect-patterns.c (vect_recog_bool_pattern)
13247 (vect_recog_mask_conversion_pattern)
13248 (vect_recog_gather_scatter_pattern): Use it.
13249 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
13250 the "dr" and "stmt" fields of dr_vec_info instead of
13251 STMT_VINFO_DATA_REF.
13253 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13255 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
13256 (is_pattern_stmt_p): Use it.
13257 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
13258 on pattern statements.
13260 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13262 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
13263 original stmt as a stmt_vec_info rather than a gimple stmt.
13264 (vect_pattern_recog_1): Take the statement directly as a
13265 stmt_vec_info, rather than via a gimple_stmt_iterator.
13266 Update call to vect_mark_pattern_stmts.
13267 (vect_pattern_recog): Update calls accordingly.
13269 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13271 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
13272 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
13273 a vect_def_type for the first argument.
13274 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
13275 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
13276 operand if it isn't defined by a vectorized statement.
13277 (vect_build_gather_load_calls): Remove the mask_dt argument and
13278 update calls to vect_get_vec_def_for_stmt_copy.
13279 (vectorizable_bswap): Likewise the dt argument.
13280 (vectorizable_call): Update calls to vectorizable_bswap and
13281 vect_get_vec_def_for_stmt_copy.
13282 (vectorizable_simd_clone_call, vectorizable_assignment)
13283 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
13284 (vectorizable_comparison): Update calls to
13285 vect_get_vec_def_for_stmt_copy.
13286 (vectorizable_store): Likewise. Remove now-unnecessary calls to
13287 vect_is_simple_use.
13288 (vect_get_loop_based_defs): Remove dt argument and update call
13289 to vect_get_vec_def_for_stmt_copy.
13290 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
13291 and vect_get_vec_def_for_stmt_copy.
13292 (vectorizable_load): Update calls to vect_build_gather_load_calls
13293 and vect_get_vec_def_for_stmt_copy.
13294 * tree-vect-loop.c (vect_create_epilog_for_reduction)
13295 (vectorizable_reduction, vectorizable_live_operation): Update calls
13296 to vect_get_vec_def_for_stmt_copy.
13298 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13300 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
13301 and gimple stmt arguments with a stmt_vec_info.
13302 (vect_record_base_alignments): Update calls accordingly.
13303 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
13304 and gimple stmt arguments with a stmt_vec_info.
13305 (vect_build_slp_tree_1): Remove vinfo argument and update call
13306 to vect_record_max_nunits.
13307 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
13308 and vect_record_max_nunits.
13310 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13312 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
13313 file and take a stmt_vec_info instead of a gimple stmt.
13314 (supportable_widening_operation, vect_finish_replace_stmt)
13315 (vect_finish_stmt_generation, vect_get_store_rhs)
13316 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
13317 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
13318 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
13319 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
13320 (vect_create_data_ref_ptr, bump_vector_ptr)
13321 (vect_permute_store_chain, vect_setup_realignment)
13322 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
13323 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
13324 (vectorizable_reduction, vectorizable_induction)
13325 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
13326 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
13327 than gimple stmts as arguments.
13328 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
13329 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
13330 (can_group_stmts_p, vect_check_gather_scatter)
13331 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
13332 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
13333 (vect_permute_load_chain, vect_shift_permute_load_chain)
13334 (vect_transform_grouped_load)
13335 (vect_record_grouped_load_vectors): Likewise.
13336 * tree-vect-loop.c (vect_fixup_reduc_chain)
13337 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
13338 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
13339 (vectorizable_reduction, vectorizable_induction)
13340 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
13341 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
13342 (vect_get_load_store_mask): Likewise.
13343 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
13344 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
13345 * tree-vect-stmts.c (vect_mark_relevant)
13346 (is_simple_and_all_uses_invariant)
13347 (exist_non_indexing_operands_for_use_p, process_use)
13348 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
13349 (vect_get_vec_def_for_operand, vect_get_vec_defs)
13350 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
13351 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
13352 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
13353 (get_negative_load_store_type, get_load_store_type)
13354 (vect_check_load_store_mask, vect_check_store_rhs)
13355 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
13356 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13357 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13358 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13359 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
13360 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
13361 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
13362 (supportable_widening_operation): Likewise.
13364 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13366 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
13367 a stmt_vec_info instead of a gcall.
13368 (vect_check_gather_scatter): Update call accordingly.
13369 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
13371 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
13372 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
13373 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
13374 instead of a gimple stmt.
13375 (vect_transform_loop): Update calls accordingly.
13376 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
13377 stmt_vec_infos instead of gimple stmts.
13378 (vect_analyze_slp_instance): Update use accordingly.
13379 * tree-vect-stmts.c (read_vector_array, write_vector_array)
13380 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
13381 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
13382 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
13383 (vect_gen_widened_results_half, vect_get_loop_based_defs)
13384 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
13385 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
13386 down to subroutines.
13388 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13390 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
13391 of the worklist from a vector of gimple stmts to a vector of
13393 * tree-vect-stmts.c (vect_mark_relevant, process_use)
13394 (vect_mark_stmts_to_be_vectorized): Likewise
13396 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13398 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
13399 statement before passing it to vect_analyze_stmt.
13400 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
13401 the chain of phi vector definitions. Track the exit phi via its
13403 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
13404 STMT_VINFO_REDUC_DEF.
13405 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
13406 stmt_vec_infos to handle the statement chains.
13407 (vect_get_slp_defs): Record the first statement in the node
13408 using a stmt_vec_info.
13409 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
13410 statements here and pass their stmt_vec_info down to subroutines.
13411 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
13412 down to vect_finish_stmt_generation.
13413 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
13414 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
13415 stmt_vec_infos to subroutines.
13416 (vect_remove_stores): Use stmt_vec_infos to handle the statement
13419 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13421 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
13422 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
13423 (vect_permute_store_chain, vect_setup_realignment)
13424 (vect_permute_load_chain, vect_shift_permute_load_chain)
13425 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
13426 stmts internally, and when passing values to other vectorizer routines.
13427 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
13428 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
13429 (vect_analyze_loop_operations, get_initial_def_for_reduction)
13430 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
13431 (vectorizable_reduction, vectorizable_induction)
13432 (vectorizable_live_operation, vect_transform_loop_stmt)
13433 (vect_transform_loop): Likewise.
13434 * tree-vect-patterns.c (vect_reassociating_reduction_p)
13435 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
13436 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
13437 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13438 (vect_slp_analyze_node_operations_1): Likewise.
13439 * tree-vect-stmts.c (vect_mark_relevant, process_use)
13440 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
13441 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
13442 (vect_finish_stmt_generation_1, get_group_load_store_type)
13443 (get_load_store_type, vect_build_gather_load_calls)
13444 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13445 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13446 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13447 (vectorizable_store, vectorizable_load, vectorizable_condition)
13448 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
13449 (supportable_widening_operation): Likewise.
13450 (vect_get_vector_types_for_stmt): Likewise.
13451 * tree-vectorizer.h (vect_dr_behavior): Likewise.
13453 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13455 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
13456 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
13457 (vect_permute_store_chain, vect_permute_load_chain)
13458 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
13459 repeated stmt_vec_info lookups.
13460 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
13461 (vect_update_ivs_after_vectorizer): Likewise.
13462 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
13463 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13464 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
13465 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13466 (vect_bb_slp_scalar_cost): Likewise.
13467 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
13469 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13471 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
13472 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
13473 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
13474 to get gassigns and gcalls, rather than operating on generc gimple
13476 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
13477 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
13478 (vectorizable_load, vect_analyze_stmt): Likewise.
13479 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
13481 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13483 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
13484 return stmt_vec_infos rather than gimple stmts. Do not accept
13486 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
13488 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
13489 Update use of get_later_stmt.
13490 (vect_get_constant_vectors): Update call accordingly.
13491 (vect_schedule_slp_instance): Likewise
13492 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
13493 (vect_slp_analyze_instance_dependence): Likewise.
13494 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
13496 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13498 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
13499 (stmt_info_for_cost::stmt_info): ...this new field.
13500 (add_stmt_costs): Update accordingly.
13501 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13502 (vect_get_known_peeling_cost): Likewise.
13503 (vect_estimate_min_profitable_iters): Likewise.
13504 * tree-vect-stmts.c (record_stmt_cost): Likewise.
13506 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13508 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
13509 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13510 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
13512 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
13514 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13516 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
13517 a gimple stmt to a stmt_vec_info.
13518 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
13520 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13522 * tree-vectorizer.h (vec_info::grouped_stores): Change from
13523 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13524 (_loop_vec_info::reduction_chains): Likewise.
13525 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
13527 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13529 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13531 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
13532 a gimple stmt to a stmt_vec_info.
13533 (_stmt_vec_info::next_element): Likewise.
13534 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
13535 (vect_slp_analyze_and_verify_node_alignment)
13536 (vect_analyze_group_access_1, vect_analyze_group_access)
13537 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
13538 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
13539 (vect_supportable_dr_alignment): Update accordingly.
13540 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
13541 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
13542 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
13543 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
13544 * tree-vect-slp.c (vect_build_slp_tree_1)
13545 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
13546 (vect_split_slp_store_group, vect_analyze_slp_instance)
13547 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
13548 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
13549 (get_group_load_store_type, get_load_store_type)
13550 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
13551 (vect_transform_stmt, vect_remove_stores): Likewise.
13553 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13555 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
13556 than a gimple stmt.
13557 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
13558 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
13559 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
13560 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
13561 (vector_alignment_reachable_p, vect_get_data_access_cost)
13562 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
13563 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
13564 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
13565 (vect_analyze_group_access_1, vect_analyze_group_access)
13566 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
13567 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
13568 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
13569 result of vect_dr_stmt and use the stmt_vec_info instead of
13570 the associated gimple stmt.
13571 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13572 (vect_gen_prolog_loop_niters): Likewise.
13573 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13575 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13577 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
13578 to a vec<stmt_vec_info>.
13579 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
13580 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
13581 vec<stmt_vec_info>.
13582 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
13583 to a vec<stmt_vec_info>.
13584 (bst_traits::value_type, bst_traits::value_type): Likewise.
13585 (bst_traits::hash): Update accordingly.
13586 (vect_get_and_check_slp_defs): Change the stmts parameter from
13587 a vec<gimple *> to a vec<stmt_vec_info>.
13588 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
13589 (vect_build_slp_tree): Likewise.
13590 (vect_build_slp_tree_2): Likewise. Update uses of
13591 SLP_TREE_SCALAR_STMTS.
13592 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
13593 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
13594 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
13595 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
13596 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
13597 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
13598 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
13599 (vect_get_constant_vectors, vect_get_slp_defs)
13600 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
13601 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
13602 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
13603 instead of gimple stmts.
13604 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
13605 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
13606 (vect_slp_analyze_instance_dependence): Update uses of
13607 SLP_TREE_SCALAR_STMTS.
13608 (vect_slp_analyze_and_verify_node_alignment): Likewise.
13609 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
13610 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
13611 (get_initial_defs_for_reduction): Likewise.
13612 (vect_create_epilog_for_reduction): Likewise.
13613 (vectorize_fold_left_reduction): Likewise.
13614 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
13615 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
13616 (can_vectorize_live_stmts): Likewise.
13618 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13620 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
13621 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13622 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
13624 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
13626 (gather_scalar_reductions): Update after above interface changes.
13627 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
13628 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
13630 (vect_force_simple_reduction): Likewise.
13631 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
13632 LOOP_VINFO_REDUCTIONS.
13633 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13635 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13637 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
13638 a gimple stmt to a stmt_vec_info.
13639 * tree-vect-loop.c (vect_active_double_reduction_p)
13640 (vect_force_simple_reduction, vectorizable_reduction): Update
13643 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13645 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
13646 vec<gimple *> to a vec<stmt_vec_info>.
13647 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
13648 the reduction_phis argument from a vec<gimple *> to a
13649 vec<stmt_vec_info>.
13650 (vectorizable_reduction): Likewise the phis local variable that
13651 is passed to vect_create_epilog_for_reduction. Update for new type
13652 of SLP_TREE_VEC_STMTS.
13653 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
13654 (vectorizable_live_operation): Likewise.
13655 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
13656 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
13658 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13660 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
13661 a gimple stmt to a stmt_vec_info.
13662 (vectorizable_condition, vectorizable_live_operation)
13663 (vectorizable_reduction, vectorizable_induction): Pass back the
13664 vectorized statement as a stmt_vec_info.
13665 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
13666 use of STMT_VINFO_VEC_STMT.
13667 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
13668 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
13669 as stmt_vec_infos rather than gimple stmts.
13670 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
13671 to a stmt_vec_info.
13672 (vectorizable_live_operation): Likewise.
13673 (vectorizable_reduction, vectorizable_induction): Likewise,
13674 updating use of STMT_VINFO_VEC_STMT.
13675 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
13676 of STMT_VINFO_VEC_STMT.
13677 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
13678 (vectorizable_simd_clone_call, vectorizable_conversion)
13679 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13680 (vectorizable_store, vectorizable_load, vectorizable_condition)
13681 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
13682 from a gimple stmt to a stmt_vec_info.
13683 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
13684 pointer to a stmt_vec_info to the vectorizable_* routines.
13686 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13688 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
13689 a gimple stmt to a stmt_vec_info.
13690 (is_pattern_stmt_p): Update accordingly.
13691 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
13692 (vect_record_grouped_load_vectors): Likewise.
13693 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
13694 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
13695 (vect_model_reduction_cost): Likewise.
13696 (vect_create_epilog_for_reduction): Likewise.
13697 (vectorizable_reduction, vectorizable_induction): Likewise.
13698 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13699 Return the stmt_vec_info for the pattern statement.
13700 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
13701 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
13702 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13703 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
13704 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
13705 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
13706 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
13707 (free_stmt_vec_info, vect_is_simple_use): Likewise.
13709 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13711 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
13712 (vect_finish_stmt_generation): Likewise.
13713 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13714 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
13715 (vect_build_gather_load_calls): Use the return value of the above
13716 functions instead of a separate call to vinfo_for_stmt. Use narrow
13717 scopes for the input gimple stmt and wider scopes for the associated
13718 stmt_vec_info. Use vec_info::lookup_def when setting these
13719 stmt_vec_infos from an SSA_NAME definition.
13720 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13721 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13722 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13723 (vectorizable_store, vectorizable_load, vectorizable_condition)
13724 (vectorizable_comparison): Likewise.
13725 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
13726 (vectorizable_reduction): Likewise.
13728 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13730 * tree-vectorizer.h (vect_is_simple_use): Add an optional
13731 stmt_vec_info * parameter before the optional gimple **.
13732 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13733 (process_use, vect_get_vec_def_for_operand_1): Update callers.
13734 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
13735 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13736 (vectorizable_live_operation): Likewise.
13737 * tree-vect-patterns.c (type_conversion_p): Likewise.
13738 (vect_look_through_possible_promotion): Likewise.
13739 (vect_recog_rotate_pattern): Likewise.
13740 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
13742 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13744 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
13745 a typedef to a wrapper class.
13746 (NULL_STMT_VEC_INFO): New macro.
13747 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
13748 (stmt_vec_info::operator*): New function.
13749 (stmt_vec_info::operator gimple *): Likewise.
13750 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
13751 (add_stmt_costs): Likewise.
13752 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
13753 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13754 (vect_get_known_peeling_cost): Likewise.
13755 (vect_estimate_min_profitable_iters): Likewise.
13756 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13757 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
13758 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13759 (vectorizable_store, free_stmt_vec_infos): Likewise.
13760 (new_stmt_vec_info): Change return type of xcalloc to
13763 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13765 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
13766 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
13767 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
13768 a single_imm_use-based sequence.
13769 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
13771 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13773 * tree-vectorizer.h (vec_info::lookup_def): Declare.
13774 * tree-vectorizer.c (vec_info::lookup_def): New function.
13775 * tree-vect-patterns.c (vect_get_internal_def): Use it.
13776 (vect_widened_op_tree): Likewise.
13777 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13778 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
13779 (vectorizable_reduction): Likewise.
13780 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
13782 (vect_is_slp_reduction): Update calls accordingly. Use
13783 vec_info::lookup_def.
13784 (vect_is_simple_reduction): Likewise
13785 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
13787 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13789 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
13790 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
13791 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
13793 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
13794 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
13795 (vect_update_vf_for_slp, vect_analyze_loop_operations)
13796 (vect_is_slp_reduction, vectorizable_induction)
13797 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
13798 * tree-vect-patterns.c (vect_init_pattern_stmt):
13799 (vect_determine_min_output_precision_1, vect_determine_precisions)
13800 (vect_pattern_recog): Likewise.
13801 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
13802 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
13803 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
13804 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13805 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
13806 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
13807 info field from a loop to a loop_vec_info.
13809 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13811 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
13812 (vec_info::add_stmt): Declare.
13813 * tree-vectorizer.c (vec_info::add_stmt): New function.
13814 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
13815 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
13816 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13817 (vectorizable_induction): Likewise.
13818 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
13819 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13820 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
13821 (vectorizable_load): Likewise.
13822 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13823 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
13824 (vect_recog_gather_scatter_pattern): Likewise.
13825 (append_pattern_def_seq): Likewise. Remove a check that is
13826 performed by add_stmt itself.
13828 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13830 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
13831 which make_ssa_name was called with new_stmt before new_stmt
13834 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13836 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
13838 (vect_is_slp_reduction): ...here...
13839 (vect_is_simple_reduction): ...and here. Remove repetition of tests
13840 that are already known to be false.
13842 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13844 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
13845 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
13846 STMT_VINFO_NUM_SLP_USES when it's true.
13847 (vect_free_slp_instance): Add a final_p parameter and pass it to
13848 vect_free_slp_tree.
13849 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
13850 (vect_analyze_slp_instance): Likewise.
13851 (vect_slp_analyze_operations): Likewise.
13852 (vect_slp_analyze_bb_1): Likewise.
13853 * tree-vectorizer.c (vec_info): Likewise.
13854 * tree-vect-loop.c (vect_transform_loop): Likewise.
13856 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13858 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13859 function is not called for second and subsequent members of
13862 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13864 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
13865 cases for nested loops from here to ...
13866 (vect_create_epilog_for_reduction): ...here. Only call
13867 vect_is_simple_use for inner-loop reductions.
13869 2018-07-31 Martin Liska <mliska@suse.cz>
13871 PR gcov-profile/85338
13872 PR gcov-profile/85350
13873 PR gcov-profile/85372
13874 * profile.c (struct location_triplet): New.
13875 (struct location_triplet_hash): Likewise.
13876 (output_location): Do not output a BB that
13877 is already recorded for a line.
13878 (branch_prob): Use streamed_locations.
13880 2018-07-31 Martin Liska <mliska@suse.cz>
13882 PR gcov-profile/85370
13883 * coverage.c (coverage_begin_function): Do not mark target
13884 clones as artificial functions.
13886 2018-07-31 Martin Liska <mliska@suse.cz>
13888 PR gcov-profile/83813
13889 PR gcov-profile/84758
13890 PR gcov-profile/85217
13891 PR gcov-profile/85332
13892 * profile.c (branch_prob): Do not record GOTO expressions
13893 for GIMPLE statements which locations are already streamed.
13895 2018-07-31 Olivier Hainque <hainque@adacore.com>
13897 * gcc.c (handle_spec_function): Accept a soft_matched_part
13898 argument, as do_spec_1. Pass it down to ...
13899 (eval_spec_function): Accept a soft_matched_part argument,
13900 and pass it down to ...
13901 (do_spec_2): Accept a soft_matched_part argument, and pass
13902 it down to do_spec_1.
13903 (do_spec_1): Pass soft_matched_part to handle_spec_function.
13904 (handle_braces): Update call to handle_spec_function.
13905 (driver::set_up_specs): Update calls to do_spec_2.
13906 (compare_debug_dump_opt_spec_function): Likewise.
13907 (compare_debug_self_opt_spec_function): Likewise.
13909 2018-07-31 Olivier Hainque <hainque@adacore.com>
13911 * common.opt (nolibc): New option.
13912 * doc/invoke.texi (Link Options): Document it.
13913 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
13914 * config/alpha/linux.h: Likewise.
13915 * config/arc/elf.h: Likewise.
13916 * config/arm/uclinux-elf.h: Likewise.
13917 * config/arm/unknown-elf.h: Likewise.
13918 * config/avr/avrlibc.h: Likewise.
13919 * config/bfin/bfin.h: Likewise.
13920 * config/bfin/linux.h: Likewise.
13921 * config/bfin/uclinux.h: Likewise.
13922 * config/darwin.h: Likewise.
13923 * config/darwin10.h: Likewise.
13924 * config/darwin12.h: Likewise.
13925 * config/gnu-user.h: Likewise.
13926 * config/lm32/uclinux-elf.h: Likewise.
13927 * config/pa/pa-hpux11.h: Likewise.
13928 * config/pa/pa64-hpux.h: Likewise.
13929 * config/sparc/sparc.h: Likewise.
13931 2018-07-31 Olivier Hainque <hainque@adacore.com>
13933 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
13934 undefined variables.
13936 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13939 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
13940 instead of GEN_INT.
13942 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
13944 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
13945 terminated string literal.
13947 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13949 PR rtl-optimization/85160
13950 * combine.c (is_just_move): New function.
13951 (try_combine): Allow combining two instructions into two if neither of
13952 the original instructions was a move.
13954 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
13957 * doc/extend.texi (Global Register Variables): Discourage use of type
13959 (Local Register Variables): Likewise.
13961 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
13963 PR tree-optimization/86506
13964 * hwint.h (ceil_log2): Resync with hwint.c implementation.
13966 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
13969 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
13970 hard_regno, make sure no insn between `from` and `to` clobbers it.
13972 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
13973 Tom de Vries <tdevries@suse.de>
13975 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
13976 (PTX_DEFAULT_RUNTIME_DIM): ... this.
13977 (nvptx_goacc_validate_dims): Set default worker and gang dims to
13978 PTX_DEFAULT_RUNTIME_DIM.
13979 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
13981 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
13983 * config/pa/pa.c (pa_output_addr_vec): Align address table.
13984 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
13985 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
13987 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
13989 * config/rs6000/constraints.md (wG constraint): Delete, no longer
13991 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
13992 predicate to reflect toc fusion has been deleted.
13993 (toc_fusion_mem_raw): Delete, no longer used.
13994 (toc_fusion_mem_wrapped): Likewise.
13995 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
13997 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
13998 Delete, no longer used.
13999 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
14000 meant to be used for toc fusion.
14001 (rs6000_debug_print_mode): Delete toc fusion debugging.
14002 (rs6000_debug_reg_global): Likewise.
14003 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
14004 fusion and secondary reload support that were never used.
14005 (rs6000_option_override_internal): Delete TOC fusion, that was only
14006 partially defined, and it did not work unless you also used the
14008 (rs6000_legitimate_address_p): Delete TOC fusion support.
14009 (rs6000_opt_masks): Likewise.
14010 (fusion_wrap_memory_address): Delete function, no longer used.
14011 (fusion_split_address); Delete TOC fusion support.
14012 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
14013 longer used with toc fusion being deleted.
14014 (TARGET_TOC_FUSION_FP): Likewise.
14015 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
14017 (toc fusion spliter): Delete TOC fusion support.
14018 (toc_fusionload_<mode>): Likewise.
14019 (toc_fusionload_di): Likewise.
14020 (fusion_gpr_load_<mode>): Delete generator function, this insn no
14021 longer needs to be named. Rename predicate to delete TOC fusion.
14022 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
14023 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
14024 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
14025 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
14026 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
14028 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14030 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
14031 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
14032 __int128 in built-in function prototypes.
14033 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
14034 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
14036 2018-07-27 Martin Sebor <msebor@redhat.com>
14038 PR tree-optimization/86696
14039 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
14040 types, including enums.
14041 (handle_char_store): Be prepared for the above function to fail.
14043 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
14045 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
14046 when optimization level is lower than 2 or optimize for size.
14048 2018-07-26 Martin Sebor <msebor@redhat.com>
14050 PR tree-optimization/86043
14051 PR tree-optimization/86042
14052 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
14053 (get_string_cst_length): Rename...
14054 (get_min_string_length): ...to this. Add argument.
14055 (handle_char_store): Extend to handle multi-character stores by
14057 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
14058 * tree.h (initializer_zerop): Add argument.
14060 2018-07-26 Jakub Jelinek <jakub@redhat.com>
14062 PR middle-end/86660
14063 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
14064 declare target to variables if they have always,{to,from,tofrom} map
14067 2018-07-26 Martin Liska <mliska@suse.cz>
14070 * lto-wrapper.c: Add linker_output as prefix
14071 for ltrans_output_file.
14073 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
14075 PR rtl-optimization/85805
14076 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
14077 value for hard registers if that was written in the same mode.
14079 2018-07-26 Martin Liska <mliska@suse.cz>
14081 PR gcov-profile/86536
14082 * gcov.c (format_gcov): Use printf format %.*f directly
14083 and do not handle special values.
14085 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14087 * common/config/arc/arc-common.c (arc_option_optimization_table):
14088 Update default optimizations for size.
14090 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14092 * config/arc/arc.md (movsf_insn): Add short instruction selection.
14093 * config/arc/constraints.md (CfZ): New constraint.
14094 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
14095 (subsf3_fpu): Likewise.
14096 (cmpsf_fpu): Likewise.
14097 (cmpsf_fpu_uneq): Likewise.
14099 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14101 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
14103 (arc_is_uncached_mem_p): uncached applies to both the variable and
14106 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14108 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
14111 2018-07-25 David Malcolm <dmalcolm@redhat.com>
14113 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
14114 field "m_scopes" from vec to auto_vec.
14116 2018-07-25 Martin Liska <mliska@suse.cz>
14118 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
14121 2018-07-25 Richard Biener <rguenther@suse.de>
14124 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
14125 special wrt context_die late.
14126 (gen_subprogram_die): Re-use DIEs in local scope.
14128 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
14130 PR tree-optimization/86644
14131 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
14133 2018-07-25 Martin Liska <mliska@suse.cz>
14135 PR middle-end/86645
14136 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
14137 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
14139 2018-07-25 Martin Liska <mliska@suse.cz>
14142 * params.def: Explain ASan abbreviation and provide
14143 a documentation link.
14145 2018-07-24 Martin Sebor <msebor@redhat.com>
14147 PR tree-optimization/86622
14148 PR tree-optimization/86532
14149 * builtins.h (string_length): Declare.
14150 * builtins.c (c_strlen): Correct handling of non-constant offsets.
14151 (check_access): Be prepared for non-constant length ranges.
14152 (string_length): Make extern.
14153 * expr.c (string_constant): Only handle the minor non-constant
14154 array index. Use string_constant to compute the length of
14155 a generic string constant.
14157 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
14159 PR tree-optimization/86618
14160 * tree-vect-stmts.c (vectorizable_call): Don't take the address
14161 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
14163 2018-07-24 David Malcolm <dmalcolm@redhat.com>
14165 PR tree-optimization/86636
14166 * json.cc (json::object::set): Fix comment. Add assertions.
14167 (json::array::append): Move here from json.h. Add comment and an
14169 (json::string::string): Likewise.
14170 * json.h (json::array::append): Move to json.cc.
14171 (json::string::string): Likewise.
14172 * optinfo-emit-json.cc
14173 (optrecord_json_writer::impl_location_to_json): Assert that we
14174 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
14175 wrapper around it. Expand the location once, rather than three
14177 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
14178 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
14180 (optrecord_json_writer::optinfo_to_json): Likewise, in four
14181 places. Fix some overlong lines.
14183 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
14185 * config/aarch64/aarch64-simd.md
14186 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
14187 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
14188 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
14189 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
14190 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
14191 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
14192 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
14193 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
14194 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
14196 2018-07-24 Jakub Jelinek <jakub@redhat.com>
14198 PR middle-end/86627
14199 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
14200 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
14201 and abs_d == d, do the power of two handling if profitable.
14203 2018-07-24 Richard Biener <rguenther@suse.de>
14205 * match.pd: Add BIT_FIELD_REF canonicalizations.
14207 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14210 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
14212 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14214 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
14215 terminated STRING_CST object.
14217 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14219 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
14221 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
14223 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
14224 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
14225 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
14226 the elements into a register.
14227 (rs6000_split_v4si_init_di_reg): Delete.
14228 (rs6000_split_v4si_init): Delete.
14229 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
14230 (vsx_init_v4si): Rewrite as a define_expand.
14232 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
14234 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
14235 zero_extend argument from memory): New.
14237 2018-07-22 Martin Sebor <msebor@redhat.com>
14240 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
14241 diagnosing calls with unknown arguments unless -Walloca-larger-than
14242 is restricted to less than PTRDIFF_MAX bytes.
14244 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
14246 * doc/gcov.texi (Invoking Gcov): Editorial changes.
14248 2018-07-20 David Malcolm <dmalcolm@redhat.com>
14250 * pretty-print.c (text_info::set_location): Remove redundant
14251 "line_table" parameter from call to rich_location::set_range.
14253 2018-07-20 Martin Sebor <msebor@redhat.com>
14255 PR middle-end/82063
14256 * builtins.c (expand_builtin_alloca): Adjust.
14257 * calls.c (alloc_max_size): Simplify.
14258 * cgraphunit.c (cgraph_node::expand): Adjust.
14259 * common.opt (larger_than_size, warn_frame_larger_than): Remove
14261 (frame_larger_than_size): Same.
14262 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
14263 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
14265 * doc/invoke.texi (GCC Command Options): Document option arguments.
14266 Explain byte-size arguments and suffixes.
14267 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
14268 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
14269 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
14270 * doc/options.texi (UInteger): Expand.
14271 (Host_Wide_Int, ByteSize): Document new properties.
14272 * final.c (final_start_function_1): Include sizes in an error message.
14273 * function.c (frame_offset_overflow): Same.
14274 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
14275 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
14276 Diagnose unbounded alloca calls only for limits of less than
14278 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
14279 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
14281 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
14282 only for limits of less than PTRDIFF_MAX.
14283 * langhooks-def.h (lhd_handle_option): Change function argument
14285 * langhooks.c (lhd_handle_option): Same.
14286 * langhooks.h (handle_option): Same.
14287 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
14289 (var_type, var_type_struct): Same.
14290 (var_set): Handle ByteSize flag.
14291 * optc-gen.awk: Add comments to output to ease debugging. Make
14292 use of HOST_WIDE_INT where appropriate.
14293 * opts-gen-save.awk: Use %lx to format unsigned long.
14294 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
14295 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
14296 arguments. Parse bytes-size suffixes.
14297 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
14298 (enum_value_to_arg): Same.
14299 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
14300 (handle_option): Adjust.
14301 (generate_option): Change function argument to HOST_WIDE_INT.
14302 (cmdline_handle_error): Adjust.
14303 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
14304 (set_option): Change function argument to HOST_WIDE_INT.
14305 (option_enabled): Handle cl_host_wide_int.
14306 (get_option_state): Handle CLVC_SIZE.
14307 (control_warning_option): Same.
14308 * opts.c (common_handle_option): Change function argument to
14309 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
14310 OPT_Wvla_larger_than_.
14311 * opts.h (enum cl_var_type): Add an enumerator.
14312 * stor-layout.c (layout_decl): Print a more meaningful warning.
14313 * toplev.c (output_stack_usage): Adjust.
14315 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
14317 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
14318 call to inline_expand_builtin_string_cmp.
14319 (expand_builtin_strcmp): Likewise.
14320 (expand_builtin_strncmp): Likewise.
14321 (inline_string_cmp): Delete the last parameter, change char_type_node
14322 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
14324 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
14325 the inlining expansion on target where the type of the call has same or
14326 narrower precision than unsigned char.
14328 2018-07-20 David Malcolm <dmalcolm@redhat.com>
14330 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
14331 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
14332 * common.opt (fsave-optimization-record): New option.
14333 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
14334 * doc/invoke.texi (-fsave-optimization-record): New option.
14335 * dumpfile.c: Include "optinfo-emit-json.h".
14336 (struct kv_pair): Move to coretypes.h.
14337 (optgroup_options): Make non-static.
14338 (dump_context::end_scope): Call
14339 optimization_records_maybe_pop_dump_scope.
14340 * dumpfile.h (optgroup_options): New decl.
14341 * json.cc: New file.
14342 * json.h: New file.
14343 * optinfo-emit-json.cc: New file.
14344 * optinfo-emit-json.h: New file.
14345 * optinfo.cc: Include "optinfo-emit-json.h".
14346 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
14347 (optinfo_enabled_p): Check optimization_records_enabled_p.
14348 (optinfo_wants_inlining_info_p): Likewise.
14349 * optinfo.h: Update comment.
14350 * profile-count.c (profile_quality_as_string): New function.
14351 * profile-count.h (profile_quality_as_string): New decl.
14352 (profile_count::quality): New accessor.
14353 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
14354 and optinfo_emit_json_cc_tests.
14355 * selftest.h (selftest::json_cc_tests): New decl.
14356 (selftest::optinfo_emit_json_cc_tests): New decl.
14357 * toplev.c: Include "optinfo-emit-json.h".
14358 (compile_file): Call optimization_records_finish.
14359 (do_compile): Call optimization_records_start.
14360 * tree-ssa-live.c: Include optinfo.h.
14361 (remove_unused_scope_block_p): Retain inlining information if
14362 optinfo_wants_inlining_info_p returns true.
14364 2018-07-20 Richard Biener <rguenther@suse.de>
14367 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
14368 to cover -flto-partition=none.
14370 2018-07-20 Martin Liska <mliska@suse.cz>
14372 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
14373 (get_decl_source_range): Remove unused function.
14375 2018-07-20 Richard Biener <rguenther@suse.de>
14377 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
14378 (struct vn_phi_s): Likewise.
14379 (struct vn_reference_s): Likewise.
14380 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
14381 for searching the slot of an entry known to be in the hash itself.
14382 (vn_phi_hasher::equal): Likewise.
14383 (vn_reference_hasher::equal): Likewise.
14384 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
14386 (optimistic_info, current_info): Remove, keeping only valid_info.
14387 (vn_reference_lookup_1): Remove fallback lookup.
14388 (vn_reference_lookup_2): Likewise.
14389 (vn_nary_op_lookup_1): Likewise.
14390 (vn_phi_lookup): Likewise.
14391 (vn_nary_build_or_lookup_1): Make sure to not chain the built
14393 (vn_reference_insert): Adjust, chain the inserted hash element
14394 at last_inserted_ref.
14395 (vn_reference_insert_pieces): Likewise.
14396 (visit_reference_op_call): Likewise.
14397 (vn_nary_op_insert_into): Chain the inserted hash element at
14398 last_inserted_nary.
14399 (vn_nary_op_insert_pieces): Adjust.
14400 (vn_nary_op_insert): Likewise.
14401 (vn_nary_op_insert_stmt): Likewise.
14402 (vn_phi_insert): Adjust, chain the inserted hash element at
14404 (process_scc): Remove clearing and copying the optimistic
14405 table. Instead remove elements inserted during an optimistic
14406 iteration from the single table we maintain.
14407 (init_scc_vn): Adjust.
14408 (free_scc_vn): Likewise.
14409 (sccvn_dom_walker::record_cond): Likewise.
14410 (sccvn_dom_walker::after_dom_children): Likewise.
14412 2018-07-19 Martin Sebor <msebor@redhat.com>
14414 PR tree-optimization/84047
14415 PR tree-optimization/83776
14416 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
14417 (check_array_bounds): Call it.
14419 2018-07-19 Martin Sebor <msebor@redhat.com>
14421 * align.h (align_flags): Use member initialization.
14423 2018-07-19 David Malcolm <dmalcolm@redhat.com>
14425 * Makefile.in (OBJS): Add optinfo.o.
14426 * coretypes.h (class symtab_node): New forward decl.
14427 (struct cgraph_node): New forward decl.
14428 (class varpool_node): New forward decl.
14429 * dump-context.h: New file.
14430 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
14432 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
14433 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
14434 (set_alt_dump_file): Likewise.
14435 (dump_context::~dump_context): New dtor.
14436 (dump_gimple_stmt): Move implementation to...
14437 (dump_context::dump_gimple_stmt): ...this new member function.
14438 Add the stmt to any pending optinfo, creating one if need be.
14439 (dump_gimple_stmt_loc): Move implementation to...
14440 (dump_context::dump_gimple_stmt_loc): ...this new member function.
14441 Start a new optinfo and add the stmt to it.
14442 (dump_gimple_expr): Move implementation to...
14443 (dump_context::dump_gimple_expr): ...this new member function.
14444 Add the stmt to any pending optinfo, creating one if need be.
14445 (dump_gimple_expr_loc): Move implementation to...
14446 (dump_context::dump_gimple_expr_loc): ...this new member function.
14447 Start a new optinfo and add the stmt to it.
14448 (dump_generic_expr): Move implementation to...
14449 (dump_context::dump_generic_expr): ...this new member function.
14450 Add the tree to any pending optinfo, creating one if need be.
14451 (dump_generic_expr_loc): Move implementation to...
14452 (dump_context::dump_generic_expr_loc): ...this new member
14453 function. Add the tree to any pending optinfo, creating one if
14455 (dump_printf): Move implementation to...
14456 (dump_context::dump_printf_va): ...this new member function. Add
14457 the text to any pending optinfo, creating one if need be.
14458 (dump_printf_loc): Move implementation to...
14459 (dump_context::dump_printf_loc_va): ...this new member function.
14460 Start a new optinfo and add the stmt to it.
14461 (dump_dec): Move implementation to...
14462 (dump_context::dump_dec): ...this new member function. Add the
14463 value to any pending optinfo, creating one if need be.
14464 (dump_context::dump_symtab_node): New member function.
14465 (dump_context::get_scope_depth): New member function.
14466 (dump_context::begin_scope): New member function.
14467 (dump_context::end_scope): New member function.
14468 (dump_context::ensure_pending_optinfo): New member function.
14469 (dump_context::begin_next_optinfo): New member function.
14470 (dump_context::end_any_optinfo): New member function.
14471 (dump_context::s_current): New global.
14472 (dump_context::s_default): New global.
14473 (dump_scope_depth): Delete global.
14474 (dumpfile_ensure_any_optinfo_are_flushed): New function.
14475 (dump_symtab_node): New function.
14476 (get_dump_scope_depth): Reimplement in terms of dump_context.
14477 (dump_begin_scope): Likewise.
14478 (dump_end_scope): Likewise.
14479 (selftest::temp_dump_context::temp_dump_context): New ctor.
14480 (selftest::temp_dump_context::~temp_dump_context): New dtor.
14481 (selftest::verify_item): New function.
14482 (ASSERT_IS_TEXT): New macro.
14483 (ASSERT_IS_TREE): New macro.
14484 (ASSERT_IS_GIMPLE): New macro.
14485 (selftest::test_capture_of_dump_calls): New test.
14486 (selftest::dumpfile_c_tests): Call it.
14487 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
14488 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
14489 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
14490 descriptive comment.
14491 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
14492 (dump_node, dump_bb): Move these unrelated decls.
14493 (class dump_manager): Add leading comment.
14494 * optinfo.cc: New file.
14495 * optinfo.h: New file.
14497 2018-07-19 Michael Collison <michael.collison@arm.com>
14498 Richard Henderson <rth@redhat.com>
14500 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
14501 (subti): Handle op1 zero.
14502 (subvti4, usub4ti4): New.
14503 (*sub<GPI>3_compare1_imm): New.
14504 (sub<GPI>3_carryinCV): New.
14505 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
14506 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
14508 2018-07-19 Michael Collison <michael.collison@arm.com>
14509 Richard Henderson <rth@redhat.com>
14511 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
14512 (addti3): Create simpler code if low part is already known to be 0.
14513 (addvti4, uaddvti4): New.
14514 (*add<GPI>3_compareC_cconly_imm): New.
14515 (*add<GPI>3_compareC_cconly): New.
14516 (*add<GPI>3_compareC_imm): New.
14517 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
14518 handle constants within this pattern..
14519 (*add<GPI>3_compareV_cconly_imm): New.
14520 (*add<GPI>3_compareV_cconly): New.
14521 (*add<GPI>3_compareV_imm): New.
14522 (add<GPI>3_compareV): New.
14523 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
14524 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
14525 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
14526 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
14527 with 'comparison' operator.
14528 (*add<GPI>3_compareV_cconly_imm): Ditto.
14529 (*add<GPI>3_compareV_cconly): Ditto.
14530 (*add<GPI>3_compareV_imm): Ditto.
14531 (add<GPI>3_compareV): Ditto.
14532 (add<mode>3_carryinC): Ditto.
14533 (*add<mode>3_carryinC_zero): Ditto.
14534 (*add<mode>3_carryinC): Ditto.
14535 (add<mode>3_carryinV): Ditto.
14536 (*add<mode>3_carryinV_zero): Ditto.
14537 (*add<mode>3_carryinV): Ditto.
14539 2018-07-19 Michael Collison <michael.collison@arm.com>
14540 Richard Henderson <rth@redhat.com>
14542 * config/aarch64/aarch64-modes.def (CC_V): New.
14543 * config/aarch64/aarch64-protos.h
14544 (aarch64_addti_scratch_regs): Declare
14545 (aarch64_subvti_scratch_regs): Declare.
14546 (aarch64_expand_subvti): Declare.
14547 (aarch64_gen_unlikely_cbranch): Declare
14548 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
14549 for signed overflow using CC_Vmode.
14550 (aarch64_get_condition_code_1): Handle CC_Vmode.
14551 (aarch64_gen_unlikely_cbranch): New function.
14552 (aarch64_addti_scratch_regs): New function.
14553 (aarch64_subvti_scratch_regs): New function.
14554 (aarch64_expand_subvti): New function.
14556 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14558 * config/aarch64/aarch64-option-extensions.def: New entry for profile
14560 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
14561 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
14564 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14567 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14568 address check not strict.
14570 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14572 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
14574 (store_pair_lanes<mode>): Likewise.
14575 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
14576 enum value 'ADDR_QUERY_LDP_STP_N'.
14577 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
14578 (aarch64_print_address_internal): Add declaration.
14579 (aarch64_print_ldpstp_address): Remove.
14580 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
14581 (aarch64_print_operand): Change printing of 'y'.
14582 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
14583 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
14584 'true' rather than '1'.
14585 * config/aarch64/constraints.md (Uml): Likewise.
14586 (Uml): Rename to Umn.
14589 2018-07-19 Richard Biener <rguenther@suse.de>
14591 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
14593 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
14594 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
14595 (vn_reference_hasher): Likewise.
14596 (struct vn_tables_s): Remove obstack and alloc-pool members.
14597 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
14598 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
14599 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
14600 (vn_reference_insert_pieces): Likewise.
14601 (alloc_vn_nary_op_noinit): Adjust.
14602 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
14603 (vn_phi_eq): Adjust.
14604 (shared_lookup_phiargs): Remove.
14605 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
14606 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
14607 (visit_reference_op_call): Likewise.
14608 (copy_nary, copy_phi, copy_reference): Remove.
14609 (process_scc): Rewind the obstack when iterating. Do not
14610 copy the elements to valid_info but just move them from one
14611 hashtable to the other.
14612 (allocate_vn_table): Adjust.
14613 (free_vn_table): Likewise.
14614 (init_scc_vn): Likewise.
14615 (free_scc_vn): Likewise.
14617 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
14620 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
14621 indirect_return as function type attribute.
14622 (ix86_attribute_table): Change indirect_return to function
14624 * doc/extend.texi: Update indirect_return attribute.
14626 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
14628 * wide-int.h (widest2_int): New.
14629 * gimple-fold.c (arith_overflowed_p): Use it.
14630 * tree.h (widest2_int_cst): New.
14631 * tree-vrp.c (wide_int_binop_overflow): Rename from
14632 vrp_int_const_binop.
14633 Rewrite to work on trees.
14634 (extract_range_from_multiplicative_op_1): Abstract code to...
14635 (wide_int_range_min_max): ...here.
14636 (wide_int_range_cross_product): ...and here.
14637 (extract_range_from_binary_expr_1): Abstract overflow code to...
14638 (wide_int_range_mult_wrapping): ...here.
14639 * tree-vrp.h (wide_int_range_cross_product): New.
14640 (wide_int_range_mult_wrapping): New.
14642 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
14643 Julia Koval <julia.koval@intel.com>
14645 * config/i386/x86-tune-costs.h (skylake_memcpy,
14646 skylake_memset): Replace rep_prefix with unrolling for size 512.
14648 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
14650 PR middle-end/86544
14651 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
14652 comparision with EQ_EXPR in last stmt.
14654 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
14656 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
14657 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
14658 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
14659 previously known as "PowerPC AltiVec Built-in Functions". Move
14660 some material to new subsubsections "PowerPC AltiVec Built-in
14661 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
14663 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
14664 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
14665 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
14666 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
14668 2018-07-18 Richard Biener <rguenther@suse.de>
14670 PR tree-optimization/86557
14671 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
14674 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
14676 * config/s390/s390.c (s390_function_profiler): Generate CFI.
14678 2018-07-17 Jeff Law <law@redhat.com>
14680 * config/arm/arm.c (get_label_padding): Update for recent
14681 changes to label_to_alignment.
14683 PR tree-optimization/86010
14684 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
14686 * config/mips/mips.c (vr4130_align_insns): Update for recent
14687 changes to label_to_alignment.
14689 * config/frv/frv.c (frv_label_align): Update for recent changes
14690 to label_to_alignment.
14692 * config/nios2/nios2.c (nios2_label_align): Update for recent
14693 changes which dropped ALIGN_LABELS_LOG.
14695 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
14697 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
14698 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
14700 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14702 * config/arc/arc.c (arc_label_align): Use align_labels instead of
14703 deprecated align_labels_log.
14705 2018-07-17 Richard Biener <rguenther@suse.de>
14708 * dwarf2out.c (init_sections_and_labels): Always generate
14709 a debug_line_str_section for early LTO debug.
14710 (dwarf2out_finish): Reset debug_line_str_hash output early.
14711 Bump counter for extra dwarf5 .debug_loc labels to not conflict
14713 (dwarf2out_early_finish): Output debug_line_str.
14715 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14717 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
14718 index register on z196 or later.
14720 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14722 * config/s390/s390.c (s390_default_align): Set default function
14724 (s390_override_options_after_change): Call s390_default align.
14725 (s390_option_override_internal): Call s390_default_align.
14726 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
14728 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14730 PR middle-end/86542
14731 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
14732 to _looptemp_ clauses, other than the first two.
14734 2018-07-17 Martin Liska <mliska@suse.cz>
14736 * opts.c: Do not enable OPT_falign_* for -Os.
14738 2018-07-17 Martin Liska <mliska@suse.cz>
14740 * align.h (MAX_CODE_ALIGN): New.
14741 (MAX_CODE_ALIGN_VALUE): New.
14742 * common/config/i386/i386-common.c (ix86_handle_option):
14743 (MAX_CODE_ALIGN): Moved to align.h.
14744 * final.c (MAX_CODE_ALIGN): Likewise.
14745 * opts.c (parse_and_check_align_values):
14746 (MAX_CODE_ALIGN): Likewise.
14747 (MAX_CODE_ALIGN_VALUE): Likewise.
14749 2018-07-17 Martin Liska <mliska@suse.cz>
14751 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
14752 in order to fulfil coding style.
14753 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
14754 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14755 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14756 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
14757 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
14758 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
14759 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
14760 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
14761 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14763 2018-07-17 Martin Liska <mliska@suse.cz>
14765 * align.h: New file.
14766 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
14768 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
14769 align_flags of label_to_alignment.
14770 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
14772 * config/m68k/m68k.c: Do not use removed align_labels_value and
14774 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
14775 (LOOP_ALIGN): Likewise.
14776 (LABEL_ALIGN): Likewise.
14777 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14778 Remove not used macro.
14779 (rs6000_loop_align): Change return type to align_flags.
14780 (rs6000_loop_align_max_skip): Remove.
14781 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
14782 Change return type to align_flags.
14783 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14784 Remove not used macro.
14785 (rs6000_loop_align): Change return type to align_flags.
14786 (rs6000_loop_align_max_skip): Remove.
14787 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
14788 * config/rx/rx-protos.h (rx_align_for_label): Make it
14790 * config/rx/rx.c (rx_align_for_label): Change return type
14792 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
14794 into align_flags class.
14795 (LABEL_ALIGN): Likewise.
14796 (LOOP_ALIGN): Likewise.
14797 * config/s390/s390.c (s390_label_align): Use align_flags
14799 (s390_asm_output_function_label): Likewise.
14800 * config/sh/sh.c (sh_override_options_after_change):
14801 Use align_flags class directly without macros.
14802 (find_barrier): Likewise.
14803 (barrier_align): Likewise.
14804 (sh_loop_align): Likewise.
14805 * config/spu/spu.c (spu_option_override):
14806 Use align_flags_tuple::get_value instead of removed macros.
14807 (spu_sched_init): Likewise.
14808 * config/spu/spu.h (GTY): Likewise.
14809 * config/visium/visium.c (visium_option_override):
14810 Set "8" as default secondary alignment.
14811 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
14812 in order to guarantee secondary alignment of 8.
14813 * coretypes.h: Include align.h header file.
14814 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
14815 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
14816 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
14817 * doc/tm.texi.in: Likewise.
14818 * final.c (struct label_alignment): Remove not used structure.
14819 (LABEL_ALIGN): Change type to align_flags.
14820 (LOOP_ALIGN): Likewise.
14821 (JUMP_ALIGN): Likewise.
14822 (default_loop_align_max_skip): Remove.
14823 (default_label_align_max_skip): Likewise.
14824 (default_jump_align_max_skip): Likewise.
14825 (default_label_align_after_barrier_max_skip):
14826 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
14827 (LABEL_TO_MAX_SKIP): Remove.
14828 (label_to_alignment): Return align_flags type instead of integer.
14829 (label_to_max_skip): Remove.
14830 (align_fuzz): Use align_flags type.
14831 (compute_alignments): Use align_flags type and use align_flags::max
14832 to combine multiple alignments.
14833 (grow_label_align): Grow vec instead of C array.
14834 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
14835 (shorten_branches): Use align_flags type and use align_flags::max
14836 to combine multiple alignments.
14837 (final_scan_insn_1): Remove usage of secondary alignment that comes
14838 from label alignment, but instead use proper secondary alignment
14839 which is computed in grow_label_align.
14840 * flags.h (struct align_flags_tuple): Move to align.h.
14841 (struct align_flags): Likewise.
14842 (state_align_loops): Rename to align_loops.
14843 (state_align_jumps): Rename to align_jumps.
14844 (state_align_labels): Rename to align_labels.
14845 (state_align_functions): Rename to align_functions.
14846 (align_loops_log): Remove.
14847 (align_jumps_log): Remove.
14848 (align_labels_log): Remove.
14849 (align_functions_log): Remove.
14850 (align_loops_max_skip): Remove.
14851 (align_jumps_max_skip): Remove.
14852 (align_labels_max_skip): Remove.
14853 (align_functions_max_skip): Remove.
14854 (align_loops_value): Remove.
14855 (align_jumps_value): Remove.
14856 (align_labels_value): Remove.
14857 (align_functions_value): Remove.
14858 * output.h (label_to_alignment): Change return type to align_flags.
14859 (label_to_max_skip): Remove.
14860 * target.def: Remove loop_align_max_skip, label_align_max_skip,
14861 jump_align_max_skip macros.
14862 * targhooks.h (default_loop_align_max_skip): Remove.
14863 (default_label_align_max_skip): Likewise.
14864 (default_jump_align_max_skip): Likewise.
14865 (default_label_align_after_barrier_max_skip): Remove.
14866 * toplev.c (read_log_maxskip): Use ::normalize function.
14867 (parse_N_M): Remove not used argument and also call ::normalize.
14868 (parse_alignment_opts): Do not pass unused arguments.
14869 * varasm.c (assemble_start_function): Use directly align_functions
14870 instead of removed macros.
14871 * system.h: Do not poison removed macros.
14873 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14875 PR middle-end/86539
14876 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
14877 and cond temporaries don't have reference type if iterator has
14878 pointer type. For init use &for_pre_body instead of pre_p if
14879 for_pre_body is non-empty.
14881 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14883 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
14884 double-double modes to SFmode directly directly.
14885 (trunc<mode>sf2_fprs): Delete.
14887 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14889 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
14890 for conversions between IFmode and the decimal floating point modes.
14891 (init_float128_ieee): Use the correct names for conversions between
14892 KFmode and the decimal floating point modes.
14894 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14896 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
14897 for the conversions between TDmode and IFmode.
14898 (init_float128_ieee): Use more correct names for the conversions
14899 between TDmode and KFmode.
14901 2018-07-16 Jakub Jelinek <jakub@redhat.com>
14903 PR tree-optimization/86526
14904 * builtins.c (expand_builtin_memcmp): Formatting fixes.
14905 (inline_expand_builtin_string_cmp): Likewise.
14906 (inline_string_cmp): Likewise. Use c_readstr instead of
14907 builtin_memcpy_read_str. Add unit_mode temporary.
14909 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
14911 PR middle-end/86528
14912 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
14913 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
14915 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
14917 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
14918 Alphabetize prototypes of built-in functions, separating out
14919 built-in functions that are listed in this section but should be
14920 described elsewhere.
14922 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
14925 * expmed.c (emit_store_flag): Do not emit setcc followed by a
14926 conditional move when trapping comparison was split to a
14927 non-trapping one (and vice versa).
14929 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14931 * config/s390/s390.c (s390_function_profiler): Generate nops
14932 instead of profiler call sequences.
14933 * config/s390/s390.opt: Add the new option.
14935 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14937 * config/s390/s390.c (s390_function_profiler): Generate
14938 __mcount_loc section.
14939 * config/s390/s390.opt: Add the new option.
14941 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14943 * common.opt: Add the new warning.
14944 * config/s390/s390.c (s390_function_profiler): Emit "brasl
14945 %r0,__fentry__" when -mfentry is specified.
14946 (s390_option_override_internal): Disallow -mfentry for 31-bit
14948 * config/s390/s390.opt: Add the new option.
14950 2018-07-16 Richard Biener <rguenther@suse.de>
14953 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
14954 for function-local FUNCTION_DECL and RESULT_DECL immediately.
14956 2018-07-16 Martin Liska <mliska@suse.cz>
14959 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
14962 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14964 * config/arc/arcHS.md: Update ARCHS scheduling rules.
14966 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14968 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
14970 * config/arc/arc-cpus.def (hs4x): New cpu.
14972 * config/arc/arc-tables.opt: Regenerate.
14973 * config/arc/arc.c (arc_sched_issue_rate): New function.
14974 (TARGET_SCHED_ISSUE_RATE): Define.
14975 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
14976 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
14978 (attr tune): Add ARCHS4x tune values.
14979 (attr tune_dspmpy): Define.
14980 (*tst): Correct instruction type.
14981 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
14982 * config/arc/arcHS4x.md: New file.
14983 * config/arc/fpu.md: Update instruction type attributes.
14984 * config/arc/t-multilib: Regenerate.
14986 2018-07-16 Tom de Vries <tdevries@suse.de>
14989 * var-tracking.c (vt_initialize): Fix pre_dec handling.
14991 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14993 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
14996 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
14998 PR tree-optimization/86514
14999 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
15000 conversion to a boolean type from a type with greater precision.
15002 2018-07-16 Tom de Vries <tdevries@suse.de>
15004 * var-tracking.c (vt_initialize): Print adjusted insn slim if
15005 dump_flags request TDF_SLIM.
15007 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
15009 * fold-const.c (int_const_binop_1): Abstract...
15010 (wide_int_binop): ...wide int code here.
15011 (poly_int_binop): ...poly int code here.
15012 Abstract the rest of int_const_binop_1 into int_const_binop.
15013 * fold-const.h (wide_int_binop): New.
15014 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
15015 Remove useless PLUS/MINUS_EXPR case.
15016 (zero_nonzero_bits_from_vr): Move wide int code...
15017 (zero_nonzero_bits_from_bounds): ...here.
15018 (extract_range_from_binary_expr_1): Move mask optimization code...
15019 (range_easy_mask_min_max): ...here.
15020 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
15021 (range_easy_mask_min_max): New.
15023 2018-07-15 Jeff Law <law@redhat.com>
15026 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
15029 2018-07-14 Jim Wilson <jimw@sifive.com>
15031 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
15033 2018-07-14 Paul Koning <ni1d@arrl.net>
15035 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
15037 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
15039 * lto-streamer-out.c (copy_function_or_variable): Dump info about
15042 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
15043 Steve Munroe <munroesj52@gmail.com>
15045 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
15046 (_mm_andnot_si128): Likewise.
15047 (_mm_or_si128): Likewise.
15048 (_mm_xor_si128): Likewise.
15050 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
15052 PR middle-end/78809
15053 * builtins.c (expand_builtin_memcmp): Inline the calls first
15054 when result_eq is false.
15055 (expand_builtin_strcmp): Inline the calls first.
15056 (expand_builtin_strncmp): Likewise.
15057 (inline_string_cmp): New routine. Expand a string compare
15058 call by using a sequence of char comparison.
15059 (inline_expand_builtin_string_cmp): New routine. Inline expansion
15060 a call to str(n)cmp/memcmp.
15061 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
15063 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
15065 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
15067 * config/arm/driver-arm.c: Include arm-native.h.
15068 (host_detect_local_cpu): Use auto-generated data tables.
15069 (vendors, arm_cpu_table): Delete. Move part information to ...
15070 * config/arm/arm-cpus.in: ... here.
15071 * config/arm/parsecpu.awk (gen_native): New function.
15072 (vendor, part): New CPU fields.
15073 (END): Add support for building the native CPU detection tables.
15074 * config/arm/t-arm (arm-native.h): Add build rule.
15075 (driver-arm.o): Add dependency on arm-native.h.
15077 2018-07-13 Richard Biener <rguenther@suse.de>
15079 PR middle-end/85974
15080 * match.pd (addr1 - addr2): Allow either of the operand to
15083 2018-07-13 Tom de Vries <tdevries@suse.de>
15085 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
15088 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
15090 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
15091 arrays instead of numbered variables.
15093 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
15095 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
15096 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
15098 2018-07-13 Richard Biener <rguenther@suse.de>
15101 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
15102 instead of get_context_die.
15104 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15105 Richard Biener <rguenther@suse.de>
15107 PR middle-end/86489
15108 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
15109 that the loop latch destination where phi is defined.
15111 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
15113 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
15114 (riscv_expand_epilogue): Add assertion to check interrupt mode.
15115 (riscv_set_current_function): Extract getting interrupt type to new
15117 (riscv_get_interrupt_type): New function.
15118 (riscv_merge_decl_attributes): New function, checking interrupt type is
15120 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
15122 2018-07-12 Paul Koning <ni1d@arrl.net>
15124 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
15127 2018-07-12 Paul Koning <ni1d@arrl.net>
15129 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
15130 zero reference, add doloop_end instead.
15131 * doc/md.texi (decrement_and_branch_until_zero): Remove.
15132 (Looping patterns): Remove decrement_and_branch_until_zero. Add
15133 detail for doloop_end.
15135 2018-07-12 Martin Sebor <msebor@redhat.com>
15138 * attribs.c (decl_attributes): Reject conflicting attributes before
15139 calling attribute handlers.
15141 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
15143 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
15145 (gcc::dump_manager::get_dump_file_name): likewise.
15146 (dump_begin): Likewise.
15147 * dumpfile.h (dump_begin): Update prototype.
15148 (gcc::dump_manager::get_dump_file_name,
15149 gcc::dump_manager::get_dump_file_name): Update prototype.
15151 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15153 * internal-fn.h (vectorizable_internal_fn_p): New function.
15154 * tree-vect-slp.c (compatible_calls_p): Likewise.
15155 (vect_build_slp_tree_1): Remove nops argument. Handle calls
15156 to internal functions.
15157 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
15159 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15161 * fold-const.h (inverse_conditions_p): Declare.
15162 * fold-const.c (inverse_conditions_p): New function.
15163 * match.pd: Use inverse_conditions_p. Add folds of view_converts
15164 that test the inverse condition of a conditional internal function.
15165 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
15166 * internal-fn.c (internal_fn_mask_index): Handle conditional
15167 internal functions.
15168 (vectorized_internal_fn_supported_p): New function.
15169 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
15170 (any_pred_load_store): Replace with...
15171 (need_to_predicate): ...this new variable.
15172 (redundant_ssa_names): New variable.
15173 (ifcvt_can_use_mask_load_store): Move initial checks to...
15174 (ifcvt_can_predicate): ...this new function. Handle tree codes
15175 for which a conditional internal function exists.
15176 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
15177 instead of ifcvt_can_use_mask_load_store. Update after variable
15179 (predicate_load_or_store): New function, split out from
15180 predicate_mem_writes.
15181 (check_redundant_cond_expr): New function.
15182 (value_available_p): Likewise.
15183 (predicate_rhs_code): Likewise.
15184 (predicate_mem_writes): Rename to...
15185 (predicate_statements): ...this. Use predicate_load_or_store
15186 and predicate_rhs_code.
15187 (combine_blocks, tree_if_conversion): Update after above name changes.
15188 (ifcvt_local_dce): Handle redundant_ssa_names.
15189 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
15190 general conditional functions.
15191 * tree-vect-stmts.c (vectorizable_call): Likewise.
15193 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15194 Alan Hayward <alan.hayward@arm.com>
15195 David Sherwood <david.sherwood@arm.com>
15197 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
15198 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
15199 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
15200 plus and minus and convert them into IFN_COND_FMA-based sequences.
15201 (convert_mult_to_fma): Handle conditional plus and minus.
15203 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15205 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
15206 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
15207 (cond_fnms_optab): New optabs.
15208 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
15209 internal functions.
15210 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
15211 * internal-fn.h (get_conditional_internal_fn): Declare.
15212 (get_unconditional_internal_fn): Likewise.
15213 * internal-fn.c (cond_ternary_direct): New macro.
15214 (expand_cond_ternary_optab_fn): Likewise.
15215 (direct_cond_ternary_optab_supported_p): Likewise.
15216 (FOR_EACH_COND_FN_PAIR): Likewise.
15217 (get_conditional_internal_fn): New function.
15218 (get_unconditional_internal_fn): Likewise.
15219 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
15220 (gimple_match_op::gimple_match_op): Add a new overload for 5
15222 (gimple_match_op::set_op): Likewise.
15223 (gimple_resimplify5): Declare.
15224 * genmatch.c (decision_tree::gen): Generate simplifications for
15226 * gimple-match-head.c (gimple_simplify): Define an overload for
15227 5 operands. Handle calls with 5 arguments in the top-level overload.
15228 (convert_conditional_op): Handle conversions from unconditional
15229 internal functions to conditional ones.
15230 (gimple_resimplify5): New function.
15231 (build_call_internal): Pass a fifth operand.
15232 (maybe_push_res_to_seq): Likewise.
15233 (try_conditional_simplification): Try converting conditional
15234 internal functions to unconditional internal functions.
15235 Handle 3-operand unconditional forms.
15236 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
15237 Define ternary equivalents of the current rules for binary conditional
15238 internal functions.
15239 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
15240 ternary operations.
15241 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
15242 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
15243 (optab): Handle them.
15244 (SVE_COND_FP_TERNARY): New int iterator.
15245 (sve_fmla_op, sve_fmad_op): New int attributes.
15246 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
15247 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
15248 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
15250 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15252 * target.def (preferred_else_value): New target hook.
15253 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
15254 * doc/tm.texi: Regenerate.
15255 * targhooks.h (default_preferred_else_value): Declare.
15256 * targhooks.c (default_preferred_else_value): New function.
15257 * internal-fn.h (conditional_internal_fn_code): Declare.
15258 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
15259 (get_conditional_internal_fn): Use it.
15260 (conditional_internal_fn_code): New function.
15261 * gimple-match.h (gimple_match_cond): New struct.
15262 (gimple_match_op): Add a cond member function.
15263 (gimple_match_op::gimple_match_op): Update all forms to take a
15265 * genmatch.c (expr::gen_transform): Use the same condition as res_op
15266 for the suboperation, but don't specify a particular else_value.
15267 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
15268 (visit_nary_op, visit_reference_op_load): Pass
15269 gimple_match_cond::UNCOND to the gimple_match_op constructor.
15270 * gimple-match-head.c: Include tree-eh.h
15271 (convert_conditional_op): New function.
15272 (maybe_resimplify_conditional_op): Likewise.
15273 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
15274 (gimple_resimplify2): Likewise.
15275 (gimple_resimplify3): Likewise.
15276 (gimple_resimplify4): Likewise.
15277 (maybe_push_res_to_seq): Return null for conditional operations.
15278 (try_conditional_simplification): New function.
15279 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
15281 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
15283 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
15285 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
15287 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
15289 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
15291 (hash_tree): Do not hash DECL_FCONTEXT
15292 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
15293 Do not stream DECL_FCONTEXT.
15294 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
15295 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
15297 2018-07-12 Richard Biener <rguenther@suse.de>
15300 * dwarf2out.c (gen_block_die): Only output blocks when they have
15301 at least one !DECL_IGNORED_P variable.
15303 2018-07-12 Richard Biener <rguenther@suse.de>
15306 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
15307 Remove -mieee-fp handling.
15309 2018-07-12 Richard Biener <rguenther@suse.de>
15311 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
15312 left-over from last patch.
15314 2018-07-12 Jakub Jelinek <jakub@redhat.com>
15316 PR tree-optimization/86492
15317 * gimple-ssa-store-merging.c
15318 (imm_store_chain_info::coalesce_immediate_stores): Call
15319 check_no_overlap even for the merge_overlapping case. Formatting fix.
15321 2018-07-12 Richard Biener <rguenther@suse.de>
15323 PR middle-end/86479
15324 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
15325 move possibly trapping operations into the conditional.
15327 2018-07-12 Richard Biener <rguenther@suse.de>
15329 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
15330 (vn_lookup_simplify_result): Remove recursion limit applied
15332 (vn_nary_build_or_lookup_1): Adjust.
15333 (try_to_simplify): Likewise.
15334 * gimple-match-head.c (gimple_resimplify1): Instead apply one
15336 (gimple_resimplify2): Likewise.
15337 (gimple_resimplify3): Likewise.
15338 (gimple_resimplify4): Likewise.
15340 2018-07-11 Jakub Jelinek <jakub@redhat.com>
15342 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
15343 Use __mmask64 type instead of __mmask8 for __M argument.
15344 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
15345 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
15347 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
15348 __mmask16 for __M argument.
15349 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
15350 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
15351 to __mmask16 instead of __mmask8.
15352 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
15353 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
15354 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
15355 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
15356 instead of __mmask16 for __U argument.
15357 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
15358 __mmask16 instead of __mmask8 for __U argument.
15359 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
15361 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
15363 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
15365 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
15367 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
15369 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
15371 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
15373 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
15375 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
15377 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
15379 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
15380 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
15381 return type as well as __M argument type and all casts from __mmask8
15383 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
15384 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
15385 return type as well as __M argument type and all casts from __mmask8
15387 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
15388 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
15389 return type as well as __M argument type and all casts from __mmask8
15391 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
15392 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
15393 return type as well as __M argument type and all casts from __mmask8
15395 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
15396 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
15399 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
15401 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
15402 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
15405 2018-07-11 Paul Koning <ni1d@arrl.net>
15407 * doc/md.texi (define_subst): Document how multiple occurrences of
15408 the same argument in the replacement pattern are handled.
15410 2018-07-11 Paul Koning <ni1d@arrl.net>
15412 * doc/extend.texi (Common Variable Attributes): Move "mode" into
15413 alphabetical order.
15414 (Common Type Attributes): Add "mode" attribute.
15416 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
15418 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
15419 stream DECL_ORIGINAL_TYPE.
15420 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
15421 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
15422 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
15423 Do not walk original type.
15424 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
15426 (write_ts_decl_non_common_tree_pointers): Do not stream
15428 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
15429 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
15431 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
15433 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
15434 thread twice from the same starting edge.
15436 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
15438 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
15439 * gimple.c (gimple_call_nonnull_result_p): ...here...
15440 (gimple_call_nonnull_arg): ...and here.
15441 * gimple.h (gimple_call_nonnull_result_p): New.
15442 (gimple_call_nonnull_arg): New.
15444 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
15446 * config/arm/arm-cpus.in: Move information from fpu field of each
15447 cpu definition to the isa field.
15448 * config/arm/parsecpu.awk (fpu): Delete match rule.
15449 (gen_comm_data): Don't add bits from the CPU's FPU entry.
15451 2018-07-11 Richard Biener <rguenther@suse.de>
15454 * dwarf2out.c (init_sections_and_labels): Use
15455 output_asm_line_debug_info consistently.
15456 (dwarf2out_early_finish): Likewise.
15457 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
15460 2018-07-11 Richard Biener <rguenther@suse.de>
15462 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
15463 Rework father_bb setting in a way to avoid propagating constants
15464 multiple times on a loop body.
15466 2018-07-10 Mark Wielaard <mark@klomp.org>
15469 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
15471 2018-07-10 Richard Biener <rguenther@suse.de>
15473 * hash-map.h (hash_map::iterator::operator*): Return
15474 references to key and value.
15476 2018-07-10 Jakub Jelinek <jakub@redhat.com>
15479 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
15480 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
15481 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
15482 (gimplify_omp_for): For composite loops, move outer
15483 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
15484 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
15485 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
15486 TREE_LIST for both the original class iterator and the "last" helper
15487 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
15488 loop, remember has_decl_expr from outer composite loops for the
15489 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
15491 2018-07-09 Martin Sebor <msebor@redhat.com>
15493 PR middle-end/77357
15494 PR middle-end/86428
15495 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
15496 accessing implicitly initialized array elements.
15497 * expr.c (string_constant): Handle string initializers of
15498 character arrays within aggregates.
15499 * gimple-fold.c (fold_array_ctor_reference): Add argument.
15500 Store element offset. As a special case, handle zero size.
15501 (fold_nonarray_ctor_reference): Same.
15502 (fold_ctor_reference): Add argument. Store subobject offset.
15503 * gimple-fold.h (fold_ctor_reference): Add argument.
15505 2018-07-09 Paul Koning <ni1d@arrl.net>
15507 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
15508 (pdp11_insn_cost): New function.
15509 (pdp11_md_asm_adjust): New function.
15510 (TARGET_INVALID_WITHIN_DOLOOP): Define.
15511 (pdp11_rtx_costs): Update to match machine better.
15512 (output_addr_const_pdp11): Correct format mismatch warnings.
15513 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
15514 * config/pdp11/pdp11.md: General change to add base_cost and/or
15515 length attributes for use by new pdp11_insn_cost function.
15516 (MIN_BRANCH): Correct definition.
15518 (doloop_end): Use standard pattern name for looping pattern.
15519 (doloop_end_nocc): New.
15520 (movsf): Add another constraint alternative.
15521 (zero_extendqihi2): Add constraint alternatives for not in place
15523 (zero_extendhisi2): Remove.
15524 (shift patterns): Add CC handling variants.
15528 (define_peephole2): New peephole to recognize mov that sets CC for
15531 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15533 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
15534 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
15537 2018-07-09 Jakub Jelinek <jakub@redhat.com>
15540 * real.c (real_nextafter): Return true if result is denormal.
15542 2018-07-09 Martin Liska <mliska@suse.cz>
15544 * common.opt: Add back wrongly removed attribute.
15546 2018-07-09 Richard Biener <rguenther@suse.de>
15549 * dwarf2out.c (gen_block_die): For an early generated DIE
15550 always output high/low PC attributes.
15552 2018-07-09 Tom de Vries <tdevries@suse.de>
15554 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
15555 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
15556 onto VAR_DECL with abstract origin.
15558 2018-07-07 Jim Wilson <jimw@sifive.com>
15560 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
15562 2018-07-07 Tom de Vries <tdevries@suse.de>
15564 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
15566 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
15568 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
15570 (combine_bound): Use wide-int overflow calculation instead of
15572 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
15574 * fold-const.c (int_const_binop_2): Same.
15575 (extract_muldiv_1): Same.
15576 (fold_div_compare): Same.
15577 (fold_abs_const): Same.
15579 * poly-int.h (add): Same.
15583 * predict.c (predict_iv_comparison): Same.
15584 * profile-count.c (slow_safe_scale_64bit): Same.
15585 * simplify-rtx.c (simplify_const_binary_operation): Same.
15586 * tree-chrec.c (tree_fold_binomial): Same.
15587 * tree-data-ref.c (split_constant_offset_1): Same.
15588 * tree-if-conv.c (idx_within_array_bound): Same.
15589 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
15590 * tree-ssa-phiopt.c (minmax_replacement): Same.
15591 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
15592 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
15593 * vr-values.c (vr_values::adjust_range_with_scev): Same.
15594 * wide-int.cc (wi::add_large): Same.
15595 (wi::mul_internal): Same.
15596 (wi::sub_large): Same.
15597 (wi::divmod_internal): Same.
15598 * wide-int.h: Change overflow type to overflow_type for neg, add,
15599 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
15600 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
15601 mul_internal, divmod_internal.
15602 (overflow_type): New enum.
15603 (accumulate_overflow): New.
15605 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
15607 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
15608 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
15610 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
15612 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
15613 argument is checked for zero before entering loop, avoid checking again.
15615 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
15617 * gimplify.h (generic_expr_could_trap_p): Set as global function.
15618 * gimplify.h (generic_expr_could_trap_p): Likwise.
15619 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
15621 2018-07-06 Jakub Jelinek <jakub@redhat.com>
15623 PR tree-optimization/86401
15624 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
15625 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
15626 (fold_bit_and_mask): ... here. New helper function for match.pd.
15627 * fold-const.h (fold_bit_and_mask): Declare.
15628 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
15630 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
15633 * target.def (translate_mode_attribute): New hook.
15634 * targhooks.h (default_translate_mode_attribute): Declare.
15635 * targhooks.c (default_translate_mode_attribute): New function.
15636 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
15637 * doc/tm.texi: Regenerate.
15638 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
15639 (rs6000_translate_mode_attribute): New function.
15641 2018-07-06 Paul Koning <ni1d@arrl.net>
15643 * doc/md.texi (define_split): Document DONE and FAIL.
15644 (define_peephole2): Ditto.
15646 2018-07-05 Jeff Law <law@redhat.com>
15648 PR tree-optimization/86010
15649 * tree-ssa-dse.c (compute_trims): More aggressively trim at
15650 both the head and tail of mem* and str* calls.
15652 2018-07-05 Jim Wilson <jimw@sifive.com>
15654 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
15656 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
15658 * config/aarch64/aarch64-simd.md: correct flags text for
15659 MIN_EXPR replacement.
15661 2018-07-05 James Clarke <jrtc27@jrtc27.com>
15663 * configure: Regenerated.
15665 2018-07-05 Carl Love <cel@us.ibm.com>
15667 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
15668 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
15669 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
15670 VSX_BUILTIN_DOUBLEL_V4SF.
15672 2018-07-05 Martin Sebor <msebor@redhat.com>
15675 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
15676 than its domain to compute its the upper bound of a char array.
15678 2018-07-05 Nathan Sidwell <nathan@acm.org>
15680 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
15681 * doc/cpp.texi: Update comment.
15682 * doc/tm.texi: Rebuilt.
15683 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
15684 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
15685 * doc/extend.texi (Backwards Compatibility): Clarify it is system
15686 headers affected by extern "C".
15687 * system.h: Poison NO_IMPLICIT_EXTERN_C.
15688 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
15689 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
15690 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
15691 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
15692 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
15693 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
15694 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
15695 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
15696 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
15697 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
15698 config/sparc/sp64-elf.h, config/spu/spu.h,
15699 config/stormy16/stormy16.h, config/v850/v850.h,
15700 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
15701 define NO_IMPLICIT_EXTERN_C.
15702 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
15704 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15707 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
15708 instead of GET_MODE_SIZE when comparing Units.
15710 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15713 * rtlanal.c (set_noop_p): Constrain on mode change,
15714 include hard-reg-set.h
15716 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15718 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
15720 2018-07-05 Jakub Jelinek <jakub@redhat.com>
15723 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15726 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15728 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15730 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15733 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15735 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15737 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
15739 PR middle-end/86380
15740 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
15742 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
15744 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
15745 neg_*_op* variables.
15747 2018-07-04 Martin Liska <mliska@suse.cz>
15749 * tree-switch-conversion.c: Define
15750 max_ratio_for_speed and max_ratio_for_size constants.
15752 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15753 Martin Liska <mliska@suse.cz>
15755 PR middle-end/66240
15758 * common.opt: Rename align options with 'str_' prefix.
15759 * common/config/i386/i386-common.c (set_malign_value): New
15761 (ix86_handle_option): Use it to set -falign-* options/
15762 * config/aarch64/aarch64-protos.h (struct tune_params): Change
15763 type from int to string.
15764 * config/aarch64/aarch64.c: Update default values from int
15766 * config/alpha/alpha.c (alpha_override_options_after_change):
15768 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
15769 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15770 max skip conditionally.
15771 * config/i386/freebsd.h (SUBALIGN_LOG): New.
15772 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15773 max skip conditionally.
15774 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15775 max skip conditionally.
15776 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
15777 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15778 max skip conditionally.
15779 * config/i386/i386.c (struct ptt): Change type from int to
15781 (ix86_default_align): Set default values.
15782 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
15783 max skip conditionally.
15784 * config/i386/iamcu.h (SUBALIGN_LOG): New.
15785 (ASM_OUTPUT_MAX_SKIP_ALIGN):
15786 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
15787 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15788 max skip conditionally.
15789 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
15790 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
15791 * config/i386/x86-64.h (SUBALIGN_LOG): New.
15792 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15793 max skip conditionally.
15794 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15795 * config/ia64/ia64.c (ia64_option_override): Set default values
15796 for alignment options.
15797 * config/m68k/m68k.c: Handle new str_align_* options.
15798 * config/mips/mips.c (mips_set_compression_mode): Change
15800 (mips_option_override): Set default values for options.
15801 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
15803 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15805 * config/rx/rx.c (rx_option_override): Likewise.
15806 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
15807 (LABEL_ALIGN): Use align_labels_log.
15808 (LOOP_ALIGN): Use align_loops_align.
15809 * config/s390/s390.c (s390_asm_output_function_label): Use new
15811 * config/sh/sh.c (sh_override_options_after_change):
15812 Change type of constants.
15813 * config/spu/spu.c (spu_sched_init): Likewise.
15814 * config/sparc/sparc.c (sparc_option_override): Set default
15815 values for options.
15816 * config/visium/visium.c (visium_option_override): Likewise.
15817 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
15818 emit p2align format with last argument if it's not needed.
15819 * doc/invoke.texi: Document extended format of -falign-*.
15820 * final.c: Use align_labels alignment.
15821 * flags.h (struct target_flag_state): Change type to use
15823 (struct align_flags_tuple): New.
15824 (struct align_flags): Likewise.
15825 (align_loops_log): Redefine macro to use new types.
15826 (align_loops_max_skip): Redefine macro to use new types.
15827 (align_jumps_log): Redefine macro to use new types.
15828 (align_jumps_max_skip): Redefine macro to use new types.
15829 (align_labels_log): Redefine macro to use new types.
15830 (align_labels_max_skip): Redefine macro to use new types.
15831 (align_functions_log): Redefine macro to use new types.
15832 (align_loops): Redefine macro to use new types.
15833 (align_jumps): Redefine macro to use new types.
15834 (align_labels): Redefine macro to use new types.
15835 (align_functions): Redefine macro to use new types.
15836 (align_functions_max_skip): Redefine macro to use new types.
15837 (align_loops_value): New macro.
15838 (align_jumps_value): New macro.
15839 (align_labels_value): New macro.
15840 (align_functions_value): New macro.
15841 * function.c (invoke_set_current_function_hook): Propagate
15842 alignment values from flags to global variables default in
15844 * ipa-icf.c (sem_function::equals_wpa): Use
15845 cl_optimization_option_eq instead of memcmp.
15846 * lto-streamer.h (cl_optimization_stream_out): Support streaming
15848 (cl_optimization_stream_in): Likewise.
15849 * optc-save-gen.awk: Support strings in cl_optimization.
15850 * opth-gen.awk: Likewise.
15851 * opts.c (finish_options): Remove error checking of invalid
15853 (MAX_CODE_ALIGN): Remove.
15854 (MAX_CODE_ALIGN_VALUE): Likewise.
15855 (parse_and_check_align_values): New function.
15856 (check_alignment_argument): Likewise.
15857 (common_handle_option): Use check_alignment_argument.
15858 * opts.h (parse_and_check_align_values): Declare.
15859 * toplev.c (init_alignments): Remove.
15860 (read_log_maxskip): New.
15861 (parse_N_M): Likewise.
15862 (parse_alignment_opts): Likewise.
15863 (backend_init_target): Remove usage of init_alignments.
15864 * toplev.h (parse_alignment_opts): Declare.
15865 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
15867 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
15868 * tree.c (cl_option_hasher::equal): New.
15869 * varasm.c: Use new global macros.
15871 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15873 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
15874 Use a simpler align directive also if MAXSKIP = ALIGN-1.
15875 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15876 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15877 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15878 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15879 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
15880 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
15881 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
15882 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15883 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15884 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15885 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15887 2018-07-04 Martin Liska <mliska@suse.cz>
15888 Jonathan Wakely <jwakely@redhat.com>
15890 * coverage.c: Use correct type.
15891 * doc/invoke.texi: Language correction.
15893 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
15896 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
15897 ENDBRANCH for non-tail call which may return via indirect branch.
15898 * doc/extend.texi: Document indirect_return attribute.
15900 2018-07-03 Martin Sebor <msebor@redhat.com>
15902 PR tree-optimization/86274
15903 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
15905 (format_floating): Correct handling of infinities and NaNs.
15907 2018-07-03 Martin Sebor <msebor@redhat.com>
15909 * print-tree.c (print_real_cst): New function.
15910 (print_node_brief): Call it.
15911 (print_node): Ditto.
15913 2018-07-03 Jeff Law <law@redhat.com>
15915 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
15916 into a single pattern.
15918 * config/h8300/h8300.md (ors code_iterator): New.
15919 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
15920 a single pattern and single splitter.
15921 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
15922 (iorqi3_1, xorqi3_1): Likewise.
15923 (iorqi3, xorqi3 expanders): Similarly.
15925 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
15926 (movmd_internal) into a single pattern using the P mode iterator.
15927 (movmd splitters): Similarly.
15928 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
15929 (movsd splitters): Similarly.
15931 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
15932 ADDB, ADDW and ADDL into a single ADD attribute which selects the
15933 right table based on the size of the operand.
15934 * config/h8300/h8300.md (length_table): Corresponding changes. All
15935 references to "addb", "addw" and "addl" changed to "add".
15936 (btst patterns): Merge two variants into a single pattern.
15937 (tstqi, tsthi): Likewise.
15938 (addhi3_incdec, addsi3_incdec): Likewise.
15939 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
15940 (mulhi3, mulsi3): Likewise.
15941 (udivhi3, udivsi3): Likewise.
15942 (divhi3, divsi3): Likewise.
15943 (andorqi3, andorhi3, andorsi3): Likewise.
15945 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
15948 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
15949 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
15951 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15953 PR tree-optimization/85694
15954 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
15955 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
15957 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
15958 (<u>avg<mode>3_ceil): New patterns.
15960 2018-07-03 David Malcolm <dmalcolm@redhat.com>
15962 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
15963 scan-tree-dump directive.
15964 * gcc.dg/vect/slp-perm-2.c: Likewise.
15965 * gcc.dg/vect/slp-perm-3.c: Likewise.
15966 * gcc.dg/vect/slp-perm-5.c: Likewise.
15967 * gcc.dg/vect/slp-perm-6.c: Likewise.
15968 * gcc.dg/vect/slp-perm-7.c: Likewise.
15969 * gcc.dg/vect/slp-perm-8.c: Likewise.
15971 2018-07-03 Marek Polacek <polacek@redhat.com>
15973 PR middle-end/86202
15974 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
15976 2018-07-03 Richard Biener <rguenther@suse.de>
15979 * tree-ssa-structalias.c (find_func_clobbers): Properly
15980 handle indirect calls.
15982 2018-07-03 Jeff Law <law@redhat.com>
15984 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
15985 (shifts): New code iterator.
15986 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
15987 expander. Fix HImode handling on H8/SX.
15988 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
15989 (subqi3, subhi3, subsi3 expanders): Likewise.
15990 (andqi3, andhi3, andsi3 expanders): Likewise.
15991 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
15992 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
15993 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
15994 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
15995 (zero_extendqihi2, zero_extendqisi2): Likewise.
15996 (extendqihi2, extendqisi2): Likewise.
15997 (rotlqi3, rotlhi3, rotlsi3): Likewise.
15998 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
15999 (rotlqi3_1, rotlhi3_1): Likewise.
16000 (logicalhi3_sn, logicalsi3_sn): Likewise.
16001 (logicalhi3, logicalsi3): Likewise.
16003 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16005 * tree-vect-patterns.c (vect_recog_rotate_pattern)
16006 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
16007 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
16008 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
16009 type to append_pattern_def_seq instead of creating a stmt_vec_info
16011 (build_mask_conversion): Likewise. Remove vinfo argument.
16012 (vect_add_conversion_to_patterm): Likewise, renaming to...
16013 (vect_add_conversion_to_pattern): ...this.
16014 (vect_recog_mask_conversion_pattern): Update call to
16015 build_mask_conversion. Pass the vector type to
16016 append_pattern_def_seq here too.
16017 (vect_recog_gather_scatter_pattern): Update call to
16018 vect_add_conversion_to_pattern.
16020 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16022 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
16023 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16024 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
16025 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
16026 STMT_VINFO_PATTERN_DEF_SEQ to null here.
16027 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
16028 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
16029 append_pattern_def_seq instead of new_pattern_def_seq.
16030 (vect_recog_divmod_pattern): Do both of the above.
16031 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
16034 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16036 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
16037 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
16038 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
16039 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
16040 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
16041 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
16042 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
16043 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
16044 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
16045 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
16046 parameter with a single stmt_vec_info.
16047 (vect_recog_func_ptr): Likewise.
16048 (vect_recog_gather_scatter_pattern): Likewise, folding in...
16049 (vect_try_gather_scatter_pattern): ...this.
16050 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
16051 the stmt_vec_info of the statement to be matched. Don't clear
16052 STMT_VINFO_RELATED_STMT.
16053 (vect_pattern_recog): Update call accordingly.
16055 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16057 PR tree-optimization/85694
16058 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
16059 (uavgM3_ceil): Document new optabs.
16060 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
16061 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
16063 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
16064 (savg_ceil_optab): New optabs.
16065 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
16066 (vect_vect_recog_func_ptrs): Add it.
16067 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
16068 constant directly from the associated lhs.
16070 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16072 * tree-vect-patterns.c (vect_split_statement): New function.
16073 (vect_convert_input): Use it to try to split an existing cast.
16075 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16077 * poly-int.h (print_hex): New function.
16078 * dumpfile.h (dump_dec, dump_hex): Declare.
16079 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
16080 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
16081 min_input_precision, operation_precision and operation_sign.
16082 * tree-vect-patterns.c (vect_get_range_info): New function.
16083 (vect_same_loop_or_bb_p, vect_single_imm_use)
16084 (vect_operation_fits_smaller_type): Delete.
16085 (vect_look_through_possible_promotion): Add an optional
16086 single_use_p parameter.
16087 (vect_recog_over_widening_pattern): Rewrite to use new
16088 stmt_vec_info infomration. Handle one operation at a time.
16089 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
16090 (vect_truncatable_operation_p, vect_set_operation_type)
16091 (vect_set_min_input_precision): New functions.
16092 (vect_determine_min_output_precision_1): Likewise.
16093 (vect_determine_min_output_precision): Likewise.
16094 (vect_determine_precisions_from_range): Likewise.
16095 (vect_determine_precisions_from_users): Likewise.
16096 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
16097 (vect_vect_recog_func_ptrs): Put over_widening first.
16099 (vect_pattern_recog): Call vect_determine_precisions.
16101 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16103 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
16104 statements that have been replaced by further pattern statements.
16105 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
16107 2018-07-03 Richard Biener <rguenther@suse.de>
16109 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
16110 always set *dt. Dump vectype in vectype overload.
16111 * dumpfile.h (dump_gimple_expr): New function.
16112 (dump_gimple_expr_loc): Likewise.
16113 * dumpfile.c (dump_gimple_expr): New function.
16114 (dump_gimple_expr_loc): Likewise.
16116 2018-07-02 Jeff Law <law@redhat.com>
16118 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
16119 the H8/300, H8/300H and H8/S variants into a single pattern.
16120 (movhi_h8300, movqi_h8300hs): Similarly.
16121 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
16122 (QHI mode iterator): New.
16124 * config/h8300/h8300.md: Remove trailing whitespace.
16126 2018-07-02 Jim Wilson <jimw@sifive.com>
16128 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
16129 instead of emit_insn for interrupt returns.
16130 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
16131 (riscv_sret, riscv_uret): Likewise.
16133 2018-07-02 David Malcolm <dmalcolm@redhat.com>
16135 * pretty-print.c (selftest::test_pp_format): Move save and restore
16136 of quotes to class auto_fix_quotes, and add an instance.
16137 * selftest.c: Include "intl.h".
16138 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
16139 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
16140 * selftest.h (selftest::auto_fix_quotes): New class.
16142 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
16144 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
16145 (aarch64_sve_prepare_conditional_op): Remove.
16146 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
16147 Allow aarch64_simd_reg_or_zero as select operand; remove
16148 the aarch64_sve_prepare_conditional_op call.
16149 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
16150 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
16151 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
16152 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
16153 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
16154 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
16155 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
16156 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
16157 and a splitters to match all of the *_any patterns.
16158 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
16160 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
16161 (SVE_COND_FP_BINARY_REV): Remove.
16162 (sve_int_op_rev, sve_fp_op_rev): New.
16163 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
16164 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
16165 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
16166 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
16167 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
16168 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
16169 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
16170 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
16171 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
16173 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
16174 Remove match_dup 1 from the inner unspec.
16175 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
16177 * config/aarch64/aarch64.md (movprfx): New attr.
16178 (length): Default movprfx to 8.
16179 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
16180 (*madd<SVE_I>, *msub<SVE_I): Likewise.
16181 (*<su>mul<SVE_I>3_highpart): Likewise.
16182 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
16183 (*v<ASHIFT><SVE_I>3): Likewise.
16184 (*<su><MAXMIN><SVE_I>3): Likewise.
16185 (*<su><MAXMIN><SVE_F>3): Likewise.
16186 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
16187 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
16188 (*div<SVE_F>4): Likewise.
16190 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
16192 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
16195 2018-07-02 Richard Biener <rguenther@suse.de>
16197 PR tree-optimization/86363
16198 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
16199 memset argument refers to a non-variable address.
16201 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
16203 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
16204 {PLUS,MINUS}_EXPR code to...
16205 (adjust_symbolic_bound): ...here,
16206 (combine_bound): ...here,
16207 (set_value_range_with_overflow): ...and here.
16209 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
16211 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
16213 (extract_range_from_abs_expr): ...here.
16215 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
16217 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
16218 -fno-omit-frame-pointer when not optimizing.
16220 2018-07-02 Martin Liska <mliska@suse.cz>
16223 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
16224 (propagate_nothrow): Likewise.
16226 2018-07-02 Martin Liska <mliska@suse.cz>
16229 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
16231 2018-07-02 David Malcolm <dmalcolm@redhat.com>
16233 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
16234 function in r262149, changing "loc" param from source_location to
16235 const dump_location_t &.
16236 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
16237 declaration, as above.
16239 2018-07-01 Paul Koning <ni1d@arrl.net>
16241 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16242 -munit-asm, -mgnu-asm, -mdec-asm.
16243 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
16244 (pdp11_output_labelref): New.
16245 (pdp11_output_def): New.
16246 (pdp11_output_addr_vec_elt): New.
16247 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
16248 %# and %@ format codes.
16249 (pdp11_option_override): New.
16250 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
16251 (pdp11_output_ident): New.
16252 (pdp11_asm_named_section): New.
16253 (pdp11_asm_init_sections): New.
16254 (pdp11_file_start): New.
16255 (pdp11_file_end): New.
16256 (output_ascii): Use .ascii/.asciz for -mdec-asm.
16257 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
16258 %o, like %c but octal.
16259 (pdp11_option_override): New.
16260 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
16262 (DATA_SECTION_ASM_OP): Ditto.
16263 (READONLY_DATA_SECTION_ASM_OP): New.
16264 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
16265 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
16266 (ASM_OUTPUT_LABELREF): Ditto.
16267 (ASM_OUTPUT_DEF): Ditto.
16268 (ASM_OUTPUT_EXTERNAL): New.
16269 (ASM_OUTPUT_SOURCE_FILENAME): New.
16270 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
16271 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
16272 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
16273 %# and %@ format codes.
16274 * config/pdp11/pdp11.opt (mgnu-asm): New.
16275 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
16276 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
16277 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
16279 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
16281 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
16282 dereferencing path[] beyond its length.
16284 (debug_all_paths): New.
16285 (rewire_first_differing_edge): New.
16286 (adjust_paths_after_duplication): New.
16287 (duplicate_thread_path): Call adjust_paths_after_duplication.
16289 (thread_through_all_blocks): Add new argument to
16290 duplicate_thread_path.
16292 2018-06-30 Jim Wilson <jimw@sifive.com>
16294 * config/riscv/predicates.md (p2m1_shift_operand): New.
16295 (high_mask_shift_operand): New.
16296 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
16297 pattern using p2m1_shift_operand.
16298 (lshsi3_zero_extend_3+2): New combiner pattern using
16299 high_mask_shift_operand.
16301 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16303 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
16305 (vect_recog_rotate_pattern): ...here.
16306 (vect_convert_input): Try to insert casts of invariants in the
16308 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
16309 preheader to be empty.
16311 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16313 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
16314 vector type. If given, install it in the new statement's
16315 STMT_VINFO_VECTYPE.
16316 (vect_element_precision): New function.
16317 (vect_unpromoted_value): New struct.
16318 (vect_unpromoted_value::vect_unpromoted_value): New function.
16319 (vect_unpromoted_value::set_op): Likewise.
16320 (vect_look_through_possible_promotion): Likewise.
16321 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
16322 (vect_widened_op_tree, vect_convert_input): Likewise.
16323 (vect_convert_inputs, vect_convert_output): Likewise.
16324 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
16325 to handle the optional cast of the multiplication result and
16326 vect_widened_op_tree to detect the widened multiplication itself.
16327 Do not require the input and output of promotion casts to have
16328 the same sign, but base the signedness of the operation on the
16329 input rather than the result. If the pattern includes two
16330 promotions, check that those promotions have the same sign.
16331 Do not restrict the MULT_EXPR handling to a double-width result;
16332 handle quadruple-width results and wider. Use vect_convert_inputs
16333 to convert the inputs to the common type.
16334 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
16335 to handle the optional cast of the ABS result. Also allow a sign
16336 change or a sign extension between the ABS and MINUS.
16337 Use vect_widened_op_tree to detect the widened subtraction and use
16338 vect_convert_inputs to convert the inputs to the common type.
16339 (vect_handle_widen_op_by_const): Delete.
16340 (vect_recog_widen_op_pattern): New function.
16341 (vect_recog_widen_mult_pattern): Use it.
16342 (vect_recog_widen_shift_pattern): Likewise.
16343 (vect_recog_widen_sum_pattern): Use
16344 vect_look_through_possible_promotion to handle the promoted
16347 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16349 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
16350 the containing gimple_seq *.
16351 * gimple-iterator.h (gsi_for_stmt): Declare it.
16352 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
16353 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
16354 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
16355 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
16356 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
16358 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
16360 (vect_mark_pattern_stmts): ...here. Handle cases in which the
16361 statement being replaced is part of an existing pattern
16362 definition sequence, inserting the new pattern statements before
16364 (vect_pattern_recog_1): Don't return a bool. If the statement
16365 is already part of a pattern, instead apply pattern matching
16366 to the pattern definition statements. Don't clear the
16367 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
16368 (vect_pattern_recog): Don't break after the first match;
16369 continue processing the pattern definition statements instead.
16370 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
16372 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16374 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
16375 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16376 (vect_recog_widen_sum_pattern): Use it.
16378 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16380 * tree-vect-loop.c (vectorizable_reduction): Assert that the
16381 phi is not a pattern statement and has not been replaced by
16382 a pattern statement.
16383 * tree-vect-patterns.c (type_conversion_p): Don't check
16384 STMT_VINFO_IN_PATTERN_P.
16385 (vect_recog_vector_vector_shift_pattern): Likewise.
16386 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
16387 the pattern statement rather than the original statement; check
16388 directly for a WIDEN_MULT_EXPR here.
16389 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
16390 vect_is_simple_use to return the pattern statement rather
16391 than the original statement; use is_pattern_stmt_p to check
16392 for such a pattern statement.
16393 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
16394 to return the pattern statement rather than the original statement;
16395 don't do the same transformation here.
16396 (vect_is_simple_use): If the defining statement has been replaced
16397 by a pattern statement, return the pattern statement instead.
16398 Remove the corresponding (local) transformation from the vectype
16401 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16403 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
16404 end and default to null.
16405 * tree-vect-loop.c (vect_create_epilog_for_reduction)
16406 (vectorizable_reduction): Update calls accordingly, dropping the
16407 gimple ** argument if the passed-back statement isn't needed.
16408 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
16409 (vect_recog_rotate_pattern): Likewise.
16410 (vect_recog_mask_conversion_pattern): Likewise.
16411 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
16412 (vect_mask_constant_operand_p): Likewise.
16413 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
16414 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
16415 (get_group_load_store_type, get_load_store_type): Likewise.
16416 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
16417 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
16418 (vectorizable_conversion, vectorizable_assignment): Likewise.
16419 (vectorizable_shift, vectorizable_operation): Likewise.
16420 (vectorizable_store, vect_is_simple_cond): Likewise.
16421 (vectorizable_condition, vectorizable_comparison): Likewise.
16422 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
16423 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
16424 and move it to the end. Cope with null def_stmt_outs.
16426 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
16428 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
16430 2018-06-29 Jeff Law <law@redhat.com>
16432 * config/v850/v850.c (v850_legitimate_address_p): Handle large
16433 displacements for TARGET_V850E2V3 and newer.
16434 (TARGET_LRA_P): Remove. Defaults to LRA now.
16435 * config/v850/v850.md (sign23byte_load): Remove.
16436 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
16437 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
16439 2018-06-29 Martin Liska <mliska@suse.cz>
16442 * coverage.c (coverage_init): Mangle full path name.
16443 * doc/invoke.texi: Document the change.
16444 * gcov-io.c (mangle_path): New.
16445 * gcov-io.h (mangle_path): Likewise.
16446 * gcov.c (mangle_name): Use mangle_path for path mangling.
16448 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16450 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
16451 if starting source register is not even.
16453 2018-06-29 Martin Liska <mliska@suse.cz>
16455 PR tree-optimization/86263
16456 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
16457 Make edge redirection.
16459 2018-06-29 David Malcolm <dmalcolm@redhat.com>
16461 * dumpfile.c (dump_loc): Add indentation based on scope depth.
16462 (dump_scope_depth): New variable.
16463 (get_dump_scope_depth): New function.
16464 (dump_begin_scope): New function.
16465 (dump_end_scope): New function.
16466 * dumpfile.h (get_dump_scope_depth): New declaration.
16467 (dump_begin_scope): New declaration.
16468 (dump_end_scope): New declaration.
16469 (class auto_dump_scope): New class.
16470 (AUTO_DUMP_SCOPE): New macro.
16471 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
16474 2018-06-29 Richard Biener <rguenther@suse.de>
16476 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
16477 compute_all_dependences succeeds.
16478 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
16479 exceed --param loop-max-datarefs-for-datadeps.
16481 2018-06-29 Jakub Jelinek <jakub@redhat.com>
16483 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
16485 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
16488 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
16489 alternative 0 in preferred_for_speed attribute.
16491 2018-06-28 Paul Koning <ni1d@arrl.net>
16493 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
16494 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
16495 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
16496 * config/pdp11/pdp11.md: Correct "length" attribute calculation
16497 for shift insn patterns.
16499 2018-06-28 David Malcolm <dmalcolm@redhat.com>
16501 * cgraph.c (cgraph_node::get_body): Replace assignments to
16502 "dump_file" with calls to set_dump_file.
16503 * dumpfile.c (alt_dump_file): Make static, and group with...
16504 (alt_flags): ...this definition.
16505 (dumps_are_enabled): New variable.
16506 (refresh_dumps_are_enabled): New function.
16507 (set_dump_file): New function.
16508 (set_alt_dump_file): New function.
16509 (gcc::dump_manager::dump_start): Replace assignments to
16510 "dump_file" and "alt_dump_file" with calls to set_dump_file and
16512 (gcc::dump_manager::dump_finish): Likewise.
16513 * dumpfile.h (alt_dump_file): Delete decl.
16514 (dumps_are_enabled): New variable decl.
16515 (set_dump_file): New function decl.
16516 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
16518 * tree-nested.c (lower_nested_functions): Replace assignments to
16519 "dump_file" with calls to set_dump_file.
16521 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
16523 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
16524 goto_locus of each outgoing edge of each basic block.
16526 2018-06-28 Richard Biener <rguenther@suse.de>
16528 * dwarf2out.c (decl_scope_table): Remove.
16529 (push_decl_scope): Likewise.
16530 (pop_decl_scope): Likewise.
16531 (gen_type_die_for_member): Do not call push/pop_decl_scope.
16532 (gen_struct_or_union_type_die): Likewise.
16533 (gen_tagged_type_die): Likewise.
16534 (dwarf2out_init): Do not initialize decl_scope_table.
16535 (dwarf2out_c_finalize): Do not free it.
16537 2018-06-28 Richard Biener <rguenther@suse.de>
16539 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
16540 deciding whether to not re-use a DIE.
16542 2018-06-28 Richard Biener <rguenther@suse.de>
16544 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
16545 DW_AT_abstract_origin attribute.
16547 2018-06-28 Martin Liska <mliska@suse.cz>
16549 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
16550 Use newly introduced constants.
16551 * tree-switch-conversion.h (struct jump_table_cluster):
16552 Define max_ratio_for_size and max_ratio_for_speed.
16554 2018-06-28 Martin Liska <mliska@suse.cz>
16556 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16557 Add new checking assert to catch invalid state.
16558 (jump_table_cluster::can_be_handled): Handle single case
16560 (jump_table_cluster::is_beneficial): Bail out for such case.
16561 (bit_test_cluster::find_bit_tests):
16562 Add new checking assert to catch invalid state.
16563 (bit_test_cluster::can_be_handled): Handle single case
16565 (bit_test_cluster::is_beneficial): Bail out for such case.
16566 (switch_decision_tree::analyze_switch_statement):
16569 2018-06-28 Martin Liska <mliska@suse.cz>
16571 * common.opt: Introduce -completion option.
16572 * gcc.c (driver_handle_option): Handle it.
16573 (driver::main): Print completions if completion
16575 * opt-suggestions.c (option_proposer::get_completions):
16577 (option_proposer::suggest_completion): Likewise.
16578 (option_proposer::find_param_completions): Likewise.
16579 (verify_autocompletions): Likewise.
16580 (test_completion_valid_options): Likewise.
16581 (test_completion_valid_params): Likewise.
16582 (in_completion_p): Likewise.
16583 (empty_completion_p): Likewise.
16584 (test_completion_partial_match): Likewise.
16585 (test_completion_garbage): Likewise.
16586 (opt_proposer_c_tests): Likewise.
16587 * opt-suggestions.h: Declare new functions.
16588 * opts.c (common_handle_option): Handle OPT__completion_.
16589 * selftest-run-tests.c (selftest::run_tests): Add
16590 opt_proposer_c_tests.
16591 * selftest.c (assert_str_startswith): New.
16592 * selftest.h (assert_str_startswith): Likewise.
16593 (opt_proposer_c_tests): New.
16594 (ASSERT_STR_STARTSWITH): Likewise.
16596 2018-06-28 Martin Liska <mliska@suse.cz>
16598 * Makefile.in: Add opt-suggestions.o.
16599 * gcc-main.c: Include opt-suggestions.h.
16600 * gcc.c (driver::driver): Likewise.
16601 (driver::~driver): Remove m_option_suggestions.
16602 (driver::build_option_suggestions): Moved to option_proposer.
16603 (driver::suggest_option): Likewise.
16604 (driver::handle_unrecognized_options): Use option_proposer.
16605 * gcc.h (class driver): Add new memver m_option_proposer.
16606 * opt-suggestions.c: New file.
16607 * opt-suggestions.h: New file.
16609 2018-06-28 Martin Liska <mliska@suse.cz>
16611 * vec.h (class auto_string_vec): New (moved from auto_argvec).
16612 (auto_string_vec::~auto_string_vec): Likewise.
16614 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
16616 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
16617 prevent_decl_creation_for_types fields up and add reset_location field.
16618 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
16619 statement if id->reset_location is true.
16620 (copy_edges_for_bb): Do not set goto_locus on the new edges if
16621 id->reset_location is true.
16622 (copy_phis_for_bb): Force input_location on the arguments if
16623 id->reset_location is true.
16624 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
16625 is set on the function to be inlined.
16627 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
16629 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
16631 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
16633 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
16634 registers for Pmode.
16635 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
16636 hard registers for the clobbered pseudo.
16638 2018-06-27 Paul Koning <ni1d@arrl.net>
16640 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16641 mutually exclusive options.
16642 * config/pdp11/constraints.md (h): New constraint.
16643 (O): Update definition to match shift code generation.
16644 (D): New constraint.
16645 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
16647 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
16649 (output_jump): Change arguments.
16650 (pdp11_fixed_cc_regs): New function.
16651 (pdp11_cc_mode): Ditto.
16652 (pdp11_expand_shift): Ditto.
16653 (pdp11_assemble_shift): Ditto.
16654 (pdp11_small_shift): Ditto.
16655 (pdp11_branch_cost): Remove.
16656 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
16658 (pdp11_register_move_cost): Update for CC registers.
16659 (pdp11_rtx_costs): Add case for LSHIFTRT.
16660 (pdp11_output_jump): Add CCNZ mode conditional branches.
16661 (notice_update_cc_on_set): Remove.
16662 (pdp11_cc_mode): New function.
16663 (simple_memory_operand): Correct pre/post decrement case.
16664 (no_side_effect_operand): New function.
16665 (pdp11_regno_reg_class): Add CC_REGS class.
16666 (pdp11_fixed_cc_regs): New function.
16667 (pdp11_small_shift): New function.
16668 (pdp11_expand_shift): New function to expand shift insns.
16669 (pdp11_assemble_shift): New function to output shifts.
16670 (pdp11_branch_cost): Remove.
16671 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
16672 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
16673 (WCHAR_TYPE): Ditto.
16674 (PTRDIFF_TYPE): Ditto.
16675 (ADJUST_INSN_LENGTH): New macro.
16676 (FIXED_REGISTERS): Add CC registers.
16677 (CALL_USED_REGISTERS): Ditto.
16678 (reg_class): Ditto.
16679 (REG_CLASS_NAMES): Ditto.
16680 (REG_CLASS_CONTENTS): Ditto.
16681 (SELECT_CC_MODE): Use new function.
16682 (TARGET_FLAGS_REGNUM): New macro.
16683 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
16684 (cc0_reg_rtx): Remove.
16685 (CC_STATUS_MDEP): Remove.
16686 (CC_STATUS_MDEFP_INIT): Remove.
16687 (CC_IN_FPU): Remove.
16688 (NOTICE_UPDATE_CC): Remove.
16689 (REGISTER_NAMES): Add CC registers.
16690 (BRANCH_COST): Change to constant 1.
16691 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
16693 * config/pdp11/pdp11.opt (mbcopy): Remove.
16694 (mbcopy-builtin): Remove.
16695 (mbranch-cheap): Remove.
16696 (mbranch-expensive): Remove.
16697 * config/pdp11/predicates.md (expand_shift_operand): Update to
16698 match shift code generation.
16699 (ccnz_operator): New predicate.
16700 * doc/invoke.texi (PDP-11 Options): Remove deleted options
16701 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
16702 Remove non-existent option -mabshi, -mno-abshi. Document mutually
16704 * doc/md.texi (PDP-11): Document new D and h constraints. Update
16705 description of O constraint.
16707 2018-06-27 Jeff Law <law@redhat.com>
16708 Austin Law <austinklaw@gmail.com>
16710 * config/v850/v850.md (addsi3_set_flags): New pattern.
16711 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
16712 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
16713 (zero_extendhisi2_v850_set_flags): Likewise.
16714 (zero_extendqisi2_v850_set_flags): Likewise.
16715 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
16716 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
16717 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
16719 * config/v850/v850-protos.h (notice_update_cc): Remove.
16720 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
16721 (v850_print_operand): Handle 'D' and "d".
16722 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
16723 Add handling of arithmetic/logical operations compared against zero.
16724 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
16725 Do not look at v850_compare_op, instead get mode from last argument.
16726 (v850_gen_compare): Remove
16727 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
16728 after reload for prologue insns.
16729 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
16731 (construct_save_jarl): Likewise.
16732 (TARGET_FLAGS_REGNUM): Define.
16733 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
16734 (NOTICE_UPDATE_CC): Remove.
16735 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
16736 than cc0. Conditionalize on reload_completed.
16737 (cmpsi_insn, setfcc_insn): Likewise.
16738 (tst1 splitter): Turn into define_and_split which sets the flags
16740 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
16741 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
16742 (cstoresf4, cstoredf4): Clobber the flags.
16743 (cmpsi, cmpsf, cmpdf): Remove expanders.
16744 (setf_insn): Remove pattern.
16745 (addsi3): Turn into define_and_split which clobbers the flags after
16746 reload and a suitable pattern (addsi3_clobber_flags) for use after
16748 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
16749 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
16750 (ashrsi3, ashrsi3_v850e2): Likewise.
16751 (bins): Clobber the flags.
16752 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
16753 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
16754 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
16755 (call_value_internal_short, call_value_internal_long): Likewise.
16756 (callt_save_interrupt, callt_return_interrupt): Likewise.
16757 (save_interrupt, return_interrupt): Likewise.
16758 (callt_save_all_interrupt, save_all_interrupt): Likewise.
16759 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
16760 (restore_all_interrupt, _restore_all_interrupt): Likewise.
16761 (All FP comparisons): Only allow after reload has completed.
16763 (divh, divhu): Tweak output template.
16764 (branch_z_normal, branch_z_invert): Remove
16765 (branch_nz_normal, branch_nz_invert): Likewise.
16766 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
16768 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
16769 * config/v850/v850.c (notice_update_cc): Remove.
16770 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
16771 (CC_NO_CARRY): Likewise.
16772 (NOTICE_UPDATE_CC): Define to nothing.
16773 * config/v850/v850.md: Remove block comment on cc0 handling
16774 Remove "cc" attribute from all patterns. Remove cc_status handling
16775 from all patterns. Minor formatting fixes.
16777 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16779 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
16780 (cortex-a76.cortex-a55): Likewise.
16781 * config/aarch64/aarch64-tune.md: Regenerate.
16782 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
16783 cortex-a76.cortex-a55.
16785 2018-06-27 Jeff Law <law@redhat.com>
16787 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
16788 (MULTILIB_DIRNAMES): Similarly.
16790 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
16792 * gimple.h (gimple_return_retbnd): Delete.
16793 (gimple_return_set_retbnd): Likewise.
16794 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
16795 gimple_return_set_retbnd.
16796 * gimple-pretty-print.c (dump_gimple_return): Remove call to
16797 gimple_return_retbnd and adjust.
16798 * tree-inline.h (struct copy_body_data): Remove retbnd field.
16799 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
16800 Explicitly return NULL in a couple more cases. Move assertion
16801 on debug statements and remove unreachable code.
16802 (reset_debug_binding): Do not test id->retbnd.
16803 (expand_call_inline): Do not set it.
16805 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
16807 * configure.ac: Add --disable-gcov option.
16808 * configure: Regenerate.
16809 * Makefile.in: Honour @enable_gcov@.
16810 * doc/install.texi: Document --disable-gcov.
16812 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16814 * config/arm/arm-cpus.in (cortex-a76): New entry.
16815 (cortex-a76.cortex-a55): Likewise.
16816 * config/arm/arm-tables.opt: Regenerate.
16817 * config/arm/arm-tune.md: Likewise.
16818 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
16819 * doc/invoke.texi (ARM Options): Document cortex-a76 and
16820 cortex-a76.cortex-a55.
16822 2018-06-27 Tamar Christina <tamar.christina@arm.com>
16825 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
16827 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
16829 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
16831 (EPILOGUE_USES): Likewise.
16833 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
16835 * tree-inline.c (remap_location): New function extracted from...
16836 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
16837 (copy_phis_for_bb): ...here. Call remap_location.
16838 (copy_cfg_body): Adjust call to copy_edges_for_bb.
16840 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
16842 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
16843 unaligned vsx for 16B memset.
16845 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
16848 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
16849 ieee128_float_type_node to long_double_type_node unless
16850 TARGET_LONG_DOUBLE_128 is set.
16852 2018-06-26 David Malcolm <dmalcolm@redhat.com>
16854 * cfgloop.c (get_loop_location): Convert return type from
16855 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
16856 by implicit construction from rtx_insn *, and using
16857 dump_user_location_t::from_function_decl for the fallback case.
16858 * cfgloop.h (get_loop_location): Convert return type from
16859 location_t to dump_user_location_t.
16860 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
16861 dump_printf_loc to pass in a dump_location_t rather than a
16862 location_t, via the gimple stmt.
16863 * coverage.c (get_coverage_counts): Update calls to
16864 dump_printf_loc to pass in dump_location_t rather than a
16866 * doc/optinfo.texi (Dump types): Convert example of
16867 dump_printf_loc from taking "locus" to taking "insn". Update
16868 description of the "_loc" calls to cover dump_location_t.
16869 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
16871 (dump_user_location_t::dump_user_location_t): New constructors,
16872 from gimple *stmt and rtx_insn *.
16873 (dump_user_location_t::from_function_decl): New function.
16874 (dump_loc): Make static.
16875 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
16876 const dump_location_t &.
16877 (dump_generic_expr_loc): Delete.
16878 (dump_printf_loc): Convert param "loc" from location_t to
16879 const dump_location_t &.
16880 (selftest::test_impl_location): New function.
16881 (selftest::dumpfile_c_tests): New function.
16882 * dumpfile.h: Include "profile-count.h".
16883 (class dump_user_location_t): New class.
16884 (struct dump_impl_location_t): New struct.
16885 (class dump_location_t): New class.
16886 (dump_printf_loc): Convert 2nd param from source_location to
16887 const dump_location_t &.
16888 (dump_generic_expr_loc): Delete.
16889 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
16890 const dump_location_t &.
16891 * gimple-fold.c (fold_gimple_assign): Update call to
16892 dump_printf_loc to pass in a dump_location_t rather than a
16893 location_t, via the gimple stmt.
16894 (gimple_fold_call): Likewise.
16895 * gimple-loop-interchange.cc
16896 (loop_cand::analyze_iloop_reduction_var): Update for change to
16897 check_reduction_path.
16898 (tree_loop_interchange::interchange): Update for change to
16899 find_loop_location.
16900 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
16901 change in return-type of find_loop_location.
16902 (graphite_regenerate_ast_isl): Likewise.
16903 * graphite-optimize-isl.c (optimize_isl): Likewise.
16904 * graphite.c (graphite_transform_loops): Likewise.
16905 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
16906 pass in a dump_location_t rather than a location_t, via the
16908 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16909 * ipa.c (walk_polymorphic_call_targets): Likewise.
16910 * loop-unroll.c (report_unroll): Convert "locus" param from
16911 location_t to dump_location_t.
16912 (decide_unrolling): Update for change to get_loop_location's
16914 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
16915 location_t to dump_user_location_t.
16916 (grid_find_single_omp_among_assignments_1): Updates calls to
16917 dump_printf_loc to pass in a dump_location_t rather than a
16918 location_t, via the gimple stmt.
16919 (grid_parallel_clauses_gridifiable): Convert "tloc" from
16920 location_t to dump_location_t. Updates calls to dump_printf_loc
16921 to pass in a dump_location_t rather than a location_t, via the
16923 (grid_inner_loop_gridifiable_p): Likewise.
16924 (grid_dist_follows_simple_pattern): Likewise.
16925 (grid_gfor_follows_tiling_pattern): Likewise.
16926 (grid_target_follows_gridifiable_pattern): Likewise.
16927 (grid_attempt_target_gridification): Convert initialization
16928 of local "grid" from memset to zero-initialization; FIXME: does
16929 this require C++11? Update call to dump_printf_loc to pass in a
16930 optinfo_location rather than a location_t, via the gimple stmt.
16931 * profile.c (read_profile_edge_counts): Updates call to
16932 dump_printf_loc to pass in a dump_location_t rather than a
16934 (compute_branch_probabilities): Likewise.
16935 * selftest-run-tests.c (selftest::run_tests): Call
16937 * selftest.h (dumpfile_c_tests): New decl.
16938 * tree-loop-distribution.c (pass_loop_distribution::execute):
16939 Update for change in return type of find_loop_location.
16940 * tree-parloops.c (parallelize_loops): Likewise.
16941 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
16942 "locus" from location_t to dump_user_location_t.
16943 (canonicalize_loop_induction_variables): Likewise.
16944 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
16945 for change in return type of find_loop_location.
16946 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
16947 to dump_printf_loc to pass in a dump_location_t rather than a
16948 location_t, via the stmt.
16949 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
16951 * tree-vect-loop-manip.c (find_loop_location): Convert return
16952 type from source_location to dump_user_location_t.
16953 (vect_do_peeling): Update for above change.
16954 (vect_loop_versioning): Update for change in type of
16956 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
16957 from location_t to dump_user_location_t.
16958 (vect_estimate_min_profitable_iters): Update for change in type
16960 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
16961 location_t to dump_location_t.
16962 (vect_slp_bb): Update for change in type of vect_location.
16963 * tree-vectorizer.c (vect_location): Convert from source_location
16964 to dump_user_location_t.
16965 (try_vectorize_loop_1): Update for change in vect_location's type.
16966 (vectorize_loops): Likewise.
16967 (increase_alignment): Likewise.
16968 * tree-vectorizer.h (vect_location): Convert from source_location
16969 to dump_user_location_t.
16970 (find_loop_location): Convert return type from source_location to
16971 dump_user_location_t.
16972 (check_reduction_path): Convert 1st param from location_t to
16973 dump_user_location_t.
16974 * value-prof.c (check_counter): Update call to dump_printf_loc to
16975 pass in a dump_user_location_t rather than a location_t; update
16976 call to error_at for change in type of "locus".
16977 (check_ic_target): Update call to dump_printf_loc to
16978 pass in a dump_user_location_t rather than a location_t, via the
16981 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
16983 * config/s390/s390.h (enum processor_flags): Do not use
16984 default tune parameter when -march was specified.
16986 2018-06-26 Jakub Jelinek <jakub@redhat.com>
16989 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
16990 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
16993 2018-06-26 Richard Biener <rguenther@suse.de>
16995 PR tree-optimization/86287
16997 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
16998 (vect_analyze_loop): Initialize n_stmts.
17000 2018-06-26 Richard Biener <rguenther@suse.de>
17002 PR middle-end/86271
17003 * fold-const.c (fold_convertible_p): Pointer extension
17006 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
17009 * dwarf2out.c (loc_list_has_views): Adjust comments.
17010 (dw_loc_list): Split single cross-partition range with
17013 2018-06-25 Jeff Law <law@redhat.com>
17015 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
17016 on -mbig-switch by default.
17018 * config/v850/predicates.md (const_float_1_operand): Fix match_code
17020 (const_float_0_operand): Remove unused predicate.
17021 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
17022 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
17023 (recipsf2): New expander. Original pattern now called
17025 (recipdf2, recipdf2_insn): Similarly.
17026 (rsqrtsf2, rsqrtsf2_insn): Similarly
17027 (rsqrtdf2, rsqrtdf2_insn): Similarly
17029 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
17031 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
17032 Simplify logic for FreeBSD (twice).
17034 2018-06-25 Martin Sebor <msebor@redhat.com>
17036 PR tree-optimization/86204
17037 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
17038 a strnlen result if it's less than the length of the string.
17040 2018-06-25 Martin Sebor <msebor@redhat.com>
17042 PR tree-optimization/85700
17043 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
17044 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
17045 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
17047 2018-06-25 Martin Sebor <msebor@redhat.com>
17049 * doc/extend.texi (Zero-length arrays): Update and clarify.
17051 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
17053 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
17054 added IEEE/IBM long double multilib support on PowerPC little
17055 endian Linux systems.
17056 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
17057 (MULTILIB_DEFAULTS): Likewise.
17058 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17060 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
17061 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
17062 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
17064 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
17066 PR middle-end/86311
17067 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
17068 (REORDER_45): Likewise.
17070 2018-06-25 Jeff Law <law@redhat.com>
17072 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
17073 dividend to 32 bits. Adjust length.
17074 (udivmodhi4): Cleanup output template. Fix length.
17076 2018-06-25 Carl Love <cel@us.ibm.com>
17078 * config/rs6000/vsx.md: Change word selector to prefered location.
17080 2018-06-25 Richard Biener <rguenther@suse.de>
17082 PR tree-optimization/86304
17083 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
17084 epilogue-if-converted loops as well.
17086 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
17088 * lto-section-out.c (lto_begin_section): Do not print section
17089 name for noaddr and unnumbered dumps.
17091 2018-06-25 Richard Biener <rguenther@suse.de>
17093 * tree-vectorizer.h (struct vec_info_shared): New structure
17094 with parts split out from struct vec_info and loop_nest from
17095 struct _loop_vec_info.
17096 (struct vec_info): Adjust accordingly.
17097 (struct _loop_vec_info): Likewise.
17098 (LOOP_VINFO_LOOP_NEST): Adjust.
17099 (LOOP_VINFO_DATAREFS): Likewise.
17100 (LOOP_VINFO_DDRS): Likewise.
17101 (struct _bb_vec_info): Likewise.
17102 (BB_VINFO_DATAREFS): Likewise.
17103 (BB_VINFO_DDRS): Likewise.
17104 (struct _stmt_vec_info): Add dr_aux member.
17105 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
17106 (DR_MISALIGNMENT_UNINITIALIZED): New.
17107 (set_dr_misalignment): Adjust.
17108 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
17109 (vect_analyze_loop): Adjust prototype.
17110 (vect_analyze_loop_form): Likewise.
17111 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17112 Compute dependences lazily.
17113 (vect_record_base_alignments): Use shared datarefs/ddrs.
17114 (vect_verify_datarefs_alignment): Likewise.
17115 (vect_analyze_data_refs_alignment): Likewise.
17116 (vect_analyze_data_ref_accesses): Likewise.
17117 (vect_analyze_data_refs): Likewise.
17118 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
17119 constructor parameter for shared part.
17120 (vect_analyze_loop_form): Pass in shared part and adjust.
17121 (vect_analyze_loop_2): Pass in storage for the number of
17122 stmts. Move loop nest finding to the caller. Compute
17124 (vect_analyze_loop): Pass in shared part.
17125 (vect_transform_loop): Verify shared datarefs are unchanged.
17126 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
17127 constructor parameter for shared part.
17128 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
17129 (vect_slp_bb): Verify shared datarefs are unchanged before
17131 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
17133 (new_stmt_vec_info): Initialize DR_AUX misalignment to
17134 DR_MISALIGNMENT_UNINITIALIZED.
17135 * tree-vectorizer.c (vec_info::vec_info): Add constructor
17136 parameter for shared part.
17137 (vec_info::~vec_info): Adjust.
17138 (vec_info_shared::vec_info_shared): New.
17139 (vec_info_shared::~vec_info_shared): Likewise.
17140 (vec_info_shared::save_datarefs): Likewise.
17141 (vec_info_shared::check_datarefs): Likewise.
17142 (try_vectorize_loop_1): Construct shared part live for analyses
17143 of a single loop for multiple vector sizes.
17144 * tree-parloops.c (gather_scalar_reductions): Adjust.
17146 2018-06-25 Richard Biener <rguenther@suse.de>
17148 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
17149 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
17150 (vect_analyze_data_refs): Remove similar code from here and
17151 simplify accordingly.
17153 2018-06-25 Richard Biener <rguenther@suse.de>
17155 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
17156 for reverse storage order accesses rather than asserting
17157 they cannot happen here.
17159 2018-06-25 Tom de Vries <tdevries@suse.de>
17162 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
17163 Use data16 instead of .byte for insn prefix.
17165 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
17168 * parser.c (make_char_string_pack): Pass this literal chars
17169 through cpp_interpret_string.
17170 (cp_parser_userdef_numeric_literal): Check the result of
17171 make_char_string_pack.
17173 2018-06-24 Maya Rashish <coypu@sdf.org>
17175 * ginclude/stddef.h: Simplify conditions around avoiding
17176 re-definition of __size_t.
17178 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17180 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
17181 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
17183 2018-06-22 Maya Rashish <coypu@sdf.org>
17185 * doc/invoke.texi (mno-fancy-math-387): Update for changes
17186 made to OpenBSD and NetBSD through the years.
17188 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
17191 behavior of vec_pack (vector double, vector double) to match
17192 behavior of vec_float2 (vector double, vector double).
17194 2018-06-22 Olivier Hainque <hainque@adacore.com>
17196 * gimplify.c (gimplify_function_tree): Prevent creation
17197 of a trampoline for the address of the current function
17198 passed to entry/exit instrumentation hooks.
17200 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
17203 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
17206 2018-06-22 Martin Liska <mliska@suse.cz>
17208 PR tree-optimization/86263
17209 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
17210 Bail out if is_enabled is false.
17211 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
17213 (jump_table_cluster::is_enabled): New function.
17215 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17217 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
17218 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
17219 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
17220 (lto_input_ts_binfo_tree_pointers): Likewise.
17221 * tree-streamer-out.c (streamer_write_tree_bitfields,
17222 write_ts_binfo_tree_pointers): Likewise.
17223 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
17225 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17227 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
17229 2018-06-22 Martin Liska <mliska@suse.cz>
17231 * symbol-summary.h (get): Make it pure and inline move
17232 functionality from ::get function.
17233 (get): Remove and inline into ::get and ::get_create.
17234 (get_create): Move code from ::get function.
17236 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17239 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
17240 -x assembler-with-cpp.
17242 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17244 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
17245 _FILE_OFFSET_BITS=64 for C++.
17247 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
17249 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
17250 conversion insn that shows up when pr85657-3.c is compiled using
17251 IEEE 128-bit long double.
17252 (neg<mode>2_internal): Use the correct mode to check whether the
17253 mode is IBM extended.
17254 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
17255 multiply and divide external functions from being created more
17258 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17260 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
17262 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
17263 the edge can be forwarded.
17264 (cfg_layout_merge_blocks): Likewise.
17266 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17268 * except.c (finish_eh_generation): Commit edge insertions only after
17269 the EH edges have been redirected from post-landing to landing pads.
17271 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17273 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
17274 create_tmp_var_for to create the FRAME decl.
17275 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
17277 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17279 * tree-inline.c (copy_edges_for_bb): Minor tweak.
17280 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
17281 debug statement when resetting its value.
17282 (expand_call_inline): Copy the locus of the call onto the assignment
17283 of the return value, if any. Use local variable in more cases.
17285 2018-06-21 Martin Liska <mliska@suse.cz>
17287 * ipa-pure-const.c (propagate_nothrow): Use
17288 funct_state_summaries->get.
17289 (dump_malloc_lattice): Likewise.
17290 (propagate_malloc): Likewise.
17292 2018-06-21 Richard Biener <rguenther@suse.de>
17294 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
17295 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
17296 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
17298 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
17299 BLOCK_ABSTRACT_ORIGIN unconditionally.
17301 2018-06-21 David Malcolm <dmalcolm@redhat.com>
17303 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
17305 * ipa-reference.c (ipa_reference_c_finalize): Delete
17306 ipa_ref_opt_sum_summaries and set it to NULL.
17308 2018-06-21 Tom de Vries <tdevries@suse.de>
17310 PR tree-optimization/85859
17311 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
17312 test with comment from bb_no_side_effects_p.
17314 2018-06-21 Richard Biener <rguenther@suse.de>
17316 PR tree-optimization/86232
17317 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
17318 max for constant niter.
17320 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
17322 * config/aarch64/aarch64-simd.md
17323 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
17325 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
17327 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
17328 Make opernads of the unspec commutative.
17330 2018-06-21 Richard Biener <rguenther@suse.de>
17332 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
17333 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17334 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
17335 (vect_analyze_data_ref_dependence): Re-order checks to deal with
17337 (vect_record_base_alignments): Do not record base alignment
17338 for gathers or scatters.
17339 (vect_compute_data_ref_alignment): Drop return value that is always
17340 true. Bail out early for gathers or scatters.
17341 (vect_enhance_data_refs_alignment): Bail out early for gathers
17343 (vect_find_same_alignment_drs): Likewise.
17344 (vect_analyze_data_refs_alignment): Remove dead code.
17345 (vect_slp_analyze_and_verify_node_alignment): Likewise.
17346 (vect_analyze_data_refs): For possible gathers or scatters do
17347 not create an alternate DR, just check their possible validity
17348 and mark them. Adjust DECL_NONALIASED handling to not rely
17349 on DR_BASE_ADDRESS.
17350 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
17351 update inits of gathers or scatters.
17352 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
17353 Also copy gather/scatter flag to pattern vinfo.
17355 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17357 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
17358 behavior of vec_packsu (vector unsigned long long, vector unsigned
17359 long long) to match behavior of vec_packs with same signature.
17361 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
17362 Thomas Schwinge <thomas@codesourcery.com>
17363 Cesar Philippidis <cesar@codesourcery.com>
17365 * gimplify.c (gimplify_scan_omp_clauses): Add support for
17366 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
17367 (gimplify_adjust_omp_clauses): Likewise.
17368 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
17369 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
17370 (gimplify_omp_target_update): Update handling of acc update and
17372 * omp-low.c (install_var_field): Remove unused parameter
17373 base_pointers_restrict.
17374 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
17375 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
17377 (omp_target_base_pointers_restrict_p): Delete.
17378 (scan_omp_target): Update call to scan_sharing_clauses.
17379 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
17381 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
17382 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
17383 (convert_local_omp_clauses): Likewise.
17384 * tree-pretty-print.c (dump_omp_clause): Likewise.
17385 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
17387 (omp_clause_code_name): Likewise.
17389 2018-06-20 Jakub Jelinek <jakub@redhat.com>
17392 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
17395 PR tree-optimization/86231
17396 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
17397 anti-range don't overwrite *vr0min before using it to compute *vr0max.
17399 2018-06-20 Tom de Vries <tdevries@suse.de>
17401 PR tree-optimization/86097
17402 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
17403 iv type if signedness of iv type is not the same as that of *nit.
17405 2018-06-20 Jakub Jelinek <jakub@redhat.com>
17407 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
17408 EDGE_EH edges, verify they are all EDGE_EH.
17410 2018-06-20 Maya Rashish <coypu@sdf.org>
17412 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
17414 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17416 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
17417 * config/aarch64/aarch64.c (xgene1_tunings): Add
17418 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
17419 (aarch64_mode_valid_for_sched_fusion_p):
17420 Allow 16-byte modes.
17421 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
17422 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
17424 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
17426 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
17427 * config/aarch64/iterators.md (VQ2): New mode iterator.
17429 2018-06-20 Martin Liska <mliska@suse.cz>
17431 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
17432 Change default ratio from 10 to 8.
17434 2018-06-20 Martin Liska <mliska@suse.cz>
17436 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
17438 (bit_test_cluster::find_bit_tests): Likewise.
17439 (switch_decision_tree::analyze_switch_statement): Find clusters.
17440 * tree-switch-conversion.h (struct jump_table_cluster): Document
17443 2018-06-20 Martin Liska <mliska@suse.cz>
17445 * tree-switch-conversion.c (switch_conversion::collect):
17446 Record m_uniq property.
17447 (switch_conversion::expand): Bail out for special conditions.
17448 (group_cluster::~group_cluster): New.
17449 (group_cluster::group_cluster): Likewise.
17450 (group_cluster::dump): Likewise.
17451 (jump_table_cluster::emit): New.
17452 (switch_decision_tree::fix_phi_operands_for_edges): New.
17453 (struct case_node): Remove struct.
17454 (jump_table_cluster::can_be_handled): New.
17455 (case_values_threshold): Moved to header.
17456 (reset_out_edges_aux): Likewise.
17457 (jump_table_cluster::is_beneficial): New.
17458 (bit_test_cluster::can_be_handled): Likewise.
17459 (add_case_node): Remove.
17460 (bit_test_cluster::is_beneficial): New.
17461 (case_bit_test::cmp): New.
17462 (bit_test_cluster::emit): New.
17463 (expand_switch_as_decision_tree_p): Remove.
17464 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
17465 (fix_phi_operands_for_edge): Likewise.
17466 (switch_decision_tree::analyze_switch_statement): New.
17467 (compute_cases_per_edge): Move ...
17468 (switch_decision_tree::compute_cases_per_edge): ... here.
17469 (try_switch_expansion): Likewise.
17470 (switch_decision_tree::try_switch_expansion): Likewise.
17471 (record_phi_operand_mapping): Likewise.
17472 (switch_decision_tree::record_phi_operand_mapping): Likewise.
17473 (emit_case_decision_tree): Likewise.
17474 (switch_decision_tree::emit): Likewise.
17475 (balance_case_nodes): Likewise.
17476 (switch_decision_tree::balance_case_nodes): Likewise.
17477 (dump_case_nodes): Likewise.
17478 (switch_decision_tree::dump_case_nodes): Likewise.
17479 (emit_jump): Likewise.
17480 (switch_decision_tree::emit_jump): Likewise.
17481 (emit_cmp_and_jump_insns): Likewise.
17482 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
17483 (emit_case_nodes): Likewise.
17484 (switch_decision_tree::emit_case_nodes): Likewise.
17485 (conditional_probability): Remove.
17486 * tree-switch-conversion.h (enum cluster_type): New.
17488 (struct cluster): Likewise.
17489 (cluster::cluster): Likewise.
17490 (struct simple_cluster): Likewise.
17491 (simple_cluster::simple_cluster): Likewise.
17492 (struct group_cluster): Likewise.
17493 (struct jump_table_cluster): Likewise.
17494 (struct bit_test_cluster): Likewise.
17495 (struct min_cluster_item): Likewise.
17496 (struct case_tree_node): Likewise.
17497 (case_tree_node::case_tree_node): Likewise.
17498 (jump_table_cluster::case_values_threshold): Likewise.
17499 (struct case_bit_test): Likewise.
17500 (struct switch_decision_tree): Likewise.
17501 (struct switch_conversion): Likewise.
17502 (switch_decision_tree::reset_out_edges_aux): Likewise.
17504 2018-06-20 Martin Liska <mliska@suse.cz>
17506 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
17507 (hoist_edge_and_branch_if_true): Likewise.
17508 (expand_switch_using_bit_tests_p): Likewise.
17509 (struct case_bit_test): Likewise.
17510 (case_bit_test_cmp): Likewise.
17511 (emit_case_bit_tests): Likewise.
17512 (switch_conversion::switch_conversion): New class.
17513 (struct switch_conv_info): Remove old struct.
17514 (collect_switch_conv_info): More to ...
17515 (switch_conversion::collect): ... this.
17516 (check_range): Likewise.
17517 (switch_conversion::check_range): Likewise.
17518 (check_all_empty_except_final): Likewise.
17519 (switch_conversion::check_all_empty_except_final): Likewise.
17520 (check_final_bb): Likewise.
17521 (switch_conversion::check_final_bb): Likewise.
17522 (create_temp_arrays): Likewise.
17523 (switch_conversion::create_temp_arrays): Likewise.
17524 (free_temp_arrays): Likewise.
17525 (gather_default_values): Likewise.
17526 (switch_conversion::gather_default_values): Likewise.
17527 (build_constructors): Likewise.
17528 (switch_conversion::build_constructors): Likewise.
17529 (constructor_contains_same_values_p): Likewise.
17530 (switch_conversion::contains_same_values_p): Likewise.
17531 (array_value_type): Likewise.
17532 (switch_conversion::array_value_type): Likewise.
17533 (build_one_array): Likewise.
17534 (switch_conversion::build_one_array): Likewise.
17535 (build_arrays): Likewise.
17536 (switch_conversion::build_arrays): Likewise.
17537 (gen_def_assigns): Likewise.
17538 (switch_conversion::gen_def_assigns): Likewise.
17539 (prune_bbs): Likewise.
17540 (switch_conversion::prune_bbs): Likewise.
17541 (fix_phi_nodes): Likewise.
17542 (switch_conversion::fix_phi_nodes): Likewise.
17543 (gen_inbound_check): Likewise.
17544 (switch_conversion::gen_inbound_check): Likewise.
17545 (process_switch): Use the newly created class.
17546 (switch_conversion::expand): New.
17547 (switch_conversion::~switch_conversion): New.
17548 * tree-switch-conversion.h: New file.
17550 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17552 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
17553 tree-vect-patterns.c.
17554 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
17555 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
17556 (vect_recog_sad_pattern): Likewise.
17557 (vect_recog_widen_sum_pattern): Likewise.
17558 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
17559 (vect_recog_widen_shift_pattern): Remove the type_in argument.
17560 (vect_recog_rotate_pattern): Likewise.
17561 (vect_recog_mult_pattern): Likewise.
17562 (vect_recog_vector_vector_shift_pattern): Likewise.
17563 (vect_recog_divmod_pattern): Likewise.
17564 (vect_recog_mixed_size_cond_pattern): Likewise.
17565 (vect_recog_bool_pattern): Likewise.
17566 (vect_recog_mask_conversion_pattern): Likewise.
17567 (vect_try_gather_scatter_pattern): Likewise.
17568 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
17569 (vect_recog_over_widening_pattern): Likewise.
17570 (vect_recog_gather_scatter_pattern): Likewise.
17571 (vect_recog_func_ptr): Move from tree-vectorizer.h
17572 (vect_vect_recog_func_ptrs): Move further down the file.
17573 (vect_recog_func): Likewise. Remove the third argument.
17574 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
17575 (vect_pattern_recog_1): Expect the pattern function to do any
17576 necessary target tests. Also expect it to provide a vector type.
17577 Remove the type_in handling.
17579 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17581 * tree-vect-patterns.c (vect_pattern_detected): New function.
17582 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
17583 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
17584 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
17585 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
17586 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
17587 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
17588 (vect_recog_mask_conversion_pattern)
17589 (vect_try_gather_scatter_pattern): Likewise.
17591 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17593 * tree-vect-patterns.c (vect_get_internal_def): New function.
17594 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
17595 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
17596 (search_type_for_mask_1): Use it.
17598 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17600 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
17601 redundant WIDEN_SUM_EXPR handling.
17602 (vect_recog_sad_pattern): Likewise.
17604 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17606 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
17607 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
17608 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
17609 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
17610 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
17612 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17614 * tree-vect-stmts.c (vectorizable_call): Make sure that we
17615 use the stmt_vec_info of the original bb statement for the
17616 new zero assignment, even if the call is part of a pattern.
17618 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17620 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
17621 that the sequence is attached to the original statement rather
17622 than the pattern statement.
17623 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
17624 PATTERN_DEF_SEQ from the original statement rather than
17625 the main pattern statement.
17626 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
17627 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17628 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
17630 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17632 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
17633 definition statements before the early exit for statements that aren't
17635 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
17637 (vect_transform_loop): ...here. Process pattern definition
17638 statements without first checking whether the main pattern
17639 statement is live or relevant.
17641 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17643 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
17644 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
17646 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
17648 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
17649 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
17650 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
17651 (expand_block_compare): Change select_block_compare_mode call.
17652 (expand_strncmp_align_check): Use new functions, fix comment.
17653 (emit_final_str_compare_gpr): New function.
17654 (expand_strn_compare): Refactor and clean up code.
17655 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
17657 2018-06-19 Tony Reix <tony.reix@atos.com>
17658 Damien Bergamini <damien.bergamini@atos.com>
17659 David Edelsohn <dje.gcc@gmail.com>
17661 * collect2.c (static_obj): New variable.
17662 (static_libs): New variable.
17663 (is_in_list): Uncomment declaration.
17664 (main): Track AIX libraries linked statically.
17665 (is_in_list): Uncomment definition.
17666 (scan_prog_file): Don't add AIX shared libraries initializer
17667 to constructor list if linking statically.
17669 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
17671 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
17673 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
17675 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
17677 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
17680 2018-06-19 Martin Liska <mliska@suse.cz>
17682 * config/i386/i386.c (ix86_can_inline_p): Do not use
17683 ipa_fn_summaries::get_create.
17684 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
17686 (devirtualization_time_bonus): Likewise.
17687 (ipcp_propagate_stage): Likewise.
17688 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
17689 (edge_set_predicate): Likewise.
17690 (evaluate_conditions_for_known_args): Likewise.
17691 (evaluate_properties_for_edge): Likewise.
17692 (ipa_call_summary::reset): Tranform to ...
17693 (ipa_call_summary::~ipa_call_summary): ... this.
17694 (ipa_fn_summary::reset): Transform to ...
17695 (ipa_fn_summary::~ipa_fn_summary): ... this.
17696 (ipa_fn_summary_t::remove): Rename to ...
17697 (ipa_fn_summary_t::remove_callees): ... this.
17698 (ipa_fn_summary_t::duplicate): Use placement new
17699 instead of memory copy.
17700 (ipa_call_summary_t::duplicate): Likewise.
17701 (ipa_call_summary_t::remove): Remove.
17702 (dump_ipa_call_summary): Change get_create to get.
17703 (ipa_dump_fn_summary): Dump only when summary exists.
17704 (analyze_function_body): Use symbol_summary::get instead
17706 (compute_fn_summary): Likewise.
17707 (estimate_edge_devirt_benefit): Likewise.
17708 (estimate_edge_size_and_time): Likewise.
17709 (inline_update_callee_summaries): Likewise.
17710 (remap_edge_change_prob): Likewise.
17711 (remap_edge_summaries): Likewise.
17712 (ipa_merge_fn_summary_after_inlining): Likewise.
17713 (write_ipa_call_summary): Likewise.
17714 (ipa_fn_summary_write): Likewise.
17715 (ipa_free_fn_summary): Likewise.
17716 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
17717 (struct ipa_call_summary): Likewise.
17718 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
17720 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
17721 (estimate_size_after_inlining): Likewise.
17722 (estimate_growth): Likewise.
17723 (growth_likely_positive): Likewise.
17724 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
17725 (inline_call): Likewise.
17726 * ipa-inline.c (caller_growth_limits): Likewise.
17727 (can_inline_edge_p): Likewise.
17728 (can_inline_edge_by_limits_p): Likewise.
17729 (compute_uninlined_call_time): Likewise.
17730 (compute_inlined_call_time): Likewise.
17731 (want_inline_small_function_p): Likewise.
17732 (edge_badness): Likewise.
17733 (update_caller_keys): Likewise.
17734 (update_callee_keys): Likewise.
17735 (inline_small_functions): Likewise.
17736 (inline_to_all_callers_1): Likewise.
17737 (dump_overall_stats): Likewise.
17738 (early_inline_small_functions): Likewise.
17739 (early_inliner): Likewise.
17740 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17741 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17742 * ipa-pure-const.c (malloc_candidate_p): Likewise.
17743 * ipa-split.c (execute_split_functions): Likewise.
17744 * symbol-summary.h: Likewise.
17745 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
17747 2018-06-19 Richard Biener <rguenther@suse.de>
17749 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
17750 (vectorize_loops): ... here. Fix dbgcnt handling.
17751 (try_vectorize_loop): Wrap try_vectorize_loop_1.
17753 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
17756 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
17757 ieee128 argument takes up only one (vector) register, not two (floating
17760 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17762 * gimplify.c (gimplify_init_constructor): Really never clear for an
17763 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
17765 2018-06-19 Richard Biener <rguenther@suse.de>
17767 PR tree-optimization/86179
17768 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
17769 after failed recognition.
17771 2018-06-18 Martin Sebor <msebor@redhat.com>
17773 PR middle-end/85602
17774 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
17775 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
17776 Handle integer subtraction.
17777 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
17778 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
17780 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17782 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
17783 param from rtx to rtx_insn *.
17784 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
17786 (frv_ifcvt_modify_insn): Likwise.
17787 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
17788 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
17789 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
17790 as_a <rtx_insn *> cast to local "unprotected_region" once
17791 it's been established that it's not NULL or pc_rtx.
17792 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
17793 param "sethi" from rtx to rtx_insn *.
17794 (nds32_group_float_insns): Likewise for param "insn".
17795 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
17797 (vax_output_int_subtract): Likewise.
17798 * config/vax/vax.c (vax_output_int_add): Likewise for param
17800 (vax_output_int_subtract): Likewise.
17801 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
17802 (emit_pattern_after): Likewise for param "after".
17803 (emit_insn_after): Likewise.
17804 (emit_jump_insn_after): Likewise.
17805 (emit_call_insn_after): Likewise.
17806 (emit_debug_insn_after): Likewise.
17807 (emit_pattern_before): Likewise for param "before".
17808 (emit_insn_before): Likewise.
17809 (emit_jump_insn_before): Likewise.
17810 * final.c (get_insn_template): Likewise for param "insn", removing
17812 * output.h (get_insn_template): Likewise for 2nd param.
17813 * rtl.h (emit_insn_before): Likewise.
17814 (emit_jump_insn_before): Likewise.
17815 (emit_debug_insn_before_noloc): Likewise.
17816 (emit_insn_after): Likewise.
17817 (emit_jump_insn_after): Likewise.
17818 (emit_call_insn_after): Likewise.
17819 (emit_debug_insn_after): Likewise.
17820 (set_insn_deleted): Likewise for param.
17822 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
17825 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
17826 floating point modes, so that IFmode is numerically greater than
17827 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
17828 to declare the ordering. This prevents IFmode from being
17829 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
17830 machine. Include rs6000-modes.h to share the fractional values
17831 between genmodes* and the rest of the compiler.
17832 (IFmode): Likewise.
17833 (KFmode): Likewise.
17834 (TFmode): Likewise.
17835 * config/rs6000/rs6000-modes.h: New file.
17836 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
17837 meaning of rs6000_long_double_size so that 126..128 selects an
17838 appropriate 128-bit floating point type.
17839 (rs6000_option_override_internal): Likewise.
17840 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
17841 (TARGET_LONG_DOUBLE_128): Change the meaning of
17842 rs6000_long_double_size so that 126..128 selects an appropriate
17843 128-bit floating point type.
17844 (LONG_DOUBLE_TYPE_SIZE): Update comment.
17845 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
17846 source and destination to match the standard usage.
17847 (truncifkf2): Likewise.
17848 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
17849 ISA 2.07 to use an explicit clobber, instead of passing in a
17851 (copysign<mode>3_soft): Likewise.
17853 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17855 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17856 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
17857 (vect_slp_analyze_instance_dependence): Likewise.
17858 (vect_enhance_data_refs_alignment): Likewise.
17859 (vect_analyze_data_refs_alignment): Likewise.
17860 (vect_slp_analyze_and_verify_instance_alignment
17861 (vect_analyze_data_ref_accesses): Likewise.
17862 (vect_prune_runtime_alias_test_list): Likewise.
17863 (vect_analyze_data_refs): Likewise.
17864 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
17865 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17866 (vect_analyze_scalar_cycles_1): Likewise.
17867 (vect_get_loop_niters): Likewise.
17868 (vect_analyze_loop_form_1): Likewise.
17869 (vect_update_vf_for_slp): Likewise.
17870 (vect_analyze_loop_operations): Likewise.
17871 (vect_analyze_loop): Likewise.
17872 (vectorizable_induction): Likewise.
17873 (vect_transform_loop): Likewise.
17874 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
17875 * tree-vect-slp.c (vect_analyze_slp): Likewise.
17876 (vect_make_slp_decision): Likewise.
17877 (vect_detect_hybrid_slp): Likewise.
17878 (vect_slp_analyze_operations): Likewise.
17879 (vect_slp_bb): Likewise.
17880 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
17881 (vectorizable_bswap): Likewise.
17882 (vectorizable_call): Likewise.
17883 (vectorizable_simd_clone_call): Likewise.
17884 (vectorizable_conversion): Likewise.
17885 (vectorizable_assignment): Likewise.
17886 (vectorizable_shift): Likewise.
17887 (vectorizable_operation): Likewise.
17888 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
17890 2018-06-18 Martin Sebor <msebor@redhat.com>
17892 PR tree-optimization/81384
17893 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
17894 * builtins.c (expand_builtin_strnlen): New function.
17895 (expand_builtin): Call it.
17896 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
17897 * builtins.def (BUILT_IN_STRNLEN): New.
17898 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
17899 Warn for bounds in excess of maximum object size.
17900 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
17901 single-value ranges. Handle strnlen.
17902 (handle_builtin_strlen): Handle strnlen.
17903 (strlen_check_and_optimize_stmt): Same.
17904 * doc/extend.texi (Other Builtins): Document strnlen.
17906 2018-06-18 Maya Rashish <coypu@sdf.org>
17908 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
17909 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
17910 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
17912 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
17914 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
17916 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17918 * tree.c (escaped_string::escape): Replace cast to char * by
17919 const_cast<char *> (unescaped).
17921 2018-06-18 Nick Clifton <nickc@redhat.com>
17924 * tree.c (escaped_string): New class. Converts an unescaped
17925 string into its escaped equivalent.
17926 (warn_deprecated_use): Use the new class to convert the
17927 deprecation message, if present.
17928 (test_escaped_strings): New self test.
17929 (test_c_tests): Add test_escaped_strings.
17930 * doc/extend.texi (deprecated): Add a note that the
17931 deprecation message is affected by the -fmessage-length
17932 option, and that control characters will be escaped.
17933 (#pragma GCC error): Document this pragma.
17934 (#pragma GCC warning): Likewise.
17935 * doc/invoke.texi (-fmessage-length): Document this option's
17936 effect on the #warning and #error preprocessor directives and
17937 the deprecated attribute.
17939 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
17941 * tree.c (decl_value_expr_lookup): Revert latest change.
17942 (decl_value_expr_insert): Likewise.
17944 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
17946 * gimplify.c (nonlocal_vlas): Delete.
17947 (nonlocal_vla_vars): Likewise.
17948 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
17950 (gimplify_body): Do not create and destroy nonlocal_vlas.
17951 * tree-nested.c: Include diagnostic.h.
17952 (use_pointer_in_frame): Tweak.
17953 (lookup_field_for_decl): Add assertion and declare the transformation.
17954 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
17955 internal error when the reference is in a wrong context. Do not
17956 create a debug decl by default.
17957 (note_nonlocal_block_vlas): Delete.
17958 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
17959 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
17960 create a debug decl by default.
17961 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
17962 call is in a wrong context.
17963 (fixup_vla_decls): New function.
17964 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
17965 debug variables were created.
17966 * tree.c (decl_value_expr_lookup): Add checking assertion.
17967 (decl_value_expr_insert): Likewise.
17969 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17971 PR middle-end/82479
17972 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
17973 * tree-scalar-evolution.c (interpret_expr): Likewise.
17974 (expression_expensive_p): Likewise.
17975 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
17976 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
17977 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
17978 (ssa_defined_by_minus_one_stmt_p): New.
17980 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17982 PR middle-end/64946
17983 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
17984 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
17985 * dojump.c (do_jump): Likewise.
17986 * expr.c (expand_expr_real_2): Check operand type's sign.
17987 * fold-const.c (const_unop): Handle ABSU_EXPR.
17988 (fold_abs_const): Likewise.
17989 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
17990 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
17991 (strip_sign_op_1): Likesise.
17992 * match.pd: Add new pattern to generate ABSU_EXPR.
17993 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
17994 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17995 * tree-eh.c (operation_could_trap_helper_p): Likewise.
17996 * tree-inline.c (estimate_operator_cost): Likewise.
17997 * tree-pretty-print.c (dump_generic_node): Likewise.
17998 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
17999 * tree.def (ABSU_EXPR): New.
18001 2018-06-16 Jakub Jelinek <jakub@redhat.com>
18003 PR middle-end/86095
18004 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
18005 documented as preserved for backward compatibility only.
18006 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
18008 PR rtl-optimization/86108
18009 * bb-reorder.c (create_forwarder_block): Renamed to ...
18010 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
18011 jump from new landing pad to the second part.
18012 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
18015 2018-06-15 Jakub Jelinek <jakub@redhat.com>
18017 PR middle-end/85878
18018 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
18019 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
18020 Only call store_expr for halves if the mode is the same.
18022 PR middle-end/86123
18023 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
18024 Fix up comment formatting.
18026 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
18028 * typed-splay-tree.h (typed_splay_tree::remove): New function.
18029 (typed_splay_tree::closure,
18030 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
18031 (typed_splay_tree::typed_splay_tree,
18032 typed_splay_tree::operator =): Declared private.
18033 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
18034 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
18035 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
18036 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
18037 typed_splay_tree::splay_tree_splay,
18038 typed_splay_tree::splay_tree_foreach_helper,
18039 typed_splay_tree::splay_tree_insert,
18040 typed_splay_tree::splay_tree_remove,
18041 typed_splay_tree::splay_tree_lookup,
18042 typed_splay_tree::splay_tree_predecessor,
18043 typed_splay_tree::splay_tree_successor,
18044 typed_splay_tree::splay_tree_min,
18045 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
18046 (typed_splay_tree::root, typed_splay_tree::comp,
18047 typed_splay_tree::delete_key,
18048 typed_splay_tree::delete_value): New data members.
18049 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
18050 typed_splay_tree::remove.
18052 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
18054 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
18055 -mginv and -mno-ginv to the assembler.
18056 * config/mips/mips.opt (-mcrc): New option.
18057 (-mginv): Likewise.
18058 * doc/invoke.text (-mcrc): Document.
18059 (-mginv): Likewise.
18061 2018-06-15 Nick Clifton <nickc@redhat.com>
18064 * tree.c (escaped_string): New class. Converts an unescaped
18065 string into its escaped equivalent.
18066 (warn_deprecated_use): Use the new class to convert the
18067 deprecation message, if present.
18068 (test_escaped_strings): New self test.
18069 (test_c_tests): Add test_escaped_strings.
18070 * doc/extend.texi (deprecated): Add a note that the
18071 deprecation message is affected by the -fmessage-length
18072 option, and that control characters will be escaped.
18073 (#pragma GCC error): Document this pragma.
18074 (#pragma GCC warning): Likewise.
18075 * doc/invoke.texi (-fmessage-length): Document this option's
18076 effect on the #warning and #error preprocessor directives and
18077 the deprecated attribute.
18079 2018-06-15 Richard Biener <rguenther@suse.de>
18081 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
18082 here, also noting vector size used.
18083 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
18084 size used in MSG_OPTIMIZED_LOCATIONS dump.
18085 (pass_slp_vectorize::execute): Adjust.
18087 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
18090 * config/arc/arc.c (arc_return_address_register): Fix
18093 2018-06-15 Richard Biener <rguenther@suse.de>
18095 PR middle-end/86159
18096 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
18097 leave useless conversion stripping to force_gimple_operand_gsi.
18098 (gimplify_build2): Likewise.
18099 (gimplify_build1): Likewise.
18101 2018-06-15 Richard Biener <rguenther@suse.de>
18103 PR middle-end/86076
18104 * tree-cfg.c (move_stmt_op): unshare invariant addresses
18105 before adjusting their block.
18107 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
18109 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
18110 multilibs for *-*-rtems*.
18111 * config/riscv/t-rtems: New file.
18113 2018-06-14 Jakub Jelinek <jakub@redhat.com>
18115 PR middle-end/86122
18116 * match.pd ((A +- CST1) +- CST2): Punt if last resort
18117 unsigned_type_for returns NULL.
18120 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
18121 subregs of multi-word pseudos unless the float mode has word size.
18123 2018-06-14 Richard Biener <rguenther@suse.de>
18125 PR middle-end/86139
18126 * tree-vect-generic.c (build_word_mode_vector_type): Remove
18127 duplicate and harmful type_hash_canon.
18128 * tree.c (type_hash_canon): Assert we didn't find ourselves.
18130 2018-06-14 Richard Biener <rguenther@suse.de>
18133 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
18136 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
18138 * config/rtems.h (STDINT_LONG32): Define.
18140 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
18141 Prachi Godbole <prachi.godbole@imgtec.com>
18143 * config/mips/mips-cpus.def: Define P6600.
18144 * config/mips/mips-tables.opt: Regenerate.
18145 * config/mips/mips.c (mips_ucbranch_type): New enum.
18146 (mips_rtx_cost_data): Add support for P6600.
18147 (mips_issue_rate): Likewise.
18148 (mips_multipass_dfa_lookahead): Likewise.
18149 (mips_avoid_hazard): Likewise.
18150 (mips_reorg_process_insns): Likewise.
18151 (mips_classify_branch_p6600): New function.
18152 * config/mips/mips.h (TUNE_P6600): New define.
18153 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
18154 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
18155 * config/mips/mips.md: Include p6600.md.
18156 (processor): Add p6600.
18157 * config/mips/p6600.md: New file.
18158 * doc/invoke.texi: Add p6600 to supported architectures.
18160 2018-06-13 Martin Sebor <msebor@redhat.com>
18162 PR tree-optimization/86114
18163 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
18165 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
18167 2018-06-13 Richard Biener <rguenther@suse.de>
18169 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
18170 Properly set vector type of the intermediate stmt.
18171 * tree-vect-stmts.c (vectorizable_operation): The destination
18172 var always has vectype_out type.
18174 2018-06-13 Jeff Law <law@redhat.com>
18176 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
18177 integer 0 for argument to print_rtl_with_bb.
18178 (rl78_reorg): Likewise.
18180 2018-06-13 David Malcolm <dmalcolm@redhat.com>
18182 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
18183 from rtx to rtx_insn *.
18184 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
18186 (add_sched_insns_for_speculation): Likewise for local "target",
18187 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
18188 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
18189 from rtx_insn ** to rtx_code_label **.
18190 (reorg_emit_nops): Likewise.
18191 (c6x_reorg): Likewise for local "call_labels".
18192 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
18194 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
18195 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
18196 the loops over LABEL_REFS.
18197 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
18199 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
18200 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
18201 (split_branches): Strengthen local "olabel" from rtx to
18202 rtx_insn *, adding a safe_as_a cast.
18203 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
18205 (add_insn_after): Likewise for first two params.
18206 (add_insn_before): Likewise.
18207 (remove_insn): Likewise for param.
18208 (emit_pattern_before_noloc): Likewise for second and third params.
18209 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
18210 (emit_call_insn_before_noloc): Likewise.
18211 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
18213 (emit_barrier_before): Likewise.
18214 (emit_label_before): Strengthen "label" param from "rtx" to
18215 "rtx_code_label *". Strengthen "before" param from "rtx" to
18217 (emit_insn_after_1): Strengthen "after" param from "rtx" to
18219 (emit_pattern_after_noloc): Likewise.
18220 (emit_insn_after_noloc): Likewise.
18221 (emit_jump_insn_after_noloc): Likewise.
18222 (emit_call_insn_after_noloc): Likewise.
18223 (emit_debug_insn_after_noloc): Likewise.
18224 (emit_barrier_after): Likewise.
18225 (emit_label_after): Likewise for both params.
18226 (emit_pattern_after_setloc): Likewise for "after" param. Convert
18227 "loc" param from "int" to "location_t".
18228 (emit_insn_after_setloc): Likewise.
18229 (emit_jump_insn_after_setloc): Likewise.
18230 (emit_call_insn_after_setloc): Likewise.
18231 (emit_debug_insn_after_setloc): Likewise.
18232 (emit_pattern_before_setloc): Likewise for "before" param. Convert
18233 "loc" param from "int" to "location_t".
18234 (emit_pattern_before): Convert NULL_RTX to NULL.
18235 (emit_insn_before_setloc): Convert "loc" param from "int" to
18237 (emit_jump_insn_before_setloc): Likewise.
18238 (emit_call_insn_before_setloc): Likewise.
18239 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
18240 rtx_insn *. Convert "loc" param from "int" to "location_t".
18241 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
18242 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
18243 Convert 3rd param from "int" to "location_t".
18244 (emit_barrier_before, emit_barrier_after, next_real_insn):
18245 Strengthen param from rtx to rtx_insn *.
18246 (emit_label_before): Strengthen 1st param from "rtx" to
18247 "rtx_code_label *". Strengthen 2nd param from "rtx" to
18249 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
18250 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
18251 Strengthen 2nd param from "rtx" to "rtx_insn *".
18252 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
18253 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
18254 Likewise. Convert 3rd param from "int" to "location_t".
18255 (emit_label_after): Strengthen 1st param from "rtx" to
18256 "rtx_code_label *".
18257 (next_real_insn, remove_insn): Strengthen param from "rtx" to
18259 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
18260 from "rtx" to "rtx_insn *".
18262 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
18264 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
18265 bodies streamed in with -Q.
18266 * dumpfile.c (dump_files): Add lto-stream-out dump file.
18267 * dumpfile.h (tree_dump_index): Add lto_stream_out.
18268 * gimple-streamer-out.c: Include gimple-pretty-print.h
18269 (output_bb): Dump stmts streamed.
18270 * lto-section-out.c: Include print-tree.h
18271 (lto_begin_section): Dump sections created.
18272 (lto_output_decl_index): Dump decl encoded.
18273 * lto-streamer-out.c: Include print-tree.h
18274 (create_output_block): Dump output block created.
18275 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
18276 (output_function): Dump function output.
18277 (output_constructor): Dump constructor streamed.
18278 (write_global_stream): Output indexes encoded.
18279 (produce_asm_for_decls): Dump streams encoded.
18280 * lto-streamer.c (streamer_dump_file): New global var.
18281 * lto-streamer.h (streamer_dump_file): Declare.
18282 * passes.c (ipa_write_summaries): Initialize streamer dump.
18283 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
18286 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
18289 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
18290 offsets for register save directives. Emit a second batch of save
18291 directives, if need be, when the function accesses prior frames.
18293 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18295 * config/arc/fpu.md (fmasf4): Force operand to register.
18296 (fnmasf4): Likewise.
18298 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18300 * config/arc/arc-protos.h (arc_pad_return): Remove.
18301 * config/arc/arc.c (machine_function): Remove force_short_suffix
18303 (arc_print_operand): Adjust printing of '&'.
18304 (arc_verify_short): Remove conditional printing of short suffix.
18305 (arc_final_prescan_insn): Remove reference to size_reason.
18306 (pad_return): New function.
18307 (arc_reorg): Call pad_return.
18308 (arc_pad_return): Remove.
18309 (arc_init_machine_status): Remove reference to force_short_suffix.
18310 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
18311 (attr length): When attribute iscompact is true force to 2
18312 regardless; in the case of maybe check if we want to force the
18313 instruction to have 4 bytes length.
18314 (nopv): Change it to generate 4 byte long nop as well.
18315 (blockage): New pattern.
18316 (simple_return): Remove call to arc_pad_return.
18317 (p_return_i): Likewise.
18319 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18321 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
18323 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18325 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
18328 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18330 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
18331 for ARC700 and ARCv2.
18333 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
18336 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
18337 operands[2] instead of operands[1].
18340 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18342 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
18343 case, check whether the outer register overlaps an unallocatable
18344 register, not just whether it fits the required class.
18346 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18348 * poly-int.h (can_div_trunc_p): Add new overload in which all values
18350 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
18351 (memrefs_conflict_p): Likewise.
18352 (init_alias_analysis): Likewise.
18353 * cfgexpand.c (expand_debug_expr): Likewise.
18354 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
18355 * cse.c (fold_rtx): Likewise.
18356 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
18357 * expr.c (emit_block_move_hints): Likewise.
18358 (clear_storage_hints, push_block, emit_push_insn): Likewise.
18359 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
18360 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
18361 (emit_group_store): Likewise.
18362 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
18363 to read the PRE/POST_MODIFY increment.
18364 * calls.c (store_one_arg): Use strip_offset.
18365 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
18367 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
18369 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
18370 (simplify_binary_operation_1): Extend CONST_INT handling to
18372 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
18373 than a HOST_WIDE_INT.
18374 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
18376 (adjust_mems, add_stores): Update accodingly.
18377 (vt_canonicalize_addr): Track polynomial offsets.
18378 (emit_note_insn_var_location): Likewise.
18379 (vt_add_function_parameter): Likewise.
18380 (vt_initialize): Likewise.
18382 2018-06-12 Jeff Law <law@redhat.com>
18384 * config.gcc (alpha*-*-freebsd*): Remove.
18385 * config/alpha/freebsd.h: Remove.
18387 2018-06-12 David Malcolm <dmalcolm@redhat.com>
18390 * spellcheck-tree.c (levenshtein_distance): Rename to...
18391 (get_edit_distance): ...this, and update for underlying renaming.
18392 * spellcheck-tree.h (levenshtein_distance): Rename to...
18393 (get_edit_distance): ...this.
18394 * spellcheck.c (levenshtein_distance): Rename to...
18395 (get_edit_distance): ...this. Convert from Levenshtein distance
18396 to Damerau-Levenshtein distance by supporting transpositions of
18397 adjacent characters. Rename "v1" to "v_next" and "v0" to
18399 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
18400 (selftest::test_edit_distance_unit_test_oneway): ...this, and
18401 update for underlying renaming.
18402 (selftest::levenshtein_distance_unit_test): Rename to...
18403 (selftest::test_get_edit_distance_unit): ...this, and update for
18404 underlying renaming.
18405 (selftest::test_find_closest_string): Add example from PR 69968
18406 where transposition helps
18407 (selftest::test_metric_conditions): Update for renaming.
18408 (selftest::test_metric_conditions): Likewise.
18409 (selftest::spellcheck_c_tests): Likewise.
18410 * spellcheck.h (levenshtein_distance): Rename both overloads to...
18411 (get_edit_distance): ...this.
18412 (best_match::consider): Update for renaming.
18414 2018-06-12 Martin Sebor <msebor@redhat.com>
18416 PR tree-optimization/85259
18417 * builtins.c (compute_objsize): Handle constant offsets.
18418 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
18419 true iff a warning has been issued.
18420 * gimple.h (gimple_nonartificial_location): New function.
18421 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
18422 gimple_nonartificial_location and handle -Wno-system-headers.
18423 (handle_builtin_stxncpy): Same.
18425 2018-06-12 Martin Sebor <msebor@redhat.com>
18428 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
18430 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18432 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18433 BUILTIN_VEC_XST entries for pointer to double and long long.
18435 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
18438 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
18440 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
18443 2018-06-12 Martin Liska <mliska@suse.cz>
18445 * doc/options.texi: Document IntegerRange.
18447 2018-06-12 Martin Liska <mliska@suse.cz>
18449 * config/i386/i386.opt: Make MPX-related options as Deprecated.
18450 * opt-functions.awk: Handle Deprecated flag.
18451 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
18453 (read_cmdline_option): Report warning for a deprecated option.
18454 * opts.h (struct cl_option): Add new field cl_deprecated.
18455 (CL_ERR_DEPRECATED): New.
18457 2018-06-12 Martin Liska <mliska@suse.cz>
18459 * doc/options.texi: Document Deprecated option flag.
18461 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18463 * config/arc/arc-arch.h (arc_extras): New enum.
18464 (arc_cpu_t):Add field extra.
18465 (arc_cpu_types): Consider the extras.
18466 * config/arc/arc-cpus.def: Add extras info.
18467 * config/arc/arc-opts.h (processor_type): Consider extra field.
18468 * config/arc/arc.c (arc_override_options): Handle extra field.
18470 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18472 * config/arc/arc-arch.h: Update ARC_OPTX macro.
18473 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
18475 * config/arc/arc.c (arc_init): Update pic warning.
18476 (irq_range): Update irq range parsing warnings.
18477 (arc_override_options): Update various warning messages.
18478 (arc_handle_aux_attribute): Likewise.
18480 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
18482 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
18484 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18486 * doc/sourcebuild.texi: Document usage of line number 0 in verify
18487 compiler messages directives.
18489 2018-06-12 Matthew Fortune <mfortune@gmail.com>
18491 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
18492 * config/mips/mips-tables.opt: Regenerate.
18493 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
18495 * doc/invoke.texi: Document -march=i6500.
18497 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
18499 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
18500 (i6400_gpmul): Add cpu_unit.
18501 (i6400_gpdiv): Likewise.
18502 (i6400_msa_add_d): Update reservations.
18503 (i6400_msa_int_add) Likewise.
18504 (i6400_msa_short_logic3) Likewise.
18505 (i6400_msa_short_logic2) Likewise.
18506 (i6400_msa_short_logic) Likewise.
18507 (i6400_msa_move) Likewise.
18508 (i6400_msa_cmp) Likewise.
18509 (i6400_msa_short_float2) Likewise.
18510 (i6400_msa_div_d) Likewise.
18511 (i6400_msa_long_logic1) Likewise.
18512 (i6400_msa_long_logic2) Likewise.
18513 (i6400_msa_mult) Likewise.
18514 (i6400_msa_long_float2) Likewise.
18515 (i6400_msa_long_float4) Likewise.
18516 (i6400_msa_long_float5) Likewise.
18517 (i6400_msa_long_float8) Likewise.
18518 (i6400_fpu_fadd): Include frint type.
18519 (i6400_fpu_store): New define_insn_reservation.
18520 (i6400_fpu_load): Likewise.
18521 (i6400_fpu_move): Likewise.
18522 (i6400_fpu_fcmp): Likewise.
18523 (i6400_fpu_fmadd): Likewise.
18524 (i6400_int_mult): Include imul3nc type and update reservation.
18525 (i6400_int_div): Include idiv3 type and update reservation.
18526 (i6400_int_load): Update to check type not move_type.
18527 (i6400_int_store): Likewise.
18528 (i6400_int_prefetch): Set zero latency.
18530 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
18532 * gcc.c: Document new %@{...} sequence.
18533 (LINK_COMMAND_SPEC): Use it for the -L switches.
18534 (cpp_unique_options): Use it for the -I switches.
18535 (at_file_argbuf): New global variable.
18536 (in_at_file): Likewise.
18537 (alloc_args): Create at_file_argbuf.
18538 (clear_args): Truncate at_file_argbuf.
18539 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
18540 (open_at_file): New function.
18541 (close_at_file): Likewise.
18542 (create_at_file): Delete.
18543 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
18546 (validate_switches_from_spec): Deal with %@{...} sequence.
18547 (validate_switches): Likewise.
18548 (driver::finalize): Call clear_args.
18550 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
18552 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
18554 2018-06-11 Martin Sebor <msebor@redhat.com>
18556 * doc/invoke.texi (-Wall): List -Wc++17-compat.
18557 (Wno-class-memaccess): Add @opindex.
18558 (Wno-templates, Wno-multiple-inheritance): Same.
18559 (Wno-virtual-inheritance, Wno-namespaces): Same.
18560 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
18561 (Wno-format-overflow, Wno-format-truncation): Same.
18562 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
18563 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
18564 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
18565 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
18566 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
18567 (Wno-misspelled-isr): Same.
18569 2018-06-11 Martin Sebor <msebor@redhat.com>
18571 * PR tree-optimization/86083
18572 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
18574 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
18576 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
18578 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
18581 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
18582 on the correct operand.
18583 (*movdi_internal64): Ditto.
18585 2018-06-11 Martin Liska <mliska@suse.cz>
18587 PR tree-optimization/86089
18588 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
18590 2018-06-11 Julia Koval <julia.koval@intel.com>
18592 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
18593 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
18594 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
18596 2018-06-11 Olivier Hainque <hainque@adacore.com>
18598 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
18599 for Ada with strict dwarf2.
18601 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
18604 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
18607 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
18609 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
18611 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
18613 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
18616 2018-06-08 Martin Liska <mliska@suse.cz>
18618 * tree-cfg.h (debug_function): Fix argument type to match
18621 2018-06-08 Martin Liska <mliska@suse.cz>
18623 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
18624 Remove usage of MPX-related (and removed) fields.
18625 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
18627 2018-06-08 David Malcolm <dmalcolm@redhat.com>
18629 * cfg.c (debug): Use TDF_NONE rather than 0.
18630 * cfghooks.c (debug): Likewise.
18631 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
18632 (struct dump_option_value_info): Convert to...
18633 (struct kv_pair): ...this template type.
18634 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
18636 (optinfo_verbosity_options): Likewise.
18637 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
18639 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
18640 than int for "optgroup_flags" param.
18641 (dump_generic_expr_loc): Use dump_flags_t rather than int for
18643 (dump_dec): Likewise.
18644 (dump_finish): Use TDF_NONE rather than 0.
18645 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
18646 rather than int for "optgroup_flags" param. Use TDF_NONE rather
18647 than 0. Update for change to option_ptr.
18648 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
18649 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
18650 0. Update for changes to optinfo_verbosity_options and
18652 (opt_info_switch_p): Convert optgroup_flags from int to
18654 (dump_basic_block): Use dump_flags_t rather than int
18655 for "dump_kind" param.
18656 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
18657 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
18658 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
18659 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
18660 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
18661 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
18662 TDF_NONE): Convert from macros to...
18663 (enum dump_flag): ...this new enum.
18664 (dump_flags_t): Update to use enum.
18665 (operator|, operator&, operator~, operator|=, operator&=):
18666 Implement for dump_flags_t.
18667 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
18668 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
18669 Convert from macros to...
18670 (enum optgroup_flag): ...this new enum.
18671 (optgroup_flags_t): New typedef.
18672 (operator|, operator|=): Implement for optgroup_flags_t.
18673 (struct dump_file_info): Convert field "alt_flags" to
18674 dump_flags_t. Convert field "optgroup_flags" to
18676 (dump_basic_block): Use dump_flags_t rather than int for param.
18677 (dump_generic_expr_loc): Likewise.
18678 (dump_dec): Likewise.
18679 (dump_register): Convert param "optgroup_flags" to
18681 (opt_info_enable_passes): Likewise.
18682 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
18684 * gimple-pretty-print.c (debug): Likewise.
18685 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
18686 (merged_store_group::apply_stores): Likewise.
18687 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
18688 * gimple.c (verify_gimple_pp): Likewise.
18689 * graphite-poly.c (print_pbb_body): Likewise.
18690 * passes.c (pass_manager::register_one_dump_file): Convert
18691 local "optgroup_flags" to optgroup_flags_t.
18692 * print-tree.c (print_node): Use TDF_NONE rather than 0.
18694 (debug_body): Likewise.
18695 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
18696 to optgroup_flags_t.
18697 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
18699 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
18700 (convert_mult_to_fma): Likewise.
18701 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
18702 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
18703 * tree-vect-data-refs.c (dump_lower_bound): Convert param
18704 "dump_kind" to dump_flags_t.
18706 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18708 * config/rs6000/rs6000.c (min, max): Delete.
18710 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18712 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
18713 -mabi=spe and -mabi=no-spe.
18715 2018-06-08 Martin Liska <mliska@suse.cz>
18717 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
18718 where we expect an existing summary.
18720 2018-06-08 Martin Liska <mliska@suse.cz>
18722 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
18723 * ipa-inline.h (estimate_edge_growth): Likewise.
18725 2018-06-08 Martin Liska <mliska@suse.cz>
18727 * cgraph.c (function_version_hasher::hash): Use
18728 cgraph_node::get_uid ().
18729 (function_version_hasher::equal):
18730 * cgraph.h (cgraph_node::get_uid): New method.
18731 * ipa-inline.c (update_caller_keys): Use
18732 cgraph_node::get_uid ().
18733 (update_callee_keys): Likewise.
18734 * ipa-utils.c (searchc): Likewise.
18735 (ipa_reduced_postorder): Likewise.
18736 * lto-cgraph.c (input_node): Likewise.
18737 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
18738 * symbol-summary.h (symtab_insertion): Likewise.
18739 (symtab_removal): Likewise.
18740 (symtab_duplication): Likewise.
18741 * tree-pretty-print.c (dump_function_header): Likewise.
18742 * tree-sra.c (convert_callers_for_node): Likewise.
18744 2018-06-08 Martin Liska <mliska@suse.cz>
18746 * cgraph.c (symbol_table::create_edge): Always assign a new
18748 (symbol_table::free_edge): Do not recycle numbers.
18749 * cgraph.h (cgraph_edge::get): New method.
18750 * symbol-summary.h (symtab_removal): Use it.
18751 (symtab_duplication): Likewise.
18752 (call_summary::hashable_uid): Remove.
18754 2018-06-08 Martin Liska <mliska@suse.cz>
18756 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
18757 (initialize_growth_caches): Remove.
18758 (free_growth_caches): Likewise.
18759 (do_estimate_edge_time): Use edge_growth_cache.
18760 (do_estimate_edge_size): Likewise.
18761 (do_estimate_edge_hints): Likewise.
18762 * ipa-inline.c (reset_edge_caches): Likewise.
18763 (recursive_inlining): Likewise.
18764 (inline_small_functions): Likewise.
18765 * ipa-inline.h (initialize_growth_caches): Remove.
18766 (estimate_edge_size): Likewise.
18767 (estimate_edge_time): Likewise.
18768 (estimate_edge_hints): Likewise.
18769 (reset_edge_growth_cache): Likewise.
18770 * symbol-summary.h (call_summary::remove): New method.
18772 2018-06-08 Martin Liska <mliska@suse.cz>
18774 * ipa-cp.c (class edge_clone_summary): New summary.
18775 (grow_edge_clone_vectors): Remove.
18776 (ipcp_edge_duplication_hook): Remove.
18777 (class edge_clone_summary_t): New call_summary class.
18778 (ipcp_edge_removal_hook): Remove.
18779 (edge_clone_summary_t::duplicate): New function.
18780 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
18781 (create_specialized_node): Likewise.
18782 (ipcp_driver): Initialize edge_clone_summaries and do not
18785 2018-06-08 Martin Liska <mliska@suse.cz>
18787 * symbol-summary.h (get): New function.
18788 (call_summary::m_initialize_when_cloning): New class member.
18790 2018-06-08 Martin Liska <mliska@suse.cz>
18792 * cgraph.c (cgraph_node::remove): Do not recycle uid.
18793 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
18794 (symbol_table::allocate_cgraph_symbol): Do not set uid.
18795 * passes.c (uid_hash_t): Record removed_nodes by their uids.
18796 (remove_cgraph_node_from_order): Use the removed_nodes set.
18797 (do_per_function_toporder): Likwise.
18798 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
18799 instead of summary_uid.
18800 (symtab_removal): Likewise.
18801 (symtab_duplication): Likewise.
18803 2018-06-08 Martin Liska <mliska@suse.cz>
18805 * ipa-cp.c (ipcp_store_bits_results): Use
18806 ipcp_transformation_sum.
18807 (ipcp_store_vr_results): Likewise.
18808 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
18810 (ipcp_transformation_initialize): ... this.
18811 (ipa_set_node_agg_value_chain):
18812 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
18813 (write_ipcp_transformation_info): Likewise.
18814 (read_ipcp_transformation_info): Likewise.
18815 (ipcp_update_bits): Likewise.
18816 (ipcp_update_vr): Likewise.
18817 (ipcp_transform_function): Likewise.
18818 * ipa-prop.h: Rename ipcp_transformation_summary to
18819 ipcp_transformation.
18820 (class ipcp_transformation_t): New function summary.
18821 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
18822 (ipa_get_agg_replacements_for_node): Likewise.
18824 2018-06-08 Martin Liska <mliska@suse.cz>
18826 * ipa-pure-const.c (struct funct_state_d): Do it class instead
18828 (class funct_state_summary_t): New function_summary class.
18829 (has_function_state): Remove.
18830 (get_function_state): Likewise.
18831 (set_function_state): Likewise.
18832 (add_new_function): Likewise.
18833 (funct_state_summary_t::insert): New function.
18834 (duplicate_node_data): Remove.
18835 (remove_node_data): Remove.
18836 (funct_state_summary_t::duplicate): New function.
18837 (register_hooks): Create new funct_state_summaries.
18838 (pure_const_generate_summary): Use it.
18839 (pure_const_write_summary): Likewise.
18840 (pure_const_read_summary): Likewise.
18841 (propagate_pure_const): Likewise.
18842 (propagate_nothrow): Likewise.
18843 (dump_malloc_lattice): Likewise.
18844 (propagate_malloc): Likewise.
18845 (execute): Do not register hooks, just remove summary
18847 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
18850 2018-06-08 Martin Liska <mliska@suse.cz>
18852 * ipa-reference.c (remove_node_data): Remove.
18853 (duplicate_node_data): Likewise.
18854 (class ipa_ref_var_info_summary_t): New class.
18855 (class ipa_ref_opt_summary_t): Likewise.
18856 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
18857 (get_reference_optimization_summary): Use
18858 ipa_ref_opt_sum_summaries.
18859 (set_reference_vars_info): Remove.
18860 (set_reference_optimization_summary): Likewise.
18861 (ipa_init): Create summaries.
18862 (init_function_info): Use function summary.
18863 (ipa_ref_opt_summary_t::duplicate): New function.
18864 (ipa_ref_opt_summary_t::remove): New function.
18865 (get_read_write_all_from_node): Fix GNU coding style.
18866 (propagate): Use function summary.
18867 (write_node_summary_p): Fix GNU coding style.
18868 (stream_out_bitmap): Likewise.
18869 (ipa_reference_read_optimization_summary): Use function summary.
18870 (ipa_reference_c_finalize): Do not release hooks.
18872 2018-06-08 Martin Liska <mliska@suse.cz>
18874 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
18875 (analyze_function_body): Extract multiple calls of get_create.
18876 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18877 * ipa-inline.c (recursive_inlining): Use ::get method.
18878 * ipa-inline.h (estimate_edge_growth): Likewise.
18880 2018-06-08 Martin Liska <mliska@suse.cz>
18882 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
18884 (hsa_function_summary::hsa_function_summary): Use the new enum
18886 (hsa_gpu_implementation_p): Use hsa_summaries::get.
18887 * hsa-gen.c (hsa_get_host_function): Likewise.
18888 (get_brig_function_name): Likewise.
18889 * ipa-hsa.c (process_hsa_functions): Likewise.
18890 (ipa_hsa_write_summary): Likewise.
18891 * symbol-summary.h (symtab_duplication): Use ::get function/
18892 (get): New function.
18894 2018-06-08 Martin Liska <mliska@suse.cz>
18896 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
18898 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
18899 (hsa_register_kernel): Likewise.
18900 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
18901 * hsa-gen.c (hsa_get_host_function): Likewise.
18902 (get_brig_function_name): Likewise.
18903 (generate_hsa): Likewise.
18904 (pass_gen_hsail::execute): Likewise.
18905 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
18906 (devirtualization_time_bonus): Likewise.
18907 (ipcp_propagate_stage): Likewise.
18908 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
18909 (edge_set_predicate): Likewise.
18910 (evaluate_conditions_for_known_args): Likewise.
18911 (evaluate_properties_for_edge): Likewise.
18912 (ipa_fn_summary::reset): Likewise.
18913 (ipa_fn_summary_t::duplicate): Likewise.
18914 (dump_ipa_call_summary): Likewise.
18915 (ipa_dump_fn_summary): Likewise.
18916 (analyze_function_body): Likewise.
18917 (compute_fn_summary): Likewise.
18918 (estimate_edge_devirt_benefit): Likewise.
18919 (estimate_edge_size_and_time): Likewise.
18920 (estimate_calls_size_and_time): Likewise.
18921 (estimate_node_size_and_time): Likewise.
18922 (inline_update_callee_summaries): Likewise.
18923 (remap_edge_change_prob): Likewise.
18924 (remap_edge_summaries): Likewise.
18925 (ipa_merge_fn_summary_after_inlining): Likewise.
18926 (ipa_update_overall_fn_summary): Likewise.
18927 (read_ipa_call_summary): Likewise.
18928 (inline_read_section): Likewise.
18929 (write_ipa_call_summary): Likewise.
18930 (ipa_fn_summary_write): Likewise.
18931 (ipa_free_fn_summary): Likewise.
18932 * ipa-hsa.c (process_hsa_functions): Likewise.
18933 (ipa_hsa_write_summary): Likewise.
18934 (ipa_hsa_read_section): Likewise.
18935 * ipa-icf.c (sem_function::merge): Likewise.
18936 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18937 (do_estimate_edge_time): Likewise.
18938 (estimate_size_after_inlining): Likewise.
18939 (estimate_growth): Likewise.
18940 (growth_likely_positive): Likewise.
18941 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
18942 (inline_call): Likewise.
18943 * ipa-inline.c (caller_growth_limits): Likewise.
18944 (can_inline_edge_p): Likewise.
18945 (can_inline_edge_by_limits_p): Likewise.
18946 (compute_uninlined_call_time): Likewise.
18947 (compute_inlined_call_time): Likewise.
18948 (want_inline_small_function_p): Likewise.
18949 (edge_badness): Likewise.
18950 (update_caller_keys): Likewise.
18951 (update_callee_keys): Likewise.
18952 (recursive_inlining): Likewise.
18953 (inline_small_functions): Likewise.
18954 (inline_to_all_callers_1): Likewise.
18955 (dump_overall_stats): Likewise.
18956 (early_inline_small_functions): Likewise.
18957 (early_inliner): Likewise.
18958 * ipa-inline.h (estimate_edge_growth): Likewise.
18959 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18960 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18961 * ipa-prop.h (IPA_NODE_REF): Likewise.
18962 (IPA_EDGE_REF): Likewise.
18963 * ipa-pure-const.c (malloc_candidate_p): Likewise.
18964 (propagate_malloc): Likewise.
18965 * ipa-split.c (execute_split_functions): Likewise.
18966 * symbol-summary.h: Rename get to get_create.
18968 (get_create): Likewise.
18969 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
18971 2018-06-08 Martin Liska <mliska@suse.cz>
18973 * symbol-summary.h (release): Move definition out of class
18975 (symtab_removal): Likewise.
18976 (symtab_duplication): Likewise.
18978 2018-06-08 Martin Liska <mliska@suse.cz>
18980 * symbol-summary.h (function_summary): Move constructor
18981 implementation out of class declaration.
18982 (release): Likewise.
18983 (symtab_insertion): Likewise.
18984 (symtab_removal): Likewise.
18985 (symtab_duplication): Likewise.
18988 2018-06-08 Martin Liska <mliska@suse.cz>
18990 * Makefile.in: Remove support for MPX (macros, related functions,
18991 fields in cgraph_node, ...).
18992 * builtin-types.def (BT_BND): Likewise.
18993 (BT_FN_BND_CONST_PTR): Likewise.
18994 (BT_FN_CONST_PTR_BND): Likewise.
18995 (BT_FN_VOID_PTR_BND): Likewise.
18996 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
18997 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
18998 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
18999 (expand_builtin_mempcpy_with_bounds): Likewise.
19000 (expand_builtin_memset_with_bounds): Likewise.
19001 (expand_builtin_memset_args): Likewise.
19002 (std_expand_builtin_va_start): Likewise.
19003 (expand_builtin): Likewise.
19004 (expand_builtin_with_bounds): Likewise.
19005 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
19006 (DEF_LIB_BUILTIN_CHKP): Likewise.
19007 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
19008 (DEF_CHKP_BUILTIN): Likewise.
19009 (BUILT_IN_MEMCPY): Likewise.
19010 (BUILT_IN_MEMMOVE): Likewise.
19011 (BUILT_IN_MEMPCPY): Likewise.
19012 (BUILT_IN_MEMSET): Likewise.
19013 (BUILT_IN_STPCPY): Likewise.
19014 (BUILT_IN_STRCAT): Likewise.
19015 (BUILT_IN_STRCHR): Likewise.
19016 (BUILT_IN_STRCPY): Likewise.
19017 (BUILT_IN_STRLEN): Likewise.
19018 (BUILT_IN_MEMCPY_CHK): Likewise.
19019 (BUILT_IN_MEMMOVE_CHK): Likewise.
19020 (BUILT_IN_MEMPCPY_CHK): Likewise.
19021 (BUILT_IN_MEMSET_CHK): Likewise.
19022 (BUILT_IN_STPCPY_CHK): Likewise.
19023 (BUILT_IN_STRCAT_CHK): Likewise.
19024 (BUILT_IN_STRCPY_CHK): Likewise.
19025 * calls.c (store_bounds): Likewise.
19026 (emit_call_1): Likewise.
19027 (special_function_p): Likewise.
19028 (maybe_warn_nonstring_arg): Likewise.
19029 (initialize_argument_information): Likewise.
19030 (finalize_must_preallocate): Likewise.
19031 (compute_argument_addresses): Likewise.
19032 (expand_call): Likewise.
19033 * cfgexpand.c (expand_call_stmt): Likewise.
19034 (expand_return): Likewise.
19035 (expand_gimple_stmt_1): Likewise.
19036 (pass_expand::execute): Likewise.
19037 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19038 (cgraph_node::remove): Likewise.
19039 (cgraph_node::dump): Likewise.
19040 (cgraph_node::verify_node): Likewise.
19041 * cgraph.h (chkp_function_instrumented_p): Likewise.
19042 (symtab_node::get_alias_target): Likewise.
19043 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
19044 (cgraph_local_p): Likewise.
19045 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
19046 (cgraph_edge::rebuild_references): Likewise.
19047 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
19048 (walk_polymorphic_call_targets): Likewise.
19049 (cgraph_node::expand_thunk): Likewise.
19050 (symbol_table::output_weakrefs): Likewise.
19051 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
19052 (ix86_handle_option): Likewise.
19053 * config/i386/constraints.md: Likewise.
19054 * config/i386/i386-builtin-types.def (BND): Likewise.
19058 * config/i386/i386-builtin.def (BDESC_END): Likewise.
19059 (BDESC_FIRST): Likewise.
19061 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
19062 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
19063 * config/i386/i386.c (enum reg_class): Likewise.
19064 (ix86_target_string): Likewise.
19065 (ix86_option_override_internal): Likewise.
19066 (ix86_conditional_register_usage): Likewise.
19067 (ix86_valid_target_attribute_inner_p): Likewise.
19068 (ix86_set_indirect_branch_type): Likewise.
19069 (ix86_set_current_function): Likewise.
19070 (ix86_function_arg_regno_p): Likewise.
19071 (init_cumulative_args): Likewise.
19072 (ix86_function_arg_advance): Likewise.
19073 (ix86_function_arg): Likewise.
19074 (ix86_pass_by_reference): Likewise.
19075 (ix86_function_value_regno_p): Likewise.
19076 (ix86_function_value_1): Likewise.
19077 (ix86_function_value_bounds): Likewise.
19078 (ix86_return_in_memory): Likewise.
19079 (ix86_setup_incoming_vararg_bounds): Likewise.
19080 (ix86_va_start): Likewise.
19081 (indirect_thunk_need_prefix): Likewise.
19082 (print_reg): Likewise.
19083 (ix86_print_operand): Likewise.
19084 (ix86_expand_call): Likewise.
19085 (ix86_output_function_return): Likewise.
19086 (reg_encoded_number): Likewise.
19087 (BDESC_VERIFYS): Likewise.
19088 (ix86_init_mpx_builtins): Likewise.
19089 (ix86_init_builtins): Likewise.
19090 (ix86_emit_cmove): Likewise.
19091 (ix86_emit_move_max): Likewise.
19092 (ix86_expand_builtin): Likewise.
19093 (ix86_builtin_mpx_function): Likewise.
19094 (ix86_get_arg_address_for_bt): Likewise.
19095 (ix86_load_bounds): Likewise.
19096 (ix86_store_bounds): Likewise.
19097 (ix86_load_returned_bounds): Likewise.
19098 (ix86_store_returned_bounds): Likewise.
19099 (ix86_class_likely_spilled_p): Likewise.
19100 (ix86_hard_regno_mode_ok): Likewise.
19101 (x86_order_regs_for_local_alloc): Likewise.
19102 (ix86_mitigate_rop): Likewise.
19103 (ix86_bnd_prefixed_insn_p): Likewise.
19104 (ix86_mpx_bound_mode): Likewise.
19105 (ix86_make_bounds_constant): Likewise.
19106 (ix86_initialize_bounds): Likewise.
19107 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
19108 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
19109 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
19110 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
19111 (TARGET_CHKP_BOUND_MODE): Likewise.
19112 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
19113 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
19114 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
19115 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
19116 * config/i386/i386.h (TARGET_MPX): Likewise.
19117 (TARGET_MPX_P): Likewise.
19118 (VALID_BND_REG_MODE): Likewise.
19119 (FIRST_BND_REG): Likewise.
19120 (LAST_BND_REG): Likewise.
19121 (enum reg_class): Likewise.
19122 (BND_REG_P): Likewise.
19123 (BND_REGNO_P): Likewise.
19124 (BNDmode): Likewise.
19125 (ADJUST_INSN_LENGTH): Likewise.
19126 * config/i386/i386.md: Likewise.
19127 * config/i386/i386.opt: Likewise.
19128 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
19129 (defined): Likewise.
19130 (LINK_MPX): Likewise.
19131 (MPX_SPEC): Likewise.
19132 (LIBMPX_SPEC): Likewise.
19133 (LIBMPXWRAPPERS_SPEC): Likewise.
19134 (CHKP_SPEC): Likewise.
19135 * config/i386/predicates.md: Likewise.
19136 * dbxout.c (dbxout_type): Likewise.
19137 * doc/extend.texi: Likewise.
19138 * doc/invoke.texi: Likewise.
19139 * doc/md.texi: Likewise.
19140 * doc/tm.texi: Likewise.
19141 * doc/tm.texi.in: Likewise.
19142 * dwarf2out.c (is_base_type): Likewise.
19143 (gen_formal_types_die): Likewise.
19144 (gen_subprogram_die): Likewise.
19145 (gen_type_die_with_usage): Likewise.
19146 (gen_decl_die): Likewise.
19147 (dwarf2out_late_global_decl): Likewise.
19148 * expr.c (expand_assignment): Likewise.
19149 (emit_storent_insn): Likewise.
19150 (store_expr_with_bounds): Likewise.
19151 (store_expr): Likewise.
19152 (expand_expr_real_1): Likewise.
19153 * expr.h (store_expr_with_bounds): Likewise.
19154 * function.c (use_register_for_decl): Likewise.
19155 (struct bounds_parm_data): Likewise.
19156 (assign_parms_augmented_arg_list): Likewise.
19157 (assign_parm_find_entry_rtl): Likewise.
19158 (assign_parm_is_stack_parm): Likewise.
19159 (assign_parm_load_bounds): Likewise.
19160 (assign_bounds): Likewise.
19161 (assign_parms): Likewise.
19162 (expand_function_start): Likewise.
19163 * gcc.c (CHKP_SPEC): Likewise.
19164 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
19165 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
19166 (wrestrict_dom_walker::check_call): Likewise.
19167 * gimple.c (gimple_build_call_from_tree): Likewise.
19168 * gimple.h (enum gf_mask): Likewise.
19169 (gimple_call_with_bounds_p): Likewise.
19170 (gimple_call_set_with_bounds): Likewise.
19171 * gimplify.c (gimplify_init_constructor): Likewise.
19172 * ipa-cp.c (initialize_node_lattices): Likewise.
19173 (propagate_constants_across_call): Likewise.
19174 (find_more_scalar_values_for_callers_subset): Likewise.
19175 * ipa-hsa.c (process_hsa_functions): Likewise.
19176 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
19177 * ipa-icf.c (sem_function::merge): Likewise.
19178 * ipa-inline.c (early_inliner): Likewise.
19179 * ipa-pure-const.c (warn_function_noreturn): Likewise.
19180 (warn_function_cold): Likewise.
19181 (propagate_pure_const): Likewise.
19182 * ipa-ref.h (enum GTY): Likewise.
19183 * ipa-split.c (find_retbnd): Likewise.
19184 (consider_split): Likewise.
19185 (split_function): Likewise.
19186 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
19187 * ipa.c (walk_polymorphic_call_targets): Likewise.
19188 (symbol_table::remove_unreachable_nodes): Likewise.
19189 (process_references): Likewise.
19190 (cgraph_build_static_cdtor_1): Likewise.
19191 * lto-cgraph.c (lto_output_node): Likewise.
19192 (output_refs): Likewise.
19193 (compute_ltrans_boundary): Likewise.
19194 (input_overwrite_node): Likewise.
19195 (input_node): Likewise.
19196 (input_cgraph_1): Likewise.
19197 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
19198 * passes.c (pass_manager::execute_early_local_passes): Likewise.
19199 (class pass_chkp_instrumentation_passes): Likewise.
19200 (make_pass_chkp_instrumentation_passes): Likewise.
19201 * passes.def: Likewise.
19202 * rtl.h (struct GTY): Likewise.
19203 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
19204 * stor-layout.c (layout_type): Likewise.
19205 * symtab.c: Likewise.
19206 * target.def: Likewise.
19207 * targhooks.c (default_chkp_bound_type): Likewise.
19208 (default_chkp_bound_mode): Likewise.
19209 (default_builtin_chkp_function): Likewise.
19210 (default_chkp_function_value_bounds): Likewise.
19211 (default_chkp_make_bounds_constant): Likewise.
19212 (default_chkp_initialize_bounds): Likewise.
19213 * targhooks.h (default_chkp_bound_type): Likewise.
19214 (default_chkp_bound_mode): Likewise.
19215 (default_builtin_chkp_function): Likewise.
19216 (default_chkp_function_value_bounds): Likewise.
19217 (default_chkp_make_bounds_constant): Likewise.
19218 (default_chkp_initialize_bounds): Likewise.
19219 * toplev.c (compile_file): Likewise.
19220 (process_options): Likewise.
19221 * tree-core.h (DEF_BUILTIN): Likewise.
19222 (DEF_BUILTIN_CHKP): Likewise.
19223 * tree-inline.c (declare_return_variable): Likewise.
19224 (remap_gimple_stmt): Likewise.
19225 (copy_bb): Likewise.
19226 (initialize_inlined_parameters): Likewise.
19227 (expand_call_inline): Likewise.
19228 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
19229 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
19230 (make_pass_ipa_chkp_produce_thunks): Likewise.
19231 (make_pass_chkp): Likewise.
19232 (make_pass_chkp_opt): Likewise.
19233 (make_pass_chkp_instrumentation_passes): Likewise.
19234 * tree-pretty-print.c (dump_generic_node): Likewise.
19235 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
19236 * tree-ssa-dce.c (propagate_necessity): Likewise.
19237 (eliminate_unnecessary_stmts): Likewise.
19238 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
19239 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
19240 * tree-ssa-sccvn.h: Likewise.
19241 * tree-ssa-strlen.c (get_string_length): Likewise.
19242 (valid_builtin_call): Likewise.
19243 (adjust_last_stmt): Likewise.
19244 (handle_builtin_strchr): Likewise.
19245 (handle_builtin_strcpy): Likewise.
19246 (handle_builtin_stxncpy): Likewise.
19247 (handle_builtin_memcpy): Likewise.
19248 (handle_builtin_strcat): Likewise.
19249 (strlen_check_and_optimize_stmt): Likewise.
19250 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
19251 * tree-streamer-in.c: Likewise.
19252 * tree-streamer.c (record_common_node): Likewise.
19253 * tree.c (tree_code_size): Likewise.
19254 (wide_int_to_tree_1): Likewise.
19255 (type_contains_placeholder_1): Likewise.
19256 (build_common_tree_nodes): Likewise.
19257 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
19258 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
19259 (POINTER_BOUNDS_P): Likewise.
19260 (BOUNDED_TYPE_P): Likewise.
19261 (BOUNDED_P): Likewise.
19262 (CALL_WITH_BOUNDS_P): Likewise.
19263 (pointer_bounds_type_node): Likewise.
19264 * value-prof.c (gimple_ic): Likewise.
19265 * var-tracking.c (vt_add_function_parameters): Likewise.
19266 * varasm.c (make_decl_rtl): Likewise.
19267 (assemble_start_function): Likewise.
19268 (output_constant): Likewise.
19269 (maybe_assemble_visibility): Likewise.
19270 * varpool.c (ctor_for_folding): Likewise.
19271 * chkp-builtins.def: Remove.
19272 * ipa-chkp.c: Remove.
19273 * ipa-chkp.h: Remove.
19274 * rtl-chkp.c: Remove.
19275 * rtl-chkp.h: Remove.
19276 * tree-chkp-opt.c: Remove.
19277 * tree-chkp.c: Remove.
19278 * tree-chkp.h: Remove.
19280 2018-06-07 Carl Love <cel@us.ibm.com>
19282 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19283 vextract_fp_from_shortl): Add BE support.
19285 2018-06-07 Paul Koning <ni1d@arrl.net>
19287 * compare-elim.c (try_merge_compare): Don't merge compare if
19288 address contains a side effect.
19289 (try_eliminate_compare): Likewise.
19291 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
19293 * config.gcc: Support "tremont".
19294 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
19295 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19297 * config/i386/i386.c (m_TREMONT): Define.
19298 (processor_target_table): Add "tremont".
19299 (PTA_TREMONT): Define.
19300 (ix86_lea_outperforms): Add TARGET_TREMONT.
19301 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
19302 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
19303 and M_INTEL_GOLDMONT_PLUS.
19304 (fold_builtin_cpu): Add "tremont".
19305 (ix86_add_stmt_cost): Add TARGET_TREMONT.
19306 (ix86_option_override_internal): Add "tremont".
19307 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
19308 (processor_type): Add PROCESSOR_TREMONT.
19309 * config/i386/x86-tune.def: Add m_TREMONT.
19310 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
19312 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19314 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
19315 symbol defined for msp430i* devices to be lower case.
19317 2018-06-07 Richard Biener <rguenther@suse.de>
19319 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
19320 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
19321 Properly wrap signed arithmetic if overflow wraps.
19323 2018-06-07 Jakub Jelinek <jakub@redhat.com>
19325 PR tree-optimization/69615
19326 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
19327 of a cast from a same precision integral SSA_NAME in a bb dominated
19328 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
19329 cast to utype if rhs2 has already a compatible type.
19331 2018-06-07 Richard Biener <rguenther@suse.de>
19333 PR tree-optimization/85935
19334 * graphite-scop-detection.c (find_params_in_bb): Analyze
19335 condition operands with respect to the correct loop. Assert
19336 the analysis doesn't fail.
19338 2018-06-04 Carl Love <cel@us.ibm.com>
19340 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
19341 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
19342 as it is slightly cheaper.
19343 (first_match_or_eos_index_<mode>):
19344 Calculate index using natural element order.
19345 (first_match_index_<mode>):
19346 Calculate index using natural element order.
19347 (first_match_or_eos_index_<mode>):
19348 Calculate index using natural order.
19349 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
19350 for BE and LE modes.
19351 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
19352 P9V_BUILTIN_VCLZLSBB_V16QI.
19353 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
19356 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
19358 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
19359 indentation and line wrap for many prototypes. Add missing
19360 @smallexample directives around block of prototypes for vec_xl and
19363 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
19365 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
19366 track if we pass or return IEEE 128-bit floating point.
19367 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
19368 C++ mangling that is compatible with GCC 8.1.
19369 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
19370 (init_cumulative_args): Note if we pass or return IEEE 128-bit
19371 floating point types.
19372 (rs6000_function_arg_advance_1): Likewise.
19373 (rs6000_mangle_type): Optionally generate mangled names that match
19374 what GCC 8.1 generated for IEEE 128-bit floating point types.
19375 (rs6000_globalize_decl_name): If we have an external function that
19376 passes or returns IEEE 128-bit types, generate a weak reference
19377 from the mangled name used in GCC 8.1 to the current mangled
19379 (rs6000_init_builtins): Make __ibm128 use the long double type if
19380 long double is IBM extended double. Make __float128 use the long
19381 double type if long double is IEEE 128-bit.
19384 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
19385 macro for __ibm128 built-in functions.
19386 (PACK_IF): Add __ibm128 pack/unpack functions.
19387 (UNPACK_IF): Likewise.
19388 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
19389 enable long double built-in functions if long double is IEEE
19390 128-bit floating point.
19391 (rs6000_invalid_builtin): Update long double built-in function
19393 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
19394 functions, adjust the built-in function to use the long double
19395 built-in function if __ibm128 and long double are the same type.
19396 * doc/extend.texi (PowerPC builtins): Update documention for
19397 __builtin_{,un}pack_longdouble. Add documentation for
19398 __builtin_{,un}pack_ibm128.
19400 2018-06-06 Jim Wilson <jimw@sifive.com>
19402 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
19403 (struct machine_function): New field interrupt_mode.
19404 (riscv_handle_type_attribute): New function. Add forward declaration.
19405 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
19406 (riscv_expand_epilogue): Check interrupt_mode field.
19407 (riscv_set_current_function): Check interrupt attribute args and
19408 set interrupt_mode field.
19409 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
19410 (riscv_sret, riscv_uret): New.
19411 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
19412 new arguments to interrupt attribute.
19414 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
19417 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
19418 Don't handle -mcpu=power8 if -mpower9-vector is also used.
19420 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
19422 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
19423 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
19424 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19425 several redundant entries.
19427 2018-06-06 David Malcolm <dmalcolm@redhat.com>
19429 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
19430 type from "rtx" to "rtx_insn *".
19431 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
19432 for local "call_insn", removing cast.
19433 (ix86_expand_call): Likewise, introducing a "call_insn" local.
19435 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
19437 PR tree-optimization/86066
19438 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
19439 for BIT_INSERT_EXPR stores.
19441 2018-06-06 Richard Biener <rguenther@suse.de>
19443 PR tree-optimization/86062
19444 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
19445 component refs ontop
19446 of to be offsetted base.
19448 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19450 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
19451 to be static and remove check on interrupt attribute name.
19453 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
19455 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
19456 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
19458 2018-06-05 Steve Ellcey <sellcey@cavium.com>
19461 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
19463 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
19464 Remove second argument, change how error is called.
19465 (aarch64_layout_arg): Remove second argument from
19466 aarch64_err_no_fpadvsimd call.
19467 (aarch64_init_cumulative_args): Ditto.
19468 (aarch64_gimplify_va_arg_expr): Ditto.
19469 * config/aarch64/aarch64.md (mov<mode>): Ditto.
19471 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
19473 * config/i386/i386.md (simple_return_indirect_internal): New expander.
19474 (*simple_return_indirect_internal<mode>): Rename from
19475 simple_return_indirect_internal. Use W mode iterator.
19476 (rstorssp): New expander.
19477 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
19478 (clrssbsy): New expander.
19479 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
19481 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19483 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
19485 (cmse_check_pointed_object): Likewise.
19487 2018-06-05 Martin Liska <mliska@suse.cz>
19489 PR gcov-profile/47618
19490 * doc/invoke.texi: Document how -fprofile-dir format
19493 2018-06-05 Richard Biener <rguenther@suse.de>
19495 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
19496 removal pretend DOM info isn't available so we do not update
19497 it and only remove edges, not dominated blocks. Actually free
19498 DOM info in case we removed something. Remove unreachable blocks.
19499 (mfb_keep_latches): Work with either DOM info or marked backedges.
19500 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
19501 first. Mark backedges if DOM info isn't available.
19502 (Re-)compute DOM info after cleanup_control_flow_pre.
19504 2018-06-05 Richard Biener <rguenther@suse.de>
19506 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
19507 (locus_discrim_hasher::hash): Adjust.
19508 (locus_discrim_hasher::equal): Likewise.
19509 (next_discriminator_for_locus): Work on line directly.
19510 (same_line_p): Pass in expanded locus1 as well.
19511 (assign_discriminators): Avoid redundant location expansions.
19513 2018-06-05 Richard Biener <rguenther@suse.de>
19515 PR tree-optimization/86046
19516 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
19517 if required after clearing TREE_ADDRESSABLE.
19519 2018-06-05 Richard Biener <rguenther@suse.de>
19521 PR tree-optimization/86047
19522 * tree-ssa-loop.c (for_each_index): Glob handling of all
19523 decls and constants and really handle all of them.
19525 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19528 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19529 qualifier_void_pointer and qualifier_const_void_pointer.
19530 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
19531 (arm_init_builtins): Handle the above.
19532 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
19533 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
19534 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
19537 2018-06-05 Martin Liska <mliska@suse.cz>
19539 * auto-profile.c (read_autofdo_file): Do not use
19540 gcov_ctr_summary struct.
19541 (afdo_callsite_hot_enough_for_early_inline): Likewise.
19542 * coverage.c (struct counts_entry): Likewise.
19543 (read_counts_file): Read just single summary entry.
19544 (get_coverage_counts): Use gcov_summary struct.
19545 * coverage.h (get_coverage_counts): Likewise.
19546 * gcov-dump.c (dump_working_sets): Likewise.
19547 (tag_summary): Dump just single summary.
19548 * gcov-io.c (gcov_write_summary): Write just histogram
19550 (gcov_read_summary): Read just single summary.
19551 (compute_working_sets): Use gcov_summary struct.
19552 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
19553 of GCOV_COUNTERS_SUMMABLE.
19554 (GCOV_COUNTERS_SUMMABLE): Remove.
19555 (GCOV_FIRST_VALUE_COUNTER): Replace with
19556 GCOV_COUNTER_V_INTERVAL.
19557 (struct gcov_ctr_summary): Remove.
19558 (struct gcov_summary): Directly use fields of former
19560 (compute_working_sets): Use gcov_summary struct.
19561 * gcov.c (read_count_file): Do not use ctrs fields.
19562 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
19564 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
19566 * profile.c: Likewise.
19567 * profile.h: Likewise.
19569 2018-06-05 Martin Liska <mliska@suse.cz>
19571 PR gcov-profile/84846
19572 * gcov.c (output_lines): Print working directory only
19573 in intermediate format.
19575 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
19577 * config/s390/s390-builtin-types.def: Add void function type.
19578 * config/s390/s390-builtins.def: Use the function type for the
19581 2018-06-04 Jim Wilson <jimw@sifive.com>
19583 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
19585 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
19586 and maybe_eh_return. Change regno to unsigned int. Use new args to
19587 handle EH_RETURN_DATA_REGNO registers properly.
19588 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
19589 (riscv_expand_epilogue): Update comment. Change argument name and
19590 type. Update code to use new name and type. Pass new args to
19591 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
19593 * config/riscv/riscv.md (NORMAL_RETURN): New.
19594 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
19595 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
19596 (eh_return): Call gen_eh_return_internal and emit barrier.
19597 (eh_return_internal): Call riscv_expand_epilogue.
19599 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
19601 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
19602 bit_insertion field and declare can_be_merged_into method.
19603 (merged_store_group::can_be_merged_into): New method.
19604 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
19605 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
19606 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
19608 2018-06-04 Richard Biener <rguenther@suse.de>
19610 PR tree-optimization/85955
19611 * builtins.c (fold_builtin_sincos): Convert pointers to
19612 destination to appropriate type before dereferencing.
19614 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
19616 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
19618 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
19620 * expr.c (expand_expr_real_1): Force the operand into memory if
19621 its TYPE_MODE is BLKmode and if there is no integer mode for
19622 the number of bits being extracted.
19624 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19628 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
19629 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
19631 2018-06-04 Richard Biener <rguenther@suse.de>
19633 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
19634 (cleanup_tree_cfg_noloop): ... single caller. Do
19635 start_recording_case_labels later.
19637 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
19639 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
19640 to _IMMINTRIN_H_INCLUDED.
19641 * config/i386/pconfigintrin.h: Ditto.
19642 * config/i386/waitpkgintrin.h: Ditto.
19643 * config/i386/immintrin.h: Add includes for sgxintrin.h,
19644 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
19645 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
19646 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
19647 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
19648 waitpkgintrin.h and cldemoteintrin.h.
19650 2018-06-04 Richard Biener <rguenther@suse.de>
19652 PR tree-optimization/86038
19653 * tracer.c (find_best_successor): Check probability for
19654 being initialized, bail out if not.
19656 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
19659 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
19660 of bits to ignore when comparing architectures.
19662 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19664 PR tree-optimization/69615
19665 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
19666 maximum or minimum of the type, try to merge it also as if
19667 range1 is + [-, x - 1] or + [x + 1, -].
19670 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
19672 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
19674 PR tree-optimization/86034
19675 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
19676 the unsigned bitfield type in a bit insertion sequence if it does not
19677 have a larger precision than the bitfield size.
19678 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
19680 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
19682 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
19684 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19686 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
19687 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
19688 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
19689 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
19691 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19693 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
19694 Disable -fdelete-null-pointer-checks for ELF toolchain.
19696 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19697 Kito Cheng <kito.cheng@gmail.com>
19699 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
19700 (nds32le-*-*, nds32be-*-*): Integrate checking process.
19701 (nds32*-*-*): Add glibc and uclibc conditions.
19702 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
19703 (TARGET_EXCEPT_UNWIND_INFO): Define.
19704 * config/nds32/elf.h: New file.
19705 * config/nds32/linux.h: New file.
19706 * config/nds32/nds32-elf.opt: New file.
19707 * config/nds32/nds32-linux.opt: New file.
19708 * config/nds32/nds32-fp-as-gp.c
19709 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
19710 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
19712 (nds32_asm_file_end): Ditto.
19713 (nds32_print_operand): Ditto.
19714 (nds32_insert_attributes): Ditto.
19715 (nds32_init_libfuncs): New function.
19716 (TARGET_HAVE_TLS): Define.
19717 (TARGET_INIT_LIBFUNCS): Define.
19718 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
19720 (TARGET_ELF): Apply different mcmodel setting.
19721 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
19722 been migrated into elf.h and linux.h files.
19723 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
19724 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
19725 (mcmodel): The content has been migrated into nds32-elf.opt and
19726 nds32-linux.opt files.
19727 * config/nds32/t-elf: New file.
19728 * config/nds32/t-linux: New file.
19730 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19731 Shiva Chen <shiva0217@gmail.com>
19733 * config/nds32/constants.md (unspec_volatile_element): Add
19734 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
19735 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
19737 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
19738 (make_pass_nds32_fp_as_gp): Declare.
19739 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
19741 (nds32_asm_function_end_prologue): Remove unused asm output.
19742 (nds32_asm_function_begin_epilogue): Remove unused asm output.
19743 (nds32_asm_file_start): Output necessary fp_as_gp information.
19744 (nds32_option_override): Adjust register usage.
19745 (nds32_expand_prologue): Consider fp_as_gp situation.
19746 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
19747 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
19750 (simple_return): Ditto.
19751 (omit_fp_begin): Output special directive for fp_as_gp.
19752 (omit_fp_end): Output special directive for fp_as_gp.
19753 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
19754 mforbid-fp-as-gp): New options.
19756 2018-06-01 Mark Wielaard <mark@klomp.org>
19758 * dwarf2out.c (dwarf2out_finish): Remove generation of
19759 DW_AT_loclists_base.
19761 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
19763 * gimple-ssa-store-merging.c: Include gimple-fold.h.
19764 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
19765 (struct merged_store_group): Add bit_insertion field.
19766 (dump_char_array): Use standard hexadecimal format.
19767 (merged_store_group::merged_store_group): Set bit_insertion to false.
19768 (merged_store_group::apply_stores): Use optimal buffer size. Deal
19769 with BIT_INSERT_EXPR stores. Move up code updating the mask and
19770 also print the mask in the dump file.
19771 (pass_store_merging::gate): Minor tweak.
19772 (imm_store_chain_info::coalesce_immediate): Fix wrong association
19773 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
19774 stores with INTEGER_CST stores.
19775 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
19776 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
19777 and use it throughout. Generate bit insertion sequences if need be.
19778 (pass_store_merging::process_store): Remove redundant condition.
19779 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
19781 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
19783 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
19784 the 128-bit floating point types. Fix function comment.
19786 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19788 * config/aarch64/aarch64-simd.md
19789 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
19791 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
19794 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19796 PR tree-optimization/85989
19797 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
19799 (backprop::intersect_uses): Check it when deciding whether this
19800 is a backedge reference.
19801 (backprop::process_block): Add each phi to m_visited_phis
19802 after visiting it, then clear it at the end.
19804 2018-06-01 Richard Biener <rguenther@suse.de>
19806 * tree-vectorizer.h (vect_dr_stmt): New function.
19807 (vect_get_load_cost): Adjust.
19808 (vect_get_store_cost): Likewise.
19809 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
19810 Use vect_dr_stmt instead of DR_SMTT.
19811 (vect_record_base_alignments): Likewise.
19812 (vect_calculate_target_alignment): Likewise.
19813 (vect_compute_data_ref_alignment): Likewise and make static.
19814 (vect_update_misalignment_for_peel): Likewise.
19815 (vect_verify_datarefs_alignment): Likewise.
19816 (vector_alignment_reachable_p): Likewise.
19817 (vect_get_data_access_cost): Likewise. Pass down
19818 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
19819 (vect_get_peeling_costs_all_drs): Likewise.
19820 (vect_peeling_hash_get_lowest_cost): Likewise.
19821 (vect_enhance_data_refs_alignment): Likewise.
19822 (vect_find_same_alignment_drs): Likewise.
19823 (vect_analyze_data_refs_alignment): Likewise.
19824 (vect_analyze_group_access_1): Likewise.
19825 (vect_analyze_group_access): Likewise.
19826 (vect_analyze_data_ref_access): Likewise.
19827 (vect_analyze_data_ref_accesses): Likewise.
19828 (vect_vfa_segment_size): Likewise.
19829 (vect_small_gap_p): Likewise.
19830 (vectorizable_with_step_bound_p): Likewise.
19831 (vect_prune_runtime_alias_test_list): Likewise.
19832 (vect_analyze_data_refs): Likewise.
19833 (vect_supportable_dr_alignment): Likewise.
19834 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
19835 (vect_gen_prolog_loop_niters): Likewise.
19836 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
19837 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
19839 (vect_recog_mask_conversion_pattern): Likewise.
19840 (vect_try_gather_scatter_pattern): Likewise.
19841 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
19842 to vect_get_store_cost.
19843 (vect_get_store_cost): Get stmt_info instead of DR.
19844 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
19845 (vect_get_load_cost): Get stmt_info instead of DR.
19847 2018-06-01 Richard Biener <rguenther@suse.de>
19849 PR middle-end/86017
19850 * gimple-fold.c (var_decl_component_p): Also allow offsetted
19851 vars wrapped in MEM_REFs.
19853 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19855 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19856 Fix subreg tests so that we only return a choice between
19857 GENERAL_REGS and FP_REGS if the original classes included both.
19859 2018-06-01 Richard Biener <rguenther@suse.de>
19862 * tree-ssa-structalias.c (get_function_part_constraint):
19863 Handle NULL fi->decl.
19864 (find_func_aliases_for_call): Properly handle indirect
19865 fi from direct call.
19866 (find_func_clobbers): Likewise.
19867 (ipa_pta_execute): Likewise.
19868 (create_variable_info_for): For functions that are ifunc_resolver
19869 resolve to a varinfo that contains the result of the resolver call.
19870 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
19873 2018-05-31 Michael Collison <michael.collison@arm.com>
19875 * config/aarch64/aarch64.md:
19876 (*fix_to_zero_extenddfdi2): New pattern.
19877 * gcc.target/aarch64/fix_extend1.c: New testcase.
19879 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
19881 PR middle-end/78809
19882 PR middle-end/83026
19883 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
19884 and BUILT_IN_STRNCMP_EQ.
19885 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
19886 BUILT_IN_STRNCMP_EQ.
19887 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
19888 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19889 (gimple_fold_builtin): Likewise.
19890 * tree-ssa-strlen.c (compute_string_length): New function.
19891 (determine_min_obsize): New function.
19892 (handle_builtin_string_cmp): New function to handle calls to
19893 string compare functions.
19894 (strlen_optimize_stmt): Add handling to builtin string compare
19896 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19897 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19898 * tree.c (build_common_builtin_nodes): Add new defines of
19899 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
19901 2018-05-31 Jakub Jelinek <jakub@redhat.com>
19904 * bb-reorder.c (pass_partition_blocks::gate): Return false for
19905 functions with naked attribute.
19907 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
19909 * config/i386/sse.md (avx_vec_concat<mode>):
19910 Substitute concat_tg_mode mode attribute with xtg_mode.
19911 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
19912 (concat_tg_mode): Remove mode attribute.
19914 2018-05-31 Martin Sebor <msebor@redhat.com>
19917 * calls.c (alloc_max_size): Correct a logic error/typo.
19918 Treat excessive arguments as infinite. Warn for invalid arguments.
19919 * doc/invoke.texi (-Walloc-size-larger-than): Update.
19921 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
19924 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
19925 and movx for Haswell.
19927 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
19928 Cesar Philippidis <cesar@codesourcery.com>
19930 PR middle-end/85879
19931 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
19932 when emitting error on private/firstprivate reductions.
19933 * omp-low.c (lower_omp_target): Avoid reference-type processing
19934 on pointers for firstprivate clause.
19936 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
19938 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
19941 * config/aarch64/aarch64-simd.md
19942 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
19943 (aarch64_ld1_x3_<mode>): Likewise
19944 (aarch64_st1x2<VALLDIF:mode>): Likewise
19945 (aarch64_st1_x2_<mode>): Likewise
19946 (aarch64_st1x3<VALLDIF:mode>): Likewise
19947 (aarch64_st1_x3_<mode>): Likewise
19948 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
19949 (vld1_s8_x3): Likewise.
19950 (vld1_u16_x3): Likewise.
19951 (vld1_s16_x3): Likewise.
19952 (vld1_u32_x3): Likewise.
19953 (vld1_s32_x3): Likewise.
19954 (vld1_u64_x3): Likewise.
19955 (vld1_s64_x3): Likewise.
19956 (vld1_f16_x3): Likewise.
19957 (vld1_f32_x3): Likewise.
19958 (vld1_f64_x3): Likewise.
19959 (vld1_p8_x3): Likewise.
19960 (vld1_p16_x3): Likewise.
19961 (vld1_p64_x3): Likewise.
19962 (vld1q_u8_x3): Likewise.
19963 (vld1q_s8_x3): Likewise.
19964 (vld1q_u16_x3): Likewise.
19965 (vld1q_s16_x3): Likewise.
19966 (vld1q_u32_x3): Likewise.
19967 (vld1q_s32_x3): Likewise.
19968 (vld1q_u64_x3): Likewise.
19969 (vld1q_s64_x3): Likewise.
19970 (vld1q_f16_x3): Likewise.
19971 (vld1q_f32_x3): Likewise.
19972 (vld1q_f64_x3): Likewise.
19973 (vld1q_p8_x3): Likewise.
19974 (vld1q_p16_x3): Likewise.
19975 (vld1q_p64_x3): Likewise.
19976 (vst1_s64_x2): Likewise.
19977 (vst1_u64_x2): Likewise.
19978 (vst1_f64_x2): Likewise.
19979 (vst1_s8_x2): Likewise.
19980 (vst1_p8_x2): Likewise.
19981 (vst1_s16_x2): Likewise.
19982 (vst1_p16_x2): Likewise.
19983 (vst1_s32_x2): Likewise.
19984 (vst1_u8_x2): Likewise.
19985 (vst1_u16_x2): Likewise.
19986 (vst1_u32_x2): Likewise.
19987 (vst1_f16_x2): Likewise.
19988 (vst1_f32_x2): Likewise.
19989 (vst1_p64_x2): Likewise.
19990 (vst1q_s8_x2): Likewise.
19991 (vst1q_p8_x2): Likewise.
19992 (vst1q_s16_x2): Likewise.
19993 (vst1q_p16_x2): Likewise.
19994 (vst1q_s32_x2): Likewise.
19995 (vst1q_s64_x2): Likewise.
19996 (vst1q_u8_x2): Likewise.
19997 (vst1q_u16_x2): Likewise.
19998 (vst1q_u32_x2): Likewise.
19999 (vst1q_u64_x2): Likewise.
20000 (vst1q_f16_x2): Likewise.
20001 (vst1q_f32_x2): Likewise.
20002 (vst1q_f64_x2): Likewise.
20003 (vst1q_p64_x2): Likewise.
20004 (vst1_s64_x3): Likewise.
20005 (vst1_u64_x3): Likewise.
20006 (vst1_f64_x3): Likewise.
20007 (vst1_s8_x3): Likewise.
20008 (vst1_p8_x3): Likewise.
20009 (vst1_s16_x3): Likewise.
20010 (vst1_p16_x3): Likewise.
20011 (vst1_s32_x3): Likewise.
20012 (vst1_u8_x3): Likewise.
20013 (vst1_u16_x3): Likewise.
20014 (vst1_u32_x3): Likewise.
20015 (vst1_f16_x3): Likewise.
20016 (vst1_f32_x3): Likewise.
20017 (vst1_p64_x3): Likewise.
20018 (vst1q_s8_x3): Likewise.
20019 (vst1q_p8_x3): Likewise.
20020 (vst1q_s16_x3): Likewise.
20021 (vst1q_p16_x3): Likewise.
20022 (vst1q_s32_x3): Likewise.
20023 (vst1q_s64_x3): Likewise.
20024 (vst1q_u8_x3): Likewise.
20025 (vst1q_u16_x3): Likewise.
20026 (vst1q_u32_x3): Likewise.
20027 (vst1q_u64_x3): Likewise.
20028 (vst1q_f16_x3): Likewise.
20029 (vst1q_f32_x3): Likewise.
20030 (vst1q_f64_x3): Likewise.
20031 (vst1q_p64_x3): Likewise.
20033 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
20035 * config/msp430/msp430.c (msp430_output_labelref): Prepend
20036 user_label_prefix to name.
20038 * tree-core.h: Update comment about the format of NAME string
20039 passed to handler in attribute_spec.
20041 * config/msp430/msp430.md: Remove erroneous subreg expression from
20042 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
20043 zero_extend{q,h}isi2.
20045 2018-05-30 Borislav Petkov <bp@suse.de>
20047 * doc/extend.texi: Document some architecture specific
20048 constraints and sort entries.
20050 2018-05-30 Martin Sebor <msebor@redhat.com>
20052 PR middle-end/85369
20053 * builtins.c (expand_builtin_stpcpy_1): New function.
20054 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
20055 only if the former succeeds.
20057 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
20059 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
20062 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20064 * doc/invoke.texi (-flinker-output): Document
20066 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20068 * passes.c (ipa_write_summaries): Only modify statements if body
20070 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
20071 incrementally linking.
20072 (ipa_passes): Likewise.
20073 * lto-cgraph.c (lto_output_node): When incrementally linking do not
20074 pass down resolution info.
20075 * common.opt (flag_incremental_link): Update info.
20076 * gcc.c (plugin specs): Turn flinker-output=* to
20077 -plugin-opt=-linker-output-known
20078 * toplev.c (compile_file): Also cut compilation when doing incremental
20080 * flag-types. (enum lto_partition_model): Add
20081 LTO_LINKER_OUTPUT_NOLTOREL.
20082 (invoke.texi): Add -flinker-output docs.
20083 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
20084 link same way as WPA; do not stream in dead initializers.
20086 * dwarf2out.c (dwarf2out_die_ref_for_decl,
20087 darf2out_register_external_decl): Support incremental link.
20089 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20091 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
20093 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20095 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
20096 it down to simple_object_copy_lto_debug_sections.
20097 (run_gcc): Determine incremental LTO link time and configure
20098 lto1 into non-wpa mode, disable renaming of debug sections.
20100 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
20102 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
20103 descriptions of various incorrectly documented functions.
20105 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
20108 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20109 address check not strict.
20111 2018-05-30 Richard Biener <rguenther@suse.de>
20113 PR tree-optimization/85964
20114 * tracer.c (better_p): Drop initialized count check, we only
20115 call the function with initialized counts now.
20116 (find_best_successor): Do find a best edge if one
20117 has uninitialized count.
20118 (find_best_predecessor): Likewise. Do BB frequency check only
20119 if count is initialized.
20121 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
20123 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
20124 (aarch64_ldrstr_offset_compare): New.
20125 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
20126 load/store orderings.
20127 (aarch64_gen_adjusted_ldpstp): Likewise.
20129 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20131 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20132 Check for subset of GENERAL_REGS and FP_REGS.
20133 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
20136 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
20138 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
20139 and wi::to_poly_offset. Add the current offset and then check
20140 whether the sum fits, rather than using an unchecked addition of
20141 a checked term. Check for a shwi rather than a uhwi.
20142 * expr.c (get_bit_range): Use tree_to_poly_uint64.
20143 (store_constructor): Use poly_int_tree_p.
20144 (expand_expr_real_1): Likewise.
20145 * function.c (assign_temp): Likewise.
20146 * fold-const.c (const_binop): Use poly_int_tree_p and
20147 wi::to_poly_offset.
20148 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
20150 * ipa-icf-gimple.c (func_checker::compare_operand): Use
20151 to_poly_offset for MEM offsets.
20152 * ipa-icf.c (sem_variable::equals): Likewise.
20153 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
20154 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
20155 wi::to_poly_offset for BIT_FIELD_REF offsets.
20156 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
20157 wi::to_poly_offset.
20158 * var-tracking.c (emit_note_insn_var_location): Use
20159 tree_to_poly_uint64.
20161 2018-05-29 Jim Wilson <jimw@sifive.com>
20163 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
20165 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
20168 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
20169 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
20171 (sse4_1_round<mode>2): Use nonimmediate_operand
20172 for operand 1 predicate.
20174 2018-05-29 Martin Sebor <msebor@redhat.com>
20175 Richard Biener <rguenther@suse.de>
20178 * calls.c (get_size_range): Call determine_value_range instead
20179 of get_value_range..
20180 * tree-vrp.h (determine_value_range): Declared new function.
20181 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
20183 2018-05-29 Richard Biener <rguenther@suse.de>
20185 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
20186 sure to use non-pattern stmts for get_earlier_stmt arguments.
20187 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
20188 called on pattern stmts.
20189 (get_later_stmt): Likewise.
20191 2018-05-29 Martin Liska <mliska@suse.cz>
20193 PR gcov-profile/85759
20194 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
20197 2018-05-29 Jakub Jelinek <jakub@redhat.com>
20199 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
20201 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
20205 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
20206 VEC_PACK_FLOAT_EXPR): New tree codes.
20207 * tree-pretty-print.c (op_code_prio): Handle
20208 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
20209 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
20210 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
20211 * tree-inline.c (estimate_operator_cost): Likewise.
20212 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
20213 * fold-const.c (const_binop): Likewise.
20214 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
20215 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
20216 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
20217 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
20218 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
20219 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
20220 * expr.c (expand_expr_real_2): Likewise.
20221 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
20222 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
20223 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
20225 * optabs.c (expand_widen_pattern_expr): For
20226 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
20227 sign from result type rather than operand's type.
20228 (expand_binop_directly): For vec_packu_float_optab and
20229 vec_packs_float_optab allow result type to be different from operand's
20231 * optabs-tree.c (optab_for_tree_code): Handle
20232 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
20233 VEC_PACK_FLOAT_EXPR. Formatting fixes.
20234 * tree-vect-generic.c (expand_vector_operations_1): Handle
20235 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
20236 VEC_PACK_FLOAT_EXPR.
20237 * tree-vect-stmts.c (supportable_widening_operation): Handle
20239 (supportable_narrowing_operation): Handle FLOAT_EXPR.
20240 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
20241 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
20242 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
20243 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
20245 (vec_pack<floatprefix>_float_<mode>): New expander.
20246 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
20248 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
20249 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
20250 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
20251 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
20252 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
20254 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
20255 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
20256 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
20257 VEC_PACK_FLOAT_EXPR): Document.
20259 2018-05-29 Richard Biener <rguenther@suse.de>
20261 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
20263 (stmt_vec_info_vec): Make pointer.
20264 (init_stmt_vec_info_vec): Remove.
20265 (free_stmt_vec_info_vec): Likewise.
20266 (set_stmt_vec_info_vec): New function.
20267 (free_stmt_vec_infos): Likewise.
20268 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
20269 (set_vinfo_for_stmt): Likewise.
20270 (get_earlier_stmt): Likewise.
20271 (get_later_stmt): Likewise.
20272 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
20273 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
20274 (vec_info::~vec_info): Free stmt_vec_infos.
20275 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
20276 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
20277 (pass_slp_vectorize::execute): Likewise.
20278 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
20279 (free_stmt_vec_info_vec): Likewise.
20280 (set_stmt_vec_info_vec): New function.
20281 (free_stmt_vec_infos): Likewise.
20282 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
20283 the global stmt_vec_info_vec.
20284 * tree-parloops.c (gather_scalar_reductions): Use
20285 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
20288 2018-05-29 Richard Biener <rguenther@suse.de>
20290 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
20292 2018-05-29 Martin Liska <mliska@suse.cz>
20293 David Malcolm <dmalcolm@redhat.com>
20295 * vec.c (test_reverse): New.
20296 (vec_c_tests): Add new test.
20297 * vec.h (vl_ptr>::reverse): New function.
20299 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
20301 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
20303 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
20306 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20308 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
20310 2018-05-28 Richard Biener <rguenther@suse.de>
20312 PR tree-optimization/85933
20313 * tree-vect-data-refs.c (vect_record_base_alignments): Only
20314 look at stmts marked as vectorizable.
20316 2018-05-28 Richard Biener <rguenther@suse.de>
20318 PR tree-optimization/85934
20319 * tree-vect-generic.c (expand_vector_operations_1): Hoist
20320 vector boolean check before scalar optimization.
20322 2018-05-28 Jakub Jelinek <jakub@redhat.com>
20324 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
20327 2018-05-28 Mark Wielaard <mark@klomp.org>
20329 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
20330 if it is an expression containing a minus sign.
20332 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
20334 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
20336 2018-05-27 Paul Koning <ni1d@arrl.net>
20338 * config/pdp11/pdp11.md (truncsihi2): Remove.
20340 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
20341 Chung-Ju Wu <jasonwucj@gmail.com>
20343 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
20345 (unaligned_store_dw): Ditto.
20346 * config/nds32/nds32-memory-manipulation.c
20347 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
20348 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
20349 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
20350 (emit_setmem_word_loop): Rename to ...
20351 (emit_setmem_doubleword_loop): ... this.
20352 (nds32_gen_dup_4_byte_to_word_value): New function.
20353 (nds32_gen_dup_8_byte_to_double_word_value): New function.
20354 (nds32_expand_setmem_loop): Refine implementation.
20355 (nds32_expand_setmem_loop_v3m): Ditto.
20356 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
20359 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
20361 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
20363 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
20365 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
20366 (nds32_init_machine_status): Initialize machine->attr_naked_p and
20367 machine->attr_no_prologue_p.
20368 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
20369 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
20370 (nds32_expand_epilogue): Consider attr_naked_p.
20371 (nds32_expand_epilogue_v3pop): Likewise.
20372 (nds32_can_use_return_insn): Likewise.
20373 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
20374 attr_no_prologue_p fields.
20375 * config/nds32/nds32.opt (mret-in-naked-func): New option.
20377 2018-05-27 Jakub Jelinek <jakub@redhat.com>
20380 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
20382 * config/i386/sse.md
20383 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
20385 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
20387 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
20389 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
20391 (*<floatsuffix>floatv2div2sf2): Rename to ...
20392 (*float<floatunssuffix>v2div2sf2): ... this.
20393 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
20394 (float<floatunssuffix>v2div2sf2_mask): ... this.
20395 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
20396 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
20397 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
20399 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
20401 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
20403 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
20405 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
20407 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
20409 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
20410 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
20411 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
20412 gen_ufix_truncv8dfv8si2.
20413 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
20414 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
20415 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
20416 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
20417 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
20418 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
20419 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
20420 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
20422 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20426 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
20428 2018-05-25 Jim Wilson <jimw@sifive.com>
20430 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
20431 * config/riscv/riscv.c (struct machine_function): Add
20432 interrupt_handler_p and attribute_checked_p fields.
20433 (riscv_attribute_table): Add interrupt.
20434 (riscv_interrupt_type_p): New.
20435 (riscv_save_reg_p): Save extra regs for interrupt handler.
20436 (riscv_use_save_libcall): Return false for interrupt handler.
20437 (riscv_first_stack_step): Add forward declaration.
20438 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
20439 for interrupt handler with large frame. Use it for saved reg list.
20440 (riscv_expand_prologue): Move flag_stack_usage_info support to
20441 eliminate duplication.
20442 (riscv_expand_epilogue): Generate mret for interrupt handler.
20443 (riscv_epilogue_uses): New.
20444 (riscv_can_use_return_insn): Return false for interrupt handler.
20445 (riscv_function_ok_for_sibcall): Likewise.
20446 (riscv_set_current_function): Add interrupt handler support.
20447 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
20448 * config/riscv/riscv.md (UNSPECV_MRET): New.
20450 (riscv_frflags, riscv_fsflags): Use tab after opcode.
20452 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
20454 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
20456 PR tree-optimization/85712
20457 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
20458 this candidate has already been replaced in-situ by a copy.
20460 2018-05-25 Jason Merrill <jason@redhat.com>
20462 PR c++/80485 - inline function non-zero address.
20463 * symtab.c (nonzero_address): Check DECL_COMDAT.
20465 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
20468 * config/alpha/alpha.md (ashlsi3): New insn pattern.
20469 (*ashlsi_se): Rename from *ashldi_se. Define as sign
20470 extension of SImode operation. Use const123_operand predicate.
20471 (*saddsi_1): Remove.
20472 (*saddl_se_1): Ditto.
20473 (*ssubsi_1): Ditto.
20474 (*ssubl_se_1): Ditto.
20475 * config/alpha/predicates.md (const123_operand): New predicate.
20476 * config/alpha/constraints.md (P): Use IN_RANGE.
20478 2018-05-25 Richard Biener <rguenther@suse.de>
20480 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
20482 (ref_maybe_used_by_stmt_p): Likewise.
20483 (stmt_may_clobber_ref_p): Likewise.
20484 (stmt_may_clobber_ref_p_1): Likewise.
20485 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
20487 (ref_maybe_used_by_stmt_p): Likewise.
20488 (stmt_may_clobber_ref_p): Likewise.
20489 (stmt_may_clobber_ref_p_1): Likewise.
20490 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
20491 the alias oracle to disambiguate DRs with stmts DR analysis
20493 (vect_analyze_data_refs): Do not give up on not analyzable
20494 DRs for BB vectorization. Remove code truncating the dataref
20497 2018-05-25 Jakub Jelinek <jakub@redhat.com>
20500 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
20501 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
20502 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
20504 2018-05-25 Richard Biener <rguenther@suse.de>
20506 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
20507 function, combining stmt data ref gathering and fatal analysis
20509 (vect_analyze_data_refs): Remove now redudnant code and simplify.
20510 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
20511 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
20512 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
20513 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
20515 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20517 PR tree-optimization/85720
20518 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
20519 SCC if all partitions are builtins.
20520 (version_loop_by_alias_check): New parameter. Generate cancelable
20521 runtime alias check if all partitions are builtins.
20522 (distribute_loop): Update call to above function.
20524 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20526 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
20527 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
20528 (parm_default_def_partition_arg): Ditto.
20529 (set_parm_default_def_partition): Ditto.
20530 (get_parm_default_def_partitions): Ditto and make it static.
20531 (get_undefined_value_partitions): Ditto and make it static.
20532 (remove_ssa_form): Refactor call to init_var_map here.
20533 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
20534 computation for loop region.
20535 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
20536 (register_default_def): Delete.
20537 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
20538 (parm_default_def_partition_arg): Ditto.
20539 (set_parm_default_def_partition): Ditto.
20540 (get_parm_default_def_partitions): Ditto and make it static.
20541 (get_undefined_value_partitions): Ditto and make it static.
20542 (coalesce_with_default, coalesce_with_default): Update comment.
20543 (create_coalesce_list_for_region): New func factored out from
20544 create_outofssa_var_map.
20545 (populate_coalesce_list_for_outofssa): New func factored out from
20546 create_outofssa_var_map and coalesce_ssa_name.
20547 (create_outofssa_var_map): Delete.
20548 (coalesce_ssa_name): Refactor to support live range computation.
20549 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
20550 (get_parm_default_def_partitions): Delete.
20551 (get_undefined_value_partitions): Ditto.
20552 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
20553 computation for loop region.
20554 (new_tree_live_info, loe_visit_block): Ditto.
20555 (live_worklist, set_var_live_on_entry): Ditto.
20556 (calculate_live_on_exit, verify_live_on_entry): Ditto.
20557 * tree-ssa-live.h (struct _var_map): New fields.
20558 (init_var_map): Change decl.
20559 (region_contains_p): New.
20561 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20563 * tree-ssa-live.h (live_merge_and_clear): Delete.
20565 2018-05-25 Richard Biener <rguenther@suse.de>
20568 * tree-dump.c (dequeue_and_dump): Remove access to removed
20569 operand 2 of a SWITCH_EXPR.
20571 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20573 * doc/sourcebuild.texi (vect_double_cond_arith): Include
20574 multiplication and division.
20575 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
20576 (cond_udiv@var{m}, cond_umod@var{m}): Document.
20577 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
20578 (cond_udiv_optab, cond_umod_optab): New optabs.
20579 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
20580 (IFN_COND_RDIV): New internal functions.
20581 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
20582 TRUNC_MOD_EXPR and RDIV_EXPR.
20583 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
20584 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
20586 (SVE_INT_BINARY): Include mult.
20587 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
20588 (optab, sve_int_op): Handle mult.
20589 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
20591 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
20592 for SVE_INT_BINARY_SD.
20594 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20596 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
20597 (optab, sve_int_op): Handle div and udiv.
20598 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
20599 for SVE_INT_BINARY_SD.
20600 (*<optab><mode>3): New insn for the same.
20602 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20604 * tree-vect-patterns.c: Include predict.h.
20605 (vect_recog_divmod_pattern): Restrict check for division support
20606 to when optimizing for size.
20608 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20610 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
20611 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
20612 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
20613 (gimple_match_op::set_op): Likewise.
20614 (gimple_resimplify4): Declare.
20615 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
20616 (expr::gen_transform): Likewise.
20617 (decision_tree::gen): Generate a simplification routine for 4 operands.
20618 * gimple-match-head.c (gimple_simplify): Add an overload for
20619 4 operands. In the top-level function, handle up to 4 call
20620 arguments and call gimple_resimplify4.
20621 (gimple_resimplify4): New function.
20622 (build_call_internal): Pass a fourth operand.
20623 (maybe_push_to_seq): Likewise.
20624 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
20625 Fold VEC_COND_EXPRs of an operation and a default value into
20626 an IFN_COND_* function if possible.
20627 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
20629 (SVE_COND_FP_BINARY): Include them.
20630 (optab, sve_fp_op): Handle them.
20631 (SVE_INT_BINARY_REV): New code iterator.
20632 (SVE_COND_FP_BINARY_REV): New int iterator.
20633 (commutative): New int attribute.
20634 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
20636 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
20638 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
20639 (*cond_<optab><mode>): New patterns for reversed operands.
20641 2018-05-25 Richard Biener <rguenther@suse.de>
20643 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
20644 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
20645 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
20646 (STMT_VINFO_GROUPED_ACCESS): Adjust.
20647 * tree-vect-data-refs.c (everywhere): Adjust users.
20648 * tree-vect-loop.c (everywhere): Likewise.
20649 * tree-vect-slp.c (everywhere): Likewise.
20650 * tree-vect-stmts.c (everywhere): Likewise.
20651 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
20653 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20655 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
20657 (gcc_cv_as_section_exclude): ... this.
20658 Try Solaris as #exclude syntax.
20659 * configure: Regenerate.
20660 * config.in: Regenerate.
20661 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
20663 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
20664 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
20666 * varasm.c (default_elf_asm_named_section): Don't check if
20667 HAVE_GAS_SECTION_EXCLUDE is defined.
20669 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20671 * doc/md.texi: Update the documentation of the cond_* optabs
20672 to mention the new final operand. Fix GET_MODE_NUNITS call.
20673 Describe the scalar case too.
20674 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
20675 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
20677 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
20678 (get_conditional_internal_fn): Update comment.
20679 * tree-vect-loop.c (vectorizable_reduction): Pass the original
20680 accumulator value as a final argument to conditional functions.
20681 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
20682 a define_expand and add an "else" operand. Assert for now that
20683 the else operand is equal to operand 2. Use SVE_INT_BINARY and
20684 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
20685 (*cond_<optab><mode>): New patterns.
20686 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
20687 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
20688 (UNSPEC_COND_EOR): Delete.
20689 (optab): Remove associated mappings.
20690 (SVE_INT_BINARY): New code iterator.
20691 (sve_int_op): Remove int attribute and add "minus" to the code
20693 (SVE_COND_INT_OP): Delete.
20694 (SVE_COND_FP_OP): Rename to...
20695 (SVE_COND_FP_BINARY): ...this.
20697 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20699 * optabs.c (can_reuse_operands_p): New function.
20700 (maybe_legitimize_operands): Try to reuse the results for
20703 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20705 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
20706 Add {q} suffix to insn mnemonic.
20708 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
20710 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
20711 (msp430_warn_func_return): New.
20713 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
20715 * fold-const.c (tree_nonzero_bits): New function.
20716 * fold-const.h (tree_nonzero_bits): Likewise.
20717 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
20718 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
20720 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20724 * varasm.c (assemble_alias): Check ifunc_resolver only on
20727 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20730 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
20731 when memory input operand is handled.
20733 2018-05-24 Luis Machado <luis.machado@linaro.org>
20735 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
20737 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
20739 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20741 * match.pd: Delay FMA folds until after vectorization.
20743 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
20746 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20747 address check not strict.
20749 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20751 * gimple-match.h (gimple_match_op): New class.
20752 (mprts_hook): Replace parameters with a gimple_match_op *.
20753 (maybe_build_generic_op): Likewise.
20754 (gimple_simplified_result_is_gimple_val): Replace parameters with
20755 a const gimple_match_op *.
20756 (gimple_simplify): Replace code_helper * and tree * parameters with
20757 a gimple_match_op * parameter.
20758 (gimple_resimplify1): Replace code_helper *, tree and tree *
20759 parameters with a gimple_match_op * parameter.
20760 (gimple_resimplify2): Likewise.
20761 (gimple_resimplify3): Likewise.
20762 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
20763 parameters with a gimple_match_op * parameter.
20764 * gimple-match-head.c (gimple_simplify): Change prototypes of
20765 auto-generated functions to take a gimple_match_op * instead of
20766 separate code_helper * and tree * parameters. Make the same
20767 change in the top-level overload and update calls to the
20768 gimple_resimplify routines. Update calls to the auto-generated
20769 functions and to maybe_push_res_to_seq in the publicly-facing
20770 operation-specific gimple_simplify overloads.
20771 (gimple_match_op::MAX_NUM_OPS): Define.
20772 (gimple_resimplify1): Replace rcode and ops with a single res_op
20773 parameter. Update call to gimple_simplify.
20774 (gimple_resimplify2): Likewise.
20775 (gimple_resimplify3): Likewise.
20776 (mprts_hook): Replace parameters with a gimple_match_op *.
20777 (maybe_build_generic_op): Likewise.
20778 (build_call_internal): Replace type, nargs and ops with
20779 a gimple_match_op *.
20780 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
20781 with a single gimple_match_op *. Update calls to mprts_hook,
20782 build_call_internal and gimple_simplified_result_is_gimple_val.
20783 Factor out code that is common to the tree_code and combined_fn cases.
20784 * genmatch.c (expr::gen_transform): Replace tem_code and
20785 tem_ops with a gimple_match_op called tem_op. Update calls
20786 to the gimple_resimplify functions and maybe_push_res_to_seq.
20787 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
20788 res_ops. Update call to the gimple_resimplify functions.
20789 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
20790 (decision_tree::gen): Make the functions take a gimple_match_op *
20791 called res_op instead of separate res_code and res_ops parameters.
20792 Update call accordingly.
20793 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
20794 and ops with a single res_op parameter. Update calls to
20795 maybe_build_generic_op and maybe_push_res_to_seq.
20796 (fold_stmt_1): Update calls to gimple_simplify and
20797 replace_stmt_with_simplification.
20798 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
20799 and gimple_simplified_result_is_gimple_val.
20800 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
20802 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
20803 with a gimple_match_op *.
20804 (vn_nary_build_or_lookup): Likewise. Update call to
20805 vn_nary_build_or_lookup_1.
20806 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
20807 gimple_match_op *. Update calls to the gimple_resimplify routines
20808 and to gimple_simplified_result_is_gimple_val.
20809 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
20810 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
20811 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
20812 (visit_nary_op): Likewise.
20813 (visit_reference_op_load): Likewise.
20815 2018-05-23 Luis Machado <luis.machado@linaro.org>
20817 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
20818 modifier for printing the step amount.
20820 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20823 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
20826 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
20828 * doc/sourcebuild.texi (Endianness): New subsubsection.
20830 2018-05-23 Luis Machado <luis.machado@linaro.org>
20832 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20833 <prefetch_dynamic_strides>: New const bool field.
20834 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20835 prefetch_dynamic_strides.
20836 (exynosm1_prefetch_tune): Likewise.
20837 (thunderxt88_prefetch_tune): Likewise.
20838 (thunderx_prefetch_tune): Likewise.
20839 (thunderx2t99_prefetch_tune): Likewise.
20840 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
20842 (aarch64_override_options_internal): Update to set
20843 PARAM_PREFETCH_DYNAMIC_STRIDES.
20844 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20845 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20846 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20847 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20848 prefetch-dynamic-strides setting.
20850 2018-05-23 Luis Machado <luis.machado@linaro.org>
20852 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20853 <minimum_stride>: New const int field.
20854 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20855 minimum_stride field defaulting to -1.
20856 (exynosm1_prefetch_tune): Likewise.
20857 (thunderxt88_prefetch_tune): Likewise.
20858 (thunderx_prefetch_tune): Likewise.
20859 (thunderx2t99_prefetch_tune): Likewise.
20860 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
20861 <default_opt_level>: Set to 3.
20862 (aarch64_override_options_internal): Update to set
20863 PARAM_PREFETCH_MINIMUM_STRIDE.
20864 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20865 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20866 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20867 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20868 stride is constant and is below the minimum stride threshold.
20870 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20872 * config/arm/arm-cpus.in (mode26): Delete.
20873 (armv4): Delete mode26 reference.
20874 * config/arm/arm.c (arm_configure_build_target): Delete use of
20877 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20879 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
20881 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
20882 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
20884 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
20885 (floatunsdidf2): Ditto.
20887 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20889 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
20890 (fixuns_trunc<mode>si2_avx512f): Ditto.
20891 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
20892 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
20893 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
20895 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20897 PR rtl-optimization/79985
20898 * df-scan.c (df_insn_refs_collect): Remove special case for
20899 global registers and asm statements.
20901 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20903 * extend.texi (Global Register Variables): Rewrite the bullet list.
20904 Note that the register is available for allocation. Note that access
20905 via inline asm must use constraints. Add note about async-signal
20906 handlers. Remove paragraph about automagic register selection.
20908 2018-05-23 Richard Biener <rguenther@suse.de>
20910 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
20911 of fixed offset from memset VN.
20913 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
20915 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
20916 first_interp field.
20917 (alloc_cand_and_find_basis): Initialize first_interp field.
20918 (slsr_process_mul): Modify first_interp field.
20919 (slsr_process_add): Likewise.
20920 (slsr_process_cast): Modify first_interp field for each new
20922 (slsr_process_copy): Likewise.
20923 (dump_candidate): Dump first_interp field.
20924 (replace_mult_candidate): Process all interpretations, not just
20926 (replace_rhs_if_not_dup): Likewise.
20927 (replace_one_candidate): Likewise.
20929 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
20931 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
20933 (aarch64_needs_frame_chain): New function.
20934 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
20936 2018-05-23 Sudakshina Das <sudi.das@arm.com>
20939 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
20940 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
20941 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
20942 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
20943 as true for strict-align.
20944 (aarch64_can_inline_p): Perform checks even when callee has no
20945 attributes to check for strict alignment.
20946 * doc/extend.texi (AArch64 Function Attributes): Document
20948 * doc/invoke.texi: (AArch64 Options): Likewise.
20950 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
20952 PR tree-optimization/85853
20953 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
20954 the handling of the root of the node to...
20955 (vect_slp_analyze_node_operations_1): ...this new function,
20956 and run the whole thing with the child nodes' def types
20957 set according to their SLP node's def type.
20959 2018-05-23 Richard Biener <rguenther@suse.de>
20961 PR middle-end/85874
20962 * tree-data-ref.c (create_runtime_alias_checks): Defer
20963 and ignore overflow warnings.
20965 2018-05-23 Yury Gribov <tetra2005@gmail.com>
20967 PR tree-optimization/85822
20968 * tree-vrp.c (is_masked_range_test): Fix handling of negative
20971 2018-05-23 Richard Biener <rguenther@suse.de>
20973 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
20974 memset constants via native_interpret_expr.
20976 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
20979 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
20981 (cgraph_node::create_alias): Likewise.
20982 (cgraph_node::get_availability): Check ifunc_resolver instead
20983 of looking up ifunc attribute.
20984 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
20985 * varasm.c (do_assemble_alias): Likewise.
20986 (assemble_alias): Likewise.
20987 (default_binds_local_p_3): Likewise.
20988 * cgraph.h (cgraph_node): Add ifunc_resolver.
20989 (cgraph_node::only_called_directly_or_aliased_p): Return false
20990 for IFUNC resolver.
20991 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
20993 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
20994 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
20995 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
20996 instead of looking up ifunc attribute.
20998 2018-05-22 Luis Machado <luis.machado@linaro.org>
21000 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
21002 2018-05-22 Martin Sebor <msebor@redhat.com>
21004 PR middle-end/85359
21005 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
21006 only when expasion succeeds.
21007 (expand_builtin_strcmp): Same.
21008 (expand_builtin_strncmp): Same.
21010 2018-05-22 Martin Sebor <msebor@redhat.com>
21012 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
21014 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
21015 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21017 * config/aarch64/aarch64-ldpstp.md: Replace uses of
21018 aarch64_mem_pair_operand with memory_operand and delete operand swapping
21020 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
21021 Add check for legitimate_address.
21022 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
21023 (aarch64_swap_ldrstr_operands): New.
21024 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
21027 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
21028 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21030 * config/aarch64/aarch64.md: New patterns to generate stp
21032 (store_pair_sw, store_pair_dw): New patterns to generate stp for
21033 single words and double words.
21034 (load_pair_sw, load_pair_dw): Likewise.
21035 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
21037 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
21039 * config/aarch64/aarch64-ldpstp.md: Modify peephole
21040 for different mode ldpstp and add peephole for merged zero stores.
21041 Likewise for loads.
21042 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
21044 (aarch64_gen_store_pair): Rename calls to match new patterns.
21045 (aarch64_gen_load_pair): Rename calls to match new patterns.
21046 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
21047 (load_pair<DREG:mode><DREG2:mode>): ... This.
21048 (store_pair<mode>): Rename to...
21049 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
21050 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
21051 New mode iterators.
21052 (V_INT_EQUIV): Handle SImode.
21053 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
21056 2018-05-22 Martin Sebor <msebor@redhat.com>
21059 * calls.c (maybe_warn_nonstring_arg): Use string length to set
21060 or ajust the presumed bound on an operation to avoid unnecessary
21063 2018-05-22 Martin Sebor <msebor@redhat.com>
21065 PR tree-optimization/85826
21066 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
21067 assuming that a DECL necesarily has a constant size.
21069 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
21071 PR middle-end/85862
21072 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
21074 2018-05-22 Richard Biener <rguenther@suse.de>
21076 PR tree-optimization/85834
21077 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
21078 non-constant and non-zero memset arguments.
21080 2018-05-22 Martin Liska <mliska@suse.cz>
21083 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
21085 2018-05-22 Richard Biener <rguenther@suse.de>
21087 PR tree-optimization/85863
21088 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
21089 comparisons when vectype is specified.
21090 (vectorizable_condition): Do not specify vectype for
21091 vect_is_simple_cond when SLP vectorizing.
21093 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
21096 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
21097 define __ibm128 as long double.
21098 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
21099 as a distinct type when IEEE 128-bit support is enabled.
21100 (init_float128_ieee): Fix up conversions between IFmode and IEEE
21101 128-bit types to use the correct functions.
21102 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
21103 convert between 128-bit floating point types that have different
21104 modes but the same representation, instead of using gen_lowpart to
21106 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
21108 (IFKF_reg): New attributes to give the register constraints for
21110 (extend<mode>tf2_internal): New insns to mark an explicit
21111 conversion between 128-bit floating point types that have a
21112 different mode but share the same representation.
21114 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
21116 PR tree-optimization/85814
21117 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
21118 a null return from get_strinfo when unsharing the next
21119 strinfo in the chain.
21121 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21124 * varasm.c (weak_finish): Clean up weak_decls.
21126 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21128 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
21129 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
21130 UNSPEC_UADALP values.
21131 * config/aarch64/iterators.md (ABAL): New int iterator.
21134 (sur): Add mappings for the above.
21135 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
21137 (aarch64_<sur>abal<mode>_4): Likewise.
21138 (aarch64_<sur>adalp<mode>_3): Likewise.
21139 (<sur>sadv16qi): New define_expand.
21141 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
21143 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
21144 (*movdf_internal): Ditto.
21145 (*rcpsf2_sse): Ditto.
21146 (*rsqrtsf2_sse): Ditto.
21147 (*sqrt<mode>2_sse): Ditto.
21149 2018-05-21 Tamar Christina <tamar.christina@arm.com>
21151 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
21153 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
21154 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
21155 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
21156 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
21158 * config/aarch64/arm_neon.h: Likewise.
21159 * config/aarch64/iterators.md (VQ_I): New.
21161 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
21163 * config.gcc: Add arc/t-multilib-linux to tmake_file for
21165 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
21168 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
21170 * config/nds32/constraints.md (S): New constraint.
21171 * config/nds32/nds32.md (call_internal): Use constraint S.
21172 (call_value_internal): Likewise.
21173 (sibcall_internal): Likewise.
21174 (sibcall_value_internal): Likewise.
21176 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
21177 Chung-Ju Wu <jasonwucj@gmail.com>
21179 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
21180 into consideration.
21182 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
21183 Chung-Ju Wu <jasonwucj@gmail.com>
21185 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
21186 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
21187 (nds32_rtx_costs_impl): Simplify.
21188 (nds32_address_cost_impl): Simplify.
21189 (nds32_init_rtx_costs): New function.
21190 (nds32_rtx_costs_speed_prefer): Likewise.
21191 (nds32_rtx_costs_size_prefer): Likewise.
21192 (nds32_address_cost_speed_prefer): Likewise.
21193 (nds32_address_cost_speed_fwprop): Likewise.
21194 (nds32_address_cost_size_prefer): Likewise.
21195 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
21196 * config/nds32/nds32.c (nds32_option_override): Use
21197 nds32_init_rtx_costs function.
21199 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
21201 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
21202 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
21203 (TARGET_PIPELINE_N8): Likewise.
21204 (TARGET_PIPELINE_N10): Likewise.
21205 (TARGET_PIPELINE_N13): Likewise.
21206 (TARGET_PIPELINE_GRAYWOLF): Likewise.
21208 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
21210 * config/nds32/nds32-fpu.md: Update copyright year.
21212 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21214 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
21216 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21218 * config/nds32/nds32.c
21219 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
21220 * config/nds32/nds32.opt (minline-asm-r15): New option.
21222 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21224 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
21226 * config/nds32/nds32.md (abssi2): New pattern.
21228 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
21230 * config/i386/i386.md (rex64namesuffix): New mode attribute.
21231 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
21232 Merge insn pattern from sse_cvtsi2ss<round_name> and
21233 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
21234 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
21235 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
21236 using SWI48 mode iterator.
21237 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
21238 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
21239 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
21240 pattern from sse_cvttss2si<round_saeonly_name>
21241 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
21242 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
21243 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
21244 using SWI48 mode iterator.
21245 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
21246 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
21247 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
21248 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
21249 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
21250 using SWI48 mode iterator.
21251 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
21252 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
21253 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
21254 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
21255 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
21256 SWI48 mode iterator.
21257 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
21258 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
21259 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
21260 pattern from sse_cvttsd2si<round_saeonly_name>
21261 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
21263 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21265 * config/nds32/nds32-md-auxiliary.c
21266 (nds32_valid_smw_lwm_base_p): Refine.
21267 (nds32_output_smw_single_word): Refine.
21268 (nds32_output_smw_double_word): New.
21269 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
21271 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21273 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
21274 (nds32_output_stack_pop): Refine.
21275 (nds32_expand_unaligned_load): Refine.
21276 (nds32_expand_unaligned_store): Refine.
21278 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21279 Chung-Ju Wu <jasonwucj@gmail.com>
21281 * config/nds32/constants.md: Add TP_REGNUM constant.
21282 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
21283 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
21285 * config/nds32/nds32-doubleword.md: Consider flag_pic.
21286 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
21287 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
21288 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
21289 and PIC code generation.
21290 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
21292 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
21294 * config/nds32/nds32.md: Support TLS and PIC.
21295 * config/nds32/nds32.c: Support TLS and PIC.
21296 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
21297 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
21300 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21302 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
21303 mode with E_ prefix.
21305 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21306 Chung-Ju Wu <jasonwucj@gmail.com>
21308 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
21309 * config/nds32/nds32-md-auxiliary.c
21310 (symbolic_reference_mentioned_p): New.
21311 (nds32_legitimize_ict_address): New.
21312 (nds32_expand_ict_move): New.
21313 (nds32_indirect_call_referenced_p): New.
21314 (nds32_symbol_binds_local_p): Delete.
21315 (nds32_long_call_p): Modify.
21316 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
21317 * config/nds32/nds32-protos.h
21318 (symbolic_reference_mentioned_p): Declare.
21319 (nds32_legitimize_ict_address): Declare.
21320 (nds32_expand_ict_move): Declare.
21321 (nds32_indirect_call_referenced_p): Declare.
21322 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
21323 (nds32_relax_group): Use nds32_ict_const_p as condition.
21324 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
21325 (nds32_asm_file_start): Output ict_model directive in asm code.
21326 (nds32_legitimate_address_p): Consider indirect call.
21327 (nds32_print_operand): Consider indirect call.
21328 (nds32_print_operand_address): Consider indirect call.
21329 (nds32_insert_attributes): Handle "indirect_call" attribute.
21330 (TARGET_LEGITIMATE_ADDRESS_P): Define.
21331 (TARGET_LEGITIMATE_CONSTANT_P): Define.
21332 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
21333 (TARGET_DELEGITIMIZE_ADDRESS): Define.
21334 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
21335 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
21336 (TARGET_ICT_MODEL_SMALL): Define.
21337 (TARGET_ICT_MODEL_LARGE): Define.
21338 * config/nds32/nds32.md (movsi): Consider ict model.
21339 (call, call_value): Consider ict model.
21340 (sibcall, sibcall_value): Consider ict model.
21341 * config/nds32/nds32.opt (mict-model): New option.
21342 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
21345 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
21346 Monk Chiang <sh.chiang04@gmail.com>
21347 Jim Wilson <jimw@sifive.com>
21349 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
21350 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
21351 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
21352 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
21353 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
21354 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
21355 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
21356 compute save_libcall_adjustment properly.
21357 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
21358 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
21359 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
21360 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
21361 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
21362 (ABI_SPEC): Handle mabi=ilp32e.
21363 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
21364 (RVE): Add RVE mask.
21365 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
21366 <-march>: Add rv32e as an example.
21368 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
21371 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
21372 (intra_create_variable_infos): Handle C++ constructors.
21374 2018-05-18 Martin Liska <mliska@suse.cz>
21376 * passes.def: Remove a redundant pass.
21378 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
21381 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
21383 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21385 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
21387 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
21389 (armv2, armv2a, armv3, armv3m): Delete architectures.
21390 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
21391 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
21392 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
21394 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
21395 (*mulsidi3adddi): Likewise.
21396 (mulsidi3): Likewise.
21397 (*mulsidi3_nov6): Likewise.
21398 (umulsidi3): Likewise.
21399 (umulsidi3_nov6): Likewise.
21400 (umaddsidi4): Likewise.
21401 (*umulsidi3adddi): Likewise.
21402 (smulsi3_highpart): Likewise.
21403 (*smulsi3_highpart_nov6): Likewise.
21404 (umulsi3_highpart): Likewise.
21405 (*umulsi3_highpart_nov6): Likewise.
21406 * config/arm/arm.h (arm_arch3m): Delete.
21407 * config/arm/arm.c (arm_arch3m): Delete.
21408 (arm_option_override_internal): Update armv3-related comment.
21409 (arm_configure_build_target): Delete use of isa_bit_mode32.
21410 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
21411 (arm_rtx_costs_internal): Delete check of arm_arch3m.
21412 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
21413 (mulsa3): Likewise.
21414 (mulusa3): Likewise.
21415 * config/arm/arm-protos.h (arm_arch3m): Delete.
21416 * config/arm/arm-tables.opt: Regenerate.
21417 * config/arm/arm-tune.md: Likewise.
21418 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
21419 deleted architectures.
21421 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21423 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
21424 (armv5t, armv5te): New features.
21425 (ARMv5, ARMv5e): Delete fgroups.
21426 (ARMv5t, ARMv5te): Adjust for above changes.
21427 (ARMv6m): Likewise.
21428 (armv5, armv5e): Delete arches.
21429 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
21431 (*call_reg_arm): Likewise.
21432 (*call_value_reg_armv5): Likewise.
21433 (*call_value_reg_arm): Likewise.
21434 (*call_symbol): Likewise.
21435 (*call_value_symbol): Likewise.
21436 (*sibcall_insn): Likewise.
21437 (*sibcall_value_insn): Likewise.
21438 (clzsi2): Likewise.
21439 (prefetch): Likewise.
21440 (define_split and define_peephole2 dependent on arm_arch5):
21442 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
21444 (TARGET_ARM_QBIT): Likewise.
21445 (TARGET_DSP_MULTIPLY): Likewise.
21446 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
21447 (arm_arch5, arm_arch5e): Delete.
21448 (arm_arch5t, arm_arch5te): Declare.
21449 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
21450 (arm_arch5t): Declare.
21451 (arm_option_reconfigure_globals): Update for the above.
21452 (arm_options_perform_arch_sanity_checks): Update comment, replace
21453 use of arm_arch5 with arm_arch5t.
21454 (use_return_insn): Likewise.
21455 (arm_emit_call_insn): Likewise.
21456 (output_return_instruction): Likewise.
21457 (arm_final_prescan_insn): Likewise.
21458 (arm_coproc_builtin_available): Likewise.
21459 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
21460 arm_arch5e with arm_arch5t and arm_arch5te.
21461 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
21462 (arm_arch5t, arm_arch5te): Declare.
21463 * config/arm/arm-tables.opt: Regenerate.
21464 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
21465 * config/arm/t-multilib: Likewise.
21466 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
21467 instead of arm_arch5.
21468 (*call_reg_thumb1): Likewise.
21469 (*call_value_reg_thumb1_v5): Likewise.
21470 (*call_value_reg_thumb1): Likewise.
21471 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
21473 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
21475 2018-05-18 Martin Liska <mliska@suse.cz>
21477 PR gcov-profile/84846
21478 * doc/gcov.texi: Document -t option of gcov tool.
21480 2018-05-18 Martin Liska <mliska@suse.cz>
21482 PR gcov-profile/84846
21483 * gcov.c (print_usage): Add new -t option.
21484 (process_args): Handle the option.
21485 (generate_results): Use stdout as output when requested by
21488 2018-05-18 Martin Liska <mliska@suse.cz>
21490 PR gcov-profile/84846
21491 * coverage.c (coverage_init): Write PWD to .gcno file.
21492 * doc/gcov.texi: Document how working directory is printed.
21493 * gcov-dump.c (dump_gcov_file): Print PWD.
21494 * gcov.c (output_intermediate_file): Likewise.
21495 (read_graph_file): Read PWD string.
21496 (output_lines): Print PWD.
21498 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21500 PR middle-end/85817
21501 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
21502 for retval and return false if all args to phi are zero.
21504 2018-05-18 Richard Biener <rguenther@suse.de>
21506 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
21508 (evrp_dom_walker::before_dom_children): Call it.
21510 2018-05-18 Richard Biener <rguenther@suse.de>
21512 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
21513 results when processing array refs with variable index.
21515 2018-05-18 Toon Moene <toon@moene.org>
21517 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
21518 directly after that of -floop-interchange. Indicate that both
21519 options are enabled by default when specifying -O3.
21521 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21523 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
21524 iterator. Delete separate integer-mode vec_set<mode> expander.
21525 (aarch64_simd_vec_setv2di): Delete.
21526 (vec_setv2di): Delete.
21527 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
21528 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
21529 the "w, r" alternative.
21531 2018-05-18 Martin Liska <mliska@suse.cz>
21533 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
21534 * tree-pass.h (make_pass_lower_switch_O0): New function.
21535 * tree-switch-conversion.c (node_has_low_bound): Remove.
21536 (node_has_high_bound): Likewise.
21537 (node_is_bounded): Likewise.
21538 (class pass_lower_switch): Make it a template type and create
21540 (pass_lower_switch::execute): Add template argument.
21541 (make_pass_lower_switch): New function.
21542 (make_pass_lower_switch_O0): New function.
21543 (do_jump_if_equal): Remove.
21544 (emit_case_nodes): Simplify to just handle all 3 cases and leave
21545 all the hard work to tree optimization passes.
21547 2018-05-18 Martin Liska <mliska@suse.cz>
21549 * dbgcnt.c (limit_low): Renamed from limit.
21550 (limit_high): New variable.
21551 (dbg_cnt_is_enabled): Check for upper limit.
21552 (dbg_cnt): Adjust dumping.
21553 (dbg_cnt_set_limit_by_index): Add new argument for high
21555 (dbg_cnt_set_limit_by_name): Likewise.
21556 (dbg_cnt_process_single_pair): Parse new format.
21557 (dbg_cnt_process_opt): Use strtok.
21558 (dbg_cnt_list_all_counters): Remove 'value' and add
21560 * doc/invoke.texi: Document changes.
21562 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
21564 * doc/sourcebuild.texi (scalar_all_fma): Document.
21565 * tree.def (FMA_EXPR): Delete.
21566 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
21567 * internal-fn.c (ternary_direct): New macro.
21568 (expand_ternary_optab_fn): Likewise.
21569 (direct_ternary_optab_supported_p): Likewise.
21570 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
21571 * builtins.c (fold_builtin_fma): Delete.
21572 (fold_builtin_3): Don't call it.
21573 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
21574 * expr.c (expand_expr_real_2): Likewise.
21575 * fold-const.c (operand_equal_p): Likewise.
21576 (fold_ternary_loc): Likewise.
21577 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
21578 * gimple.c (DEFTREECODE): Likewise.
21579 * gimplify.c (gimplify_expr): Likewise.
21580 * optabs-tree.c (optab_for_tree_code): Likewise.
21581 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
21582 * tree-eh.c (operation_could_trap_p): Likewise.
21583 (stmt_could_throw_1_p): Likewise.
21584 * tree-inline.c (estimate_operator_cost): Likewise.
21585 * tree-pretty-print.c (dump_generic_node): Likewise.
21586 (op_code_prio): Likewise.
21587 * tree-ssa-loop-im.c (stmt_cost): Likewise.
21588 * tree-ssa-operands.c (get_expr_operands): Likewise.
21589 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
21590 * fold-const-call.h (fold_fma): Delete.
21591 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
21592 CFN_FNMA and CFN_FNMS.
21593 (fold_fma): Delete.
21594 * genmatch.c (combined_fn): New enum.
21595 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
21596 (commutative_op): New function.
21597 (commutate): Use it. Handle more than 2 operands.
21598 (dt_operand::gen_gimple_expr): Use commutative_op.
21599 (parser::parse_expr): Allow :c to be used with non-binary
21600 operators if the commutative operand is known.
21601 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
21602 CFN_FMS, CFN_FNMA and CFN_FNMS.
21603 (backprop::process_assign_use): Remove FMA_EXPR handling.
21604 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
21605 (gen_hsa_fma): New function.
21606 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
21607 IFN_FNMA and IFN_FNMS.
21608 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
21609 * gimple-fold.h (follow_all_ssa_edges): Declare.
21610 * gimple-fold.c (follow_all_ssa_edges): New function.
21611 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
21612 gimple_build interface and use follow_all_ssa_edges to fold the result.
21613 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
21614 instead of checking for optabs directly.
21615 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
21616 rather than FMA_EXPRs.
21617 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
21618 call to IFN_FMA instead of an FMA_EXPR.
21620 2018-05-17 Jim Wilson <jimw@sifive.com>
21622 * expr.c (do_tablejump): When converting index to Pmode, if we have a
21623 sign extended promoted subreg, and the range does not have the sign bit
21624 set, then do a sign extend.
21626 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
21627 test, check for sign extended subreg and/or constant operands, and
21628 do a sign extend in that case.
21630 2018-05-17 Steve Ellcey <sellcey@cavium.com>
21632 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
21633 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
21635 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
21636 Change logics_shift_reg to logics_shift_imm.
21637 (thunderx2t99_fp_loadpair_basic): Delete.
21638 (thunderx2t99_fp_storepair_basic): Delete.
21639 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
21640 (thunderx2t99_asimd_polynomial): Delete.
21641 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
21642 and neon_fp_mul_d_scalar_q.
21643 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
21644 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
21645 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
21646 (thunderx2t99_asimd_lut): Add missing tbl types.
21647 (thunderx2t99_asimd_ext): Delete.
21648 (thunderx2t99_asimd_load1_1_mult): Delete.
21649 (thunderx2t99_asimd_load1_2_mult): Delete.
21650 (thunderx2t99_asimd_load1_ldp): New.
21651 (thunderx2t99_asimd_load1): New.
21652 (thunderx2t99_asimd_load2): Add missing *load2* types.
21653 (thunderx2t99_asimd_load3): New.
21654 (thunderx2t99_asimd_load4): New.
21655 (thunderx2t99_asimd_store1_1_mult): Delete.
21656 (thunderx2t99_asimd_store1_2_mult): Delete.
21657 (thunderx2t99_asimd_store2_mult): Delete.
21658 (thunderx2t99_asimd_store2_onelane): Delete.
21659 (thunderx2t99_asimd_store_stp): New.
21660 (thunderx2t99_asimd_store1): New.
21661 (thunderx2t99_asimd_store2): New.
21662 (thunderx2t99_asimd_store3): New.
21663 (thunderx2t99_asimd_store4): New.
21665 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
21667 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
21668 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
21670 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
21671 Segher Boessenkool <segher@kernel.crashing.org>
21674 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
21677 2018-05-17 Richard Biener <rguenther@suse.de>
21679 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
21680 for pruning loop and prune defs feeding only already visited PHIs.
21682 2018-05-17 Richard Biener <rguenther@suse.de>
21684 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
21686 2018-05-17 Bin Cheng <bin.cheng@arm.com>
21687 Richard Biener <rguenther@suse.de>
21689 PR tree-optimization/85793
21690 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
21691 for VMAT_ELEMENTWISE.
21693 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21695 * internal-fn.h (lookup_internal_fn): Declare
21696 * internal-fn.c (lookup_internal_fn): New function.
21697 * gimple.c (gimple_build_call_from_tree): Handle calls to
21698 internal functions.
21699 * gimple-pretty-print.c (dump_gimple_call): Print "." before
21700 internal function names.
21701 * tree-pretty-print.c (dump_generic_node): Likewise.
21702 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
21704 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21706 * gimple-fold.h (gimple_build): Make the function forms take
21707 combined_fn rather than built_in_function.
21708 (gimple_simplify): Likewise.
21709 * gimple-match-head.c (gimple_simplify): Likewise.
21710 * gimple-fold.c (gimple_build): Likewise.
21711 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
21712 rather than gimple_build_call_internal.
21713 (get_initial_defs_for_reduction): Likewise.
21714 (vect_create_epilog_for_reduction): Likewise.
21715 (vectorizable_live_operation): Likewise.
21717 2018-05-17 Martin Liska <mliska@suse.cz>
21719 * gimple-ssa-sprintf.c (format_directive): Do not use
21720 space in between 'G_' and '('.
21722 2018-05-17 Jakub Jelinek <jakub@redhat.com>
21725 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
21726 even if the mask is not all ones.
21729 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
21731 (ix86_gimple_fold_builtin): Likewise.
21734 * config/i386/i386.c: Include tree-vector-builder.h.
21735 (ix86_vector_shift_count): New function.
21736 (ix86_fold_builtin): Fold shift builtins by scalar count.
21737 (ix86_gimple_fold_builtin): Likewise.
21739 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
21740 _mm512_setzero): New intrinsics.
21742 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
21743 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21745 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
21746 code generation for cases where splatting a value is not useful.
21747 * simplify-rtx.c (simplify_ternary_operation): Simplify
21748 vec_merge across a vec_duplicate and a paradoxical subreg forming
21749 a vector mode to a vec_concat.
21751 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
21753 * config.gcc: Support "goldmont-plus".
21754 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21756 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21757 PROCESSOR_GOLDMONT_PLUS.
21758 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
21759 (processor_target_table): Add "goldmont-plus".
21760 (PTA_GOLDMONT_PLUS): Define.
21761 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
21762 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
21763 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
21764 (fold_builtin_cpu): Add "goldmont-plus".
21765 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
21766 (ix86_option_override_internal): Add "goldmont-plus".
21767 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
21768 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
21769 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
21770 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
21772 2018-05-17 Richard Biener <rguenther@suse.de>
21774 PR tree-optimization/85757
21775 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
21776 remove defs that only feed that PHI from further processing.
21778 2018-05-16 Jim Wilson <jimw@sifive.com>
21780 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
21782 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
21784 2018-05-16 Mark Wielaard <mark@klomp.org>
21786 * dwarf2out.c (count_index_strings): New function.
21787 (output_indirect_strings): Call count_index_strings and generate
21788 header for dwarf_version >= 5.
21790 2018-05-16 Mark Wielaard <mark@klomp.org>
21792 * dwarf2out.c (dwarf_FORM): New function.
21793 (set_indirect_string): Use dwarf_FORM.
21794 (reset_indirect_string): Likewise.
21795 (size_of_die): Likewise.
21796 (value_format): Likewise.
21797 (output_die): Likewise.
21798 (add_skeleton_AT_string): Likewise.
21799 (output_macinfo_op): Likewise.
21800 (index_string): Likewise.
21801 (output_index_string_offset): Likewise.
21802 (output_index_string): Likewise.
21803 (count_index_strings): Likewise.
21805 2018-05-16 Carl Love <cel@us.ibm.com>
21807 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
21808 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
21810 2018-05-16 Martin Jambor <mjambor@suse.cz>
21812 * ipa-prop.c (ipa_free_all_edge_args): Remove.
21813 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
21815 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
21817 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
21818 (fnma<mode>4): Likewise.
21819 (fms<mode>4): Likewise.
21820 (fnms<mode>4): Likewise.
21821 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
21822 (aarch64_fnma<mode>4): Likewise.
21823 (aarch64_fms<mode>4): Likewise.
21824 (aarch64_fnms<mode>4): Likewise.
21825 (aarch64_fnmadd<mode>4): Likewise.
21827 2018-05-16 Jason Merrill <jason@redhat.com>
21829 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
21831 2018-05-16 Richard Biener <rguenther@suse.de>
21833 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
21834 (dump_stmt_cost): Declare.
21835 (add_stmt_cost): Dump cost we add.
21836 (add_stmt_costs): New function.
21837 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
21838 No longer exported.
21839 (vect_analyze_stmt): Adjust prototype.
21840 (vectorizable_condition): Likewise.
21841 (vectorizable_live_operation): Likewise.
21842 (vectorizable_reduction): Likewise.
21843 (vectorizable_induction): Likewise.
21844 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
21845 cost vector to pass to vectorizable_ and record afterwards.
21846 (vect_model_reduction_cost): Take cost vector argument and adjust.
21847 (vect_model_induction_cost): Likewise.
21848 (vectorizable_reduction): Likewise.
21849 (vectorizable_induction): Likewise.
21850 (vectorizable_live_operation): Likewise.
21851 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
21852 SLP_TREE_NUMBER_OF_VEC_STMTS.
21853 (vect_analyze_slp_cost_1): Remove.
21854 (vect_analyze_slp_cost): Likewise.
21855 (vect_slp_analyze_node_operations): Take visited args and
21856 a target cost vector. Avoid processing already visited stmt sets.
21857 (vect_slp_analyze_operations): Use a local cost vector to gather
21858 costs and register those of non-discarded instances.
21859 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
21860 (vect_schedule_slp_instance): Remove copying of
21861 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
21863 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
21864 adding cost. Record cost entry location.
21865 (vect_prologue_cost_for_slp_op): Function to compute cost of
21866 a constant or invariant generated for SLP vect in the prologue,
21867 split out from vect_analyze_slp_cost_1.
21868 (vect_model_simple_cost): Make static. Adjust for SLP costing.
21869 (vect_model_promotion_demotion_cost): Likewise.
21870 (vect_model_store_cost): Likewise, make static.
21871 (vect_model_load_cost): Likewise.
21872 (vectorizable_bswap): Add cost vector arg and adjust.
21873 (vectorizable_call): Likewise.
21874 (vectorizable_simd_clone_call): Likewise.
21875 (vectorizable_conversion): Likewise.
21876 (vectorizable_assignment): Likewise.
21877 (vectorizable_shift): Likewise.
21878 (vectorizable_operation): Likewise.
21879 (vectorizable_store): Likewise.
21880 (vectorizable_load): Likewise.
21881 (vectorizable_condition): Likewise.
21882 (vectorizable_comparison): Likewise.
21883 (can_vectorize_live_stmts): Likewise.
21884 (vect_analyze_stmt): Likewise.
21885 (vect_transform_stmt): Adjust calls to vectorizable_*.
21886 * tree-vectorizer.c: Include gimple-pretty-print.h.
21887 (dump_stmt_cost): New function.
21889 2018-05-16 Richard Biener <rguenther@suse.de>
21891 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
21892 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
21893 * tree-ssa-dse.c: Include tree-ssa-loop.h.
21894 (check_name): New callback.
21895 (dse_classify_store): Track cycles via a visited bitmap of PHI
21896 defs and simplify handling of in-loop and across loop dead stores
21897 and properly fail for loop-variant refs. Handle byte-tracking with
21898 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
21901 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
21903 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
21904 (vect_get_mask_type_for_stmt): Likewise.
21905 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
21907 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
21908 to determine the statement's vector type and the vector type that
21909 should be used for calculating nunits. Deal with cases in which
21910 the type has to be deferred.
21911 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
21912 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
21913 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
21914 (vect_determine_vf_for_stmt): New functions, split out from...
21915 (vect_determine_vectorization_factor): ...here.
21916 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
21917 (vect_get_mask_type_for_stmt): New functions, split out from
21918 vect_determine_vectorization_factor.
21920 2018-05-16 Richard Biener <rguenther@suse.de>
21922 * tree-cfg.c (verify_gimple_assign_ternary): Properly
21923 verify the [VEC_]COND_EXPR embedded comparison.
21925 2018-05-15 Martin Sebor <msebor@redhat.com>
21927 PR tree-optimization/85753
21928 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
21929 RECORD_TYPE in addition to ARRAY_TYPE.
21931 2018-05-15 Martin Sebor <msebor@redhat.com>
21933 PR middle-end/85643
21934 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
21936 2018-05-15 Richard Biener <rguenther@suse.de>
21938 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
21939 add by_clobber_p one. Change algorithm to collect all defs
21940 representing uses we need to walk and try reducing them to
21941 a single one before failing.
21942 (dse_dom_walker::dse_optimize_stmt): Adjust.
21944 2018-05-13 Mark Wielaard <mark@klomp.org>
21946 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
21947 (size_of_loc_descr): Likewise.
21948 (output_loc_operands): Likewise.
21949 (output_loc_operands_raw): Likewise.
21950 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
21951 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
21952 (hash_loc_operands): Likewise.
21953 (compare_loc_operands): Likewise.
21955 2018-05-14 Mark Wielaard <mark@klomp.org>
21957 * dwarf2out.c (count_index_addrs): New function.
21958 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
21960 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21962 PR tree-optimization/83648
21963 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
21964 return value as malloc candidate.
21966 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21969 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
21970 param as true in call to suggest_attribute.
21972 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
21974 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
21975 -mreadonly-in-sdata.
21977 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21979 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
21981 (aarch64_crypto_aesd_fused): Likewise.
21983 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
21985 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
21986 (movsi_aarch64): Likewise.
21987 (load_pairsi): Likewise.
21988 (load_pairdi): Likewise.
21989 (store_pairsi): Likewise.
21990 (store_pairdi): Likewise.
21991 (load_pairsf): Likewise.
21992 (load_pairdf): Likewise.
21993 (store_pairsf): Likewise.
21994 (store_pairdf): Likewise.
21995 (zero_extend): Likewise.
21996 (trunc): Swap alternatives.
21997 (fcvt_target): Add '?' to prefer w over r.
21999 2018-05-14 Jakub Jelinek <jakub@redhat.com>
22002 * config/i386/i386.md: Disallow non-commutative arithmetics in
22003 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
22004 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
22005 in the peephole2 before it.
22007 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
22009 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
22010 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
22011 (ix86_handle_option): Handle -mcldemote.
22012 * config.gcc: New header.
22013 * config/i386/cldemoteintrin.h: New file.
22014 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
22015 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
22017 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22018 OPTION_MASK_ISA_CLDEMOTE.
22019 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
22020 (ix86_valid_target_attribute_inner_p): Ditto.
22021 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
22022 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
22023 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
22024 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
22025 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
22027 * config/i386/i386.opt: Add -mcldemote.
22028 * config/i386/x86intrin.h: New header.
22029 * doc/invoke.texi: Add -mcldemote.
22031 2018-05-14 Richard Biener <rguenther@suse.de>
22033 * doc/match-and-simplify.texi: Adjust :s documentation.
22035 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
22037 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
22038 intended memcpy size.
22039 (REORDER_45): Likewise.
22041 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
22043 * sort.cc: New file.
22044 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
22045 * vec.c (qsort_chk): Use gcc_qsort.
22046 * Makefile.in (OBJS-libcommon): Add sort.o.
22047 (build/sort.o): New target. Use it...
22048 (BUILD_RTL): ... here, and...
22049 (build/gencfn-macros): ... here, and...
22050 (build/genmatch): ... here.
22052 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22053 Chung-Ju Wu <jasonwucj@gmail.com>
22055 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
22056 * config/nds32/nds32-graywolf.md: New file.
22057 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
22058 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
22060 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
22061 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
22062 * config/nds32/nds32.md (pipeline_model): Add graywolf.
22063 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
22064 * config/nds32/pipelines.md: Include n15 settings.
22066 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22067 Chung-Ju Wu <jasonwucj@gmail.com>
22069 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
22070 * config/nds32/nds32-n13.md: New file.
22071 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
22072 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
22074 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
22075 * config/nds32/nds32.md (pipeline_model): Add n13.
22076 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
22077 * config/nds32/pipelines.md: Include n13 settings.
22079 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22080 Chung-Ju Wu <jasonwucj@gmail.com>
22082 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
22083 * config/nds32/nds32-n10.md: New file.
22084 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
22085 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
22087 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
22088 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
22089 * config/nds32/nds32.md (pipeline_model): Add n10.
22090 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
22091 * config/nds32/pipelines.md: Include n10 settings.
22093 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
22094 Kito Cheng <kito.cheng@gmail.com>
22095 Chung-Ju Wu <jasonwucj@gmail.com>
22097 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
22098 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22099 Add enum values for DSP extension instructions.
22100 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
22102 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
22103 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
22104 New code iterators.
22105 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
22106 * config/nds32/nds32-dspext.md: New file for DSP implementation.
22107 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
22108 * config/nds32/nds32-intrinsic.md: Likewise.
22109 * config/nds32/nds32_intrinsic.h: Likewise.
22110 * config/nds32/nds32-md-auxiliary.c: Likewise.
22111 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
22112 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
22113 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
22114 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
22115 * config/nds32/nds32-protos.h: New declarations for DSP extension.
22116 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
22117 TYPE_DMAC in switch statement.
22118 * config/nds32/nds32.c: New checking and implementation for DSP
22119 extension instructions.
22120 * config/nds32/nds32.h: Likewise.
22121 * config/nds32/nds32.md: Likewise.
22122 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
22123 * config/nds32/predicates.md: Implement new predicates for DSP
22126 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22128 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
22129 Reformat alternatives and attributes so it is easier to identify
22130 which constraints/attributes go with which instruction.
22131 (mov<mode>_hardfloat32, FMOVE64): Likewise.
22132 (mov<mode>_softfloat32, FMOVE64): Likewise.
22133 (mov<mode>_hardfloat64, FMOVE64): Likewise.
22134 (mov<mode>_softfloat64, FMOVE64): Likewise.
22136 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22138 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22140 (Basic PowerPC Built-in Functions): The new name of the
22141 subsection previously known as "PowerPC Built-in Functions".
22142 (Basic PowerPC Built-in Functions Available on all Configurations):
22144 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
22145 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
22146 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
22147 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
22149 2018-05-11 Martin Jambor <mjambor@suse.cz>
22152 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
22155 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
22158 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
22160 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
22162 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
22163 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
22164 (ix86_handle_option): Handle -mwaitpkg.
22165 * config.gcc: New header.
22166 * config/i386/cpuid.h (bit_WAITPKG): New bit.
22167 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
22168 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
22170 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22171 OPTION_MASK_ISA_WAITPKG.
22172 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
22173 (ix86_option_override_internal): Add PTA_WAITPKG.
22174 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
22175 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
22176 IX86_BUILTIN_TPAUSE.
22177 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
22178 __builtin_ia32_umwait and __builtin_ia32_tpause.
22179 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
22180 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
22181 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
22182 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
22183 UNSPECV_TPAUSE): New.
22184 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
22185 * config/i386/i386.opt: Add -mwaitpkg.
22186 * config/i386/waitpkgintrin.h: New file.
22187 * config/i386/x86intrin.h: New header.
22188 * doc/invoke.texi: Add -mwaitpkg.
22190 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
22193 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
22195 (cortex-m0): Use armv6s-m isa.
22196 (cortex-m0plus): Likewise.
22197 (cortex-m1): Likewise.
22198 (cortex-m0.small-multiply): Likewise.
22199 (cortex-m0plus.small-multiply): Likewise.
22200 (cortex-m1.small-multiply): Likewise.
22202 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
22203 Jakub Jelinek <jakub@redhat.com>
22205 PR tree-optimization/85692
22206 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
22207 source permute as well.
22209 2018-05-11 Martin Liska <mliska@suse.cz>
22212 * doc/extend.texi: Document LLVM style format for no_sanitize
22215 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
22217 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
22218 mode_supports_vsx_dform_quad to mode_supports_dq_form.
22219 (mode_supports_vsx_dform_quad): Likewise.
22220 (mode_supports_vmx_dform): Move these functions to be next to the
22221 other mode_supports functions.
22222 (mode_supports_dq_form): Likewise.
22223 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
22224 mode_supports_dq_form.
22225 (reg_offset_addressing_ok_p): Likewise.
22226 (offsettable_ok_by_alignment): Likewise.
22227 (rs6000_legitimate_offset_address_p): Likewise.
22228 (legitimate_lo_sum_address_p): Likewise.
22229 (rs6000_legitimize_address): Likewise.
22230 (rs6000_legitimize_reload_address): Likewise.
22231 (rs6000_secondary_reload_inner): Likewise.
22232 (rs6000_preferred_reload_class): Likewise.
22233 (rs6000_output_move_128bit): Likewise.
22235 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
22237 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
22238 Generate SImode target register for null target.
22239 <case IX86_BUILTIN_XGETBV>: Ditto.
22240 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
22241 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
22243 2018-05-10 Carl Love <cel@us.ibm.com>
22245 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
22246 dcbtt and dcbtstt if operands[2] is 0.
22248 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
22251 * config/i386/sse.md (usadv64qi): New expander.
22253 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
22255 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
22256 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
22257 -maltivec=be support.
22258 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
22259 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
22260 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22261 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
22262 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
22263 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
22264 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
22265 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
22266 altivec_vsumsws): Adjust.
22267 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
22268 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
22269 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
22271 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
22272 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
22273 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
22274 (altivec_lve<VI_char>x): Delete expand.
22275 (*altivec_lve<VI_char>x_internal): Rename to...
22276 (altivec_lve<VI_char>x): ... this.
22277 (altivec_lvxl_<mode>): Delete expand.
22278 (*altivec_lvxl_<mode>_internal): Rename to ...
22279 (altivec_lvxl_<mode>): ... this.
22280 (altivec_stvxl_<mode>): Delete expand.
22281 (*altivec_stvxl_<mode>_internal): Rename to ...
22282 (altivec_stvxl_<mode>): ... this.
22283 (altivec_stve<VI_char>x): Delete expand.
22284 (*altivec_stve<VI_char>x_internal): Rename to ...
22285 (altivec_stve<VI_char>x): ... this.
22286 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
22287 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
22288 reduc_plus_scal_<mode>): Adjust.
22289 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
22291 (rs6000_cpu_cpp_builtins): Adjust.
22292 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
22293 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
22294 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
22295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
22296 -maltivec=be support.
22297 (rs6000_split_vec_extract_var): Adjust.
22298 (rs6000_split_v4si_init): Adjust.
22299 (swap_selector_for_mode): Delete.
22300 (altivec_expand_lvx_be, altivec_expand_stvx_be,
22301 altivec_expand_stvex_be): Delete.
22302 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
22303 -maltivec=be support.
22304 (rs6000_gimple_fold_builtin): Ditto.
22305 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
22307 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
22308 (TARGET_DIRECT_MOVE_64BIT): Adjust.
22309 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
22310 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
22311 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
22312 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
22313 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
22314 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
22315 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
22316 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
22317 anonymous split): Adjust.
22318 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
22319 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
22321 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
22323 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
22324 when --with-gxx-include-dir is also specified.
22325 * configure: Regenerate.
22327 2018-05-09 Jim Wilson <jimw@sifive.com>
22330 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
22331 * config/riscv/t-withmultilib: New.
22332 * config/riscv/withmultilib.h: New.
22333 * doc/install.texi: Document RISC-V --with-multilib-list support.
22335 2018-05-09 Richard Biener <rguenther@suse.de>
22337 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
22339 (vect_bb_vectorization_profitable_p): Adjust. Compute
22340 actual scalar cost using the cost vector and the add_stmt_cost
22343 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22345 PR rtl-optimization/85645
22346 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
22347 in the REG_CFA_REGISTER note for LR, don't leave it empty.
22349 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22351 PR rtl-optimization/85645
22352 * shrink-wrap.c (spread_components): Return a boolean saying if
22353 anything was changed.
22354 (try_shrink_wrapping_separate): Iterate spread_components until
22355 nothing changes anymore.
22357 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22359 PR rtl-optimization/85645
22360 * regrename.c (build_def_use): Also kill the chains that include the
22361 destination of a REG_CFA_REGISTER note.
22363 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22365 PR rtl-optimization/85645
22366 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
22367 insn that has a REG_CFA_REGISTER note.
22369 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
22371 * cfgexpand.c (expand_clobber): New function.
22372 (expand_gimple_stmt_1): Use it.
22373 * tree-vect-stmts.c (vect_clobber_variable): New function,
22375 (vectorizable_simd_clone_call): ...here.
22376 (vectorizable_store): Emit a clobber either side of an
22377 IFN_STORE_LANES sequence.
22378 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
22380 2018-05-09 Tom de Vries <tom@codesourcery.com>
22383 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
22384 (define_insn "trap_if_false"): Add exit after trap.
22386 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
22388 PR rtl-optimization/85638
22389 * bb-reorder.c: Include common/common-target.h.
22390 (create_forwarder_block): New function extracted from...
22391 (fix_up_crossing_landing_pad): ...here. Rename into...
22392 (dw2_fix_up_crossing_landing_pad): ...this.
22393 (sjlj_fix_up_crossing_landing_pad): New function.
22394 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
22395 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
22396 from both partitions and exit the loop after one iteration.
22398 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22401 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22403 (Basic PowerPC Built-in Functions): The new name of the
22404 subsection previously known as "PowerPC Built-in Functions".
22405 (Basic PowerPC Built-in Functions Available on all Configurations):
22407 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
22409 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
22411 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
22413 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
22416 2018-05-08 Jim Wilson <jimw@sifive.com>
22418 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
22419 (LD_EMUL_SUFFIX): New.
22420 (LINK_SPEC): Use it.
22422 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22424 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22426 (Basic PowerPC Built-in Functions): The new name of the
22427 subsection previously known as "PowerPC Built-in Functions".
22428 (Basic PowerPC Built-in Functions Available on all Configurations):
22430 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
22432 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
22434 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
22436 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
22439 2018-05-08 Jakub Jelinek <jakub@redhat.com>
22442 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
22443 after cmpelim optimization.
22445 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
22447 * config.gcc: Support "goldmont".
22448 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
22449 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22450 PROCESSOR_GOLDMONT.
22451 * config/i386/i386.c (m_GOLDMONT): Define.
22452 (processor_target_table): Add "goldmont".
22453 (PTA_GOLDMONT): Define.
22454 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
22455 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
22456 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
22457 (fold_builtin_cpu): Add "goldmont".
22458 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
22459 (ix86_option_override_internal): Add "goldmont".
22460 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
22461 (processor_type): Add PROCESSOR_GOLDMONT.
22462 * config/i386/i386.md: Add CPU "glm".
22463 * config/i386/glm.md: New file.
22464 * config/i386/x86-tune.def: Add m_GOLDMONT.
22465 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
22467 2018-05-08 Jakub Jelinek <jakub@redhat.com>
22470 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
22472 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
22473 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
22474 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
22477 * config/i386/i386.c (ix86_fold_builtin): Handle
22478 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
22481 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
22482 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
22484 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
22487 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
22488 (check_arch): Likewise.
22489 (check_fpu): Return the result rather than printing it.
22490 (end arch): Fix operator precedence.
22491 (end cpu): Likewise.
22492 (END): Print the result from check_fpu.
22494 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
22495 Alan Hayward <alan.hayward@arm.com>
22496 David Sherwood <david.sherwood@arm.com>
22498 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
22499 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
22500 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
22501 (*fcmuo<mode>_and): New patterns.
22503 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
22505 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
22506 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
22507 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
22508 (cmp_op, sve_imm_con): New code attributes.
22509 (SVE_COND_INT_CMP, imm_con): Delete.
22510 (cmp_op): Remove above unspecs from int attribute.
22511 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
22513 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
22514 comparison-specific unspecs.
22515 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
22516 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
22517 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
22518 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
22519 (*vec_fcm<cmp_op><mode>): Rename to...
22520 (*fcm<cmp_op><mode>): ...this and adjust likewise.
22521 (*vec_fcmuo<mode>): Rename to...
22522 (*fcmuo<mode>): ...this and adjust likewise.
22523 (*pred_fcm<cmp_op><mode>): New pattern.
22524 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
22525 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
22527 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
22529 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
22530 (aarch64_emit_sve_predicated_cond): New function.
22531 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
22532 (aarch64_emit_unspec_cond_or): Replace with...
22533 (aarch64_emit_sve_or_conds): ...this new function. Use
22534 aarch64_emit_sve_ptrue_op for the individual comparisons and
22535 aarch64_emit_binop to OR them together.
22536 (aarch64_emit_inverted_unspec_cond): Replace with...
22537 (aarch64_emit_sve_inverted_cond): ...this new function. Use
22538 aarch64_emit_sve_ptrue_op for the comparison and
22539 aarch64_emit_unop to invert the result.
22540 (aarch64_expand_sve_vec_cmp_float): Update after the above
22541 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
22543 2018-05-07 Nathan Sidwell <nathan@acm.org>
22545 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
22546 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
22547 (Backwards Compatibility): Likewise.
22549 2018-05-07 Luis Machado <luis.machado@linaro.org>
22553 2018-05-07 Luis Machado <luis.machado@linaro.org>
22555 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22556 <prefetch_dynamic_strides>: New const bool field.
22557 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22558 prefetch_dynamic_strides.
22559 (exynosm1_prefetch_tune): Likewise.
22560 (thunderxt88_prefetch_tune): Likewise.
22561 (thunderx_prefetch_tune): Likewise.
22562 (thunderx2t99_prefetch_tune): Likewise.
22563 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
22565 (aarch64_override_options_internal): Update to set
22566 PARAM_PREFETCH_DYNAMIC_STRIDES.
22567 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
22568 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
22569 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
22570 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
22571 prefetch-dynamic-strides setting.
22573 2018-05-07 Luis Machado <luis.machado@linaro.org>
22575 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22576 <minimum_stride>: New const int field.
22577 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22578 minimum_stride field.
22579 (exynosm1_prefetch_tune): Likewise.
22580 (thunderxt88_prefetch_tune): Likewise.
22581 (thunderx_prefetch_tune): Likewise.
22582 (thunderx2t99_prefetch_tune): Likewise.
22583 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
22584 (aarch64_override_options_internal): Update to set
22585 PARAM_PREFETCH_MINIMUM_STRIDE.
22586 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
22587 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
22588 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22589 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22590 stride is constant and is below the minimum stride threshold.
22592 2018-05-07 Luis Machado <luis.machado@linaro.org>
22594 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
22597 2018-05-07 Luis Machado <luis.machado@linaro.org>
22599 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22600 <prefetch_dynamic_strides>: New const bool field.
22601 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22602 prefetch_dynamic_strides.
22603 (exynosm1_prefetch_tune): Likewise.
22604 (thunderxt88_prefetch_tune): Likewise.
22605 (thunderx_prefetch_tune): Likewise.
22606 (thunderx2t99_prefetch_tune): Likewise.
22607 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
22609 (aarch64_override_options_internal): Update to set
22610 PARAM_PREFETCH_DYNAMIC_STRIDES.
22611 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
22612 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
22613 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
22614 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
22615 prefetch-dynamic-strides setting.
22617 2018-05-07 Luis Machado <luis.machado@linaro.org>
22619 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22620 <minimum_stride>: New const int field.
22621 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22622 minimum_stride field.
22623 (exynosm1_prefetch_tune): Likewise.
22624 (thunderxt88_prefetch_tune): Likewise.
22625 (thunderx_prefetch_tune): Likewise.
22626 (thunderx2t99_prefetch_tune): Likewise.
22627 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
22628 (aarch64_override_options_internal): Update to set
22629 PARAM_PREFETCH_MINIMUM_STRIDE.
22630 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
22631 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
22632 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22633 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22634 stride is constant and is below the minimum stride threshold.
22636 2018-05-06 Jakub Jelinek <jakub@redhat.com>
22639 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
22640 the type is addressable. Don't force op into register if it has
22643 2018-05-05 Roland McGrath <mcgrathr@google.com>
22646 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
22647 any section for which we don't know a specific type it should have,
22648 regardless of name. Previously this was done only for the exact
22649 names ".init_array", ".fini_array", and ".preinit_array".
22650 (default_elf_asm_named_section): Add comment about
22651 relationship with default_section_type_flags and SECTION_NOTYPE.
22652 (get_section): Don't consider it a type conflict if one side has
22653 SECTION_NOTYPE and the other doesn't, as long as neither has the
22654 SECTION_BSS et al used in the default_section_type_flags logic.
22656 2018-05-05 Tom de Vries <tom@codesourcery.com>
22659 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
22660 (workaround_barsyncs): New function.
22661 (nvptx_reorg): Use workaround_barsyncs.
22662 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22663 (define_expand "nvptx_membar_cta"): New define_expand.
22664 (define_insn "*nvptx_membar_cta"): New insn.
22666 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22668 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
22669 To improve optimization opportunities.
22670 * builtin-types.def: The new needed builtin types for the above.
22672 2018-05-04 Richard Biener <rguenther@suse.de>
22674 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
22675 * gimple-ssa-store-merging.c
22676 (imm_store_chain_info::output_merged_store): Remove redundant create,
22677 release split_store vector contents on failure.
22678 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
22679 scalar stmt vector on cache hit.
22681 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
22683 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22685 * config.gcc (powerpc-xilinx-eabi*): Remove.
22686 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
22688 (fusion_addis_mem_combo_load): Ditto.
22689 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
22691 (rs6000_cpu_cpp_builtins): Ditto.
22692 * config/rs6000/rs6000-linux.c
22693 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
22694 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
22695 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
22697 (rs6000_setup_reg_addr_masks): Ditto.
22698 (rs6000_init_hard_regno_mode_ok): Ditto.
22699 (rs6000_option_override_internal): Ditto.
22700 (legitimate_lo_sum_address_p): Ditto.
22701 (rs6000_legitimize_address): Ditto.
22702 (rs6000_legitimize_reload_address): Ditto.
22703 (rs6000_legitimate_address_p): Ditto.
22704 (abi_v4_pass_in_fpr): Ditto.
22705 (setup_incoming_varargs): Ditto.
22706 (rs6000_gimplify_va_arg): Ditto.
22707 (rs6000_split_multireg_move): Ditto.
22708 (rs6000_savres_strategy): Ditto.
22709 (rs6000_emit_prologue_components): Ditto.
22710 (rs6000_emit_epilogue_components): Ditto.
22711 (rs6000_emit_prologue): Ditto.
22712 (rs6000_emit_epilogue): Ditto.
22713 (rs6000_elf_file_end): Ditto.
22714 (rs6000_function_value): Ditto.
22715 (rs6000_libcall_value): Ditto.
22716 * config/rs6000/rs6000.h: Ditto.
22717 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
22718 (TARGET_MINMAX): ... this. New.
22719 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
22720 * config/rs6000/rs6000.md: Remove Xilinx FP support.
22721 (*movsi_internal1_single): Delete.
22722 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
22723 mfpu=, mxilinx-fpu): Delete.
22724 * config/rs6000/singlefp.h: Delete.
22725 * config/rs6000/sysv4.h: Remove Xilinx FP support.
22726 * config/rs6000/t-rs6000: Ditto.
22727 * config/rs6000/t-xilinx: Delete.
22728 * config/rs6000/titan.md: Adjust for fp_type removal.
22729 * config/rs6000/vsx.md: Remove Xilinx FP support.
22730 (VStype_simple): Delete.
22731 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
22732 * config/rs6000/xfpu.h: Delete.
22733 * config/rs6000/xfpu.md: Delete.
22734 * config/rs6000/xilinx.h: Delete.
22735 * config/rs6000/xilinx.opt: Delete.
22736 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
22737 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
22739 2018-05-04 Tom de Vries <tom@codesourcery.com>
22742 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
22745 2018-05-04 Richard Biener <rguenther@suse.de>
22747 PR middle-end/85627
22748 * tree-complex.c (update_complex_assignment): We are always in SSA form.
22749 (expand_complex_div_wide): Likewise.
22750 (expand_complex_operations_1): Likewise.
22751 (expand_complex_libcall): Preserve EH info of the original stmt.
22752 (tree_lower_complex): Handle removed blocks.
22753 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
22754 on complex multiplication and division libcall builtins.
22756 2018-05-04 Richard Biener <rguenther@suse.de>
22758 PR middle-end/85574
22759 * fold-const.c (negate_expr_p): Restrict negation of operand
22760 zero of a division to when we know that can happen without
22762 (fold_negate_expr_1): Likewise.
22764 2018-05-04 Jakub Jelinek <jakub@redhat.com>
22767 * real.h (real_nextafter): Declare.
22768 * real.c (real_nextafter): New function.
22769 * fold-const-call.c (fold_const_nextafter): New function.
22770 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
22771 CASE_CFN_NEXTTOWARD.
22772 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
22773 even when arg1_mode is different from arg0_mode.
22775 2018-05-03 Nathan Sidwell <nathan@acm.org>
22777 * doc/extend.texi (Deprecated Features): Remove
22778 -ffriend-injection.
22779 (Backwards Compatibility): Likewise.
22780 * doc/invoke.texi (C++ Language Options): Likewise.
22781 (C++ Dialect Options): Likewise.
22783 2018-05-03 Jakub Jelinek <jakub@redhat.com>
22786 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
22787 _mm512_mask_mullox_epi64): New intrinsics.
22789 2018-05-03 Tom de Vries <tom@codesourcery.com>
22792 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22793 dump files): Add offload-tree.
22795 2018-05-03 Richard Biener <rguenther@suse.de>
22797 PR tree-optimization/85615
22798 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
22799 to loops not nested in BBs loop father to avoid creating multi-entry
22802 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22804 PR tree-optimization/70291
22805 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
22806 arguments. Change return type to tree. Emit libcall as a new
22807 statement rather than replacing existing one when inplace_p is true.
22808 (expand_complex_multiplication_components): New function.
22809 (expand_complex_multiplication): Expand floating-point complex
22810 multiplication using the above.
22811 (expand_complex_division): Rename inner_type parameter to type.
22812 Update expand_complex_libcall call-site.
22813 (expand_complex_operations_1): Update expand_complex_multiplication
22814 and expand_complex_division call-sites.
22816 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22819 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22820 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22821 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
22822 the highest significant bit of the shift count mask is clear. In
22823 check whether and[sq]i3 is needed verify that all significant bits
22824 of the shift count other than the highest are set.
22826 2018-05-02 Tom de Vries <tom@codesourcery.com>
22829 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
22830 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
22831 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
22832 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
22833 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
22834 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
22835 __builtin_goacc_parlevel_size.
22837 2018-05-02 Richard Biener <rguenther@suse.de>
22839 PR tree-optimization/85597
22840 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
22841 do not use split vect_get_vec_defs call but call vect_get_slp_defs
22844 2018-05-02 Tom de Vries <tom@codesourcery.com>
22847 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22848 dump files): Add ltrans-tree.
22850 2018-05-02 Tom de Vries <tom@codesourcery.com>
22853 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22854 dump files): Add wpa-ipa.
22856 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
22858 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
22859 powerpc*-*-linux*paired* target.
22860 * config/rs6000/750cl.h: Delete.
22861 * config/rs6000/paired.h: Delete.
22862 * config/rs6000/paired.md: Delete.
22863 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
22865 * config/rs6000/rs6000-builtin.def: Remove paired float support.
22866 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
22867 comment. Remove paired float support.
22868 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
22869 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
22871 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
22872 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
22874 * config/rs6000/rs6000.c: Remove paired float support.
22875 (paired_expand_vector_init, paired_expand_vector_move,
22876 paired_emit_vector_compare, paired_emit_vector_cond_expr,
22877 (paired_expand_lv_builtin, paired_expand_stv_builtin,
22878 paired_expand_builtin, paired_expand_predicate_builtin,
22879 paired_init_builtins): Delete.
22880 * config/rs6000/rs6000.h: Remove paired float support.
22881 * config/rs6000/rs6000.md: Remove paired float support.
22882 (move_from_CR_ov_bit): Delete.
22883 * config/rs6000/rs6000.opt (mpaired): Delete.
22884 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
22885 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
22887 2018-05-02 Richard Biener <rguenther@suse.de>
22889 PR middle-end/85567
22890 * gimplify.c (gimplify_save_expr): When in SSA form allow
22891 SAVE_EXPRs to compute to SSA vars.
22893 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22896 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22897 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22898 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
22899 clobber operands[2], instead use a new pseudo. Formatting fixes.
22901 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
22903 PR tree-optimization/85586
22904 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
22905 exit early for statements in the same group if the accesses are
22908 2018-05-02 Tom de Vries <tom@codesourcery.com>
22911 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
22914 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
22916 PR tree-optimization/85143
22917 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
22919 2018-05-01 Tom de Vries <tom@codesourcery.com>
22922 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
22923 not found" error message.
22925 2018-05-01 Tom de Vries <tom@codesourcery.com>
22928 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
22929 * vec.c (test_ordered_remove_if): New function.
22930 (vec_c_tests): Call test_ordered_remove_if.
22931 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
22932 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
22933 * tree-vect-patterns.c (vect_pattern_recog_1): Use
22934 VEC_ORDERED_REMOVE_IF.
22936 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22938 PR tree-optimization/82665
22939 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
22940 pointer subtraction where arguments come from a memchr call.
22942 2018-05-01 Jakub Jelinek <jakub@redhat.com>
22944 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
22945 --push-state --as-needed and --pop-state instead of --as-needed and
22946 --no-as-needed if ld supports it.
22947 * configure: Regenerated.
22950 * doc/install.texi2html: Replace _002d with - and _002a with * in
22951 generated html files using sed.
22953 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22956 * gcc-rich-location.c (blank_line_before_p): New function.
22957 (use_new_line): New function.
22958 (gcc_rich_location::add_fixit_insert_formatted): New function.
22959 * gcc-rich-location.h
22960 (gcc_rich_location::add_fixit_insert_formatted): New function.
22962 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22964 * selftest.c (assert_streq): Rename "expected" and "actual" to
22965 "val1" and "val2". Extend NULL-handling to cover both inputs
22966 symmetrically, while still requiring both to be non-NULL for a pass.
22967 * selftest.h (assert_streq): Rename "expected" and "actual" to
22969 (ASSERT_EQ): Likewise.
22970 (ASSERT_EQ_AT): Likewise.
22971 (ASSERT_KNOWN_EQ): Likewise.
22972 (ASSERT_KNOWN_EQ_AT): Likewise.
22973 (ASSERT_NE): Likewise.
22974 (ASSERT_MAYBE_NE): Likewise.
22975 (ASSERT_MAYBE_NE_AT): Likewise.
22976 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
22977 the assertion to pass.
22978 (ASSERT_STREQ_AT): Likewise.
22980 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
22982 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
22983 interaction with -pie.
22985 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22987 * selftest.h: Fix alphabetization of per-source-file selftest
22990 2018-04-30 Jason Merrill <jason@redhat.com>
22992 PR c++/61982 - dead stores to destroyed objects.
22993 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
22996 2018-04-30 Jason Merrill <jason@redhat.com>
22998 * tree.c (build_clobber): New.
22999 * tree.h: Declare it.
23000 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
23002 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23004 * diagnostic-show-locus.c (layout::layout): Update for
23005 location_get_source_line returning a char_span.
23006 (struct char_span): Move to input.h.
23007 (struct correction): Update for fields in char_span becoming
23009 (struct source_line): Update for location_get_source_line
23010 returning a char_span.
23011 (layout::print_line): Likewise.
23012 * edit-context.c (edited_file::print_content): Likewise.
23013 (edited_file::print_diff_hunk): Likewise.
23014 (edited_file::print_run_of_changed_lines): Likewise.
23015 (edited_file::get_num_lines): Likewise.
23016 (edited_line::edited_line): Likewise.
23017 * final.c (asm_show_source): Likewise.
23018 * input.c (location_get_source_line): Convert return type
23019 from const char * to char_span, losing the final "line_len"
23021 (dump_location_info): Update for the above.
23022 (get_substring_ranges_for_loc): Likewise. Use a char_span
23023 when handling the literal within the line.
23024 (test_reading_source_line): Update for location_get_source_line
23025 returning a char_span.
23026 * input.h (class char_span): Move here from
23027 diagnostic-show-locus.c, converting from a struct to a class.
23028 Make data members private.
23029 (char_span::operator bool): New.
23030 (char_span::length): New.
23031 (char_span::get_buffer): New.
23032 (char_span::operator[]): New.
23033 (char_span::subspan): Make const.
23034 (char_span::xstrdup): New.
23035 (location_get_source_line): Convert return type from const char *
23036 to char_span, losing the final "line_size" param.
23038 2018-04-30 Jan Hubicka <jh@suse.cz>
23040 * lto-wrapper.c (ltrans_priorities): New static var.
23041 (cmp_priority): New.
23042 (run_gcc): Read priorities and if doing parallel build order
23043 the Makefile by them.
23045 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23047 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
23049 2018-04-30 Richard Biener <rguenther@suse.de>
23051 * tree-cfg.c (verify_address): Remove base argument, add
23052 flag whether to check TREE_ADDRESSABLE and do that.
23053 (verify_expr): Remove.
23054 (verify_types_in_gimple_reference): Add pieces from verify_expr.
23055 (verify_gimple_assign_single): Likewise.
23056 (verify_gimple_switch): Likewise.
23057 (verify_expr_location_1): Dereference tp once. Add (disabled)
23058 piece from verify_expr.
23059 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
23061 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23063 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
23065 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23067 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
23068 (small_data_pattern): Likewise.
23069 (arc_rewrite_small_data): Likewise.
23070 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
23071 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
23072 (get_symbol_alignment): New function.
23073 (legitimate_small_data_address_p): Likewise.
23074 (legitimate_scaled_address): Update, call
23075 legitimate_small_data_address_p.
23076 (output_sdata): New static variable.
23077 (arc_print_operand): Update how we handle small data operands.
23078 (arc_print_operand_address): Likewise.
23079 (arc_legitimate_address_p): Update, use
23080 legitimate_small_data_address_p.
23081 (arc_rewrite_small_data_p): Remove.
23082 (arc_rewrite_small_data_1): Likewise.
23083 (arc_rewrite_small_data): Likewise.
23084 (small_data_pattern): Likewise.
23085 (compact_sda_memory_operand): Update to use
23086 legitimate_small_data_address_p and get_symbol_alignment.
23087 (prepare_move_operands): Don't rewite sdata pattern.
23088 (prepare_extend_operands): Remove.
23089 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
23091 (zero_extendqisi2): Likewise.
23092 (zero_extendhisi2): Likewise.
23093 (extendqihi2): Likewise.
23094 (extendqisi2): Likewise.
23095 (extendhisi2): Likewise.
23096 (addsi3): Likewise.
23097 (subsi3): Likewise.
23098 (andsi3): Likewise.
23099 * config/arc/constraints.md (Usd): Change it to memory constraint.
23101 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23103 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
23104 as source of std instructions.
23105 * config/arc/arc.md (movsi_insn): Update pattern predicate to
23106 allow 6-bit constants as source for store instructions.
23107 (movdi_insn): Update instruction pattern to allow 6-bit constants
23108 as source for store instructions.
23110 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
23112 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
23114 2018-04-30 Nathan Sidwell <nathan@acm.org>
23115 Sandra Loosemore <sandra@codesourcery.com>
23117 * dumpfile.c (dump_open): Allow '-' for stdout.
23118 * doc/invoke.texi (Developer Options): Document dump filename
23119 determination early. Document stdin/stdout selection.
23121 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
23123 Microblaze Target: PIC data text relative
23125 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
23126 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
23128 * config/microblaze/microblaze.h (microblaze_constant_address_p):
23129 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
23130 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
23131 New addressing mode for data-text relative position indepenedent code.
23132 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
23133 'ADDRESS_SYMBOLIC_TXT_REL'.
23134 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
23135 (microblaze_legitimate_pic_operand): Exclude function calls from
23136 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
23137 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
23139 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
23140 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
23141 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
23142 for 'address + offset'.
23143 (microblaze_expand_prologue): Add new function prologue call for
23145 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
23146 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
23147 table in case of TARGET_PIC_DATA_TEXT_REL.
23148 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
23149 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
23150 Add new macros 'UNSPEC_TEXT',
23151 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
23152 + exclude function calls from 'UNSPEC_PLT' in case of data text
23154 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
23155 new target hook for generating address diff vector tables in case of
23157 * doc/tm.texi : Regenerate.
23158 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
23159 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
23160 of addr diff vector generation.
23161 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
23162 target hook definition.
23163 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
23164 Add default function for generate_pic_addr_diff_vec -> flag_pic.
23165 * doc/invoke.texi (Add new pic option): Add new microblaze pic
23166 option for data text relative.
23168 2018-04-30 Richard Biener <rguenther@suse.de>
23170 * tree-chrec.h (evolution_function_is_constant_p): Remove
23172 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
23174 2018-04-30 Richard Biener <rguenther@suse.de>
23177 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
23179 2018-04-30 Richard Biener <rguenther@suse.de>
23181 PR tree-optimization/28364
23182 PR tree-optimization/85275
23183 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
23184 copying first exit test.
23186 2018-04-28 Mark Wielaard <mark@klomp.org>
23188 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
23189 dwarf_version >= 5.
23190 (dwarf_AT): Handle DW_AT_addr_base.
23191 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
23193 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
23196 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
23197 (*ashl<dwi>3_doubleword_mask_1): Ditto.
23198 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
23199 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
23201 2018-04-28 Richard Biener <rguenther@suse.de>
23203 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
23204 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
23205 to reflect use. Only add interesting stmts.
23207 2018-04-27 Martin Jambor <mjambor@suse.cz>
23210 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
23211 the jump function allows for passing through aggregate values.
23213 2018-04-27 David Malcolm <dmalcolm@redhat.com>
23215 * input.h (in_system_header_at): Convert from macro to inline
23217 (from_macro_expansion_at): Likewise.
23218 (from_macro_definition_at): Likewise.
23220 2018-04-27 Jeff Law <law@redhat.com>
23222 * config.gcc: Mark tile* targets as deprecated/obsolete.
23224 2018-04-27 Richard Biener <rguenther@suse.de>
23226 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
23229 2018-04-27 Richard Biener <rguenther@suse.de>
23231 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
23233 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
23235 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
23236 with Yd constraint. Set "preferred_for_speed" attribute from
23237 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
23238 with Yd constraint.
23239 (*movdi_internal): Ditto.
23240 (movti_interunit splitters): Remove
23241 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
23242 (movdi_interunit splitters): Ditto.
23243 * config/i386/constraints.md (Ye): Remove.
23244 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
23246 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23249 * config/aarch64/constraints.md (Usg): Limit to 31.
23250 (Usj): Limit to 63.
23252 2018-04-27 Jakub Jelinek <jakub@redhat.com>
23254 PR tree-optimization/85529
23255 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
23256 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
23257 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
23258 zero extension or masking of the MSB bit.
23259 (optimize_range_tests): Add FIRST_BB argument, pass it through
23260 to optimize_range_tests_var_bound.
23261 (maybe_optimize_range_tests, reassociate_bb): Adjust
23262 optimize_range_tests callers.
23264 2018-04-26 Richard Biener <rguenther@suse.de>
23265 Jakub Jelinek <jakub@redhat.com>
23267 * cgraph.h (symbol_table): Just declare debug method here.
23268 * symtab.c (symbol_table::debug): Define.
23270 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
23272 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
23274 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
23276 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
23277 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
23278 (*movdi_internal): Substitute Yi and Yj constraint with x
23279 and Ym and Yn constraint with y constraint. Update "isa"
23280 attribute and set "preferred_for_speed" attribute from
23281 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
23282 (*movsi_internal): Ditto.
23283 (*movdf_internal): Ditto.
23284 (*movsf_internal): Ditto.
23285 (*zero_extendsidi2): Ditto.
23286 * config/i386/sse.md (vec_set<mode>_0): Ditto.
23287 (sse2_loadld): Ditto.
23288 (*vec_extract<ssevecmodelower>_0): Ditto.
23289 (*vec_extractv4si_0_zext_sse4): Ditto.
23290 (vec_concatv2di): Ditto.
23291 (*vec_dup<mode>): Ditto.
23292 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
23293 * config/i386/constraints.md (Yi): Remove.
23298 2018-04-26 Nathan Sidwell <nathan@acm.org>
23300 * dumpfile.c (dump_open): New.
23301 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
23302 (dump_finish): Detect stdio/stderr by value not name.
23304 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
23306 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
23308 2018-04-26 Tom de Vries <tom@codesourcery.com>
23311 * config/nvptx/nvptx.c (verify_neutering_jumps)
23312 (verify_neutering_labels): New function
23313 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
23315 2018-04-26 Tom de Vries <tom@codesourcery.com>
23318 * config/nvptx/nvptx.c (needs_neutering_p): New function.
23319 (nvptx_single): Use needs_neutering_p to skip over insns that do not
23322 2018-04-26 Richard Biener <rguenther@suse.de>
23323 Tom de Vries <tom@codesourcery.com>
23326 * lto-streamer-out.c (output_function): Fixup loops if required to match
23327 discovery done in the reader.
23329 2018-04-26 Richard Biener <rguenther@suse.de>
23331 PR tree-optimization/85116
23332 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
23333 have a loop exit from the single latch predecessor. Remove
23334 case of header with just condition.
23335 (ch_base::copy_headers): Exclude infinite loops from any
23337 (pass_ch::execute): Record exits.
23339 2018-04-26 Richard Biener <rguenther@suse.de>
23341 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
23342 prologue cost vector and pass it to vect_get_load_cost.
23343 (vect_get_peeling_costs_all_drs): Likewise.
23344 (vect_peeling_hash_get_lowest_cost): Likewise.
23345 (vect_enhance_data_refs_alignment): Likewise.
23347 2018-04-26 Richard Biener <rguenther@suse.de>
23349 PR middle-end/85450
23350 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
23351 checking of integer<->pointer conversions.
23352 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
23353 sign-/zero-extending pointer types.
23354 (expand_omp_for_static_chunk): Likewise.
23356 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
23357 Jean Lee <xiaoyur347@gmail.com>
23359 * config/mips/mips.c (mips_asan_shadow_offset): New function.
23360 (TARGET_ASAN_SHADOW_OFFSET): Define.
23361 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
23362 true for -fsanitize=address.
23364 2018-04-25 Mark Wielaard <mark@klomp.org>
23366 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
23369 2018-04-25 Jakub Jelinek <jakub@redhat.com>
23371 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
23372 than "alu", remove explicit "memory" and "imm_disp" attributes.
23373 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
23375 PR middle-end/85414
23376 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
23377 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
23378 gen_lowpart_no_emit.
23380 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
23383 * config/i386/i386.c (ix86_expand_builtin): Change memory
23384 operand to XI, extend p0 to Pmode.
23385 * config/i386/i386.md: Change unspec volatile and operand
23386 1 mode to XI, change operand 0 mode to P.
23388 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23390 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
23391 GET_MODE_MASK before any checking.
23392 (nds32_can_use_bset_p): Likewise.
23393 (nds32_can_use_btgl_p): Likewise.
23395 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23397 * config/nds32/nds32-doubleword.md: New define_split pattern for
23398 illegal register number.
23400 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23402 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
23404 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23406 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
23408 2018-04-25 Richard Biener <rguenther@suse.de>
23410 * lto-streamer.h (LTO_major_version): Bump to 8.
23412 2018-04-25 Jakub Jelinek <jakub@redhat.com>
23414 * BASE-VER: Set to 9.0.0.
23416 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
23418 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
23419 in __abskf2 and __powikf2.
23421 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23424 * config/aarch64/constraints.md (Usg, Usj): New constraints.
23425 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
23426 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
23427 Use the above on operand 2. Reindent.
23428 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
23430 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
23433 * common/config/i386/i386-common.c (ix86_handle_option): Don't
23435 * config/i386/i386.opt (mcet): Removed.
23436 * doc/install.texi: Remove -mcet documentation.
23437 * doc/invoke.texi: Likewise.
23439 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
23442 * doc/install.texi: Remove -mcet from bootstrap-cet.
23444 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23447 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
23448 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
23452 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
23453 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
23454 containing a CONST_VECTOR.
23456 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
23458 * doc/install.texi: Update newlib dependency for nvptx.
23460 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23463 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
23464 instead of INTVAL when shifting x left.
23466 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
23468 PR tree-optimization/85478
23469 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
23470 vect_grouped_store_supported for single element vectors.
23472 2018-04-24 Richard Biener <rguenther@suse.de>
23475 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
23476 load cost increase to the case of non-constant step.
23478 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23481 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
23482 destination if any_malformed_asm.
23484 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
23486 PR middle-end/85496
23487 * expr.c (store_field): In the bitfield case, if the value comes from
23488 a function call and is returned in registers by means of a PARALLEL,
23489 do not change the mode of the temporary unless BLKmode and VOIDmode.
23491 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
23493 PR rtl-optimization/85423
23494 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
23495 dependencies to debug insns when the previous insn is non-debug.
23497 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
23499 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
23500 enums into a single definition.
23501 (fls): Fix predicates and printing.
23504 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
23506 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
23507 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
23508 and short u6 immediate.
23509 (check_if_valid_sleep_operand): Remove.
23510 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
23512 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23514 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
23515 flag_always_save_lp condition.
23516 * config/nds32/nds32.opt (malways-save-lp): New option.
23518 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23520 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
23521 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
23522 * config/nds32/nds32.h
23523 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
23524 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
23526 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23528 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
23529 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
23531 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23532 Chung-Ju Wu <jasonwucj@gmail.com>
23534 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
23536 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
23537 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
23539 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23541 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
23543 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23545 * config/nds32/nds32-protos.h (nds32_data_alignment,
23546 nds32_local_alignment): Declare.
23547 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
23548 nds32_local_alignment): New functions.
23549 (TARGET_CONSTANT_ALIGNMENT): Define.
23550 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
23552 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23554 * config/nds32/nds32.c
23555 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
23556 (TARGET_MODES_TIEABLE_P): Likewise.
23558 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23560 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
23561 level Ofast and Og.
23563 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
23564 Chung-Ju Wu <jasonwucj@gmail.com>
23566 * config/nds32/constants.md (unspec_volatile_element): Add enum values
23567 for unaligned access.
23568 * config/nds32/nds32-intrinsic.c: Implementation of expanding
23570 * config/nds32/nds32-intrinsic.md: Likewise.
23571 * config/nds32/nds32_intrinsic.h: Likewise.
23572 * config/nds32/nds32.h (nds32_builtins): Likewise.
23573 * config/nds32/nds32.opt (munaligned-access): New option.
23574 * config/nds32/nds32.c (nds32_asm_file_start): Display
23575 flag_unaligned_access status.
23577 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
23579 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
23580 -mno-relax is present.
23581 * config/riscv/linux.h (LINK_SPEC): Ditto.
23583 2018-04-20 Martin Sebor <msebor@redhat.com>
23586 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
23588 (gimple_fold_builtin_stxcpy_chk): Same.
23589 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
23591 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
23594 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
23595 __powikf2 when long double is IEEE 128-bit.
23597 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
23599 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
23600 step to make sure stack always aligned.
23602 2018-04-20 Carl Love <cel@us.ibm.com>
23605 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
23606 size check for arg0.
23608 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
23609 Tom de Vries <tom@codesourcery.com>
23612 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
23613 Emit insns for calls too.
23614 (nvptx_find_par): Always look for worker-level predecessor insn.
23615 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
23617 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
23618 (nvptx_process_pars): Propagate frames for calls.
23620 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
23623 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
23625 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
23626 (ix86_handle_option): Don't handle OPT_mibt.
23627 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
23629 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
23631 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
23632 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
23633 (ix86_target_macros): Define __CET__ with flag_cf_protection
23634 for -fcf-protection.
23635 * config/i386/i386.c (isa2_opts): Remove -mibt.
23636 * config/i386/i386.h (TARGET_IBT): Removed.
23637 (TARGET_IBT_P): Likewise.
23638 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
23639 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
23640 * config/i386/i386.opt (mcet): Update help message.
23641 (mshstk): Likewise.
23643 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
23644 -mcet as an alias for -mshstk.
23646 2018-04-20 Richard Biener <rguenther@suse.de>
23648 PR middle-end/85475
23649 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
23650 complexity by forcing a single use of the multiply operand.
23652 2018-04-20 Martin Jambor <mjambor@suse.cz>
23655 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
23656 recursion dependency to only apply to non-clones.
23658 2018-04-20 Martin Jambor <mjambor@suse.cz>
23661 * ipa-cp.c (create_specialized_node): Check that clones of
23662 self-recursive edges exist during IPA-CP.
23664 2018-04-19 Toon Moene <toon@moene.org>
23666 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
23669 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23671 PR tree-optimization/85467
23672 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
23673 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
23674 VECTOR_CST element to type.
23676 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23679 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
23680 * config/i386/i386.md (builtin_setjmp_setup): Removed.
23681 (builtin_longjmp): Likewise.
23682 (save_stack_nonlocal): New pattern.
23683 (restore_stack_nonlocal): Likewise.
23685 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23688 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23689 Replace ASM_OUTPUT_LABEL with fprintf.
23691 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23694 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23695 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
23696 * config/i386/i386-c.c (ix86_target_macros_internal): Also
23697 define __IBT__ and __SHSTK__ for -fcf-protection.
23698 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
23700 (ix86_trampoline_init): Likewise.
23701 (x86_output_mi_thunk): Likewise.
23702 (ix86_notrack_prefixed_insn_p): Likewise.
23703 (ix86_option_override_internal): Don't disallow -fcf-protection.
23704 * config/i386/i386.md (rdssp<mode>): Also enable for
23706 (incssp<mode>): Likewise.
23707 (nop_endbr): Likewise.
23708 * config/i386/i386.opt (mcet): Change help message to built-in
23711 (mshstk): Likewise.
23712 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
23713 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
23714 enable CET built-in functions.
23716 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
23718 * common/config/i386/i386-common.c
23719 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
23720 OPTION_MASK_ISA_MOVDIRI_UNSET,
23721 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
23722 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
23723 * config.gcc (movdirintrin.h): New header.
23724 * config/i386/cpuid.h (bit_MOVDIRI,
23725 bit_MOVDIR64B): New bits.
23726 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
23728 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
23729 (VOID, PVOID, PCVOID)): New function types.
23730 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
23731 __builtin_ia32_directstoreu_u64,
23732 __builtin_ia32_movdir64b): New builtins.
23733 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
23734 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
23736 (ix86_valid_target_attribute_inner_p): Ditto.
23737 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
23738 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
23739 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
23740 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
23741 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
23742 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
23743 (movdiri<mode>, movdir64b_<mode>): New.
23744 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
23745 * config/i386/immintrin.h: Include movdirintrin.h.
23746 * config/i386/movdirintrin.h: New file.
23747 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
23749 2018-04-19 Richard Biener <rguenther@suse.de>
23751 PR middle-end/85455
23752 * cfg.c (clear_bb_flags): When loop state says we have
23753 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
23755 2018-04-19 Richard Biener <rguenther@suse.de>
23757 PR tree-optimization/84737
23758 * tree-vect-data-refs.c (vect_copy_ref_info): New function
23759 copying restrict info.
23760 (vect_setup_realignment): Use it.
23761 * tree-vectorizer.h (vect_copy_ref_info): Declare.
23762 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
23763 the first DR to all generated stores.
23764 (vectorizable_load): Likewise for loads.
23766 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23768 PR tree-optimization/85446
23769 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
23770 the integral and pointer types to have the same precision.
23772 * doc/install.texi: Document --disable-cet being the default and
23775 2018-04-18 Martin Liska <mliska@suse.cz>
23777 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
23780 2018-04-18 Martin Liska <mliska@suse.cz>
23783 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
23786 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
23787 arguments if they are comparable.
23789 2018-04-18 Martin Liska <mliska@suse.cz>
23792 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23795 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23798 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
23801 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
23802 ENDBR after calling __morestack.
23804 2018-04-18 David Malcolm <dmalcolm@redhat.com>
23807 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
23808 by using gcc_base_ver to generate a gcc_driver_version, and use
23809 it when generating GCC_DRIVER_NAME.
23810 * configure: Regenerate.
23812 2018-04-18 Jakub Jelinek <jakub@redhat.com>
23815 * config.gcc: Obsolete powerpc*-*-*spe*.
23817 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23820 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
23821 (stabstr_D): Change type of unum from unsigned int to
23822 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
23825 2018-04-17 Jim Wilson <jimw@sifive.com>
23828 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
23829 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
23830 Set arg_pointer_offset after using pretend_args_size.
23832 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23834 PR rtl-optimization/85431
23835 * dse.c (record_store): Ignore zero width stores.
23838 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
23839 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
23840 __builtin_stack_restore rather than after it.
23841 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
23842 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
23843 argument instead of virtual_dynamic_stack_rtx.
23845 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
23847 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
23849 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23850 Add note to error message to explain internal mapping of overloaded
23851 built-in function name to non-overloaded built-in function name.
23852 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
23855 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
23858 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
23859 where the inputs overlap with the output.
23861 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23863 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
23864 (=v, v) alternative and explicit "memory" attribute.
23865 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
23866 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23868 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
23869 "sselog1" type instead of "sselog".
23870 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
23871 "sselog". Remove explicit "memory" attribute.
23872 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
23873 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23875 (vec_extract_hi_v32hi): Merge all alternatives into one, use
23876 "sselog1" type instead of "sselog". Remove explicit "memory"
23878 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
23879 use "sselog1" type instead of "sselog". Remove explicit "memory"
23881 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
23882 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23884 (vec_extract_hi_v64qi): Merge all alternatives into one, use
23885 "sselog1" type instead of "sselog". Remove explicit "memory"
23887 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
23888 use "sselog1" type instead of "sselog". Remove explicit "memory"
23892 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
23894 PR middle-end/85414
23895 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
23898 2018-04-17 Martin Jambor <mjambor@suse.cz>
23901 * ipa-cp.c (create_specialized_node): Call
23902 expand_all_artificial_thunks if necessary.
23904 2018-04-17 Martin Liska <mliska@suse.cz>
23907 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
23908 in message, remote space in between '_G' and '('.
23910 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23913 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
23914 avx512f_vmcmp<mode>3<round_saeonly_name>,
23915 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
23916 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
23917 avx512f_rndscale<mode><round_saeonly_name>,
23918 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
23919 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
23920 Use %<iptr>2 instead of %2 for -masm=intel.
23921 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
23922 avx512f_vcvttss2usi<round_saeonly_name>,
23923 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
23925 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
23926 avx512f_vcvttsd2usi<round_saeonly_name>,
23927 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
23928 Use %q1 instead of %1 for -masm=intel.
23929 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
23930 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
23931 of %3 for -masm=intel.
23932 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
23934 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
23936 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
23938 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
23939 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
23941 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
23943 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
23944 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
23945 %g1 and one with %0 and %1.
23946 (avx512er_vmrcp28<mode><round_saeonly_name>,
23947 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
23948 %1 for -masm=intel.
23949 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
23950 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
23951 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
23952 of %0 and %{%4%} for -masm=intel.
23953 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
23954 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
23955 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
23956 order of %0 and %{%5%}%{z%} for -masm=intel.
23958 2018-04-17 Jan Hubicka <jh@suse.cz>
23961 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
23963 2018-04-17 Martin Liska <mliska@suse.cz>
23966 * multiple_target.c (create_dispatcher_calls): Set apostrophes
23967 for target_clone error message. Make default implementation
23968 clone to be a local declaration.
23969 (separate_attrs): Add new argument and check for an empty
23971 (expand_target_clones): Handle it.
23972 (ipa_target_clone): Make redirection just for target_clones
23975 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
23976 Tom de Vries <tom@codesourcery.com>
23978 PR middle-end/84955
23979 * omp-expand.c (expand_oacc_for): Add dummy false branch for
23980 tiled basic blocks without omp continue statements.
23982 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
23985 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
23986 vec_extract expression as having side effects to make sure it gets
23989 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
23992 * config/i386/i386.c (get_builtin_code_for_version): Check
23995 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
23998 * config.gcc: Support "skylake".
23999 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
24001 * config/i386/i386.c (m_SKYLAKE): Define.
24002 (processor_target_table): Add "skylake".
24003 (ix86_option_override_internal): Add "skylake".
24004 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
24005 PROCESSOR_CANNONLAKE.
24006 (get_builtin_code_for_version): Fix priority for
24007 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
24008 PROCESSOR_SKYLAKE-AVX512.
24009 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
24010 (processor_type): Add PROCESSOR_SKYLAKE.
24012 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
24013 Jason Merrill <jason@redhat.com>
24016 * convert.c (convert_to_integer_1): Use direct recursion for
24017 enumeral types and types with a precision less than the number
24018 of bits in their mode.
24020 2018-04-16 Julia Koval <julia.koval@intel.com>
24023 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
24024 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
24026 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
24029 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
24030 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
24031 and -mno-direct-move.
24033 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
24036 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
24037 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
24038 Ensure negative shifts result in {0}.
24040 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
24042 PR rtl-optimization/79916
24043 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
24044 regs (if any) to define how to gnerate SD moves when LRA is in
24047 2018-04-13 Jakub Jelinek <jakub@redhat.com>
24049 PR rtl-optimization/85393
24050 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
24051 * except.c (expand_dw2_landing_pad_for_region): Make static.
24052 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
24053 a label and unconditional jump to old_bb, rather than
24054 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
24057 PR rtl-optimization/85376
24058 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
24059 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
24060 instead of a specific value.
24062 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
24063 Bin Cheng <bin.cheng@arm.com>
24065 PR tree-optimization/82965
24066 PR tree-optimization/83991
24067 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
24068 by_profile_only parameter.
24069 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
24070 information if the loop was predicted to iterate too many times.
24071 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
24073 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
24076 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
24079 2018-04-13 Martin Liska <mliska@suse.cz>
24080 Jakub Jelinek <jakub@redhat.com>
24082 PR middle-end/81657
24083 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
24084 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
24085 * builtins.c (expand_builtin_memory_copy_args): Use
24086 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
24087 handle dest_addr == pc_rtx.
24089 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24092 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
24093 asked to not generate direct moves.
24094 (fix_trunc<mode>si2_stfiwx): Similar.
24095 (fix_trunc<mode>si2_internal): Similar.
24097 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24100 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
24101 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
24102 lookup if dest in some wider mode is known to be const0_rtx and
24103 if so, record permanent equivalence for it to be ZERO_EXTEND of
24104 the narrower mode destination.
24106 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
24108 * lto-streamer-out.c (output_function): Revert 259346.
24109 * omp-expand.c (expand_oacc_for): Likewise.
24111 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
24113 PR rtl-optimization/85354
24114 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
24115 * sel-sched.c (sel_global_init): ... here.
24117 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
24120 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
24121 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
24122 mode for PE-COFF targets.
24123 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
24124 (i386_pe_asm_lto_end): Likewise.
24125 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
24126 (TARGET_ASM_LTO_END): Likewise.
24127 * config/i386/winnt.c (saved_debug_info_level): New static variable.
24128 (i386_pe_asm_lto_start): New function.
24129 (i386_pe_asm_lto_end): Likewise.
24131 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
24132 Richard Biener <rguenther@suse.de>
24134 PR middle-end/84955
24135 * lto-streamer-out.c (output_function): Fix CFG loop state before
24137 * omp-expand.c (expand_oacc_for): Handle calls to internal
24138 functions like regular functions.
24140 2018-04-12 Richard Biener <rguenther@suse.de>
24143 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
24144 for the early LTO debug to properly generate references to it
24145 during DIE emission. Do not re-use that for the skeleton for
24147 (dwarf2out_early_finish): Likewise.
24149 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24152 * config/i386/sse.md
24153 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
24154 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
24155 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
24156 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
24157 and output is a reg, avoid creating invalid lowpart subreg, but
24158 instead split into a 512-bit move. Don't split if not AVX512VL,
24159 input is xmm16+ reg and output is a mem.
24160 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
24161 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
24162 xmm16+ reg and output is a mem.
24164 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24166 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
24167 also for flag_dwarf2_cfi_asm.
24169 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24171 PR rtl-optimization/85342
24172 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
24173 a bool scalar var inside of the loop instead. Don't try to update
24174 recog_data.operand after failed apply_change_group.
24176 2018-04-12 Tom de Vries <tom@codesourcery.com>
24179 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
24180 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
24181 array with flexible array member as array without given dimension.
24182 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
24183 argument for undefined param to true.
24185 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
24188 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
24189 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
24190 from PowerPC section.
24191 * config/rs6000/sysv4.opt (mcall-): Improve help text.
24192 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
24193 help text that is too long.
24194 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
24195 help text that is too long.
24196 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
24197 help text that is too long.
24199 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
24201 * config/alpha/alpha.md (stack_probe_internal): Rename
24202 from "probe_stack". Update all callers.
24204 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24206 PR rtl-optimization/84566
24207 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
24208 sched_macro_fuse_insns.
24210 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24213 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
24214 (compute_block_dependences): ... from here.
24216 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24218 PR tree-optimization/85331
24219 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
24220 from int to HOST_WIDE_INT.
24222 2018-04-11 Martin Jambor <mjambor@suse.cz>
24225 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
24226 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
24227 not the same as the source val.
24228 (cgraph_edge_brings_value_p): New parameter.
24229 (gather_edges_for_value): Pass destination value to
24230 cgraph_edge_brings_value_p.
24231 (perhaps_add_new_callers): Likewise.
24232 (get_info_about_necessary_edges): Likewise and exclude values brought
24233 only by self-recursive edges.
24234 (create_specialized_node): Redirect only clones of self-calling edges.
24235 (+self_recursive_pass_through_p): New function.
24236 (find_more_scalar_values_for_callers_subset): Use it.
24237 (find_aggregate_values_for_callers_subset): Likewise.
24238 (known_aggs_to_agg_replacement_list): Removed.
24239 (decide_whether_version_node): Re-calculate known constants for all
24240 remaining context clones.
24242 2018-04-11 Richard Biener <rguenther@suse.de>
24245 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
24246 from early DWARF output.
24247 (dwarf2out_early_finish): Output line info unconditionally into
24248 early DWARF and add reference to it.
24250 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24253 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
24254 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
24255 other than V2DFmode using iptr mode attribute.
24256 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
24258 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24260 PR rtl-optimization/84659
24261 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
24263 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24266 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
24268 (output_loc_list): Pass address of a dummy size variable even in the
24269 locview handling loop.
24270 (index_location_lists): Add comment on why skip_loc_list_entry can't
24273 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
24276 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
24279 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
24282 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
24283 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
24284 and -mstring-compare-inline-limit.
24286 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24289 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
24290 for stack clash protection in a register whenever we need it to be in
24293 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24295 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
24296 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
24298 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24301 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
24303 (mlong-double-): Ditto.
24304 * config/rs6000/sysv4.opt (msdata=): Ditto.
24305 (mtls-size=): Ditto.
24307 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24309 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
24310 erroneous entries for
24311 "vector int vec_ldl (int, long int *)", and
24312 "vector unsigned int vec_ldl (int, unsigned long int *)".
24313 Add comments and entries for
24314 "vector bool char vec_ldl (int, bool char *)",
24315 "vector bool short vec_ldl (int, bool short *)",
24316 "vector bool int vec_ldl (int, bool int *)",
24317 "vector bool long long vec_ldl (int, bool long long *)",
24318 "vector pixel vec_ldl (int, pixel *)",
24319 "vector long long vec_ldl (int, long long *)",
24320 "vector unsigned long long vec_ldl (int, unsigned long long *)".
24321 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
24322 type tree bool_long_long_type_node and correct definition of
24323 bool_V2DI_type_node to make reference to this new type tree.
24324 (rs6000_mangle_type): Replace erroneous reference to
24325 bool_long_type_node with bool_long_long_type_node.
24326 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
24327 comments to emphasize sign distinctions for char and int types and
24328 replace RS6000_BTI_bool_long constant with
24329 RS6000_BTI_bool_long_long constant. Also add comment to restrict
24330 use of RS6000_BTI_pixel.
24331 (bool_long_type_node): Remove this macro definition.
24332 (bool_long_long_type_node): New macro definition
24334 2018-04-10 Jakub Jelinek <jakub@redhat.com>
24336 PR rtl-optimization/85300
24337 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
24338 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
24339 simplify_unary_operation fails.
24341 2018-04-10 Martin Liska <mliska@suse.cz>
24343 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
24344 cgraph_edge and ipa_ref.
24346 2018-04-10 Jakub Jelinek <jakub@redhat.com>
24350 * config/i386/sse.md
24351 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
24352 computation of the VEC_MERGE selector from mask.
24353 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
24354 Fix decoding of the VEC_MERGE selector into mask.
24356 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
24358 PR tree-optimization/85286
24359 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
24361 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
24363 * final.c (final_1): Set insn_last_address as well as
24364 insn_current_address.
24366 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24369 * explow.c (emit_stack_probe): Call validize_mem on memory location
24370 before passing it to gen_probe_stack. Create address operand and
24371 legitimize it for the probe_stack_address case.
24373 2018-04-09 Jan Hubicka <jh@suse.cz>
24376 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
24377 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
24378 * tree.c (free_lang_data_in_type): Fix handling of binfos;
24380 (free_lang_data): Rebuild type inheritance graph.
24382 2018-04-09 Martin Sebor <msebor@redhat.com>
24384 * invoke.texi (-finline-small-functions): Mention other optimization
24386 (-findirect-inlining, -fpartial-inlining): Same.
24387 (-finline-functions-called-once): Same.
24388 (-freorder-blocks-and-partition): Same.
24390 2018-04-09 Jan Hubicka <jh@suse.cz>
24393 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
24394 jumps; choose last target that matches the criteria (i.e.
24395 no partition changes for non-crossing jumps).
24396 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
24397 support for redirecting crossing jumps to non-crossing.
24399 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
24401 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
24402 also for naked functions.
24404 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
24406 * config/arc/arc.md (add_shift): New pattern.
24407 (add_shift2): Likewise.
24408 (sub_shift): Likewise.
24409 (sub_shift_cmp0_noout): Likewise.
24410 (compare_si_ashiftsi): Likewise.
24411 (xbfu_cmp0_noout): New combine pattern.
24412 (xbfu_cmp0"): Likewise.
24413 (movsi_set_cc_insn): Place the predicable variant first.
24414 (commutative_binary_cmp0_noout): Remove clobber.
24415 (commutative_binary_cmp0): New pattern.
24416 (noncommutative_binary_cmp0): Likewise.
24417 (noncommutative_binary_cmp0_noout): Likewise.
24418 (noncommutative_binary_comparison_result_used): Removed.
24419 (rsub_cmp0): New pattern.
24420 (rsub_cmp0_noout): Likewise.
24421 (extzvsi): Changed, keep only meaningful variants.
24422 (SQH, SEZ): New iterators.
24423 (SQH_postfix): New mode attribute.
24424 (SEZ_prefix): New code attribute.
24425 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
24426 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
24427 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
24428 of numerical value.
24429 (noncommutative_operator): Check the availability of barrel
24432 2018-04-09 Richard Biener <rguenther@suse.de>
24434 PR tree-optimization/85284
24435 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
24436 Only use the niter constraining form of simple_iv when the exit
24437 is always executed.
24439 2018-04-09 Tom de Vries <tom@codesourcery.com>
24442 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
24443 (define_expand "*memory_barrier"): New define_expand.
24444 (define_insn "memory_barrier"): New insn.
24446 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24448 PR rtl-optimization/80463
24449 PR rtl-optimization/83972
24450 PR rtl-optimization/83480
24452 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
24453 correct producer for the insn.
24454 (tidy_control_flow): Fixup seqnos in case of debug insns.
24456 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24458 PR rtl-optimization/83913
24460 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
24461 different sched-times when merging exprs.
24463 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24465 PR rtl-optimization/83962
24467 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
24468 tidy_fallthru_edge and tidy_control_flow.
24470 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24472 PR rtl-optimization/83530
24474 * sel-sched.c (force_next_insn): New global variable.
24475 (remove_insn_for_debug): When force_next_insn is true, also leave only
24476 next insn in the ready list.
24477 (sel_sched_region): When the region wasn't scheduled, make another pass
24478 over it with force_next_insn set to 1.
24480 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
24482 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
24484 * config/nds32/constants.md (unspec_volatile_element): Add enum values
24485 for interrupt control.
24486 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
24487 functions for interrupt control.
24488 * config/nds32/nds32-intrinsic.md: Likewise.
24489 * config/nds32/nds32_intrinsic.h: Likewise.
24490 * config/nds32/nds32.h (nds32_builtins): Likewise.
24492 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
24494 * config/nds32/nds32.c (nds32_init_machine_status,
24495 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
24496 strict_aligned_p field.
24497 (nds32_expand_to_rtl_hook): New function.
24498 (TARGET_EXPAND_TO_RTL_HOOK): Define.
24499 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
24501 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24502 Chung-Ju Wu <jasonwucj@gmail.com>
24504 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
24505 * config/nds32/nds32-n7.md: New file.
24506 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
24507 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
24509 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
24510 * config/nds32/nds32.md (pipeline_model): Add n7.
24511 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
24512 * config/nds32/pipelines.md: Include n7 settings.
24514 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24515 Chung-Ju Wu <jasonwucj@gmail.com>
24517 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
24518 * config/nds32/nds32-e8.md: New file.
24519 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
24520 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
24522 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
24523 * config/nds32/nds32.md (pipeline_model): Add e8.
24524 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
24525 * config/nds32/pipelines.md: Include e8 settings.
24527 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24528 Chung-Ju Wu <jasonwucj@gmail.com>
24530 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
24531 * config/nds32/nds32-n8.md: New file.
24532 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
24533 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
24535 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
24536 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
24537 * config/nds32/nds32.md (pipeline_model): Add n8.
24538 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
24539 * config/nds32/pipelines.md: Include n8 settings.
24541 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24542 Chung-Ju Wu <jasonwucj@gmail.com>
24544 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
24545 * config/nds32/nds32-n9-2r1w.md: New file.
24546 * config/nds32/nds32-n9-3r2w.md: New file.
24547 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
24548 nds32_register_ports): New or modify for cpu n9.
24549 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
24551 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
24552 * config/nds32/nds32-utils.c: New file.
24553 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
24554 TARGET_MUL_SLOW): Define.
24555 * config/nds32/nds32.md (pipeline_model): New attribute.
24556 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
24557 New options that support cpu n9.
24558 * config/nds32/pipelines.md: Include n9 settings.
24559 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
24561 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
24563 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
24564 information if necessary.
24565 (output_cond_branch_compare_zero): Likewise.
24566 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
24567 (nds32_target_alignment): Refine for alignment.
24568 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
24569 (FUNCTION_BOUNDARY): Modify.
24570 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
24572 * config/nds32/nds32.opt (malways-align, malign-functions): New.
24574 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
24576 * config/nds32/constants.md (unspec_volatile_element): Add values for
24577 TLB operation and data prefetch.
24578 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
24579 functions for TLB operation and data prefetch.
24580 * config/nds32/nds32-intrinsic.md: Likewise.
24581 * config/nds32/nds32_intrinsic.h: Likewise.
24582 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
24583 (nds32_print_operand): Likewise.
24584 * config/nds32/nds32.h (nds32_builtins): Likewise.
24586 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
24587 Andrew Pinski <pinsika@gcc.gnu.org>
24589 PR middle-end/82976
24590 * match.pd: Use constant_boolean_node of correct type instead of
24591 boolean_true_node or boolean_false_node for simplifying
24592 pointer comparisons to zero.
24594 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24596 PR tree-optimization/80021
24597 * tree.c (verify_type_variant): Make error call in verify_variant_match
24598 translatable and remove final full stop.
24600 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24602 * config/nds32/constants.md (unspec_volatile_element): Add
24603 UNSPEC_VOLATILE_EH_RETURN.
24604 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
24605 nds32_output_stack_pop): Support dwarf exception handling process.
24606 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
24607 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
24608 exception handling process.
24609 (nds32_compute_stack_frame): Likewise.
24610 (nds32_return_addr_rtx): Likewise.
24611 (nds32_initial_elimination_offset): Likewise.
24612 (nds32_expand_prologue): Likewise.
24613 (nds32_expand_epilogue): Likewise.
24614 (nds32_dynamic_chain_address): New function.
24615 * config/nds32/nds32.h (machine_function): Add fields for dwarf
24616 exception handling.
24617 (DYNAMIC_CHAIN_ADDRESS): Define.
24618 (EH_RETURN_DATA_REGNO): Define.
24619 (EH_RETURN_STACKADJ_RTX): Define.
24620 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
24621 patterns for dwarf exception handling.
24623 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24625 * config/nds32/nds32.h: Clean up obsolete macros.
24627 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24629 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24630 Add enum values for particular instructions.
24631 * config/nds32/nds32-intrinsic.c: Implementation of expanding
24632 particular intrinsic functions.
24633 * config/nds32/nds32-intrinsic.md: Likewise.
24634 * config/nds32/nds32_intrinsic.h: Likewise.
24635 * config/nds32/nds32.h (nds32_builtins): Likewise.
24636 * config/nds32/nds32.md (type): Add pbsad and pbsada.
24637 (btst, ave): New patterns for particular instructions.
24639 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24641 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24642 Add enum values for atomic load/store and memory sync.
24643 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
24645 * config/nds32/nds32-intrinsic.md: Likewise.
24646 * config/nds32/nds32_intrinsic.h: Likewise.
24647 * config/nds32/nds32.h (nds32_builtins): Likewise.
24649 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24651 PR tree-optimization/85257
24652 * fold-const.c (native_encode_vector): If not all elts could fit
24653 and off is -1, return 0 rather than offset.
24654 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
24655 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
24656 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
24657 adjust buffer in native_interpret_expr call.
24659 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24661 * config/nds32/constants.md (unspec_volatile_element): Add cache
24662 control enum values.
24663 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
24664 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
24665 * config/nds32/nds32.c (nds32_cctl_names): New.
24666 (nds32_print_operand): Handle cache control register names.
24667 * config/nds32/nds32.h (nds32_builtins): New enum values.
24668 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
24670 * config/nds32/nds32.md (type): Add mmu.
24671 * config/nds32/pipelines.md (simple_insn): Add mmu.
24673 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24675 * config/nds32/nds32.md (type): Remove call.
24676 * config/nds32/pipelines.md (simple_insn): Likewise.
24678 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24680 * config/nds32/constants.md (unspec_volatile_element): Add
24681 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
24682 UNSPEC_VOLATILE_FMFCFG.
24683 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
24684 description for fmfcfg and fmfcsr.
24685 (bdesc_1arg): Add fmtcsr.
24686 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
24687 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
24688 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
24689 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
24690 unspec_fmfcfg): New patterns.
24691 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
24692 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
24693 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
24694 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
24695 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
24696 __nds32__fmfcfg): Define.
24698 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24700 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
24701 intrinsic register names.
24702 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
24703 intrinsic register enum values and macros.
24705 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24707 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
24708 for load/store addressing form.
24709 (nds32_print_operand_address): Likewise.
24711 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
24714 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
24715 based on LABEL_REF. Remove useless assertion.
24716 (pic_address_needs_scratch): Fix formatting.
24717 (sparc_legitimize_pic_address): Minor tweaks.
24718 (sparc_delegitimize_address): Adjust assertion accordingly.
24719 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
24720 into symbolic_operand.
24721 (movsi_high_pic_label_ref): Likewise.
24722 (movsi_lo_sum_pic_label_ref): Likewise.
24723 (movdi_pic_label_ref): Likewise.
24724 (movdi_high_pic_label_ref): Likewise.
24725 (movdi_lo_sum_pic_label_ref): Likewise.
24727 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
24729 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
24730 custom LIB_SPEC setup.
24732 2018-04-06 Ruslan Bukin <br@bsdpad.com>
24733 Kito Cheng <kito.cheng@gmail.com>
24735 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
24736 * config/riscv/freebsd.h: New.
24738 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24740 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
24741 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
24744 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24745 Kito Cheng <kito.cheng@gmail.com>
24747 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
24748 nds32_output_call, nds32_symbol_binds_local_p): New functions.
24749 * config/nds32/nds32-protos.h (nds32_output_call,
24750 nds32_output_return): Declare.
24751 * config/nds32/nds32.md: Refine all the call and return patterns.
24753 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24756 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
24757 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
24759 PR rtl-optimization/84872
24760 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
24761 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
24762 EDGE_CROSSING edge.
24764 2018-04-06 Tamar Christina <tamar.christina@arm.com>
24766 * expr.c (copy_blkmode_to_reg): Revert 254862.
24767 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
24769 2018-04-06 Richard Biener <rguenther@suse.de>
24771 PR middle-end/85244
24772 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
24773 after seeing a component reference with an adjacent field. Treat
24774 refs to arrays at struct end of external decls similar to
24775 refs to unconstrained commons.
24777 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24780 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
24781 look through SAVE_EXPRs with non-side-effects argument. Adjust
24783 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
24786 2018-04-06 Richard Biener <rguenther@suse.de>
24788 PR middle-end/85180
24789 * alias.c (find_base_term): New wrapper around find_base_term
24790 unwinding CSELIB_VAL_PTR changes.
24791 (find_base_term): Do not restore CSELIB_VAL_PTR during the
24794 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24796 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
24798 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
24799 constant definitions.
24800 ("nop"): lr 0,0 -> nopr r0
24801 ("nop_lr0", "nop_lr1"): New insn definitions.
24803 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24805 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
24806 NDS32_V3PUSH_AVAILABLE_P macro.
24808 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
24809 Chung-Ju Wu <jasonwucj@gmail.com>
24811 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
24812 (nds32*-*-*): Add float and fpu_config into supported_defaults.
24813 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
24814 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
24815 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
24816 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
24817 * config/nds32/constraints.md: New constraints and checking for hard
24818 float configuration.
24819 * config/nds32/iterators.md: New mode iterator and attribute for hard
24820 float configuration.
24821 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
24823 * config/nds32/nds32-fpu.md: New file.
24824 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
24825 deal with hard float code generation.
24826 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
24828 (abi_type, float_reg_number): New enum type.
24829 * config/nds32/nds32-predicates.c: New predicates for hard float.
24830 * config/nds32/nds32-protos.h: Declare functions for hard float.
24831 * config/nds32/nds32.c: Implementation for hard float configuration.
24832 * config/nds32/nds32.h: Definitions for hard float configuration.
24833 * config/nds32/nds32.md: Include hard float machine description and
24834 modify patterns for hard float configuration.
24835 * config/nds32/nds32.opt: New options for hard float configuration.
24836 * config/nds32/predicates.md: New predicates for hard float
24839 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24841 * common/config/nds32/nds32-common.c
24842 (nds32_option_optimization_table): Enable -mreleax-hint by default.
24844 2018-04-05 Jakub Jelinek <jakub@redhat.com>
24846 PR middle-end/85195
24847 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
24848 CONSTRUCTOR_ELT (ctor, ...)->value.
24850 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
24853 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
24855 2018-04-05 Tom de Vries <tom@codesourcery.com>
24858 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
24861 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24862 Kito Cheng <kito.cheng@gmail.com>
24864 * config/nds32/constraints.md (U33): Fine-tune checking condition.
24865 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
24866 * config/nds32/nds32.h (nds32_16bit_address_type): Add
24867 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
24869 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24870 Kito Cheng <kito.cheng@gmail.com>
24872 * config/nds32/constraints.md (Ufe): New memory constraint.
24873 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
24874 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
24875 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
24877 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
24878 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
24880 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24882 * config/nds32/nds32.md: Use optimize_size in the condition for
24883 alu-shift instructions.
24885 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24887 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
24889 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24891 * config/nds32/nds32.md (negsi2): Refine pattern.
24893 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24894 Chung-Ju Wu <jasonwucj@gmail.com>
24896 * config/nds32/iterators.md (shift_rotate): New code iterator.
24897 (shift): New code attribute.
24898 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
24899 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
24900 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
24901 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
24902 bit-wise operations.
24903 (andsi3, *andsi3): Ditto.
24904 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
24905 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
24906 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
24907 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
24908 nds32_ior_operand, nds32_xor_operand): New predicates.
24910 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24912 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
24913 (addsi3, subsi3): ... this.
24915 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24917 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
24919 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24921 * config/nds32/nds32.md: Adjust indention.
24923 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24925 * config/nds32/nds32.md (feature): New attribute.
24927 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24929 * config/nds32/nds32.md (subtype): New attribute.
24931 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24934 * config/arm/arm-builtins.c (arm_expand_builtin): Change
24935 expansion to perform a bitwise AND of the argument followed by a
24936 boolean negation of the result.
24938 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
24940 PR rtl-optimization/84878
24941 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
24942 the basic block. Assert the use reference is not artificial and that
24943 it has an associated insn.
24945 2018-04-04 Michael Matz <matz@suse.de>
24947 * builtins.c (compute_objsize): Pass correct operand
24948 to array_at_struct_end_p.
24950 2018-04-04 Richard Biener <rguenther@suse.de>
24953 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
24954 from contexts for DINFO_LEVEL_TERSE and below.
24956 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24958 * config/nds32/nds32-doubleword.md (move_<mode>): Require
24959 resiter_operand condition.
24960 * config/nds32/nds32.md (*move<mode>): Ditto.
24962 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24963 Monk Chiang <sh.chiang04@gmail.com>
24965 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
24967 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24969 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
24971 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24972 Kito Cheng <kito.cheng@gmail.com>
24974 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
24975 nds32_cond_code_str, output_cond_branch,
24976 output_cond_branch_compare_zero, nds32_expand_cbranch,
24977 nds32_expand_cstore, nds32_expand_movcc,
24978 nds32_output_cbranchsi4_equality_zero,
24979 nds32_output_cbranchsi4_equality_reg,
24980 nds32_output_cbranchsi4_equality_reg_or_const_int,
24981 nds32_output_cbranchsi4_greater_less_zero: New functions.
24982 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
24983 nds32_expand_cstore, nds32_expand_movcc,
24984 nds32_output_cbranchsi4_equality_zero,
24985 nds32_output_cbranchsi4_equality_reg,
24986 nds32_output_cbranchsi4_equality_reg_or_const_int,
24987 nds32_output_cbranchsi4_greater_less_zero): Declare.
24988 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
24989 nds32_rimm11s_operand): New predicates.
24990 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
24991 * config/nds32/nds32.md: Rewrite all the branch and conditional move
24994 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24996 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
24997 * config/nds32/nds32.md: Ditto.
24998 * config/nds32/pipelines.md: Ditto.
25000 2018-04-04 Richard Biener <rguenther@suse.de>
25002 PR tree-optimization/85168
25003 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
25004 propagating abnormals.
25006 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
25008 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
25010 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
25011 Kito Cheng <kito.cheng@gmail.com>
25013 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
25014 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
25015 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
25016 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
25017 * config/nds32/nds32.md (sibcall_internal): New.
25018 (sibcall_register): Remove.
25019 (sibcall_immediate): Remove.
25020 (sibcall_value_internal): New.
25021 (sibcall_value_register): Remove.
25022 (sibcall_value_immediate): Remove.
25023 * config/nds32/predicates.md (nds32_general_register_operand): New.
25024 (nds32_call_address_operand): New.
25026 2018-04-03 Jakub Jelinek <jakub@redhat.com>
25028 PR rtl-optimization/85167
25029 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
25030 bb_defs if *split_p, instead preinitialize it to NULL.
25032 PR tree-optimization/85156
25033 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
25034 evaluating the argument multiple times.
25036 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
25038 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
25040 (_mm_cvtpd_ps): Likewise.
25041 (_mm_cvttpd_epi32): Likewise.
25042 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
25043 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
25044 vector, pixel, and bool following altivec.h include.
25046 2018-04-03 Martin Sebor <msebor@redhat.com>
25048 * doc/extend.texi (Common Function Attributes): Clarify.
25049 (const attribute): Likewise.
25050 (pure attribute): Likewise.
25052 2018-04-03 Jakub Jelinek <jakub@redhat.com>
25055 * config/i386/i386.c (ix86_expand_vector_set): Use
25056 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
25058 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
25060 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
25061 instructions when changing rounding bits to preserve precision bits
25062 in the x87 control word.
25064 2018-04-03 Martin Liska <mliska@suse.cz>
25066 PR tree-optimization/82491
25067 * rtl.h (strip_offset_and_add): Replace += suboffset with
25068 poly_uint64 () + suboffset.
25070 2018-03-29 Martin Liska <mliska@suse.cz>
25071 Martin Jambor <mjambor@suse.cz>
25074 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
25075 param_type is not an integral or pointer type.
25077 2018-04-03 Richard Biener <rguenther@suse.de>
25079 * sese.h (recompute_all_dominators): Remove.
25081 2018-04-02 Martin Sebor <msebor@redhat.com>
25083 * doc/invoke.texi (-Wrestrict): Fix typos.
25085 2018-04-02 Jim Wilson <jimw@sifive.com>
25087 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
25088 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
25089 (<optab>di3, <optab>si3_extend): Likewise.
25090 (<optab>si3_mask, <optab>si3_mask_1): New.
25091 (<optab>di3_mask, <optab>di3_mask_1): New.
25092 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
25093 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
25094 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
25096 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
25098 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
25101 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
25103 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
25104 (nds32_canonicalize_comparison): New function.
25106 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25107 Kito Cheng <kito.cheng@gmail.com>
25108 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25110 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
25111 * config/nds32/constants.md (unspec_volatile_element): Add
25112 UNSPEC_VOLATILE_RELAX_GROUP.
25113 * config/nds32/nds32-relax-opt.c: New file.
25114 * config/nds32/nds32-predicates.c
25115 (nds32_symbol_load_store_p): New function.
25116 * config/nds32/nds32-protos.h
25117 (nds32_symbol_load_store_p): Declare function.
25118 (make_pass_nds32_relax_opt): Declare new rtl pass function.
25119 * config/nds32/nds32.c
25120 (nds32_register_pass): New function to register pass.
25121 (nds32_register_passes): New function to register passes.
25122 * config/nds32/nds32.md (relax_group): New pattern.
25123 * config/nds32/nds32.opt (mrelax-hint): New option.
25124 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
25126 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
25128 * config/nds32/t-nds32: Modify files dependency.
25130 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25132 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
25133 (PROFILE_HOOK): Define its implementation.
25135 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25137 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
25138 type and 32-bit size.
25140 2018-04-01 Jakub Jelinek <jakub@redhat.com>
25142 PR middle-end/85090
25143 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
25144 (V_128_256): New mode iterator.
25145 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
25146 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
25147 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
25149 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
25150 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
25152 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25155 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
25156 NaN inputs correctly.
25158 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
25161 * config/rs6000/vsx.md (??r): New mode attribute.
25162 (*vsx_mov<mode>_64bit): Use it.
25163 (*vsx_mov<mode>_32bit): Likewise.
25165 2018-03-30 Martin Sebor <msebor@redhat.com>
25167 PR tree-optimization/84818
25168 * builtins.c (check_access): Use warning_n.
25170 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25173 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
25175 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
25178 2018-03-30 Julia Koval <julia.koval@intel.com>
25181 * x86-tune.def (movx, partial_reg_dependency): Enable for
25184 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
25186 PR inline-asm/84985
25187 * lra-constraints.c (process_alt_operands): Move setting
25188 this_alternative_matches below.
25190 2018-03-29 Martin Liska <mliska@suse.cz>
25193 * doc/invoke.texi: Document how LTO works with debug info.
25194 Describe auto-load support of binutils. Mention 'x86-64'
25195 as valid option value of -march option.
25197 2018-03-29 Jakub Jelinek <jakub@redhat.com>
25199 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
25202 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
25203 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
25204 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
25207 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
25210 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
25211 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
25212 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
25213 for RS6000_BTM_POWERPC64.
25214 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
25215 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
25216 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
25219 (DIVDEU): Likewise.
25221 2018-03-28 Carl Love <cel@us.ibm.com>
25224 2017-09-27 Carl Love <cel@us.ibm.com>
25226 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
25227 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
25228 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
25231 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25233 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
25234 instead of __vector bool.
25235 (_mm_max_pu8): Likewise.
25236 (_mm_min_pi16): Likewise.
25238 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
25241 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
25242 (DIVWEUO): Likewise.
25243 (DIVDEO): Likewise.
25244 (DIVDEUO): Likewise.
25245 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
25246 DIVWEUO and DIVDEUO.
25247 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
25248 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
25249 (div_extend): Likewise.
25250 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
25252 (__builtin_divweuo): Likewise.
25253 (__builtin_divdeo): Likewise.
25254 (__builtin_divdeuo): Likewise.
25256 2018-03-28 Jakub Jelinek <jakub@redhat.com>
25259 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
25260 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
25262 PR tree-optimization/82004
25263 * gimple-match-head.c (optimize_pow_to_exp): New function.
25264 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
25265 Don't fold to exp if optimize_pow_to_exp is false.
25267 2018-03-28 Martin Liska <mliska@suse.cz>
25270 * calls.c (initialize_argument_information): Fix trailing space.
25271 * common.opt: Fix typo and provide better explanation for
25272 -fsanitize-coverage option.
25273 * config/i386/i386.opt: Fix typo.
25275 2018-03-28 Jakub Jelinek <jakub@redhat.com>
25276 Martin Liska <mliska@suse.cz>
25279 * gimplify.c (asan_poison_variable): Don't do the check for
25280 gimplify_omp_ctxp here.
25281 (gimplify_decl_expr): Do it here.
25282 (gimplify_target_expr): Likewise.
25284 2018-03-28 Martin Liska <mliska@suse.cz>
25287 * config/i386/i386.c (ix86_function_arg_advance): Do not call
25288 chkp_type_bounds_count if MPX is not enabled.
25290 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
25292 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
25294 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25297 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
25298 function to create the function decl for complex long double
25299 multiply and divide for -mabi=ieeelongdouble.
25300 (init_float128_ieee): Call it.
25302 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
25305 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
25306 -fcf-protection=branch -mibt.
25307 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
25309 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25312 * config/arm/arm.c (arm_valid_symbolic_address): Handle
25313 arm_word_relocations.
25315 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
25318 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
25319 extern array declarations.
25321 2018-03-27 Richard Biener <rguenther@suse.de>
25323 PR middle-end/84067
25324 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
25325 explicit single_use checks.
25327 2018-03-27 Richard Biener <rguenther@suse.de>
25329 PR tree-optimization/85082
25330 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
25333 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25335 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
25336 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
25337 Turn on fasynchronous-unwind-tables and funwind-tables.
25339 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
25342 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
25343 (*bmi_blsr_<mode>_ccz): Ditto.
25345 2018-03-26 Tom de Vries <tom@codesourcery.com>
25347 PR tree-optimization/85063
25348 * omp-general.c (offloading_function_p): New function. Factor out
25350 * omp-offload.c (pass_omp_target_link::gate): ... here.
25351 * omp-general.h (offloading_function_p): Declare.
25352 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
25353 with attribute omp declare target for offloading functions.
25355 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
25357 PR tree-optimization/84005
25358 * tree-data-ref.h (get_base_for_alignment): Declare.
25359 * tree-data-ref.c (get_base_for_alignment_1): New function.
25360 (get_base_for_alignment): Likewise.
25361 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25362 get_base_for_alignment to find a suitable base object, instead
25363 of always using drb->base_address.
25365 2018-03-23 Jakub Jelinek <jakub@redhat.com>
25367 PR inline-asm/85022
25368 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
25369 known size by default.
25371 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
25373 PR inline-asm/85030
25374 * lra-constraints.c (process_alt_operands): Don't match BLKmode
25375 and non BLKmode operands.
25377 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25380 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
25381 Clean up attributes.
25383 2018-03-23 Richard Biener <rguenther@suse.de>
25386 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
25387 we are going to emit early debug for LTO.
25389 2018-03-23 Jakub Jelinek <jakub@redhat.com>
25391 PR inline-asm/85034
25392 * function.c (match_asm_constraints_1): Don't optimize if input
25393 doesn't satisfy general_operand predicate for output's mode.
25395 PR inline-asm/85022
25396 * alias.c (write_dependence_p): Don't require for x_canonicalized
25397 non-VOIDmode if x has VOIDmode.
25400 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
25401 just don't try to optimize it rather than assert it never happens.
25403 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
25405 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
25406 macro expansions for definition of ST_INTERNAL_<mode> and
25407 LD_INTERNAL_<mode> builtins.
25408 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
25410 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
25412 (altivec_expand_st_builtin): Likewise.
25413 (altivec_expand_builtin): Remove calls to deleted functions.
25414 (rs6000_address_for_altivec): Delete this function.
25415 * config/rs6000/vector.md: Remove expands for
25416 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
25418 2018-03-22 Sudakshina Das <sudi.das@arm.com>
25421 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
25422 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
25423 re-computing once computed.
25424 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
25425 (arm_init_machine_status): Initialize
25426 machine->static_chain_stack_bytes.
25428 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
25431 * doc/extend.texi: Add four new prototypes for vec_ld.
25432 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
25433 definitions for more logical presentation.
25434 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
25435 entries for V1TI variants of __builtin_altivec_ld builtin.
25436 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
25437 handling of V1TI variant of LVX icode pattern.
25438 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
25439 (rs6000_gimple_fold_builtin): Likewise.
25440 (altivec_init_builtins): Add code to define
25441 __builtin_altivec_lvx_v1ti function.
25443 2018-03-22 Jakub Jelinek <jakub@redhat.com>
25445 PR inline-asm/84941
25446 * function.c (match_asm_constraints_1): Don't do the optimization
25447 if input isn't a REG, SUBREG, MEM or constant.
25449 2018-03-22 Tom de Vries <tom@codesourcery.com>
25451 PR tree-optimization/84956
25452 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
25453 bb_has_abnormal_pred.
25455 2018-03-22 Jakub Jelinek <jakub@redhat.com>
25458 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
25459 DECL_INITIAL (decl) to decl at the end.
25460 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
25461 adjust the comment.
25463 2018-03-21 Joseph Myers <joseph@codesourcery.com>
25465 * doc/extend.texi (__builtin_tgmath): Document when complex
25466 integer types are treated as _Complex _Float64.
25468 2018-03-21 Tom de Vries <tom@codesourcery.com>
25470 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
25472 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25474 PR tree-optimization/84960
25475 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
25476 if it is ENTRY block, move them into single succ of ENTRY in that case.
25478 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
25480 PR tree-optimization/84811
25481 * poly-int.h (poly_span_traits): Remove the T3 parameter and
25482 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
25483 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
25484 (known_subrange_p): Update accordingly. Cast each value involved
25485 in the size comparison, rather than casting the result of the
25488 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25490 PR tree-optimization/84982
25491 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
25492 by flipping the least significant bit rather than all bits from
25493 bitpos to bitpos + bitsize - 1.
25495 2018-03-21 Nathan Sidwell <nathan@acm.org>
25497 * doc/extend.texi (Deprecated Features): Remove mention of
25498 long-deleted deprecations.
25500 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25503 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
25504 * configure: Regenerate.
25506 2018-03-21 Tom de Vries <tom@codesourcery.com>
25508 PR tree-optimization/83126
25509 * tree-parloops.c (num_phis): New function.
25510 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
25512 2018-03-21 Nathan Sidwell <nathan@acm.org>
25514 * doc/extend.texi (Deprecated Features): Update deprecated flags,
25515 mention anon-struct/union members and trailing attributes.
25517 2018-03-21 Bin Cheng <bin.cheng@arm.com>
25519 PR tree-optimization/84969
25520 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
25521 builtin memset partitions if they set different rhs values.
25523 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25525 PR rtl-optimization/84989
25526 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
25527 VEC_DUPLICATE with scalar result mode.
25529 2018-03-21 Martin Liska <mliska@suse.cz>
25532 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
25533 not intended return statement.
25535 2018-03-21 Martin Liska <mliska@suse.cz>
25538 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
25539 (chkp_find_bound_slots_1): Limit number of iterations.
25541 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
25544 * Minor grammar fixes for x86 options.
25546 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25549 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
25550 holding REG_CFA_RESTORE notes, instead turn them into a USE.
25552 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
25555 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
25556 (altivec_lvx_<mode>_1op): Likewise.
25557 (altivec_stvx_<mode>_2op): Likewise.
25558 (altivec_stvx_<mode>_1op): Likewise.
25559 (altivec_lvx_<VM2:mode>): New define_expand.
25560 (altivec_stvx_<VM2:mode>): Likewise.
25561 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
25562 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
25563 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
25564 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
25565 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
25566 (rs6000_gen_lvx): Likewise.
25567 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
25568 (altivec_expand_stv_builtin): Likewise.
25569 (altivec_expand_builtin): Likewise.
25570 * config/rs6000/vector.md: Likewise.
25572 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25575 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
25578 2018-03-20 Richard Biener <rguenther@suse.de>
25581 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
25582 sign-conversions as zero, fall back to standard scalar_stmt
25585 2018-03-20 Martin Liska <mliska@suse.cz>
25588 * predict.c (rebuild_frequencies): Handle case when we have
25589 PROFILE_ABSENT, but flag_guess_branch_prob is false.
25591 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25594 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
25595 flag_section_anchors.
25596 * varasm.c (use_blocks_for_decl_p): Remove hack for
25597 dw2_force_const_mem.
25600 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
25602 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
25603 be created, use lowpart_subreg of operands[0] rather than operands[0]
25605 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
25606 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
25607 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
25608 and n constraint instead of aarch64_shift_imm_di and Usd.
25609 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
25610 (*aarch64_<optab>_reg_minus<mode>3): ... this.
25612 2018-03-20 Sudakshina Das <sudi.das@arm.com>
25615 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
25616 to favor GPR over NEON registers.
25617 (<shift>di3_neon): Likewise.
25619 2018-03-20 Tom de Vries <tom@codesourcery.com>
25622 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
25623 (nvptx_process_pars): Emit bar.sync asap and alap.
25625 2018-03-20 Tom de Vries <tom@codesourcery.com>
25628 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
25629 seen_label if seen_label is already set.
25631 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25634 * config/i386/i386.c (fold_builtin_cpu): For features above 31
25635 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
25636 Use 1U instead of 1. Formatting fixes.
25639 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
25640 instead of TREE_TYPE (s1) for the return value.
25642 2018-03-19 Jakub Jelinek <jakub@redhat.com>
25644 PR tree-optimization/84946
25645 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
25646 bitsize + bitsize in poly_uint64 rather than poly_int64.
25649 * dwarf2asm.c: Include fold-const.c.
25650 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
25651 of decl rather than decl itself.
25653 PR rtl-optimization/84643
25654 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
25656 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
25659 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
25660 calling assemble_variable.
25662 2018-03-19 Sudakshina Das <sudi.das@arm.com>
25665 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
25666 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
25668 2018-03-19 Jim Wilson <jimw@sifive.com>
25671 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
25672 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
25673 (riscv_first_stack_step): Likewise.
25674 (riscv_option_override): Use STACK_BOUNDARY instead of
25675 MIN_STACK_BOUNDARY.
25676 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
25677 MIN_STACK_BOUNDARY.
25678 (BIGGEST_ALIGNMENT): Set to 128.
25679 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
25680 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
25683 2018-03-19 Richard Biener <rguenther@suse.de>
25685 PR tree-optimization/84933
25686 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
25687 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
25689 2018-03-19 Richard Biener <rguenther@suse.de>
25691 PR tree-optimization/84859
25692 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
25693 (cond_if_else_store_replacement): Perform sinking operation on
25694 single-store BBs regardless of MAX_STORES_TO_SINK setting.
25695 Generalize what a BB with a single eligible store is.
25697 2018-03-19 Richard Biener <rguenther@suse.de>
25699 PR tree-optimization/84929
25700 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
25701 chrec_is_positive against non-chrec arg.
25703 2018-03-19 Tamar Christina <tamar.christina@arm.com>
25706 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
25708 2018-03-18 Martin Liska <mliska@suse.cz>
25710 PR rtl-optimization/84635
25711 * regrename.c (build_def_use): Use matches_mode only when
25714 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
25716 PR tree-optimization/84913
25717 * tree-vect-loop.c (vectorizable_reduction): Don't try to
25718 vectorize chains of COND_EXPRs.
25720 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25722 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
25724 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25726 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
25728 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25730 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
25732 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25733 Kito Cheng <kito.cheng@gmail.com>
25735 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
25736 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
25737 (nds32_adjust_reg_alloc_order): New function.
25738 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
25740 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
25742 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
25743 nds32_print_operand, nds32_print_operand_address): Use
25744 HOST_WIDE_INT_PRINT_DEC instead.
25746 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25748 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
25750 2018-03-17 Jakub Jelinek <jakub@redhat.com>
25753 * config/i386/i386.c (initial_ix86_tune_features,
25754 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
25755 unsigned long long.
25756 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
25757 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
25758 rather than 1u << ix86_tune. Formatting fix.
25759 (ix86_option_override_internal): Change ix86_arch_mask from
25760 unsigned int to unsigned HOST_WIDE_INT, initialize to
25761 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
25762 (ix86_function_specific_restore): Likewise.
25764 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25767 * postreload.c (reload_combine_recognize_pattern): Perform
25768 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
25769 truncate_int_for_mode the result for the destination's mode.
25772 * hsa-gen.c (mem_type_for_type): Fix comment typo.
25773 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
25775 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
25778 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
25781 * lra-assigns.c (lra_split_hard_reg_for): Don't use
25782 regno_allocno_class_array and sorted_pseudos.
25783 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
25784 insns where regno is used.
25786 2018-03-16 Martin Liska <mliska@suse.cz>
25789 * multiple_target.c (create_dispatcher_calls): Redirect
25790 reference in the symbol table.
25792 2018-03-16 Martin Liska <mliska@suse.cz>
25795 * multiple_target.c (create_dispatcher_calls): Redirect also
25798 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25802 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
25803 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
25804 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
25806 2018-03-16 Julia Koval <julia.koval@intel.com>
25808 * doc/invoke.texi (Skylake Server): Add CLWB.
25809 Cannonlake): Remove CLWB.
25811 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25813 PR tree-optimization/84841
25814 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
25816 (FLOAT_ONE_CONST_TYPE): Define.
25817 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
25818 (sort_by_operand_rank): Put entries with higher constant_type last
25819 rather than first to match comments.
25821 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
25823 * config/nios2/nios2.md (movsi_internal): Fix thinko in
25826 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25829 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
25830 check and use address of target always.
25832 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
25835 * config/i386/i386.c (indirect_thunk_needed): Update comments.
25836 (indirect_thunk_bnd_needed): Likewise.
25837 (indirect_thunks_used): Likewise.
25838 (indirect_thunks_bnd_used): Likewise.
25839 (indirect_return_needed): New.
25840 (indirect_return_bnd_needed): Likewise.
25841 (output_indirect_thunk_function): Add a bool argument for
25843 (output_indirect_thunk_function): Don't generate alias for
25844 function return thunk.
25845 (ix86_code_end): Call output_indirect_thunk_function to generate
25846 function return thunks.
25847 (ix86_output_function_return): Set indirect_return_bnd_needed
25848 and indirect_return_needed instead of indirect_thunk_bnd_needed
25849 and indirect_thunk_needed.
25851 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
25853 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
25854 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
25855 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
25857 2018-03-15 David Malcolm <dmalcolm@redhat.com>
25858 Paul Hua <paul.hua.gm@gmail.com>
25861 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
25863 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
25865 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
25866 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
25867 resp. SFmode cases.
25869 2018-03-15 Tamar Christina <tamar.christina@arm.com>
25872 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
25873 instead of GET_MODE_SIZE when comparing Units.
25875 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
25878 * varasm.c (hash_section): Return an unchangeble hash value
25879 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
25880 Return !aarch64_can_use_per_function_literal_pools_p ().
25882 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25885 * optabs.c (emit_conditional_move): Pass address of cmode's copy
25886 rather than address of cmode as last argument to prepare_cmp_insn.
25888 2018-03-15 Julia Koval <julia.koval@intel.com>
25890 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
25891 F_AVX512VNNI, F_AVX512BITALG): New.
25893 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
25896 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
25897 insn for floating-point loads and stores.
25899 2018-03-14 Carl Love <cel@us.ibm.com>
25901 * config/rs6000/rs6000-c.c: Add macro definitions for
25902 ALTIVEC_BUILTIN_VEC_PERMXOR.
25903 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
25904 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
25905 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
25906 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
25908 * config/doc/extend.texi: Add prototypes for vec_permxor.
25910 2018-03-14 David Malcolm <dmalcolm@redhat.com>
25913 * diagnostic-show-locus.c (class layout_point): Convert m_line
25914 from int to linenum_type.
25915 (line_span::comparator): Use linenum "compare" function when
25916 comparing line numbers.
25917 (test_line_span): New function.
25918 (layout_range::contains_point): Convert param "row" from int to
25920 (layout_range::intersects_line_p): Likewise.
25921 (layout::will_show_line_p): Likewise.
25922 (layout::print_source_line): Likewise.
25923 (layout::should_print_annotation_line_p): Likewise.
25924 (layout::print_annotation_line): Likewise.
25925 (layout::print_leading_fixits): Likewise.
25926 (layout::annotation_line_showed_range_p): Likewise.
25927 (struct line_corrections): Likewise for field m_row.
25928 (line_corrections::line_corrections): Likewise for param "row".
25929 (layout::print_trailing_fixits): Likewise.
25930 (layout::get_state_at_point): Likewise.
25931 (layout::get_x_bound_for_row): Likewise.
25932 (layout::print_line): Likewise.
25933 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
25934 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
25935 * input.c (selftest::test_linenum_comparisons): New function.
25936 (selftest::input_c_tests): Call it.
25937 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
25938 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
25939 * selftest.h (ASSERT_GT): New macro.
25940 (ASSERT_GT_AT): New macro.
25941 (ASSERT_LT): New macro.
25942 (ASSERT_LT_AT): New macro.
25944 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
25946 PR rtl-optimization/84780
25947 * combine.c (distribute_links): Don't make a link based on pc_rtx.
25949 2018-03-14 Martin Liska <mliska@suse.cz>
25951 * tree.c (record_node_allocation_statistics): Use
25952 get_stats_node_kind.
25953 (get_stats_node_kind): New function extracted from
25954 record_node_allocation_statistics.
25955 (free_node): Use get_stats_node_kind.
25957 2018-03-14 Richard Biener <rguenther@suse.de>
25959 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
25960 that the value-set of ANTIC_IN doesn't grow.
25963 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
25965 (BB_VISITED_WITH_VISITED_SUCCS): New define.
25966 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
25968 2018-03-14 Julia Koval <julia.koval@intel.com>
25970 * config.gcc (icelake-client, icelake-server): New.
25972 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
25973 (initial_ix86_arch_features): Ditto.
25974 (PTA_SKYLAKE): Add SGX.
25975 (PTA_ICELAKE): Remove.
25976 (PTA_ICELAKE_CLIENT): New.
25977 (PTA_ICELAKE_SERVER): New.
25978 (ix86_option_override_internal): Split up icelake on icelake client and
25980 (get_builtin_code_for_version): Ditto.
25981 (fold_builtin_cpu): Ditto.
25982 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
25983 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
25984 * config/i386/i386.h (processor_type): Ditto.
25985 * doc/invoke.texi: Ditto.
25987 2018-03-14 Jakub Jelinek <jakub@redhat.com>
25990 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
25991 INTEGER_CST offset, add it together with bitpos / 8 and
25992 sign extend based on POINTER_SIZE.
25996 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25999 * config/i386/constraints.md (Yc): New register constraint.
26000 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26001 Use Yc constraint for alternative 2 of operand 0. Remove
26002 preferred_for_speed attribute.
26004 2018-03-14 Richard Biener <rguenther@suse.de>
26006 PR tree-optimization/84830
26007 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
26008 with the old one to avoid oscillations.
26010 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
26013 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
26015 (assign_by_spills): Return a flag of reload assignment failure.
26016 Do not process the reload assignment failures. Do not spill other
26017 reload pseudos if they has the same reg class. Update n if
26019 (lra_assign): Add a return arg. Set up from the result of
26020 assign_by_spills call.
26021 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26022 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26023 usage_insns if it is not NULL.
26024 (spill_hard_reg_in_range): New function.
26025 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26026 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26027 function prototypes.
26028 (lra_assign): Change prototype.
26029 * lra.c (lra): Add code to deal with fails by splitting hard reg
26032 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
26034 * config/riscv/riscv.opt (mrelax): New option.
26035 * config/riscv/riscv.c (riscv_file_start): Emit ".option
26036 "norelax" when riscv_mrelax is disabled.
26037 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
26039 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26042 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
26043 reassociation for int modes.
26045 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26047 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
26048 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
26050 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
26051 * config/aarch64/aarch64-sve.md
26052 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
26053 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
26054 (*extend<mode><Vwide>2): Rename to...
26055 (aarch64_sve_extend<mode><Vwide>2): ...this.
26056 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
26057 renaming the old pattern to...
26058 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
26060 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
26061 define_expand, renaming the old pattern to...
26062 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
26063 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
26064 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
26065 account when deciding which SVE instruction the optab should use.
26066 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
26068 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26070 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
26071 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
26072 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
26073 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
26074 (tlsdesc_small_<mode>): Turn a define_expand and use
26075 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
26076 (tlsdesc_small_advsimd_<mode>): ...this.
26077 (tlsdesc_small_sve_<mode>): New pattern.
26079 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26081 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
26082 (UNSPEC_UMUL_HIGHPART): New constants.
26083 (MUL_HIGHPART): New int iteraor.
26084 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
26085 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
26087 (*<su>mul<mode>3_highpart): New define_insn.
26089 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
26092 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
26095 2018-03-13 Martin Liska <mliska@suse.cz>
26098 * (sem_item_optimizer::sem_item_optimizer): Initialize new
26100 (sem_item_optimizer::~sem_item_optimizer): Release it.
26101 (sem_item_optimizer::merge_classes): Register variable aliases.
26102 (sem_item_optimizer::fixup_pt_set): New function.
26103 (sem_item_optimizer::fixup_points_to_sets): Likewise.
26104 * ipa-icf.h: Declare new variables and functions.
26106 2018-03-13 Jakub Jelinek <jakub@redhat.com>
26108 PR middle-end/84834
26109 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
26110 integer_pow2p@2 and test integer_pow2p in condition.
26111 (A < 0 ? C : 0): Similarly for @1.
26113 PR middle-end/84831
26114 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
26115 characters starting at p contain '\0' character, don't look beyond
26119 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
26120 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
26123 * reg-stack.c (change_stack): Change update_end var from int to
26124 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
26125 also call set_block_for_insn on the newly added insns and rescan.
26128 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
26129 on the last operand.
26132 * tree.c (stabilize_reference_1): Return save_expr (e) for
26133 STATEMENT_LIST even if it doesn't have side-effects.
26135 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
26137 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
26139 2018-03-12 Renlin Li <renlin.li@arm.com>
26141 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
26142 aarch64_output_scalar_simd_mov_immediate.
26144 2018-03-12 Martin Sebor <msebor@redhat.com>
26146 PR tree-optimization/83456
26147 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
26148 for perfectly overlapping calls to memcpy.
26149 (gimple_fold_builtin_memory_chk): Same.
26150 (gimple_fold_builtin_strcpy): Handle no-warning.
26151 (gimple_fold_builtin_stxcpy_chk): Same.
26152 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
26154 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
26156 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
26157 parameter. Use it for SFmode.
26158 (rs6000_function_arg_advance_1): Adjust.
26159 (rs6000_function_arg): Adjust.
26160 (rs6000_gimplify_va_arg): Pass false for that new parameter.
26162 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
26164 PR rtl-optimization/84169
26165 PR rtl-optimization/84780
26166 * combine.c (can_combine_p): Check for a 2-insn combination whether
26167 the destination register is used between the two insns, too.
26169 2018-03-12 Richard Biener <rguenther@suse.de>
26171 PR tree-optimization/84803
26172 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
26173 for refs DR analysis didn't process.
26175 2018-03-12 Richard Biener <rguenther@suse.de>
26177 PR tree-optimization/84777
26178 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
26179 force-vectorize loops ignore whether we are optimizing for size.
26181 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
26183 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
26184 (TARGET_MD_ASM_ADJUST): Define.
26186 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
26187 Kito Cheng <kito.cheng@gmail.com>
26188 Chung-Ju Wu <jasonwucj@gmail.com>
26190 * config/nds32/nds32.c (nds32_compute_stack_frame,
26191 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
26192 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
26193 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
26194 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
26195 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
26196 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
26197 * config/nds32/nds32.md (prologue, epilogue): Use macro
26198 NDS32_V3PUSH_AVAILABLE_P to do checking.
26200 2018-03-11 Jakub Jelinek <jakub@redhat.com>
26203 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
26204 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
26205 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
26206 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
26207 addition of most attributes on !orig_type_die or the attribute not
26208 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
26210 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26211 Chung-Ju Wu <jasonwucj@gmail.com>
26213 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
26214 __NDS32_VH__ macro.
26215 * config/nds32/nds32.opt (mvh): New option.
26217 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26218 Chung-Ju Wu <jasonwucj@gmail.com>
26220 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
26222 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
26223 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
26226 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26227 Chung-Ju Wu <jasonwucj@gmail.com>
26229 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
26231 * config/nds32/nds32-multiple.md (strlensi): New pattern.
26232 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
26234 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
26235 Kito Cheng <kito.cheng@gmail.com>
26236 Chung-Ju Wu <jasonwucj@gmail.com>
26238 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
26239 UNSPEC_FFMISM and UNSPEC_FLMISM.
26240 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
26241 for ffb, ffmism and flmism.
26242 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
26243 (unspec_ffmism): Ditto.
26244 (unspec_flmism): Ditto.
26245 (nds32_expand_builtin_impl): Check if string extension is available.
26246 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
26247 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
26249 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
26252 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
26255 * lra-assigns.c (assign_by_spills): Return a flag of reload
26256 assignment failure. Do not process the reload assignment
26257 failures. Do not spill other reload pseudos if they has the same
26259 (lra_assign): Add a return arg. Set up from the result of
26260 assign_by_spills call.
26261 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26262 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26263 usage_insns if it is not NULL.
26264 (spill_hard_reg_in_range): New function.
26265 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26266 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26267 function prototypes.
26268 (lra_assign): Change prototype.
26269 * lra.c (lra): Add code to deal with fails by splitting hard reg
26272 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
26275 * config/i386/i386.opt: Replace Enforcment with Enforcement.
26277 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
26280 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
26281 (dw_val_node): Add val_symbolic_view.
26282 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
26283 (symview_upper_bound): New.
26284 (new_line_info_table): Initialize symviews_since_reset.
26285 (dwarf2out_source_line): Count symviews_since_reset and set
26286 symview_upper_bound.
26287 (dw_val_equal_p): Handle symview.
26288 (add_AT_symview): New.
26289 (print_dw_val): Handle symview.
26290 (attr_checksum, attr_checksum_ordered): Likewise.
26291 (same_dw_val_p, size_of_die): Likewise.
26292 (value_format, output_die): Likewise.
26293 (add_high_low_attributes): Use add_AT_symview for entry_view.
26294 (dwarf2out_finish): Reset symview_upper_bound, clear
26297 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
26300 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
26301 Add strict argument and use it.
26302 (rs6000_split_multireg_move): Update for new strict argument.
26303 (mem_operand_gpr): Disallow all non-offsettable addresses.
26304 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
26306 2018-03-09 Jakub Jelinek <jakub@redhat.com>
26309 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
26310 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
26311 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
26314 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
26315 decl, use remap_type if we want to use the type.
26317 2018-03-09 Martin Sebor <msebor@redhat.com>
26319 PR tree-optimization/84526
26320 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
26322 (builtin_access::generic_overlap): Be prepared to handle non-array
26325 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
26327 PR rtl-optimization/84682
26328 * lra-constraints.c (process_address_1): Check is_address flag
26329 for address constraints.
26330 (process_alt_operands): Likewise.
26331 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
26332 preprocess_constraints.
26333 * recog.h (preprocess_constraints): Add oploc parameter.
26335 * recog.c (preprocess_constraints): Test address_operand for
26336 CT_ADDRESS constraints.
26338 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
26341 * lra-assigns.c (assign_by_spills): Return a flag of reload
26342 assignment failure. Do not process the reload assignment
26343 failures. Do not spill other reload pseudos if they has the same
26345 (lra_assign): Add a return arg. Set up from the result of
26346 assign_by_spills call.
26347 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26348 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26349 usage_insns if it is not NULL.
26350 (spill_hard_reg_in_range): New function.
26351 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26352 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26353 function prototypes.
26354 (lra_assign): Change prototype.
26355 * lra.c (lra): Add code to deal with fails by splitting hard reg
26358 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26361 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
26362 Accept complain bool parameter. Only emit errors if it is true.
26363 (arm_parse_cpu_option_name): Likewise.
26364 (arm_target_thumb_only): Adjust callers of the above.
26365 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
26366 prototype to take a default true bool parameter.
26367 (arm_parse_arch_option_name): Likewise.
26369 2018-03-09 David Malcolm <dmalcolm@redhat.com>
26370 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
26374 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
26375 * configure: Regenerate.
26376 * configure.ac ("linker --version-script option"): New.
26377 ("linker soname option"): New.
26379 2018-03-09 Richard Biener <rguenther@suse.de>
26381 PR tree-optimization/84775
26382 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
26383 immediate uses of predicate stmts and mark them modified.
26386 PR tree-optimization/84178
26387 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
26389 (version_loop_for_if_conversion): Delay update_ssa call.
26390 (tree_if_conversion): Delay update_ssa until after predicate
26393 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
26396 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
26397 when the function accesses prior frames.
26399 2018-03-08 Jakub Jelinek <jakub@redhat.com>
26402 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
26403 gen_llsym, otherwise call maybe_gen_llsym.
26405 PR inline-asm/84742
26406 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
26407 has ',' character inside of it.
26409 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26412 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
26413 as clobbering CC_REGNUM.
26415 2018-03-08 Richard Biener <rguenther@suse.de>
26417 PR middle-end/84552
26418 * tree-scalar-evolution.c: Include tree-into-ssa.h.
26419 (follow_copies_to_constant): Do not follow SSA names registered
26422 2018-03-08 Richard Biener <rguenther@suse.de>
26424 PR tree-optimization/84178
26425 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
26427 (version_loop_for_if_conversion): Delay update_ssa call.
26428 (tree_if_conversion): Delay update_ssa until after predicate
26431 2018-03-08 David Malcolm <dmalcolm@redhat.com>
26433 PR tree-optimization/84178
26434 * tree-if-conv.c (release_bb_predicate): Remove the
26435 the assertion that the stmts have NULL use_ops.
26436 Discard the statements, asserting that they haven't
26437 yet been added to a BB.
26439 2018-03-08 Richard Biener <rguenther@suse.de>
26441 PR tree-optimization/84746
26442 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
26443 (phi_translate): Pass in destination ANTIC_OUT set.
26444 (phi_translate_1): Likewise. For a simplified result lookup
26445 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
26446 (phi_translate_set): Adjust.
26447 (do_pre_regular_insertion): Likewise.
26448 (do_pre_partial_partial_insertion): Likewise.
26450 2018-03-08 Martin Liska <mliska@suse.cz>
26452 PR gcov-profile/84735
26453 * doc/gcov.texi: Document usage of profile files.
26454 * gcov-io.h: Document changes in the format.
26456 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
26460 * dwarf2out.c (struct dw_line_info_table): Update comments for
26462 (FORCE_RESET_NEXT_VIEW): New.
26463 (FORCE_RESETTING_VIEW_P): New.
26464 (RESETTING_VIEW_P): Check for -1 too.
26465 (ZERO_VIEW_P): Likewise.
26466 (new_line_info_table): Force-reset next view.
26467 (dwarf2out_begin_function): Likewise.
26468 (dwarf2out_source_line): Simplify zero_view_p initialization.
26469 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
26470 view directly. Omit view when omitting .loc at line 0.
26472 2018-03-08 Jakub Jelinek <jakub@redhat.com>
26474 PR tree-optimization/84740
26475 * tree-switch-conversion.c (process_switch): Call build_constructors
26476 only if info.phi_count is non-zero.
26478 PR tree-optimization/84739
26479 * tree-tailcall.c (find_tail_calls): Check call arguments against
26480 DECL_ARGUMENTS (current_function_decl) rather than
26481 DECL_ARGUMENTS (func) when checking for tail recursion.
26483 2018-03-07 Jakub Jelinek <jakub@redhat.com>
26485 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
26486 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
26487 Volker Reichelt's entry and add entries for people that perform
26488 GCC fuzzy testing and report numerous bugs.
26490 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26493 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
26494 readonly data in sdata, if that is disabled.
26495 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
26496 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
26497 -mreadonly-in-sdata option.
26499 2018-03-07 Martin Sebor <msebor@redhat.com>
26501 PR tree-optimization/84468
26502 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
26503 basic block when looking for nul assignment.
26505 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
26508 * except.h (output_function_exception_table): Adjust prototype.
26509 * except.c (output_function_exception_table): Remove FNNAME parameter
26510 and add SECTION parameter. Ouput one part of the table at a time.
26511 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
26512 the first part of the exception table and emit unwind directives.
26513 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
26514 (i386_pe_seh_cold_init): Likewise.
26515 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
26516 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
26517 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
26518 (ix86_output_call_insn): Emit a nop in one more case for SEH.
26519 * config/i386/winnt.c: Include except.h.
26520 (struct seh_frame_state): Add reg_offset, after_prologue and
26521 in_cold_section fields.
26522 (i386_pe_seh_end_prologue): Set seh->after_prologue.
26523 (i386_pe_seh_cold_init): New function.
26524 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
26525 to seh->in_cold_section.
26526 (seh_emit_push): Record the offset of the push.
26527 (seh_emit_save): Record the offet of the save.
26528 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
26529 Test seh->after_prologue to disregard the epilogue.
26530 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
26531 (i386_pe_end_cold_function): New function.
26533 2018-03-07 Jakub Jelinek <jakub@redhat.com>
26536 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
26537 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
26540 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
26542 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
26543 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
26545 PR middle-end/84723
26546 * multiple_target.c: Include tree-inline.h and intl.h.
26547 (expand_target_clones): Diagnose and fail if node->definition and
26548 !tree_versionable_function_p (node->decl).
26550 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
26552 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
26554 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
26555 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
26556 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
26558 2018-03-06 Jakub Jelinek <jakub@redhat.com>
26561 * combine.c (try_combine): Use reg_or_subregno instead of handling
26562 just paradoxical SUBREGs and REGs.
26564 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
26566 * config/arc/arc.c (arc_finalize_pic): Remove function.
26567 (arc_must_save_register): We use single base PIC register, remove
26568 checks to save/restore the PIC register.
26569 (arc_expand_prologue): Likewise.
26570 * config/arc/arc-protos.h (arc_set_default_type_attributes):
26572 (arc_verify_short): Likewise.
26573 (arc_attr_type): Likewise.
26574 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
26575 (walk_stores): Likewise.
26576 (arc_address_cost): Make it static.
26577 (arc_verify_short): Likewise.
26578 (branch_dest): Likewise.
26579 (arc_attr_type): Likewise.
26580 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
26581 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
26582 (arc_final_prescan_insn): Remove inserting the nops due to
26583 hardware hazards. It is done in reorg step.
26584 (insn_length_variant_t): Remove.
26585 (insn_length_parameters_t): Likewise.
26586 (arc_insn_length_parameters): Likewise.
26587 (arc_get_insn_variants): Likewise.
26588 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
26590 2018-03-06 Jakub Jelinek <jakub@redhat.com>
26592 PR inline-asm/84683
26593 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
26596 PR tree-optimization/84687
26597 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
26599 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
26601 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26603 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
26604 Rename to ppc_speculation_barrier.
26605 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
26606 __builtin_ppc_speculation_barrier.
26608 2018-03-05 Jakub Jelinek <jakub@redhat.com>
26611 * combine.c (combine_simplify_rtx): Don't try to simplify if
26612 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
26615 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
26617 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
26618 to 32 bytes when compiling for POWER9.
26620 2018-03-05 Jakub Jelinek <jakub@redhat.com>
26623 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
26624 regparm >= 3 with no arg reg available also for calls with
26625 flag_force_indirect_call. Pass decl to ix86_function_regparm.
26628 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
26630 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
26632 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
26635 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
26637 2018-03-05 Richard Biener <rguenther@suse.de>
26639 PR tree-optimization/84486
26640 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
26641 When inserting a __builtin_assume_aligned call set the LHS
26642 SSA name alignment info accordingly.
26644 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
26646 PR tree-optimization/84114
26647 * config/aarch64/aarch64.c (aarch64_reassociation_width)
26648 Avoid reassociation of FLOAT_MODE addition.
26650 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
26652 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
26653 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
26654 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
26655 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
26656 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
26657 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
26658 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
26660 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
26661 __builtin_ia32_wbinvd): New builtins.
26662 (SPECIAL_ARGS2): New.
26663 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
26664 (SPECIAL_ARGS2): New.
26665 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
26666 (ix86_valid_target_attribute_inner_p): Ditto.
26667 (ix86_init_mmx_sse_builtins): Add special_args2.
26668 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
26669 TARGET_WBNOINVD_P): New.
26670 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
26671 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
26672 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
26673 * config/i386/immintrin.h (_wbinvd): New intrinsic.
26674 * config/i386/pconfigintrin.h: New file.
26675 * config/i386/wbnoinvdintrin.h: Ditto.
26676 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
26678 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
26680 2018-03-05 Richard Biener <rguenther@suse.de>
26682 PR tree-optimization/84670
26683 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
26685 (BB_VISITED_WITH_VISITED_SUCCS): New define.
26686 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
26687 (compute_antic_aux): Only assert the number of values in ANTIC_IN
26688 doesn't grow if all successors (recursively) were visited at least
26691 2018-03-05 Richard Biener <rguenther@suse.de>
26693 PR tree-optimization/84650
26694 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
26695 if executed in the loop pipeline.
26697 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
26699 * doc/configfiles.texi (Configuration Files): Move info about
26700 conditionalizing $target-protos.h to...
26701 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
26702 differs from $target-protos.h.
26704 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
26705 Chung-Ju Wu <jasonwucj@gmail.com>
26707 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
26708 * config/nds32/nds32-multiple.md (setmemsi): Define.
26709 * config/nds32/nds32-memory-manipulation.c
26710 (nds32_gen_dup_4_byte_to_word_value): New.
26711 (emit_setmem_word_loop): New.
26712 (emit_setmem_byte_loop): New.
26713 (nds32_expand_setmem_loop): New.
26714 (nds32_expand_setmem_loop_v3m): New.
26715 (nds32_expand_setmem_unroll): New.
26716 (nds32_expand_setmem): New.
26718 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26719 Chung-Ju Wu <jasonwucj@gmail.com>
26721 * config/nds32/nds32-memory-manipulation.c
26722 (nds32_emit_load_store): New.
26723 (nds32_emit_post_inc_load_store): New.
26724 (nds32_emit_mem_move): New.
26725 (nds32_emit_mem_move_block): New.
26726 (nds32_expand_movmemsi_loop_unknown_size): New.
26727 (nds32_expand_movmemsi_loop_known_size): New.
26728 (nds32_expand_movmemsi_loop): New.
26729 (nds32_expand_movmemsi_unroll): New.
26730 (nds32_expand_movmemqi): Rename ...
26731 (nds32_expand_movmemsi): ... to this.
26732 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
26733 (movmemsi): ... to this.
26734 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
26735 (nds32_expand_movmemsi): ... to this.
26737 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26738 Monk Chiang <sh.chiang04@gmail.com>
26739 Chung-Ju Wu <jasonwucj@gmail.com>
26741 * config/nds32/nds32-protos.h
26742 (nds32_expand_load_multiple): New arguments.
26743 (nds32_expand_store_multiple): Ditto.
26744 (nds32_valid_multiple_load_store): Rename ...
26745 (nds32_valid_multiple_load_store_p): ... to this.
26746 * config/nds32/nds32-memory-manipulation.c
26747 (nds32_expand_load_multiple): Refine implementation.
26748 (nds32_expand_store_multiple): Ditto.
26749 * config/nds32/nds32-multiple.md
26750 (load_multiple): Update nds32_expand_load_multiple interface.
26751 (store_multiple): Update nds32_expand_store_multiple interface.
26752 * config/nds32/nds32-predicates.c
26753 (nds32_valid_multiple_load_store): Rename ...
26754 (nds32_valid_multiple_load_store_p): ... to this and refine
26756 * config/nds32/predicates.md
26757 (nds32_load_multiple_and_update_address_operation): New predicate.
26758 (nds32_store_multiple_and_update_address_operation): New predicate.
26760 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26761 Chung-Ju Wu <jasonwucj@gmail.com>
26763 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
26764 (combo): New attribute.
26765 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
26767 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
26769 * config/nds32/nds32.opt: Change -mcmodel= default value.
26771 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
26772 Monk Chiang <sh.chiang04@gmail.com>
26773 Chung-Ju Wu <jasonwucj@gmail.com>
26775 * config/nds32/constants.md (unspec_element): New enum.
26776 * config/nds32/constraints.md (Umw): New constraint.
26777 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
26778 * config/nds32/nds32-intrinsic.md: Likewise.
26779 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
26780 (nds32_valid_smw_lwm_base_p): New.
26781 (nds32_output_smw_single_word): New.
26782 (nds32_output_lmw_single_word): New.
26783 (nds32_expand_unaligned_load): New.
26784 (nds32_expand_unaligned_store): New.
26785 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
26786 (nds32_output_smw_single_word): Declare.
26787 (nds32_output_lmw_single_word): Declare.
26788 (nds32_expand_unaligned_load): Declare.
26789 (nds32_expand_unaligned_store): Declare.
26790 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
26791 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
26792 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
26793 NDS32_BUILTIN_UASTORE_DW.
26794 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
26797 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
26798 Kito Cheng <kito.cheng@gmail.com>
26799 Chung-Ju Wu <jasonwucj@gmail.com>
26801 * config/nds32/nds32-intrinsic.c
26802 (nds32_expand_builtin_null_ftype_reg): Delete.
26803 (nds32_expand_builtin_reg_ftype_imm): Ditto.
26804 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
26805 (nds32_read_argument): New.
26806 (nds32_legitimize_target): Ditto.
26807 (nds32_legitimize_argument): Ditto.
26808 (nds32_check_constant_argument): Ditto.
26809 (nds32_expand_unop_builtin): Ditto.
26810 (nds32_expand_unopimm_builtin): Ditto.
26811 (nds32_expand_binop_builtin): Ditto.
26812 (nds32_builtin_decl_impl): Ditto.
26813 (builtin_description): Ditto.
26814 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
26815 (nds32_init_builtins_impl): Ditto.
26816 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
26817 (nds32_builtin_decl): New.
26818 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
26819 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
26821 2018-03-02 Jeff Law <law@redhat.com>
26823 * reorg.c (stop_search_p): Handle DEBUG_INSN.
26824 (redundant_insn, fill_simple_delay_slots): Likewise.
26825 (fill_slots_from_thread): Likewise.
26826 * resource.c (mark_referenced_resources): Likewise.
26827 (mark_set_resources, find_dead_or_set_registers): Likewise.
26829 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26831 * substring-locations.h (format_warning_va): Formatting fix for
26832 ATTRIBUTE_GCC_DIAG.
26833 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
26835 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
26836 * substring-locations.c: Include intl.h.
26837 (format_warning_va): Turned into small wrapper around
26838 format_warning_n_va, renamed to ...
26839 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
26840 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
26842 (format_warning_at_substring_n): New function.
26843 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
26844 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
26845 format_warning_at_substring with just a shorter name instead of
26846 const function pointer.
26847 (fmtwarn_n): New function.
26848 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
26849 appropriate, get rid of all the fmtstr temporaries, move conditionals
26850 with G_() wrapped string literals directly into fmtwarn arguments,
26851 cast dir.len to (int), formatting fixes.
26853 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
26855 * doc/invoke.texi: Remove "Cilk Plus" references.
26857 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26858 Richard Biener <rguenther@suse.de>
26861 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
26862 for error or warning attributes if CALL_FROM_THUNK_P is set.
26865 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26868 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
26869 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
26872 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
26873 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
26875 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
26876 instead of -1U in last predictors element's probability member.
26878 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
26881 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
26882 arguments if they are comparable.
26884 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26886 PR tree-optimization/84634
26887 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
26888 masks and masked_loop_p with a single loop_masks, making sure it's
26889 null for bb vectorization.
26891 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26893 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
26894 (vect_analyze_data_ref_access): Use loop->safe_len rather than
26895 loop->force_vectorize to check whether there is no alias.
26897 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26900 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
26902 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
26904 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
26905 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
26906 instead of a loop around prev_real_insn.
26907 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
26910 PR inline-asm/84625
26911 * config/i386/i386.c (ix86_print_operand): Use conditional
26912 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
26915 2018-03-02 Richard Biener <rguenther@suse.de>
26917 PR tree-optimization/84427
26918 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
26919 (bitmap_set_subtract_values): Rewrite to handle multiple
26922 (prune_clobbered_mems): Likewise.
26923 (phi_translate): Take edge instead of pred/phiblock.
26924 (phi_translate_1): Likewise.
26925 (phi_translate_set): Likewise. Insert all translated
26926 exprs for a value into the set, keeping possibly multiple
26927 expressions per value.
26928 (compute_antic_aux): Adjust for phi_translate changes.
26929 When intersecting union the expressions and prune those
26930 not in the final value set, keeping possibly multiple
26931 expressions per value. Do not use value-insertion
26932 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
26933 all expressions. Add verification that the value-sets
26934 only shrink during iteration.
26935 (compute_partial_antic_aux): Adjust for the phi_translate changes.
26936 (do_pre_regular_insertion): Likewise.
26937 (do_pre_partial_partial_insertion): Likewise.
26939 2018-03-02 Richard Biener <rguenther@suse.de>
26942 * config/darwin.c (saved_debug_info_level): New static global.
26943 (darwin_asm_lto_start): Disable debug info generation for LTO out.
26944 (darwin_asm_lto_end): Restore debug info generation settings.
26946 2018-03-01 Martin Liska <mliska@suse.cz>
26949 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
26950 volatile arguments.
26952 2018-03-01 Richard Biener <rguenther@suse.de>
26955 * dwarf2out.c (gen_variable_die): Properly handle late VLA
26956 type annotation with LTO when debug was disabled at compile-time.
26958 2018-03-01 Matthew Fortune <mfortune@gmail.com>
26960 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
26962 (mips_final_postscan_insn): Likewise.
26964 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
26966 PR rtl-optimization/84528
26967 * alias.c (init_alias_target): Add commentary.
26968 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
26969 a unique base value if the frame pointer is not eliminated
26970 to the stack pointer.
26972 2018-03-01 Tom de Vries <tom@codesourcery.com>
26974 PR rtl-optimization/83327
26975 * lra-int.h (hard_regs_spilled_into): Declare.
26976 * lra.c (hard_regs_spilled_into): Define.
26977 (init_reg_info): Init hard_regs_spilled_into.
26978 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
26979 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
26980 (process_bb_lives): Handle hard_regs_spilled_into.
26981 (lra_create_live_ranges_1): Before doing liveness propagation, clear
26982 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
26984 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
26986 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
26987 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
26988 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
26989 * config/rs6000/aix72.h: New file.
26991 2018-02-28 Jakub Jelinek <jakub@redhat.com>
26993 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
26994 instead of warning_at with conditional singular and plural messages
26998 * stor-layout.c (update_alignment_for_field): For
26999 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
27000 && !DECL_PACKED (field), do the alignment update, just use
27001 only desired_align instead of MAX (type_align, desired_align)
27003 (place_field): Don't do known_align < desired_align handling
27004 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
27005 is non-NULL, instead do it after rli->prev_field handling and
27006 only if not within a bitfield word. For DECL_PACKED (field)
27007 use type_align of BITS_PER_UNIT.
27009 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
27011 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
27012 superfluous parentheses and trailing spaces.
27014 2018-02-28 Richard Biener <rguenther@suse.de>
27016 PR tree-optimization/84584
27017 * graphite-scop-detection.c (scop_detection::add_scop): Discard
27018 SCoPs with fake exit edge.
27020 2018-02-28 Martin Liska <mliska@suse.cz>
27023 * timevar.c (timer::print): Fix format to properly print 100%
27026 2018-02-28 Richard Biener <rguenther@suse.de>
27028 PR middle-end/84607
27029 * genmatch.c (capture_info::walk_match): Do not mark
27030 captured expressions without operands as expr_p given
27031 they act more like predicates and should be subject to
27032 "lost tail" side-effect preserving.
27034 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
27036 PR rtl-optimization/81611
27037 * auto-inc-dec.c (attempt_change): Move dead note from
27038 mem_insn if it's the next use of regno
27039 (find_address): Take address use of reg holding
27040 non-incremented value. Add parm to limit search to the named
27042 (merge_in_block): Attempt to use a mem insn that is the next
27043 use of the original regno.
27045 2018-02-27 Martin Sebor <msebor@redhat.com>
27048 * doc/invoke.texi (-Wmissing-attributes): New option.
27049 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
27051 2018-02-27 Martin Sebor <msebor@redhat.com>
27053 PR translation/84207
27054 * diagnostic-core.h (warning_n, error_n, inform_n): Change
27055 n argument to unsigned HOST_WIDE_INT.
27056 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
27057 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
27058 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
27059 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
27061 2018-02-27 Richard Biener <rguenther@suse.de>
27063 PR tree-optimization/84512
27064 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
27065 Do not use the estimate returned from record_stmt_cost for
27066 the scalar iteration cost but sum properly using add_stmt_cost.
27068 2018-02-27 Richard Biener <rguenther@suse.de>
27070 PR tree-optimization/84466
27071 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27072 Adjust last change to less strictly validate use operands.
27074 2018-02-27 Martin Liska <mliska@suse.cz>
27076 PR gcov-profile/84548
27077 * gcov.c (process_file): Allow partial overlap and consider it
27078 also as group functions.
27079 (output_lines): Properly calculate range of lines for a group.
27081 2018-02-27 Martin Liska <mliska@suse.cz>
27083 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
27084 'ggc' suffixes. Change first column width.
27085 (timer::print): Fix formatting of the column.
27087 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
27089 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
27090 preserve inline entry blocks for the sake of debug inline
27091 entry point markers alone.
27092 (remove_unused_locals): Suggest in comments a better place to
27093 force the preservation of inline entry blocks that are
27094 otherwise unused, but do not preserve them.
27096 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27098 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
27100 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27103 * config/i386/constraints.md (Bs): Replace
27104 ix86_indirect_branch_register with
27105 TARGET_INDIRECT_BRANCH_REGISTER.
27107 * config/i386/i386.md (indirect_jump): Likewise.
27108 (tablejump): Likewise.
27109 (*sibcall_memory): Likewise.
27110 (*sibcall_value_memory): Likewise.
27111 Peepholes of indirect call and jump via memory: Likewise.
27112 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
27113 (*sibcall_value_GOT_32): Likewise.
27114 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27115 (GOT_memory_operand): Likewise.
27116 (call_insn_operand): Likewise.
27117 (sibcall_insn_operand): Likewise.
27118 (GOT32_symbol_operand): Likewise.
27119 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
27121 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
27123 PR rtl-optimization/83496
27124 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
27125 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
27126 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
27127 redundant insn, if any.
27128 (relax_delay_slots): Likewise.
27129 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
27131 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
27133 PR tree-optimization/83965
27134 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
27135 that grouped statements are part of a reduction chain. Return
27136 true if the statement is not marked as a reduction itself but
27137 is part of a group.
27138 (vect_recog_dot_prod_pattern): Don't check whether the statement
27139 is part of a group here.
27140 (vect_recog_sad_pattern): Likewise.
27141 (vect_recog_widen_sum_pattern): Likewise.
27143 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
27146 * final.c (rest_of_clean_state): Also look for calls inside sequences.
27148 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27151 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
27153 (ix86_output_indirect_function_return): New prototype.
27154 (ix86_split_simple_return_pop_internal): Likewise.
27155 * config/i386/i386.c (indirect_return_via_cx): New.
27156 (indirect_return_via_cx_bnd): Likewise.
27157 (indirect_thunk_name): Handle return va CX_REG.
27158 (output_indirect_thunk_function): Create alias for
27159 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
27160 (ix86_output_indirect_jmp): Remove the bool argument.
27161 (ix86_output_indirect_function_return): New function.
27162 (ix86_split_simple_return_pop_internal): Likewise.
27163 * config/i386/i386.md (*indirect_jump): Don't pass false
27164 to ix86_output_indirect_jmp.
27165 (*tablejump_1): Likewise.
27166 (simple_return_pop_internal): Change it to define_insn_and_split.
27167 Call ix86_split_simple_return_pop_internal to split it for
27168 -mfunction-return=.
27169 (simple_return_indirect_internal): Call
27170 ix86_output_indirect_function_return instead of
27171 ix86_output_indirect_jmp.
27173 2018-02-26 Jakub Jelinek <jakub@redhat.com>
27176 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
27177 memset and value initialization afterwards.
27179 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
27181 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
27183 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27186 * common/config/aarch64/aarch64-common.c
27187 (aarch_option_optimization_table[]): Switch
27188 off fomit-frame-pointer
27190 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
27191 Chung-Ju Wu <jasonwucj@gmail.com>
27193 * config/nds32/nds32-multiple.md (load_multiple): Disallow
27195 (store_multiple): Ditto.
27197 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
27199 * config.gcc: Add --with-cpu support for nds32 target.
27200 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
27201 * config/nds32/nds32.opt: Add -mcpu= option.
27203 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
27205 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
27206 isel=yes): Warn for these deprecated options.
27208 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
27210 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
27211 ISA_2_5_MASKS_EMBEDDED.
27213 2018-02-23 Jakub Jelinek <jakub@redhat.com>
27215 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
27216 p->max as pointers rather than using iterative_hash_expr.
27218 2018-02-23 Carl Love <cel@us.ibm.com>
27220 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
27221 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
27223 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
27224 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
27225 P8V_BUILTIN_VEC_VUNSIGNED2.
27227 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
27230 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
27231 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
27233 * lra-constraints.c (curr_insn_transform): Set up
27234 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
27236 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
27237 * lra-eliminations.c (spill_pseudos): Ditto.
27238 (process_insn_for_elimination): Ditto.
27239 * lra-lives.c (reg_early_clobber_p): Use the new macros.
27240 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
27241 LRA_NON_CLOBBERED_ALT.
27243 2018-02-22 Martin Sebor <msebor@redhat.com>
27245 PR tree-optimization/84480
27246 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
27247 to maybe_diag_stxncpy_trunc. Call it.
27248 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
27249 from gimple_fold_builtin_strcpy. Print inlining stack.
27250 (handle_builtin_stxncpy): Print inlining stack.
27251 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
27253 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
27256 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
27257 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
27258 and -fcheck-pointer-bounds are used together.
27259 (indirect_thunk_prefix): New enum.
27260 (indirect_thunk_need_prefix): New function.
27261 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
27262 "_nt" instead of "_bnd" for NOTRACK prefix.
27263 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
27264 (output_indirect_thunk_function): Likewise.
27266 (ix86_code_end): Update output_indirect_thunk_function calls.
27267 (ix86_output_indirect_branch_via_reg): Replace
27268 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
27269 (ix86_output_indirect_branch_via_push): Likewise.
27270 (ix86_output_function_return): Likewise.
27271 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
27272 incompatible with -fcf-protection=branch and
27273 -fcheck-pointer-bounds.
27275 2018-02-22 Steve Ellcey <sellcey@cavium.com>
27278 * config/aarch64/aarch64.c (aarch64_print_address_internal):
27279 Change gcc_assert call to output_operand_lossage.
27281 2018-02-22 Steve Ellcey <sellcey@cavium.com>
27283 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
27285 2018-02-22 DJ Delorie <dj@redhat.com>
27286 Sebastian Perta <sebastian.perta@renesas.com>
27287 Oleg Endo <olegendo@gcc.gnu.org>
27289 * config/rx/rx.c (rx_rtx_costs): New function.
27290 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
27292 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27294 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
27296 2018-02-22 Martin Liska <mliska@suse.cz>
27299 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
27300 Add "native" as a possible value.
27302 2018-02-22 Martin Liska <mliska@suse.cz>
27305 * config/i386/i386.c (ix86_option_override_internal):
27306 Add "native" as a possible value for -march and -mtune.
27308 2018-02-22 Jakub Jelinek <jakub@redhat.com>
27311 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
27312 to all type variants.
27314 PR tree-optimization/84503
27315 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
27316 width as info->bitpos + info->bitsize - start.
27317 (merged_store_group::merge_overlapping): Simplify width computation.
27318 (check_no_overlap): New function.
27319 (imm_store_chain_info::try_coalesce_bswap): Compute expected
27320 start + width and last_order of the group, fail if check_no_overlap
27322 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
27323 to group if check_no_overlap fails.
27325 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
27327 * config/rs6000/altivec.md: Delete contraint arguments to
27328 define_expand, define_split, and define_peephole2, and in
27329 define_insn_and_split if always unused.
27330 * config/rs6000/darwin.md: Ditto.
27331 * config/rs6000/dfp.md: Ditto.
27332 * config/rs6000/rs6000.md: Ditto.
27333 * config/rs6000/sync.md: Ditto.
27334 * config/rs6000/vector.md: Ditto.
27335 * config/rs6000/vsx.md: Ditto.
27337 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
27339 * config/rs6000/altivec.md: Write output control strings as braced
27340 blocks instead of double-quoted strings.
27341 * config/rs6000/darwin.md: Ditto.
27342 * config/rs6000/rs6000.md: Ditto.
27343 * config/rs6000/vector.md: Ditto.
27344 * config/rs6000/vsx.md: Ditto.
27346 2018-02-21 Jason Merrill <jason@redhat.com>
27348 PR c++/84314 - ICE with templates and fastcall attribute.
27349 * attribs.c (build_type_attribute_qual_variant): Remove assert.
27351 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
27353 * ipa-cp.c (determine_versionability): Fix comment typos.
27355 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
27358 * ipa-cp.c (determine_versionability): Do not version functions caling
27361 2018-02-21 Martin Liska <mliska@suse.cz>
27364 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
27365 Add "native" as a possible value.
27366 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
27367 the macro when native cpu detection is available.
27369 2018-02-21 Martin Liska <mliska@suse.cz>
27372 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
27373 Add "native" as a possible value.
27374 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
27375 when native cpu detection is available.
27377 2018-02-21 Jakub Jelinek <jakub@redhat.com>
27378 Martin Sebor <msebor@redhat.com>
27380 PR tree-optimization/84478
27381 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
27383 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
27384 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
27385 support which is conservatively correct, for 2 only stay conservative
27386 for maxlen. Formatting and comment capitalization fixes. Add STRICT
27387 argument to the 2 argument get_range_strlen, adjust 6 arg
27388 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
27390 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
27391 (gimple_fold_builtin_strlen): Pass true as last argument to
27394 2018-02-20 Martin Sebor <msebor@redhat.com>
27396 PR middle-end/84095
27397 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
27398 (builtin_memref::set_base_and_offset): Same. Handle inner references.
27399 (builtin_memref::builtin_memref): Factor out parts into
27400 set_base_and_offset and call it.
27402 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
27404 PR middle-end/84406
27405 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
27406 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
27407 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
27408 search at the associated MODE_INT.
27410 2018-02-20 Jeff Law <law@redhat.com>
27412 PR middle-end/82123
27413 PR tree-optimization/81592
27414 PR middle-end/79257
27415 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
27416 for range data rather than using global data.
27417 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
27418 range data rather than using global data.
27419 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
27420 pass it to children as needed.
27421 (struct directive::fmtresult): Similarly.
27422 (struct directive::set_width): Similarly.
27423 (struct directive::set_precision): Similarly.
27424 (format_integer, format_directive, parse_directive): Similarly.
27425 (format_none): Accept unnamed vr_values parameter.
27426 (format_percent, format_floating, format_character): Similarly.
27427 (format_string, format_plain): Similarly.
27428 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
27429 the EVRP range analyzer for range data rather than using global data.
27430 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
27431 gimple-ssa-evrp-analyze.h
27432 (class sprintf_dom_walker): Add after_dom_children member function.
27433 Add evrp_range_analyzer member.
27434 (sprintf_dom_walker::before_dom_children): Call into the EVRP
27435 range analyzer as needed.
27436 (sprintf_dom_walker::after_dom_children): New member function.
27437 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
27439 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
27440 (evrp_range_analyzer::pop_to_marker): Likewise.
27442 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
27444 PR tree-optimization/84419
27445 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
27446 with the required type if its current type is compatible but
27449 2018-02-20 Jakub Jelinek <jakub@redhat.com>
27451 PR middle-end/82004
27452 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
27453 after vectorization.
27455 2018-02-20 Martin Liska <mliska@suse.cz>
27458 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
27459 possible values if we don't have a hint.
27461 2018-02-20 Martin Liska <mliska@suse.cz>
27465 * final.c (shorten_branches): Build align_tab array with one
27467 * opts.c (finish_options): Add alignment option limit check.
27468 (MAX_CODE_ALIGN): Likewise.
27469 (MAX_CODE_ALIGN_VALUE): Likewise.
27470 * doc/invoke.texi: Document maximum allowed option value for
27471 all -falign-* options.
27473 2018-02-19 Jakub Jelinek <jakub@redhat.com>
27476 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
27477 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
27478 * var-tracking.c (emit_note_insn_var_location): Remove all references
27479 to NOTE_INSN_CALL_ARG_LOCATION.
27480 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
27481 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
27482 Use copy_rtx_if_shared.
27483 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
27484 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
27485 (dwarf2out_var_location): Remove handling of
27486 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
27488 * final.c (final_scan_insn): Remove all references to
27489 NOTE_INSN_CALL_ARG_LOCATION.
27490 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
27491 before dumping final insns.
27492 * except.c (emit_note_eh_region_end): Remove all references to
27493 NOTE_INSN_CALL_ARG_LOCATION.
27494 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
27495 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
27496 * config/arc/arc.c (hwloop_optimize): Likewise.
27497 * config/arm/arm.c (create_fix_barrier): Likewise.
27498 * config/s390/s390.c (s390_chunkify_start): Likewise.
27499 * config/sh/sh.c (find_barrier): Likewise.
27500 * config/i386/i386.c (rest_of_insert_endbranch,
27501 ix86_seh_fixup_eh_fallthru): Likewise.
27502 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
27503 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
27504 * config/frv/frv.c (frv_function_prologue): Likewise.
27505 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
27507 (note_outside_basic_block_p): Remove all references to
27508 NOTE_INSN_CALL_ARG_LOCATION.
27509 * gengtype.c (adjust_field_rtx_def): Likewise.
27510 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
27512 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
27513 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
27516 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
27519 PR tree-optimization/84452
27520 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
27521 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
27524 2018-02-19 Martin Liska <mliska@suse.cz>
27527 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
27529 2018-02-19 Martin Liska <mliska@suse.cz>
27530 Richard Sandiford <richard.sandiford@linaro.org>
27532 PR tree-optimization/82491
27533 * gimple-fold.c (get_base_constructor): Make earlier bail out
27536 2018-02-19 Carl Love <cel@us.ibm.com>
27538 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
27539 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
27541 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
27542 P8V_BUILTIN_VEC_NEG.
27544 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
27546 * config/rl78/rl78.md (movdf): New define expand.
27548 2018-02-19 Martin Liska <mliska@suse.cz>
27551 * doc/invoke.texi: Fix typo.
27552 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
27554 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
27556 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
27557 handle rs6000_single_float and rs6000_double_float specially for
27560 2018-02-16 Jeff Law <law@redhat.com>
27562 * config/rx/rx.c (add_pop_cfi_notes): New function.;
27563 (pop_regs): Use it.
27565 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27568 * ipa-inline.c (inline_small_functions): Fix a typo.
27570 2018-02-16 Nathan Sidwell <nathan@acm.org>
27572 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
27574 2018-02-16 Carl Love <cel@us.ibm.com>
27576 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
27577 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
27578 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
27579 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
27580 expansion to P8V_BUILTIN_VEC_FLOAT2.
27582 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
27584 PR rtl-optimization/70023
27585 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
27586 src_regno into account.
27588 2018-02-16 Carl Love <cel@us.ibm.com>
27590 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
27591 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
27592 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
27593 * config/rs6000/rs6000.c: Remove case statements for
27594 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
27595 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
27596 and P9V_BUILTIN_VEC_VINSERT4B.
27597 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
27598 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
27599 * config/rs6000/vsx.md:
27600 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
27603 2018-02-16 Carl Love <cel@us.ibm.com>
27605 * config/rs6000/altivec.h: Add builtin names vec_extract4b
27607 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
27609 * config/rs6000/rs6000-c.c: Add the definitions for
27610 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
27611 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
27612 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
27613 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
27614 definition for insert4b and define insn *insert3b_internal.
27615 * doc/extend.texi: Add documentation for vec_extract4b.
27617 2018-02-16 Nathan Sidwell <nathan@acm.org>
27619 * doc/extend.texi (Backwards Compatibility): Mention friend
27620 injection. Note for-scope is deprecated.
27621 * doc/invoke.texi (-ffriend-injection): Deprecate.
27623 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
27625 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
27626 that moved to I2, also allow destinations that are a paradoxical
27627 subreg (instead of a normal reg).
27629 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
27632 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
27635 2018-02-16 Richard Biener <rguenther@suse.de>
27637 PR tree-optimization/84037
27638 PR tree-optimization/84016
27640 * config/i386/i386.c (ix86_builtin_vectorization_cost):
27641 Adjust vec_construct for the fact we need additional higher latency
27642 128bit inserts for AVX256 and AVX512 vector builds.
27643 (ix86_add_stmt_cost): Scale vector construction cost for
27646 2018-02-16 Richard Biener <rguenther@suse.de>
27648 PR tree-optimization/84417
27649 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
27650 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
27651 (non_rewritable_lvalue_p): Likewise, use poly-ints.
27653 2018-02-16 Martin Liska <mliska@suse.cz>
27656 * internal-fn.def (ASAN_CHECK): Set proper flags.
27657 (ASAN_MARK): Likewise.
27659 2018-02-16 Julia Koval <julia.koval@intel.com>
27661 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
27662 from PTA_CANNONLAKE.
27664 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27667 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
27668 Use ++iter rather than iter++ for std::list iterators.
27669 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
27670 defer deleting them until all nodes in the forest are processed. Do
27671 free even leaf nodes. Change to_process into auto_vec.
27674 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
27675 * vec.h (vec_default_construct): Use memset instead of placement new
27676 if BROKEN_VALUE_INITIALIZATION is defined.
27677 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
27678 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
27681 PR rtl-optimization/83723
27682 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
27683 * lra.c (lra_substitute_pseudo): Likewise. If true, use
27684 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
27686 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
27688 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
27690 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
27692 PR rtl-optimization/81443
27693 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
27694 from inner REGs to paradoxical SUBREGs.
27696 2018-02-16 Richard Biener <rguenther@suse.de>
27698 PR tree-optimization/84399
27699 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27700 For operands we can analyze at their definition make sure we can
27701 analyze them at each use as well.
27703 2018-02-16 Richard Biener <rguenther@suse.de>
27705 PR tree-optimization/84190
27706 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
27707 volatile accesses if the decl isn't volatile.
27709 2018-02-15 Jason Merrill <jason@redhat.com>
27711 PR c++/84314 - ICE with templates and fastcall attribute.
27712 * attribs.c (build_type_attribute_qual_variant): Don't clobber
27713 TYPE_CANONICAL on an existing type.
27715 2018-02-15 Jakub Jelinek <jakub@redhat.com>
27717 PR tree-optimization/84383
27718 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
27719 dstoff nor call operand_equal_p if dstbase is NULL.
27721 PR tree-optimization/84334
27722 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
27723 also a CONSTANT_CLASS_P, punt.
27725 2018-02-14 Jim Wilson <jimw@sifive.com>
27727 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
27728 first SMALL_OPERAND check. New local min_second_step. Move assert
27729 to where locals are set. Add TARGET_RVC support.
27730 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
27732 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
27734 * doc/invoke.texi: Correct -Wformat-overflow code sample.
27736 2018-02-14 Martin Sebor <msebor@redhat.com>
27738 PR tree-optimization/83698
27739 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
27740 arrays constrain the offset range to their bounds.
27741 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
27742 (builtin_access::overlap): Avoid setting the size of overlap if it's
27744 (maybe_diag_overlap): Also consider arrays when deciding what values
27745 of offsets to include in diagnostics.
27747 2018-02-14 Martin Sebor <msebor@redhat.com>
27750 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
27751 that correspond to the kind of a declaration.
27753 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
27756 * config/pa/pa.md: Load address of PIC label using the linkage table
27757 if the label is nonlocal.
27759 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
27761 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
27762 warning message if user requests -maltivec=be.
27763 * doc/invoke.texi: Document deprecation of -maltivec=be.
27765 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
27768 * config/rs6000/rs6000-c.c: Update definitions for
27769 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
27770 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
27772 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27775 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
27776 add _get_ssp intrinsics. Remove argument from
27777 __builtin_ia32_rdssp[d|q].
27778 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
27779 * config/i386/i386-builtin.def: Remove argument from
27780 __builtin_ia32_rdssp[d|q].
27781 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
27782 ix86_expand_special_args_builtin for _rdssp[d|q].
27783 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
27784 Clear register before usage.
27785 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
27786 Add documentation for new _get_ssp and _inc_ssp intrinsics.
27788 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
27790 PR tree-optimization/84357
27791 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
27792 operand 1 of an ARRAY_REF too.
27794 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
27797 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
27798 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27800 (set_of_reg): New struct.
27801 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
27802 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
27803 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27805 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
27806 Split into bitclr, bitset, bitinvert patterns if appropriate.
27807 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
27808 use rx_fuse_in_memory_bitop.
27809 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
27810 to named insn, correct maximum insn length.
27812 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
27815 * machmode.def: Define a complex mode for PARTIAL_INT.
27816 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
27818 * doc/rtl.texi: Document CSPImode.
27819 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
27821 (msp430_hard_regno_nregs_with_padding): Likewise.
27823 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
27826 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
27828 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27830 PR rtl-optimization/84169
27831 * combine.c (try_combine): New variable split_i2i3. Set it to true if
27832 we generated a parallel as new i3 and we split that to new i2 and i3
27833 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
27834 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
27835 those to i2, not i1. Partially rewrite this scan code.
27837 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27840 * stor-layout.c (place_field): For variable length fields, adjust
27841 offset_align afterwards not just based on the field's alignment,
27842 but also on the size.
27844 PR middle-end/84309
27845 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
27846 of exps and logs in the use_exp2 case.
27848 2018-02-13 Jeff Law <law@redhat.com>
27850 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
27851 entry for "vector".
27853 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
27856 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
27860 * common.opt (gas-loc-support, gas-locview-support): New.
27861 (ginline-points, ginternal-reset-location-views): New.
27862 * doc/invoke.texi: Document them. Use @itemx where intended.
27863 (gvariable-location-views): Adjust.
27864 * target.def (reset_location_view): New.
27865 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
27866 (TARGET_RESET_LOCATION_VIEW): New.
27867 * doc/tm.texi: Rebuilt.
27868 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
27869 (dwarf2out_default_as_locview_support): New.
27870 (output_asm_line_debug_info): Use option variables.
27871 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
27872 (output_loc_list): Likewise.
27873 (add_high_low_attributes): Check option variables.
27874 Don't output entry view attribute in strict mode.
27875 (gen_inlined_subroutine_die): Check option variables.
27876 (dwarf2out_inline_entry): Likewise.
27877 (init_sections_and_labels): Likewise.
27878 (dwarf2out_early_finish): Likewise.
27879 (maybe_reset_location_view): New, from...
27880 (dwarf2out_var_location): ... here. Call it.
27881 * debug.h (dwarf2out_default_as_loc_support): Declare.
27882 (dwarf2out_default_as_locview_support): Declare.
27883 * hooks.c (hook_int_rtx_insn_0): New.
27884 * hooks.h (hook_int_rtx_insn_0): Declare.
27885 * toplev.c (process_options): Take -gas-loc-support and
27886 -gas-locview-support from dwarf2out. Enable
27887 -gvariable-location-views by default only with locview
27888 assembler support. Enable -ginternal-reset-location-views by
27889 default only if the target defines the corresponding hook.
27890 Enable -ginline-points by default if location views are
27891 enabled; force it disabled if statement frontiers are
27893 * tree-inline.c (expand_call_inline): Check option variables.
27894 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
27896 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
27898 PR tree-optimization/84321
27899 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
27900 handling. Also check whether the anti-range contains any values
27901 that satisfy the mask; switch to a VR_RANGE if not.
27903 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
27906 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
27908 2018-02-13 Martin Jambor <mjambor@suse.cz>
27911 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
27912 of call statements, also set location of a load to a temporary.
27914 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
27916 * config/rl78/rl78.c (add_vector_labels): New function.
27917 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
27918 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
27919 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
27920 which checks that no arguments are passed.
27921 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
27922 * doc/extend.texi: Documentation for the new attribute.
27924 2018-02-13 Andreas Schwab <schwab@suse.de>
27926 * config/riscv/linux.h (CPP_SPEC): Define.
27928 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27931 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
27932 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
27933 OPTION_MASK_ISA_AES as first argument to def_builtin_const
27934 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
27935 instead of OPTION_MASK_ISA_PCLMUL as first argument to
27936 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
27937 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
27938 temporarily for AES and PCLMUL builtins.
27940 PR tree-optimization/84339
27941 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
27942 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
27945 PR middle-end/84309
27946 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
27947 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
27948 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
27950 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
27952 * omp-simd-clone.h: New file.
27953 * omp-simd-clone.c: Include omp-simd-clone.h.
27954 (expand_simd_clones): No longer static.
27955 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
27956 cgraph.h and omp-simd-clone.h.
27957 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
27958 (vect_recog_widen_shift_pattern): Formatting fix.
27959 (vect_pattern_recog_1): Don't check optab for calls.
27962 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
27963 operands[2] into a REG before using gen_lowpart on it.
27965 2018-02-12 Jeff Law <law@redhat.com>
27968 * config/sh/sh.c (find_barrier): Consider a sibling call
27971 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
27972 successfully back substituting a reg.
27974 2018-02-12 Richard Biener <rguenther@suse.de>
27976 PR tree-optimization/84037
27977 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
27978 parameter, move visited init to caller.
27979 (vect_slp_analyze_operations): Separate cost from validity
27980 check, initialize visited once for all instances.
27981 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
27983 * tree-vect-stmts.c (vect_model_simple_cost): Make early
27985 (vect_model_promotion_demotion_cost): Likewise.
27986 (vectorizable_bswap): Guard cost modeling with !slp_node
27987 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
27989 (vectorizable_call): Likewise.
27990 (vectorizable_conversion): Likewise.
27991 (vectorizable_assignment): Likewise.
27992 (vectorizable_shift): Likewise.
27993 (vectorizable_operation): Likewise.
27994 (vectorizable_store): Likewise.
27995 (vectorizable_load): Likewise.
27996 (vectorizable_condition): Likewise.
27997 (vectorizable_comparison): Likewise.
27999 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
28002 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
28003 (ASAN_MARK): Fix fnspec to account for return value, change pointer
28004 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
28006 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
28008 PR middle-end/83665
28009 * params.def (inline-min-speedup): Increase from 8 to 15.
28010 (max-inline-insns-auto): Decrease from 40 to 30.
28011 * ipa-split.c (consider_split): Add some buffer for function to
28012 be considered inlining candidate.
28013 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
28016 2018-02-12 Richard Biener <rguenther@suse.de>
28018 PR tree-optimization/84037
28019 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
28020 matched stmts if we cannot swap the non-matched ones.
28022 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
28024 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
28025 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
28026 _mm_maskz_scalef_round_ss): New intrinsics.
28027 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
28028 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
28029 __builtin_ia32_scalefss_round): Remove.
28030 (__builtin_ia32_scalefsd_mask_round,
28031 __builtin_ia32_scalefss_mask_round): New intrinsics.
28032 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
28033 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
28034 ((match_operand:VF_128 2 "<round_nimm_predicate>"
28035 "<round_constraint>")): Changed to ...
28036 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
28037 "<round_scalar_constraint>")): ... this.
28038 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
28039 %0, %1, %2<round_op3>}"): Changed to ...
28040 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
28041 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
28042 %2<round_scalar_mask_op3>}"): ... this.
28043 * config/i386/subst.md (round_scalar_nimm_predicate): New.
28045 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
28047 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
28048 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
28049 (_mm_maskz_sqrt_round_ss): New intrinsics.
28050 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
28051 (__builtin_ia32_sqrtsd_mask_round)
28052 (__builtin_ia32_sqrtss_mask_round): New builtins.
28053 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
28054 (__builtin_ia32_sqrtss_round): Remove.
28055 (__builtin_ia32_sqrtsd_mask_round)
28056 (__builtin_ia32_sqrtss_mask_round): New builtins.
28057 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
28058 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
28059 ((match_operand:VF_128 1 "vector_operand"
28060 "xBm,<round_constraint>")): Changed to ...
28061 ((match_operand:VF_128 1 "vector_operand"
28062 "xBm,<round_scalar_constraint>")): ... this.
28063 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
28064 %0, %2, %<iptr>1<round_op3>}): Changed to ...
28065 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
28066 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
28067 %<iptr>1<round_scalar_mask_op3>}): ... this.
28068 ((set_attr "prefix" "<round_prefix>")): Changed to ...
28069 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
28071 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
28074 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
28075 Cast vec_cmpeq result to correct type.
28076 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
28077 Cast vec_cmpgt result to correct type.
28079 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
28081 * final.c (final_scan_insn_1): Renamed from...
28082 (final_scan_insn): ... this. New wrapper, to recover
28083 seen from the outermost call in recursive ones.
28084 * config/sparc/sparc.c (output_return): Drop seen from call.
28085 (output_sibcall): Likewise.
28086 * config/visium/visium.c (output_branch): Likewise.
28088 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
28090 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
28093 2018-02-10 Alan Modra <amodra@gmail.com>
28096 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
28097 Specify LR as an input.
28099 2018-02-10 Jakub Jelinek <jakub@redhat.com>
28102 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
28103 remove_member_access_dummy_vars): New functions.
28104 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
28105 lower_omp_1, execute_lower_omp): Use them.
28107 PR rtl-optimization/84308
28108 * shrink-wrap.c (spread_components): Release todo vector.
28110 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
28112 PR rtl-optimization/57193
28113 * ira-color.c (struct allocno_color_data): Add member
28114 conflict_allocno_hard_prefs.
28115 (update_conflict_allocno_hard_prefs): New.
28116 (bucket_allocno_compare_func): Add a preference based on
28117 conflict_allocno_hard_prefs.
28118 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
28119 (color_allocnos): Remove a dead code. Initiate
28120 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
28122 2018-02-09 Jakub Jelinek <jakub@redhat.com>
28125 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
28126 constraint from =wa to wa. Avoid a subreg on the output operand,
28127 instead use a pseudo and subreg it in a move.
28128 (p9_xxbrd_<mode>): Changed to ...
28129 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
28130 (p9_xxbrd_v2df): New expander.
28131 (p9_xxbrw_<mode>): Changed to ...
28132 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
28133 (p9_xxbrw_v4sf): New expander.
28135 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
28137 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
28139 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
28142 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
28143 multiply in 32-bit mode.
28144 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
28145 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
28148 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
28150 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
28151 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
28152 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
28153 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
28155 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
28158 * dwarf2out.c (is_trivial_indirect_ref): New function.
28159 (dwarf2out_late_global_decl): Do not generate a location
28160 attribute for variables that have a non-trivial DECL_VALUE_EXPR
28161 and that are not defined in the current unit.
28163 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
28165 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
28166 instead of a libcall for UNORDERED.
28168 2018-02-09 Tamar Christina <tamar.christina@arm.com>
28171 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
28172 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
28174 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28177 * config/s390/s390.c (s390_set_current_function): Invoke
28178 s390_indirect_branch_settings also if fndecl didn't change.
28180 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
28182 * config/rs6000/rs6000.md (blockage): Set length to zero.
28184 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
28186 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
28188 2018-02-09 Jakub Jelinek <jakub@redhat.com>
28191 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
28192 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
28196 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
28197 PARALLEL incoming that failed vt_get_decl_and_offset check.
28199 PR middle-end/84237
28200 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
28201 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
28203 (get_variable_section): For decls in named .bss* sections pass true as
28204 second argument to bss_initializer_p.
28206 2018-02-09 Marek Polacek <polacek@redhat.com>
28207 Jakub Jelinek <jakub@redhat.com>
28210 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
28211 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
28212 Sync some changes from cxx_fold_indirect_ref.
28214 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
28216 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
28218 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
28219 (BLOCK_INLINE_ENTRY_LABEL): New.
28220 (dwarf2out_var_location): Disregard inline entry markers.
28221 (inline_entry_data): New struct.
28222 (inline_entry_data_hasher): New hashtable type.
28223 (inline_entry_data_hasher::hash): New.
28224 (inline_entry_data_hasher::equal): New.
28225 (inline_entry_data_table): New variable.
28226 (add_high_low_attributes): Add DW_AT_entry_pc and
28227 DW_AT_GNU_entry_view attributes if a pending entry is found
28228 in inline_entry_data_table. Add old entry_pc attribute only
28229 if debug nonbinding markers are disabled.
28230 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
28231 markers are enabled.
28232 (block_within_block_p, dwarf2out_inline_entry): New.
28233 (dwarf2out_finish): Check that no entries remained in
28234 inline_entry_data_table.
28235 * final.c (reemit_insn_block_notes): Handle inline entry notes.
28236 (final_scan_insn, notice_source_line): Likewise.
28237 (rest_of_clean_state): Skip inline entry markers.
28238 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
28240 * gimple.c (gimple_build_debug_inline_entry): New.
28241 * gimple.h (enum gimple_debug_subcode): Add
28242 GIMPLE_DEBUG_INLINE_ENTRY.
28243 (gimple_build_debug_inline_entry): Declare.
28244 (gimple_debug_inline_entry_p): New.
28245 (gimple_debug_nonbind_marker_p): Adjust.
28246 * insn-notes.def (INLINE_ENTRY): New.
28247 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
28248 inline entry marker notes.
28249 (print_insn): Likewise.
28250 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
28251 (INSN_DEBUG_MARKER_KIND): Likewise.
28252 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
28253 * tree-inline.c (expand_call_inline): Build and insert
28254 debug_inline_entry stmt.
28255 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
28256 inline entry blocks early, if nonbind markers are enabled.
28257 (dump_scope_block): Dump fragment info.
28258 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
28259 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
28260 (gimple_build_debug_inline_entry): New.
28261 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
28262 Enable/disable inline entry points too.
28263 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
28264 (DEBUG_INSN): Describe inline entry markers.
28266 * common.opt (gvariable-location-views): New.
28267 (gvariable-location-views=incompat5): New.
28268 * config.in: Rebuilt.
28269 * configure: Rebuilt.
28270 * configure.ac: Test assembler for view support.
28271 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
28272 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
28273 * dwarf2out.c (var_loc_view): New typedef.
28274 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
28275 (dwarf2out_locviews_in_attribute): New.
28276 (dwarf2out_locviews_in_loclist): New.
28277 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
28278 (enum dw_line_info_opcode): Add LI_adv_address.
28279 (struct dw_line_info_table): Add view.
28280 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
28281 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
28282 (zero_view_p): New variable.
28283 (ZERO_VIEW_P): New macro.
28284 (output_asm_line_debug_info): New.
28285 (struct var_loc_node): Add view.
28286 (add_AT_view_list, AT_loc_list): New.
28287 (add_var_loc_to_decl): Add view param. Test it against last.
28288 (new_loc_list): Add view params. Record them.
28289 (AT_loc_list_ptr): Handle loc and view lists.
28290 (view_list_to_loc_list_val_node): New.
28291 (print_dw_val): Handle dw_val_class_view_list.
28292 (size_of_die): Likewise.
28293 (value_format): Likewise.
28294 (loc_list_has_views): New.
28295 (gen_llsym): Set vl_symbol too.
28296 (maybe_gen_llsym, skip_loc_list_entry): New.
28297 (dwarf2out_maybe_output_loclist_view_pair): New.
28298 (output_loc_list): Output view list or entries too.
28299 (output_view_list_offset): New.
28300 (output_die): Handle dw_val_class_view_list.
28301 (output_dwarf_version): New.
28302 (output_compilation_unit_header): Use it.
28303 (output_skeleton_debug_sections): Likewise.
28304 (output_rnglists, output_line_info): Likewise.
28305 (output_pubnames, output_aranges): Update version comments.
28306 (output_one_line_info_table): Output view numbers in asm comments.
28307 (dw_loc_list): Determine current endview, pass it to new_loc_list.
28308 Call maybe_gen_llsym.
28309 (loc_list_from_tree_1): Adjust.
28310 (add_AT_location_description): Create view list attribute if
28311 needed, check it's absent otherwise.
28312 (convert_cfa_to_fb_loc_list): Adjust.
28313 (maybe_emit_file): Call output_asm_line_debug_info for test.
28314 (dwarf2out_var_location): Reset views as needed. Precompute
28315 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
28316 attribute. Set view.
28317 (new_line_info_table): Reset next view.
28318 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
28319 (dwarf2out_source_line): Likewise. Output view resets and labels to
28320 the assembler, or select appropriate line info opcodes.
28321 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
28322 (optimize_string_length): Catch it. Adjust.
28323 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
28324 dw_val_class_view_list, and remove it if no longer needed.
28325 (hash_loc_list): Hash view numbers.
28326 (loc_list_hasher::equal): Compare them.
28327 (optimize_location_lists): Check whether a view list symbol is
28328 needed, and whether the locview attribute is present, and
28329 whether they match. Remove the locview attribute if no longer
28331 (index_location_lists): Call skip_loc_list_entry for test.
28332 (dwarf2out_finish): Call output_asm_line_debug_info for test.
28333 Use output_dwarf_version.
28334 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
28335 (struct dw_val_node): Add val_view_list.
28336 * final.c (SEEN_NEXT_VIEW): New.
28337 (set_next_view_needed): New.
28338 (clear_next_view_needed): New.
28339 (maybe_output_next_view): New.
28340 (final_start_function): Rename to...
28341 (final_start_function_1): ... this. Take pointer to FIRST,
28342 add SEEN parameter. Emit param bindings in the initial view.
28343 (final_start_function): Reintroduce SEEN-less interface.
28344 (final): Rename to...
28345 (final_1): ... this. Take SEEN parameter. Output final pending
28346 next view at the end.
28347 (final): Reintroduce seen-less interface.
28348 (final_scan_insn): Output pending next view before switching
28349 sections or ending a block. Mark the next view as needed when
28350 outputting variable locations. Notify debug backend of section
28351 changes, and of location view changes.
28352 (rest_of_handle_final): Adjust.
28353 * toplev.c (process_options): Autodetect value for debug variable
28354 location views option. Warn on incompat5 without -gdwarf-5.
28355 * doc/invoke.texi (gvariable-location-views): New.
28356 (gvariable-location-views=incompat5): New.
28357 (gno-variable-location-views): New.
28359 2018-02-08 David Malcolm <dmalcolm@redhat.com>
28361 PR tree-optimization/84136
28362 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
28363 that the result of find_edge is non-NULL.
28365 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
28368 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
28369 storing integer register in SImode. Fix cost of 256 and 512
28370 byte aligned SSE register store.
28372 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
28374 * config/i386/i386.c (ix86_multiplication_cost): Fix
28375 multiplication cost for TARGET_AVX512DQ.
28377 2018-02-08 Marek Polacek <polacek@redhat.com>
28379 PR tree-optimization/84238
28380 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
28383 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28385 PR tree-optimization/84265
28386 * tree-vect-stmts.c (vectorizable_store): Don't treat
28387 VMAT_CONTIGUOUS accesses as grouped.
28388 (vectorizable_load): Likewise.
28390 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28392 PR tree-optimization/81635
28393 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
28394 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
28395 (test_round_for_mask): New functions.
28396 (wide_int_cc_tests): Call test_round_for_mask.
28397 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
28398 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
28399 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
28400 range returned by get_range_info.
28402 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
28405 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
28406 * symtab.c: Include builtins.h
28407 (symtab_node::output_to_lto_symbol_table_p): Move here
28408 from lto-streamer-out.c:output_symbol_p.
28409 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
28410 (output_symbol_p): Move all logic to symtab.c
28411 (produce_symtab): Update.
28413 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28415 * config/s390/s390-opts.h (enum indirect_branch): Define.
28416 * config/s390/s390-protos.h (s390_return_addr_from_memory)
28417 (s390_indirect_branch_via_thunk)
28418 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
28419 (enum s390_indirect_branch_type): Define.
28420 * config/s390/s390.c (struct s390_frame_layout, struct
28421 machine_function): Remove.
28422 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
28423 (indirect_branch_table_label_no, indirect_branch_table_name):
28425 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
28426 (enum s390_indirect_branch_option): Define.
28427 (s390_return_addr_from_memory): New function.
28428 (s390_handle_string_attribute): New function.
28429 (s390_attribute_table): Add new attribute handler.
28430 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
28431 (s390_indirect_branch_via_thunk): New function.
28432 (s390_indirect_branch_via_inline_thunk): New function.
28433 (s390_function_ok_for_sibcall): When jumping via thunk disallow
28434 sibling call optimization for non z10 compiles.
28435 (s390_emit_call): Force indirect branch target to be a single
28436 register. Add r1 clobber for non-z10 compiles.
28437 (s390_emit_epilogue): Emit return jump via return_use expander.
28438 (s390_reorg): Handle JUMP_INSNs as execute targets.
28439 (s390_option_override_internal): Perform validity checks for the
28440 new command line options.
28441 (s390_indirect_branch_attrvalue): New function.
28442 (s390_indirect_branch_settings): New function.
28443 (s390_set_current_function): Invoke s390_indirect_branch_settings.
28444 (s390_output_indirect_thunk_function): New function.
28445 (s390_code_end): Implement target hook.
28446 (s390_case_values_threshold): Implement target hook.
28447 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
28449 * config/s390/s390.h (struct s390_frame_layout)
28450 (struct machine_function): Move here from s390.c.
28451 (TARGET_INDIRECT_BRANCH_NOBP_RET)
28452 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
28453 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
28454 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
28455 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
28456 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
28457 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
28458 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
28459 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
28460 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
28461 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
28462 (mnemonic attribute): Add values which aren't recognized
28464 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
28465 pattern for branch conversion. Fix mnemonic attribute.
28466 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
28467 indirect branch via thunk if requested.
28468 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
28469 ("*indirect_jump"): Disable for branch conversion using out of
28471 ("indirect_jump_via_thunk<mode>_z10")
28472 ("indirect_jump_via_thunk<mode>")
28473 ("indirect_jump_via_inlinethunk<mode>_z10")
28474 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
28475 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
28476 ("casesi_jump_via_inlinethunk<mode>_z10")
28477 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
28478 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
28479 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
28480 ("*indirect2_jump"): Disable for branch conversion.
28481 ("casesi_jump"): Turn into expander and expand patterns for branch
28483 ("return_use"): New expander.
28484 ("*return"): Emit return via thunk and rename it to ...
28485 ("*return<mode>"): ... this one.
28486 * config/s390/s390.opt: Add new options and and enum for the
28489 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28491 * lra-constraints.c (match_reload): Unconditionally use
28492 gen_lowpart_SUBREG, rather than selecting between that
28493 and equivalent gen_rtx_SUBREG code.
28495 2018-02-08 Richard Biener <rguenther@suse.de>
28497 PR tree-optimization/84233
28498 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
28499 changed flag instead of boguously re-using phi_inserted.
28501 2018-02-08 Martin Jambor <mjambor@suse.cz>
28503 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
28504 static local variables.
28506 2018-02-08 Richard Biener <rguenther@suse.de>
28508 PR tree-optimization/84278
28509 * tree-vect-stmts.c (vectorizable_store): When looking for
28510 smaller vector types to perform grouped strided loads/stores
28511 make sure the mode is supported by the target.
28512 (vectorizable_load): Likewise.
28514 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
28516 * config/aarch64/aarch64.c (aarch64_components_for_bb):
28517 Increase LDP/STP opportunities by adding adjacent callee-saves.
28519 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
28521 PR rtl-optimization/84068
28522 PR rtl-optimization/83459
28523 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
28525 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
28527 PR tree-optimization/84224
28528 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
28529 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
28530 non-zero arguments.
28532 2018-02-07 Iain Sandoe <iain@codesourcery.com>
28535 * config/rs6000/altivec.md (*restore_world): Remove LR use.
28536 * config/rs6000/predicates.md (restore_world_operation): Adjust op
28537 count, remove one USE.
28539 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
28541 * doc/install.texi (Configuration): Document the
28542 --with-long-double-format={ibm,ieee} PowerPC configuration
28546 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
28547 Convert from define_expand to be define_insn_and_split. Rework
28548 float/double/_Float128 conversions to QI/HI/SImode to work with
28549 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
28550 conversions to QI/HImode types did a store and then a load to
28551 truncate the value. For conversions to VSX registers, don't split
28552 the insn, instead emit the code directly. Use the code iterator
28553 any_fix to combine signed and unsigned conversions.
28554 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
28555 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
28556 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
28557 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
28558 (fix_<mode>di2_hw): Likewise.
28559 (fixuns_<mode>di2_hw): Likewise.
28560 (fix_<mode>si2_hw): Likewise.
28561 (fixuns_<mode>si2_hw): Likewise.
28562 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
28563 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
28564 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
28565 fix<uns>_trunc<SFDF:mode>si2_p8.
28566 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
28567 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
28568 (fix<uns>_<mode>_mem): Likewise.
28569 (fctiw<u>z_<mode>_mem): Likewise.
28570 (fix<uns>_<mode>_mem): Likewise.
28571 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
28572 the register allocator from doing a direct move to the GPRs to do
28573 a store, and instead use the ISA 3.0 store byte/half-word from
28574 vector register instruction. For IEEE 128-bit floating point,
28575 also optimize stores of 32-bit ints.
28576 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
28578 2018-02-07 Alan Hayward <alan.hayward@arm.com>
28580 * genextract.c (push_pathstr_operand): New function to support
28582 (walk_rtx): Call push_pathstr_operand.
28583 (print_path): Support [a-zA-Z].
28585 2018-02-07 Richard Biener <rguenther@suse.de>
28587 PR tree-optimization/84037
28588 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
28589 (cse_and_gimplify_to_preheader): Declare.
28590 (vect_get_place_in_interleaving_chain): Likewise.
28591 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28593 (_loop_vec_info::~_loop_vec_info): Delete it.
28594 (cse_and_gimplify_to_preheader): New function.
28595 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
28596 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
28597 (vectorizable_load): Likewise. For grouped stores always base
28598 the IV on the first element.
28599 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
28600 condition before gimplifying.
28602 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28604 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
28605 *DIV_EXPR and *MOD_EXPR.
28607 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
28610 * config/i386/i386.c (ix86_option_override_internal): Mask out
28611 the CF_SET bit when checking -fcf-protection.
28613 2018-02-07 Tom de Vries <tom@codesourcery.com>
28616 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
28619 2018-02-07 Richard Biener <rguenther@suse.de>
28621 PR tree-optimization/84204
28622 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
28625 PR tree-optimization/84205
28626 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
28627 special-case isl_ast_op_zdiv_r.
28629 PR tree-optimization/84223
28630 * graphite-scop-detection.c (gather_bbs::before_dom_children):
28631 Only add conditions from within the region.
28632 (gather_bbs::after_dom_children): Adjust.
28634 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
28637 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
28638 * config/avr/avr.md: Only post-reload split REG-REG moves if
28639 either register is GENERAL_REG_P.
28641 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28643 PR tree-optimization/84235
28644 * tree-ssa-scopedtables.c
28645 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
28646 if the subtraction is performed in floating point type where NaNs are
28647 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
28648 build 1. Formatting fix.
28650 2018-02-06 Jakub Jelinek <jakub@redhat.com>
28653 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
28654 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
28655 and skip it regardless of bb boundaries. Use CALL_P macro,
28656 don't test INSN_P (insn) together with CALL_P or JUMP_P check
28657 unnecessarily, formatting fix.
28659 2018-02-06 Michael Collison <michael.collison@arm.com>
28661 * config/arm/thumb2.md:
28662 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
28663 (*thumb_mov_notscc): Ditto.
28665 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28668 * config/rs6000/rs6000.md (su code attribute): Use "u" for
28669 unsigned_fix, not "s".
28671 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28673 * configure.ac (gcc_fn_eh_frame_ro): New function.
28674 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
28675 correct .eh_frame permissions.
28676 * configure: Regenerate.
28678 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
28680 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
28681 irrelevant options.
28683 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28685 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28686 Display warning message for -mno-speculate-indirect-jumps.
28688 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
28690 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
28692 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
28694 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
28696 PR tree-optimization/84225
28697 * tree-eh.c (find_trapping_overflow): Only call
28698 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
28700 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28703 * config/i386/i386.c: Reimplement the check of possible options
28704 -mibt/-mshstk conbination. Change error messages.
28705 * doc/invoke.texi: Fix a typo: remove extra '='.
28707 2018-02-06 Marek Polacek <polacek@redhat.com>
28709 PR tree-optimization/84228
28710 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
28712 2018-02-06 Tamar Christina <tamar.christina@arm.com>
28715 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
28716 emitted arch directives.
28717 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
28718 __ARM_FEATURE_COPROC before changing architectures.
28720 2018-02-06 Richard Biener <rguenther@suse.de>
28722 * config/i386/i386.c (print_reg): Fix typo.
28723 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
28725 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
28727 * configure: Regenerate.
28729 2018-02-05 Martin Sebor <msebor@redhat.com>
28731 PR tree-optimization/83369
28732 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
28735 2018-02-05 Martin Liska <mliska@suse.cz>
28737 * doc/invoke.texi: Cherry-pick upstream r323995.
28739 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
28741 * ira.c (ira_init_register_move_cost): Adjust comment.
28743 2018-02-05 Martin Liska <mliska@suse.cz>
28745 PR gcov-profile/84137
28746 * doc/gcov.texi: Fix typo in documentation.
28748 2018-02-05 Martin Liska <mliska@suse.cz>
28750 PR gcov-profile/83879
28751 * doc/gcov.texi: Document necessity of --dynamic-list-data when
28752 using dlopen functionality.
28754 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
28756 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
28757 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
28758 _mm_maskz_range_ss, _mm_mask_range_round_ss,
28759 _mm_maskz_range_round_ss): New intrinsics.
28760 (__builtin_ia32_rangesd128_round)
28761 (__builtin_ia32_rangess128_round): Remove.
28762 (__builtin_ia32_rangesd128_mask_round,
28763 __builtin_ia32_rangess128_mask_round): New builtins.
28764 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
28765 __builtin_ia32_rangess128_round): Remove.
28766 (__builtin_ia32_rangesd128_mask_round,
28767 __builtin_ia32_rangess128_mask_round): New builtins.
28768 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
28769 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
28770 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
28771 "<round_saeonly_constraint>")): Changed to ...
28772 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
28773 "<round_saeonly_scalar_constraint>")): ... this.
28774 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
28775 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
28776 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
28777 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
28778 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
28780 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
28782 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
28784 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
28785 Remove all values except native, 8540 and 8548.
28787 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
28789 * config/i386/i386.c (ix86_output_function_return): Pass
28790 INVALID_REGNUM, instead of -1, as invalid register number to
28791 indirect_thunk_name and output_indirect_thunk.
28793 2018-02-02 Julia Koval <julia.koval@intel.com>
28795 * config.gcc: Add -march=icelake.
28796 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
28797 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
28798 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
28799 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
28800 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
28801 (processor_target_table): Add icelake.
28802 (ix86_option_override_internal): Handle new PTAs.
28803 (get_builtin_code_for_version): Handle icelake.
28804 (M_INTEL_COREI7_ICELAKE): New.
28805 (fold_builtin_cpu): Handle icelake.
28806 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
28807 * doc/invoke.texi: Add -march=icelake.
28809 2018-02-02 Julia Koval <julia.koval@intel.com>
28811 * config/i386/i386.c (ix86_option_override_internal): Change flags type
28812 to wide_int_bitmask.
28813 * wide-int-bitmask.h: New.
28815 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28818 * config/i386/i386.md: Replace Pmode with word_mode in
28819 builtin_setjmp_setup and builtin_longjmp to support x32.
28821 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
28825 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
28827 (rs6000_supported_cpu_names): New static variable.
28828 (linux_cpu_translation_table): Likewise.
28829 (elf_platform) <cpu>: Define new static variable and use it.
28830 Translate kernel AT_PLATFORM name to canonical name if needed.
28831 Error if platform name is unknown.
28833 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
28836 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
28838 2018-02-01 Jeff Law <law@redhat.com>
28841 * config/i386/i386.c (release_scratch_register_on_entry): Add new
28842 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
28843 the scratch if RELEASE_VIA_POP is false.
28844 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
28845 If we have to save a temporary register, decrement SIZE appropriately.
28846 Pass new arguments to release_scratch_register_on_entry.
28847 (ix86_adjust_stack_and_probe): Likewise.
28848 (ix86_emit_probe_stack_range): Pass new arguments to
28849 release_scratch_register_on_entry.
28851 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
28853 PR rtl-optimization/84157
28854 * combine.c (change_zero_ext): Use REG_P predicate in
28855 front of HARD_REGISTER_P predicate.
28857 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
28859 * config/avr/avr.c (avr_option_override): Move disabling of
28860 -fdelete-null-pointer-checks to...
28861 * common/config/avr/avr-common.c (avr_option_optimization_table):
28864 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28866 PR tree-optimization/81635
28867 * tree-data-ref.c (split_constant_offset_1): For types that
28868 wrap on overflow, try to use range info to prove that wrapping
28871 2018-02-01 Renlin Li <renlin.li@arm.com>
28874 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
28875 TAILCALL_ADDR_REGS.
28876 (aarch64_register_move_cost): Likewise.
28877 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
28878 TAILCALL_ADDR_REGS.
28879 (REG_CLASS_NAMES): Likewise.
28880 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
28881 TAILCALL_ADDR_REGS. Remove IP registers.
28882 * config/aarch64/aarch64.md (Ucs): Update register constraint.
28884 2018-02-01 Richard Biener <rguenther@suse.de>
28886 * domwalk.h (dom_walker::dom_walker): Add additional constructor
28887 for specifying RPO order and allow NULL for that.
28888 * domwalk.c (dom_walker::dom_walker): Likewise.
28889 (dom_walker::walk): Handle NULL RPO order.
28890 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
28892 (rewrite_update_dom_walker): Likewise.
28893 (mark_def_dom_walker): Likewise.
28895 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28897 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
28898 (aarch64_maybe_expand_sve_subreg_move): Declare.
28899 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
28900 * config/aarch64/predicates.md (aarch64_any_register_operand): New
28902 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
28903 that are semantically a reverse operation.
28904 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
28905 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
28906 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
28908 (aarch64_can_change_mode_class): For big-endian, forbid changes
28909 between two SVE modes if they have different element sizes.
28911 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28913 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
28914 the TImode handling for big-endian targets.
28916 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28918 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
28919 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
28921 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
28922 Remove BSWAP handing for big-endian targets and use the form of
28923 LD1RQ appropariate for the mode.
28925 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28927 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
28928 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
28929 duplicated element.
28931 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28934 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
28935 check for operands that need to go through aarch64_sve_reload_be.
28937 2018-02-01 Jakub Jelinek <jakub@redhat.com>
28939 PR tree-optimization/81661
28940 PR tree-optimization/84117
28941 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
28942 * tree-eh.c: Include gimplify.h.
28943 (find_trapping_overflow, replace_trapping_overflow,
28944 rewrite_to_non_trapping_overflow): New functions.
28945 * tree-vect-loop.c: Include tree-eh.h.
28946 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
28947 * tree-data-ref.c: Include tree-eh.h.
28948 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
28950 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
28952 PR rtl-optimization/84123
28953 * combine.c (change_zero_ext): Check if hard register satisfies
28954 can_change_dest_mode before calling gen_lowpart_SUBREG.
28956 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
28959 * ira.c (ira_init_register_move_cost): Remove assert.
28961 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28963 PR rtl-optimization/84071
28964 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
28965 * doc/tm.texi: Regenerate.
28967 2018-01-31 Richard Biener <rguenther@suse.de>
28969 PR tree-optimization/84132
28970 * tree-data-ref.c (analyze_miv_subscript): Properly
28971 check whether evolution_function_is_affine_multivariate_p
28972 before calling gcd_of_steps_may_divide_p.
28974 2018-01-31 Julia Koval <julia.koval@intel.com>
28977 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28978 * config/i386/i386.md (rdpid_rex64) New.
28979 (rdpid): Make 32bit only.
28981 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
28984 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
28985 an IDENTIFIER_NODE for FUNCTION_TYPE's.
28987 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28990 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
28992 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
28994 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28996 PR rtl-optimization/84071
28997 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
28998 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
29000 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
29002 * config/arc/arc.c (arc_handle_aux_attribute): New function.
29003 (arc_attribute_table): Add 'aux' attribute.
29004 (arc_in_small_data_p): Consider aux like variables.
29005 (arc_is_aux_reg_p): New function.
29006 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
29007 (arc_get_aux_arg): New function.
29008 (prepare_move_operands): Handle aux-register access.
29009 (arc_handle_aux_attribute): New function.
29010 * doc/extend.texi (ARC Variable attributes): Add subsection.
29012 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
29014 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
29015 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
29016 (arc_attribute_table): Add 'uncached' attribute.
29017 (arc_print_operand): Print '.di' flag for uncached memory
29019 (arc_in_small_data_p): Do not consider for small data the uncached
29021 (arc_is_uncached_mem_p): New function.
29022 * config/arc/predicates.md (compact_store_memory_operand): Check
29023 for uncached memory accesses.
29024 (nonvol_nonimm_operand): Likewise.
29025 * doc/extend.texi (ARC Type Attribute): New subsection.
29027 2018-01-31 Jakub Jelinek <jakub@redhat.com>
29030 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
29031 falign-loops=): Add Optimization flag.
29033 2018-01-30 Jeff Law <law@redhat.com>
29036 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
29037 INT_REGISTERS_SAVED. Check it prior to calling
29038 get_scratch_register_on_entry.
29039 (ix86_adjust_stack_and_probe): Similarly.
29040 (ix86_emit_probe_stack_range): Similarly.
29041 (ix86_expand_prologue): Corresponding changes.
29043 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29046 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
29047 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
29049 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
29052 * lra-constraints.c (curr_insn_transform): Process AND in the
29055 2018-01-30 Jakub Jelinek <jakub@redhat.com>
29057 PR rtl-optimization/83986
29058 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
29059 dependence against last_pending_memory_flush in addition to
29060 pending_jump_insns.
29062 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
29064 PR tree-optimization/81611
29065 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
29068 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29071 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
29074 2018-01-30 Richard Biener <rguenther@suse.de>
29075 Jakub Jelinek <jakub@redhat.com>
29077 PR tree-optimization/84111
29078 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
29079 inner loops added during recursion, as they don't have up-to-date
29082 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
29085 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
29086 (can_inline_edge_by_limits_p): ... here.
29087 (can_early_inline_edge_p, check_callers,
29088 update_caller_keys, update_callee_keys, recursive_inlining,
29089 add_new_edges_to_heap, speculation_useful_p,
29090 inline_small_functions,
29091 inline_small_functions, flatten_function,
29092 inline_to_all_callers_1): Update.
29094 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
29096 * profile-count.c (profile_count::combine_with_ipa_count): Handle
29099 2018-01-30 Richard Biener <rguenther@suse.de>
29101 PR tree-optimization/83008
29102 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
29103 invariant and constant vector uses in stmts when they need
29104 more than one stmt.
29106 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29109 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
29110 * configure: Regenerate.
29112 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29114 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
29116 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
29117 Use gen_rtx_REG rather than gen_lowpart.
29119 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29121 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
29122 rather than 0 when creating partial subregs.
29124 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29126 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
29129 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
29132 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
29133 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
29134 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
29135 flags. This restores the settings used before the 2017-07-24.
29136 Turning off pre increment/decrement/modify allows IVOPTS to
29137 optimize DF/SF loops where the index is an int.
29139 2018-01-29 Richard Biener <rguenther@suse.de>
29140 Kelvin Nilsen <kelvin@gcc.gnu.org>
29143 * tree-vect-stmts.c (vectorizable_call): Don't call
29144 targetm.vectorize_builtin_md_vectorized_function if callee is
29147 2018-01-22 Carl Love <cel@us.ibm.com>
29149 * doc/extend.tex: Fix typo in second arg in
29150 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
29152 2018-01-29 Richard Biener <rguenther@suse.de>
29154 PR tree-optimization/84086
29155 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
29156 (flush_ssaname_freelist): When SSA names were released reset
29157 the SCEV hash table.
29159 2018-01-29 Richard Biener <rguenther@suse.de>
29161 PR tree-optimization/84057
29162 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
29163 removed paths when removing edges.
29165 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
29167 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
29168 -mfunction-return=@var{choice}.
29170 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
29172 PR diagnostic/84034
29173 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
29174 Handle CR like TAB.
29175 (layout::print_source_line): Likewise.
29176 (test_get_line_width_without_trailing_whitespace): Add test cases.
29178 2018-01-27 Jakub Jelinek <jakub@redhat.com>
29180 PR middle-end/84040
29181 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
29184 2018-01-26 Jim Wilson <jimw@sifive.com>
29186 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
29188 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
29191 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29193 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
29194 and CMP + SUB-immediate -> SUBS.
29196 2018-01-26 Martin Sebor <msebor@redhat.com>
29198 PR tree-optimization/83896
29199 * tree-ssa-strlen.c (get_string_len): Rename...
29200 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
29201 Avoid assuming length is constant.
29202 (handle_char_store): Use HOST_WIDE_INT for string length.
29204 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
29207 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
29208 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
29210 2018-01-26 Richard Biener <rguenther@suse.de>
29212 PR rtl-optimization/84003
29213 * dse.c (record_store): Only record redundant stores when
29214 the earlier store aliases at least all accesses the later one does.
29216 2018-01-26 Jakub Jelinek <jakub@redhat.com>
29218 PR rtl-optimization/83985
29219 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
29220 REG_CFA_RESTORE insns.
29221 (delete_unmarked_insns): Don't ignore separate shrink wrapping
29222 REG_CFA_RESTORE insns here.
29225 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
29226 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
29228 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29230 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
29231 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
29232 (arc_init): Likewise.
29233 (arc_override_options): Likewise.
29234 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
29236 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
29238 * config/arc/arc.h (TARGET_DBNZ): Define.
29239 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
29240 properly set the tune attribute.
29241 (dbnz): Use TARGET_DBNZ guard.
29242 * config/arc/arc.opt (mtune): Add core3 option.
29244 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29246 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
29247 recognize new pic like addresses.
29248 (arc_delegitimize_address): Clean up.
29250 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29252 * config/arc/arc-arches.def: Option mrf16 valid for all
29254 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
29255 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
29256 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
29257 * config/arc/arc-tables.opt: Regenerate.
29258 * config/arc/arc.c (arc_conditional_register_usage): Handle
29259 reduced register file case.
29260 (arc_file_start): Set must have build attributes.
29261 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
29262 mrf16 option value.
29263 * config/arc/arc.opt (mrf16): Add new option.
29264 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
29265 * config/arc/genmultilib.awk: Handle new mrf16 option.
29266 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
29267 * config/arc/t-multilib: Regenerate.
29268 * doc/invoke.texi (ARC Options): Document mrf16 option.
29270 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29272 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
29273 * config/arc/arc.c (arc_handle_secure_attribute): New function.
29274 (arc_attribute_table): Add 'secure_call' attribute.
29275 (arc_print_operand): Print secure call operand.
29276 (arc_function_ok_for_sibcall): Don't optimize tail calls when
29278 (arc_is_secure_call_p): New function. * config/arc/arc.md
29279 (call_i): Add support for sjli instruction.
29280 (call_value_i): Likewise.
29281 * config/arc/constraints.md (Csc): New constraint.
29283 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29284 John Eric Martin <John.Martin@emmicro-us.com>
29286 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
29287 * config/arc/arc.c (_arc_jli_section): New struct.
29288 (arc_jli_section): New type.
29289 (rc_jli_sections): New static variable.
29290 (arc_handle_jli_attribute): New function.
29291 (arc_attribute_table): Add jli_always and jli_fixed attribute.
29292 (arc_file_end): New function.
29293 (TARGET_ASM_FILE_END): Define.
29294 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
29295 (arc_add_jli_section): New function.
29296 (jli_call_scan): Likewise.
29297 (arc_reorg): Call jli_call_scan.
29298 (arc_output_addsi): Remove 'S' from printing asm operand.
29299 (arc_is_jli_call_p): New function.
29300 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
29302 (movhi_insn): Likewise.
29303 (movsi_insn): Likewise.
29304 (movsi_set_cc_insn): Likewise.
29305 (loadqi_update): Likewise.
29306 (load_zeroextendqisi_update): Likewise.
29307 (load_signextendqisi_update): Likewise.
29308 (loadhi_update): Likewise.
29309 (load_zeroextendhisi_update): Likewise.
29310 (load_signextendhisi_update): Likewise.
29311 (loadsi_update): Likewise.
29312 (loadsf_update): Likewise.
29313 (movsicc_insn): Likewise.
29314 (bset_insn): Likewise.
29315 (bxor_insn): Likewise.
29316 (bclr_insn): Likewise.
29317 (bmsk_insn): Likewise.
29318 (bicsi3_insn): Likewise.
29319 (cmpsi_cc_c_insn): Likewise.
29320 (movsi_ne): Likewise.
29321 (movsi_cond_exec): Likewise.
29322 (clrsbsi2): Likewise.
29323 (norm_f): Likewise.
29333 (call_i): Remove 'S' asm letter, add jli instruction.
29334 (call_value_i): Likewise.
29335 * config/arc/arc.op (mjli-always): New option.
29336 * config/arc/constraints.md (Cji): New constraint.
29337 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
29339 (subsf3_fpx): Likewise.
29340 (mulsf3_fpx): Likewise.
29341 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
29343 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
29344 function attrbutes.
29345 * doc/invoke.texi (ARC): Document mjli-always option.
29347 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
29349 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
29350 avoid addition with 0 and use incw and decw where possible.
29352 2018-01-26 Richard Biener <rguenther@suse.de>
29354 PR tree-optimization/81082
29355 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
29356 association if it requires casting to unsigned.
29357 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
29358 from fold_plusminus_mult_expr to catch important cases late when
29359 range info is available.
29361 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29363 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
29364 * configure.ac (hidden_linkonce): New test.
29365 * configure: Regenerate.
29366 * config.in: Regenerate.
29368 2018-01-26 Julia Koval <julia.koval@intel.com>
29370 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
29371 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
29372 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
29373 _mm_mask_bitshuffle_epi64_mask): Fix type.
29374 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
29375 USI_FTYPE_V4DI_V4DI_USI): Remove.
29376 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
29377 __builtin_ia32_vpshufbitqmb256_mask,
29378 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
29379 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
29380 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
29382 2018-01-26 Alan Modra <amodra@gmail.com>
29385 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
29386 UNSPEC_VBPERMQ. Sort other unspecs.
29388 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
29390 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
29392 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
29394 PR middle-end/83055
29395 * predict.c (drop_profile): Do not push/pop cfun; update also
29397 (handle_missing_profiles): Fix logic looking for zero profiles.
29399 2018-01-25 Jakub Jelinek <jakub@redhat.com>
29401 PR middle-end/83977
29402 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
29403 on functions with #pragma omp declare simd or functions with simd
29405 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
29406 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
29407 Remove trailing \n from warning_at calls.
29409 2018-01-25 Tom de Vries <tom@codesourcery.com>
29412 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
29413 for neutered workers.
29415 2018-01-24 Joseph Myers <joseph@codesourcery.com>
29418 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
29419 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
29421 2018-01-24 Jeff Law <law@redhat.com>
29424 * i386.c (get_probe_interval): Move to earlier point.
29425 (ix86_compute_frame_layout): If -fstack-clash-protection and
29426 the frame is larger than the probe interval, then use pushes
29427 to save registers rather than reg->mem moves.
29428 (ix86_expand_prologue): Remove conditional for int_registers_saved
29431 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
29434 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
29435 min/max for never referenced object.
29437 2018-01-24 Jakub Jelinek <jakub@redhat.com>
29439 PR middle-end/83977
29440 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
29442 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
29443 attributes from DECL_ATTRIBUTES (decl) without affecting
29444 DECL_ATTRIBUTES (current_function_decl).
29445 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
29446 functions with non-NULL DECL_ABSTRACT_ORIGIN.
29448 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
29450 PR tree-optimization/83979
29451 * fold-const.c (fold_comparison): Use constant_boolean_node
29452 instead of boolean_{true,false}_node.
29454 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
29456 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
29459 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29461 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29462 Simplify the clause that sets the length attribute.
29463 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29464 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
29465 clause that sets the length attribute.
29466 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29468 2018-01-24 Tom de Vries <tom@codesourcery.com>
29471 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
29472 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
29473 Add strict parameter.
29474 (prevent_branch_around_nothing): Insert dummy insn between branch to
29475 label and label with no ptx insn inbetween.
29476 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
29478 2018-01-24 Tom de Vries <tom@codesourcery.com>
29481 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
29482 for neutered threads in warp.
29483 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
29485 2018-01-24 Richard Biener <rguenther@suse.de>
29487 PR tree-optimization/83176
29488 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
29491 2018-01-24 Richard Biener <rguenther@suse.de>
29493 PR tree-optimization/82819
29494 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
29495 code generating pluses that are no-ops in the target precision.
29497 2018-01-24 Richard Biener <rguenther@suse.de>
29499 PR middle-end/84000
29500 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
29502 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29504 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
29505 to merge probabilities.
29506 * predict.c (probably_never_executed): Also mark as cold functions
29507 with global 0 profile and guessed local profile.
29508 * profile-count.c (profile_probability::combine_with_count): New
29510 * profile-count.h (profile_probability::operator*,
29511 profile_probability::operator*=, profile_probability::operator/,
29512 profile_probability::operator/=): Reduce precision to adjusted
29513 and set value to guessed on contradictory divisions.
29514 (profile_probability::combine_with_freq): Remove.
29515 (profile_probability::combine_wiht_count): Declare.
29516 (profile_count::force_nonzero):: Set to adjusted.
29517 (profile_count::probability_in):: Set quality to adjusted.
29518 * tree-ssa-tail-merge.c (replace_block_by): Use
29519 combine_with_count.
29521 2018-01-23 Andrew Waterman <andrew@sifive.com>
29522 Jim Wilson <jimw@sifive.com>
29524 * config/riscv/riscv.c (riscv_stack_boundary): New.
29525 (riscv_option_override): Set riscv_stack_boundary. Handle
29526 riscv_preferred_stack_boundary_arg.
29527 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
29528 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
29529 (STACK_BOUNDARY): Set to riscv_stack_boundary.
29530 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
29531 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
29532 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
29534 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
29537 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
29538 of struct ix86_frame.
29539 (ix86_expand_epilogue): Likewise. Add a local variable for
29540 the reg_save_offset field in struct ix86_frame.
29542 2018-01-23 Bin Cheng <bin.cheng@arm.com>
29544 PR tree-optimization/82604
29545 * tree-loop-distribution.c (enum partition_kind): New enum item
29546 PKIND_PARTIAL_MEMSET.
29547 (partition_builtin_p): Support above new enum item.
29548 (generate_code_for_partition): Ditto.
29549 (compute_access_range): Differentiate cases that equality can be
29550 proven at all loops, the innermost loops or no loops.
29551 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
29552 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
29553 (finalize_partitions, distribute_loop): Don't fuse partition of
29554 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
29555 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
29556 parloop is enabled.
29558 2018-01-23 Martin Liska <mliska@suse.cz>
29560 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
29561 order to ignore the predictor.
29562 (PRED_POLYMORPHIC_CALL): Likewise.
29563 (PRED_RECURSIVE_CALL): Likewise.
29565 2018-01-23 Martin Liska <mliska@suse.cz>
29567 * tree-profile.c (tree_profiling): Print function header to
29568 aware reader which function we are working on.
29569 * value-prof.c (gimple_find_values_to_profile): Do not print
29570 not interesting value histograms.
29572 2018-01-23 Martin Liska <mliska@suse.cz>
29574 * profile-count.h (enum profile_quality): Add
29575 profile_uninitialized as the first value. Do not number values
29576 as they are zero based.
29577 (profile_count::verify): Update sanity check.
29578 (profile_probability::verify): Likewise.
29580 2018-01-23 Nathan Sidwell <nathan@acm.org>
29582 * doc/invoke.texi (ffor-scope): Deprecate.
29584 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29586 PR tree-optimization/83510
29587 * domwalk.c (set_all_edges_as_executable): New function.
29588 (dom_walker::dom_walker): Convert bool param
29589 "skip_unreachable_blocks" to enum reachability. Move setup of
29590 edge flags to set_all_edges_as_executable and only do it when
29591 reachability is REACHABLE_BLOCKS.
29592 * domwalk.h (enum dom_walker::reachability): New enum.
29593 (dom_walker::dom_walker): Convert bool param
29594 "skip_unreachable_blocks" to enum reachability.
29595 (set_all_edges_as_executable): New decl.
29596 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
29597 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
29599 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
29600 but converting true to REACHABLE_BLOCKS.
29601 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
29603 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
29604 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
29605 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
29607 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
29608 if check_all_array_refs will be called.
29610 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29612 * tree.c (selftest::test_location_wrappers): Add more test
29615 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29617 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
29618 (selftest::test_bit_in_range): Likewise.
29620 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
29623 * doc/sourcebuild.texi (vect_float): Say that the selector
29624 only describes the situation when -funsafe-math-optimizations is on.
29625 (vect_float_strict): Document.
29627 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
29629 PR tree-optimization/83965
29630 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
29631 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
29632 instead of checking only for a reduction.
29633 (vect_recog_widen_sum_pattern): Likewise.
29635 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29637 * predict.c (probably_never_executed): Only use precise profile info.
29638 (compute_function_frequency): Skip after inlining hack since we now
29639 have quality checking.
29641 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29643 * profile-count.h (profile_probability::very_unlikely,
29644 profile_probability::unlikely, profile_probability::even): Set
29645 precision to guessed.
29647 2018-01-23 Richard Biener <rguenther@suse.de>
29649 PR tree-optimization/83963
29650 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29651 Properly terminate dominator walk when crossing the exit edge not
29652 when visiting its source block.
29654 2018-01-23 Jakub Jelinek <jakub@redhat.com>
29657 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
29658 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
29660 2018-01-22 Jakub Jelinek <jakub@redhat.com>
29662 PR tree-optimization/83957
29663 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
29664 semicolon after for body surrounded by braces.
29666 PR tree-optimization/83081
29667 * profile-count.h (profile_probability::split): New method.
29668 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
29669 Use profile_probability::split.
29670 (do_compare_rtx_and_jump): Fix adjustment of probabilities
29671 when splitting a single conditional jump into 2.
29673 2018-01-22 David Malcolm <dmalcolm@redhat.com>
29675 PR tree-optimization/69452
29676 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
29679 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29681 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
29682 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
29683 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
29685 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29687 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
29689 * config/rl78/rl78.md (movdi): New define_expand.
29690 * config/rl78/rl78.c (rl78_split_movdi): New function.
29692 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29695 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
29697 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
29698 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
29699 128-bit to produce an UNSPEC move to get the double word with the
29700 signbit and then a shift directly to do signbit.
29701 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
29702 implementation with a new version that just does either a direct
29703 move or a regular move. Move memory interface to separate insns.
29704 Move insns so they are next to the expander.
29705 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
29706 with signbit move. Split big and little endian case.
29707 (signbit<mode>2_dm_mem_le): Likewise.
29708 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
29709 (signbit<mode>2_dm2): Likewise.
29711 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29713 * config/rl78/rl78.md (anddi3): New define_expand.
29715 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29717 * config/rl78/rl78.md (umindi3): New define_expand.
29719 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29721 * config/rl78/rl78.md (smindi3): New define_expand.
29723 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29725 * config/rl78/rl78.md (smaxdi3): New define_expand.
29727 2018-01-22 Carl Love <cel@us.ibm.com>
29729 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
29730 LVX_V1TI): Add macro expansion.
29731 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
29732 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
29733 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
29734 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
29735 Change check to determine if the instruction is a byte reversing
29736 entry. Fix typo in comment.
29737 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
29738 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
29739 Add def_builtin calls for new builtins.
29740 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
29741 Add define_insn expansion.
29743 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29745 * config/rl78/rl78.md (umaxdi3): New define_expand.
29747 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29749 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
29750 for non-QImode registers.
29752 2018-01-22 Richard Biener <rguenther@suse.de>
29754 PR tree-optimization/83963
29755 * graphite-scop-detection.c (scop_detection::get_sese): Delay
29756 including the loop exit block.
29757 (scop_detection::merge_sese): Likewise.
29758 (scop_detection::add_scop): Do it here instead.
29760 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29762 * doc/sourcebuild.texi (arm_softfloat): Document.
29764 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
29767 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
29768 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
29769 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
29771 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29772 David Edelsohn <dje.gcc@gmail.com>
29775 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29776 Change "crset eq" to "crset 2".
29777 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29778 (*call_indirect_aix<mode>_nospec): Likewise.
29779 (*call_value_indirect_aix<mode>_nospec): Likewise.
29780 (*call_indirect_elfv2<mode>_nospec): Likewise.
29781 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
29782 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
29783 change assembly output from . to $.
29784 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29785 (indirect_jump<mode>_nospec): Change assembly output from . to $.
29786 (*tablejump<mode>_internal1_nospec): Likewise.
29788 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
29791 * config/sh/sh_optimize_sett_clrt.cc:
29792 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
29794 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
29796 PR tree-optimization/83940
29797 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
29798 offset_dt to vect_constant_def rather than vect_unknown_def_type.
29799 (vect_check_load_store_mask): Add a mask_dt_out parameter and
29800 use it to pass back the definition type.
29801 (vect_check_store_rhs): Likewise rhs_dt_out.
29802 (vect_build_gather_load_calls): Add a mask_dt argument and use
29803 it instead of a call to vect_is_simple_use.
29804 (vectorizable_store): Update calls to vect_check_load_store_mask
29805 and vect_check_store_rhs. Use the dt returned by the latter instead
29806 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
29807 instead of calls to vect_is_simple_use. Pass the scalar rather
29808 than the vector operand to vect_is_simple_use when handling
29809 second and subsequent copies of an rhs value.
29810 (vectorizable_load): Update calls to vect_check_load_store_mask
29811 and vect_build_gather_load_calls. Use the cached mask_dt and
29812 gs_info.offset_dt instead of calls to vect_is_simple_use.
29814 2018-01-20 Jakub Jelinek <jakub@redhat.com>
29816 PR middle-end/83945
29817 * tree-emutls.c: Include gimplify.h.
29818 (lower_emutls_2): New function.
29819 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
29820 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
29821 it before further processing.
29824 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
29825 UINTVAL (trueop1) instead of INTVAL (op1).
29827 2018-01-19 Jakub Jelinek <jakub@redhat.com>
29831 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
29832 INCOMING_FRAME_SP_OFFSET if not defined.
29833 (scan_trace): Add ENTRY argument. If true and
29834 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
29835 emit a note to adjust the CFA offset.
29836 (create_cfi_notes): Adjust scan_trace callers.
29837 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
29838 INCOMING_FRAME_SP_OFFSET in the CIE.
29839 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
29840 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
29842 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
29843 * doc/tm.texi: Regenerated.
29845 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29847 PR rtl-optimization/83147
29848 * lra-constraints.c (remove_inheritance_pseudos): Use
29849 lra_substitute_pseudo_within_insn.
29851 2018-01-19 Tom de Vries <tom@codesourcery.com>
29852 Cesar Philippidis <cesar@codesourcery.com>
29855 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
29857 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
29860 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
29861 spaces for function labels.
29863 2018-01-19 Martin Liska <mliska@suse.cz>
29865 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
29866 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
29867 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
29868 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
29869 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
29870 (PRED_CONST_RETURN): Change from 69 to 65.
29871 (PRED_NULL_RETURN): Change from 91 to 71.
29872 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
29873 (PRED_LOOP_GUARD): Change from 66 to 73.
29875 2018-01-19 Martin Liska <mliska@suse.cz>
29877 * predict.c (predict_insn_def): Add new assert.
29878 (struct branch_predictor): Change type to signed integer.
29879 (test_prediction_value_range): Amend test to cover
29880 PROB_UNINITIALIZED.
29881 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
29882 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
29883 (PRED_LOOP_ITERATIONS_MAX): Likewise.
29884 (PRED_LOOP_IV_COMPARE): Likewise.
29885 * predict.h (PROB_UNINITIALIZED): Define new constant.
29887 2018-01-19 Martin Liska <mliska@suse.cz>
29889 * predict.c (dump_prediction): Add new format for
29890 analyze_brprob.py script which is enabled with -details
29892 * profile-count.h (precise_p): New function.
29894 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29896 PR tree-optimization/83922
29897 * tree-vect-loop.c (vect_verify_full_masking): Return false if
29898 there are no statements that need masking.
29899 (vect_active_double_reduction_p): New function.
29900 (vect_analyze_loop_operations): Use it when handling phis that
29901 are not in the loop header.
29903 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29905 PR tree-optimization/83914
29906 * tree-vect-loop.c (vectorizable_induction): Don't convert
29907 init_expr or apply the peeling adjustment for inductions
29908 that are nested within the vectorized loop.
29910 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29912 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
29915 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29919 * function.h (gimplify_parameters): Add gimple_seq * argument.
29920 * function.c: Include gimple.h and options.h.
29921 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
29922 for the added local temporaries if needed.
29923 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
29924 if there are any parameter cleanups, wrap whole body into a
29925 try/finally with the cleanups.
29927 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29930 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
29931 Use GET_MODE_CLASS for scalar floating point.
29933 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29937 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
29938 Fix call of call_cgraph_insertion_hooks.
29940 2018-01-18 Martin Sebor <msebor@redhat.com>
29942 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
29944 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29947 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
29950 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
29953 * common.opt: (-ffile-prefix-map): New option.
29954 * opts.c (common_handle_option): Defer it.
29955 * opts-global.c (handle_common_deferred_options): Handle it.
29956 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
29957 * file-prefix-map.h: New file.
29958 (remap_debug_filename, add_debug_prefix_map): ...here.
29959 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
29960 * final.c (debug_prefix_map, add_debug_prefix_map
29961 remap_debug_filename): Move to...
29962 * file-prefix-map.c: New file.
29963 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
29964 generalize, get rid of alloca(), use strrchr() instead of strchr().
29965 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
29966 Implement in terms of add_prefix_map().
29967 (remap_macro_filename, remap_debug_filename): Implement in term of
29969 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
29970 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
29971 * dbxout.c: Include file-prefix-map.h.
29972 * varasm.c: Likewise.
29973 * vmsdbgout.c: Likewise.
29974 * xcoffout.c: Likewise.
29975 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
29976 * doc/cppopts.texi (-fmacro-prefix-map): Document.
29977 * doc/invoke.texi (-ffile-prefix-map): Document.
29978 (-fdebug-prefix-map): Update description.
29980 2018-01-18 Martin Liska <mliska@suse.cz>
29982 * config/i386/i386.c (indirect_thunk_name): Document that also
29984 (output_indirect_thunk): Document why both instructions
29985 (pause and lfence) are generated.
29987 2018-01-18 Richard Biener <rguenther@suse.de>
29989 PR tree-optimization/83887
29990 * graphite-scop-detection.c
29991 (scop_detection::get_nearest_dom_with_single_entry): Remove.
29992 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
29993 (scop_detection::merge_sese): Re-implement with a flood-fill
29994 algorithm that properly finds a SESE region if it exists.
29996 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29999 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
30000 pointer_diff optimizations use view_convert instead of convert.
30002 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30004 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
30005 Generate different code for -mno-speculate-indirect-jumps.
30006 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
30007 (*call_indirect_aix<mode>): Disable for
30008 -mno-speculate-indirect-jumps.
30009 (*call_indirect_aix<mode>_nospec): New define_insn.
30010 (*call_value_indirect_aix<mode>): Disable for
30011 -mno-speculate-indirect-jumps.
30012 (*call_value_indirect_aix<mode>_nospec): New define_insn.
30013 (*sibcall_nonlocal_sysv<mode>): Generate different code for
30014 -mno-speculate-indirect-jumps.
30015 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
30017 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
30019 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
30020 long double type, set the flags for noting the default long double
30021 type, even if we don't pass or return a long double type.
30023 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
30026 * ipa-inline.c (flatten_function): Do not overwrite final inlining
30029 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
30031 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
30032 support for merge[hl].
30033 (fold_mergehl_helper): New helper function.
30034 (tree-vector-builder.h): New #include for tree_vector_builder usage.
30035 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
30036 (altivec_vmrglw_direct): Add xxmrglw insn.
30038 2018-01-17 Andrew Waterman <andrew@sifive.com>
30040 * config/riscv/riscv.c (riscv_conditional_register_usage): If
30041 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
30043 2018-01-17 David Malcolm <dmalcolm@redhat.com>
30046 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
30047 call the lto_location_cache before reading the
30048 DECL_SOURCE_LOCATION of the types.
30050 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30051 Richard Sandiford <richard.sandiford@linaro.org>
30053 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
30054 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
30055 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
30056 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
30057 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
30059 * config/aarch64/constraints.md (aarch64_movti_operand):
30061 * config/aarch64/predicates.md (Uti): Add new constraint.
30063 2018-01-17 Carl Love <cel@us.ibm.com>
30065 * config/rs6000/vsx.md (define_expand xl_len_r,
30066 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
30067 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
30069 (define_expand, define_insn): Move the shift left from the
30070 define_insn to the define_expand for lxvl and stxvl instructions.
30071 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
30072 and XL_LEN_R definitions to PURE.
30074 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
30076 * config/i386/i386.c (indirect_thunk_name): Declare regno
30077 as unsigned int. Compare regno with INVALID_REGNUM.
30078 (output_indirect_thunk): Ditto.
30079 (output_indirect_thunk_function): Ditto.
30080 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
30081 in the call to output_indirect_thunk_function.
30083 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
30085 PR middle-end/83884
30086 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
30087 rather than the size of inner_type to determine the stack slot size
30088 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
30090 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
30093 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
30096 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30098 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
30099 endian Linux systems to optionally enable multilibs for selecting
30100 the long double type if the user configured an explicit type.
30101 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
30102 have no long double multilibs if not defined.
30103 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
30104 warn if the user used -mabi={ieee,ibm}longdouble and we built
30105 multilibs for long double.
30106 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
30107 appropriate multilib option.
30108 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
30110 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
30111 for building long double multilibs.
30112 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
30114 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
30116 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
30119 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
30121 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
30124 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
30127 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
30130 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
30132 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
30133 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
30135 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
30137 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
30138 different rtl trees depending on TARGET_64BIT.
30139 (rs6000_gen_lvx): Likewise.
30141 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
30143 * config/visium/visium.md (nop): Tweak comment.
30144 (hazard_nop): Likewise.
30146 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30148 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
30149 -mspeculate-indirect-jumps.
30150 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
30151 for -mno-speculate-indirect-jumps.
30152 (*call_indirect_elfv2<mode>_nospec): New define_insn.
30153 (*call_value_indirect_elfv2<mode>): Disable for
30154 -mno-speculate-indirect-jumps.
30155 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
30156 (indirect_jump): Emit different RTL for
30157 -mno-speculate-indirect-jumps.
30158 (*indirect_jump<mode>): Disable for
30159 -mno-speculate-indirect-jumps.
30160 (*indirect_jump<mode>_nospec): New define_insn.
30161 (tablejump): Emit different RTL for
30162 -mno-speculate-indirect-jumps.
30163 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
30164 (tablejumpsi_nospec): New define_expand.
30165 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
30166 (tablejumpdi_nospec): New define_expand.
30167 (*tablejump<mode>_internal1): Disable for
30168 -mno-speculate-indirect-jumps.
30169 (*tablejump<mode>_internal1_nospec): New define_insn.
30170 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
30173 2018-01-16 Artyom Skrobov tyomitch@gmail.com
30175 * caller-save.c (insert_save): Drop unnecessary parameter. All
30178 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30179 Richard Biener <rguenth@suse.de>
30182 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
30183 return early, inline manually is_gimple_sizepos. Make sure if we
30184 call gimplify_expr we don't end up with a gimple constant.
30185 * tree.c (variably_modified_type_p): Don't return true for
30186 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
30187 * gimplify.h (is_gimple_sizepos): Remove.
30189 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30191 PR tree-optimization/83857
30192 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
30193 vectorizable_live_operation for pure SLP statements.
30194 (vectorizable_live_operation): Handle PHIs.
30196 2018-01-16 Richard Biener <rguenther@suse.de>
30198 PR tree-optimization/83867
30199 * tree-vect-stmts.c (vect_transform_stmt): Precompute
30200 nested_in_vect_loop_p since the scalar stmt may get invalidated.
30202 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30205 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
30206 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
30207 If off is not INTEGER_CST, issue a may not be aligned warning
30208 rather than isn't aligned. Use isn%'t rather than isn't.
30209 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
30211 <case MULT_EXPR>: Improve the case when bottom and one of the
30212 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
30213 operand, in that case check if the other operand is multiple of
30214 bottom divided by the INTEGER_CST operand.
30216 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30219 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
30220 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
30221 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
30222 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
30223 * config/pa/pa.c (pa_function_arg_advance): Likewise.
30224 (pa_function_arg, pa_arg_partial_bytes): Likewise.
30225 (pa_function_arg_size): New function.
30227 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30229 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
30230 in a separate statement.
30232 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30234 PR tree-optimization/83847
30235 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
30236 group gathers and scatters.
30238 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30240 PR rtl-optimization/86620
30241 * params.def (max-sched-ready-insns): Bump minimum value to 1.
30243 PR rtl-optimization/83213
30244 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
30245 to last if both are JUMP_INSNs.
30247 PR tree-optimization/83843
30248 * gimple-ssa-store-merging.c
30249 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
30250 store_immediate_info for bswap/nop orig_stores.
30252 2018-01-15 Andrew Waterman <andrew@sifive.com>
30254 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
30256 <UDIV>: Increase cost if !TARGET_DIV.
30258 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30260 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
30261 (define_attr "cr_logical_3op"): New.
30262 (cceq_ior_compare): Adjust.
30263 (cceq_ior_compare_complement): Adjust.
30264 (*cceq_rev_compare): Adjust.
30265 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
30266 (is_cracked_insn): Adjust.
30267 (insn_must_be_first_in_group): Adjust.
30268 * config/rs6000/40x.md: Adjust.
30269 * config/rs6000/440.md: Adjust.
30270 * config/rs6000/476.md: Adjust.
30271 * config/rs6000/601.md: Adjust.
30272 * config/rs6000/603.md: Adjust.
30273 * config/rs6000/6xx.md: Adjust.
30274 * config/rs6000/7450.md: Adjust.
30275 * config/rs6000/7xx.md: Adjust.
30276 * config/rs6000/8540.md: Adjust.
30277 * config/rs6000/cell.md: Adjust.
30278 * config/rs6000/e300c2c3.md: Adjust.
30279 * config/rs6000/e500mc.md: Adjust.
30280 * config/rs6000/e500mc64.md: Adjust.
30281 * config/rs6000/e5500.md: Adjust.
30282 * config/rs6000/e6500.md: Adjust.
30283 * config/rs6000/mpc.md: Adjust.
30284 * config/rs6000/power4.md: Adjust.
30285 * config/rs6000/power5.md: Adjust.
30286 * config/rs6000/power6.md: Adjust.
30287 * config/rs6000/power7.md: Adjust.
30288 * config/rs6000/power8.md: Adjust.
30289 * config/rs6000/power9.md: Adjust.
30290 * config/rs6000/rs64.md: Adjust.
30291 * config/rs6000/titan.md: Adjust.
30293 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30295 * config/i386/predicates.md (indirect_branch_operand): Rewrite
30296 ix86_indirect_branch_register logic.
30298 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30300 * config/i386/constraints.md (Bs): Update
30301 ix86_indirect_branch_register check. Don't check
30302 ix86_indirect_branch_register with GOT_memory_operand.
30304 * config/i386/predicates.md (GOT_memory_operand): Don't check
30305 ix86_indirect_branch_register here.
30306 (GOT32_symbol_operand): Likewise.
30308 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30310 * config/i386/predicates.md (constant_call_address_operand):
30311 Rewrite ix86_indirect_branch_register logic.
30312 (sibcall_insn_operand): Likewise.
30314 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30316 * config/i386/constraints.md (Bs): Replace
30317 ix86_indirect_branch_thunk_register with
30318 ix86_indirect_branch_register.
30320 * config/i386/i386.md (indirect_jump): Likewise.
30321 (tablejump): Likewise.
30322 (*sibcall_memory): Likewise.
30323 (*sibcall_value_memory): Likewise.
30324 Peepholes of indirect call and jump via memory: Likewise.
30325 * config/i386/i386.opt: Likewise.
30326 * config/i386/predicates.md (indirect_branch_operand): Likewise.
30327 (GOT_memory_operand): Likewise.
30328 (call_insn_operand): Likewise.
30329 (sibcall_insn_operand): Likewise.
30330 (GOT32_symbol_operand): Likewise.
30332 2018-01-15 Jakub Jelinek <jakub@redhat.com>
30334 PR middle-end/83837
30335 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
30336 type rather than type addr's type points to.
30337 (expand_omp_atomic_mutex): Likewise.
30338 (expand_omp_atomic): Likewise.
30340 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30343 * config/i386/i386.c (output_indirect_thunk_function): Use
30344 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
30345 for __x86_return_thunk.
30347 2018-01-15 Richard Biener <rguenther@suse.de>
30349 PR middle-end/83850
30350 * expmed.c (extract_bit_field_1): Fix typo.
30352 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30355 * config/arm/iterators.md (VF): New mode iterator.
30356 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
30357 Remove integer-related logic from pattern.
30358 (neon_vabd<mode>_3): Likewise.
30360 2018-01-15 Jakub Jelinek <jakub@redhat.com>
30362 PR middle-end/82694
30363 * common.opt (fstrict-overflow): No longer an alias.
30364 (fwrapv-pointer): New option.
30365 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
30366 also for pointer types based on flag_wrapv_pointer.
30367 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
30368 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
30369 opts->x_flag_wrapv got set.
30370 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
30371 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
30372 POINTER_TYPE_OVERFLOW_UNDEFINED.
30373 * match.pd: Likewise in address comparison pattern.
30374 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
30376 2018-01-15 Richard Biener <rguenther@suse.de>
30379 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
30380 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
30381 Reset type names to their identifier if their TYPE_DECL doesn't
30382 have linkage (and thus is used for ODR and devirt).
30383 (save_debug_info_for_decl): Remove.
30384 (save_debug_info_for_type): Likewise.
30385 (add_tree_to_fld_list): Adjust.
30386 * tree-pretty-print.c (dump_generic_node): Make dumping of
30387 type names more robust.
30389 2018-01-15 Richard Biener <rguenther@suse.de>
30391 * BASE-VER: Bump to 8.0.1.
30393 2018-01-14 Martin Sebor <msebor@redhat.com>
30396 * builtins.c (check_access): Avoid warning when the no-warning bit
30399 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
30401 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
30402 * ira-color (allocno_hard_regs_compare): Likewise.
30404 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
30407 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
30408 Use .pushsection/.popsection.
30410 2018-01-14 Martin Sebor <msebor@redhat.com>
30413 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
30415 2018-01-14 Jakub Jelinek <jakub@redhat.com>
30417 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
30418 entry from extra_headers.
30419 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
30420 extra_headers, make the list bitwise identical to the i?86-*-* one.
30422 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30424 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
30425 -mcmodel=large with -mindirect-branch=thunk,
30426 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
30427 -mfunction-return=thunk-extern.
30428 * doc/invoke.texi: Document -mcmodel=large is incompatible with
30429 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
30430 -mfunction-return=thunk and -mfunction-return=thunk-extern.
30432 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30434 * config/i386/i386.c (print_reg): Print the name of the full
30435 integer register without '%'.
30436 (ix86_print_operand): Handle 'V'.
30437 * doc/extend.texi: Document 'V' modifier.
30439 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30441 * config/i386/constraints.md (Bs): Disallow memory operand for
30442 -mindirect-branch-register.
30444 * config/i386/predicates.md (indirect_branch_operand): Likewise.
30445 (GOT_memory_operand): Likewise.
30446 (call_insn_operand): Likewise.
30447 (sibcall_insn_operand): Likewise.
30448 (GOT32_symbol_operand): Likewise.
30449 * config/i386/i386.md (indirect_jump): Call convert_memory_address
30450 for -mindirect-branch-register.
30451 (tablejump): Likewise.
30452 (*sibcall_memory): Likewise.
30453 (*sibcall_value_memory): Likewise.
30454 Disallow peepholes of indirect call and jump via memory for
30455 -mindirect-branch-register.
30456 (*call_pop): Replace m with Bw.
30457 (*call_value_pop): Likewise.
30458 (*sibcall_pop_memory): Replace m with Bs.
30459 * config/i386/i386.opt (mindirect-branch-register): New option.
30460 * doc/invoke.texi: Document -mindirect-branch-register option.
30462 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30464 * config/i386/i386-protos.h (ix86_output_function_return): New.
30465 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
30466 set function_return_type.
30467 (indirect_thunk_name): Add ret_p to indicate thunk for function
30469 (output_indirect_thunk_function): Pass false to
30470 indirect_thunk_name.
30471 (ix86_output_indirect_branch_via_reg): Likewise.
30472 (ix86_output_indirect_branch_via_push): Likewise.
30473 (output_indirect_thunk_function): Create alias for function
30474 return thunk if regno < 0.
30475 (ix86_output_function_return): New function.
30476 (ix86_handle_fndecl_attribute): Handle function_return.
30477 (ix86_attribute_table): Add function_return.
30478 * config/i386/i386.h (machine_function): Add
30479 function_return_type.
30480 * config/i386/i386.md (simple_return_internal): Use
30481 ix86_output_function_return.
30482 (simple_return_internal_long): Likewise.
30483 * config/i386/i386.opt (mfunction-return=): New option.
30484 (indirect_branch): Mention -mfunction-return=.
30485 * doc/extend.texi: Document function_return function attribute.
30486 * doc/invoke.texi: Document -mfunction-return= option.
30488 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30490 * config/i386/i386-opts.h (indirect_branch): New.
30491 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
30492 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
30493 with local indirect jump when converting indirect call and jump.
30494 (ix86_set_indirect_branch_type): New.
30495 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
30496 (indirectlabelno): New.
30497 (indirect_thunk_needed): Likewise.
30498 (indirect_thunk_bnd_needed): Likewise.
30499 (indirect_thunks_used): Likewise.
30500 (indirect_thunks_bnd_used): Likewise.
30501 (INDIRECT_LABEL): Likewise.
30502 (indirect_thunk_name): Likewise.
30503 (output_indirect_thunk): Likewise.
30504 (output_indirect_thunk_function): Likewise.
30505 (ix86_output_indirect_branch_via_reg): Likewise.
30506 (ix86_output_indirect_branch_via_push): Likewise.
30507 (ix86_output_indirect_branch): Likewise.
30508 (ix86_output_indirect_jmp): Likewise.
30509 (ix86_code_end): Call output_indirect_thunk_function if needed.
30510 (ix86_output_call_insn): Call ix86_output_indirect_branch if
30512 (ix86_handle_fndecl_attribute): Handle indirect_branch.
30513 (ix86_attribute_table): Add indirect_branch.
30514 * config/i386/i386.h (machine_function): Add indirect_branch_type
30515 and has_local_indirect_jump.
30516 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
30518 (tablejump): Likewise.
30519 (*indirect_jump): Use ix86_output_indirect_jmp.
30520 (*tablejump_1): Likewise.
30521 (simple_return_indirect_internal): Likewise.
30522 * config/i386/i386.opt (mindirect-branch=): New option.
30523 (indirect_branch): New.
30526 (thunk-inline): Likewise.
30527 (thunk-extern): Likewise.
30528 * doc/extend.texi: Document indirect_branch function attribute.
30529 * doc/invoke.texi: Document -mindirect-branch= option.
30531 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
30534 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
30536 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
30538 * ipa-inline.c (want_inline_small_function_p): Return false if
30539 inlining has already failed with CIF_FINAL_ERROR.
30540 (update_caller_keys): Call want_inline_small_function_p before
30542 (update_callee_keys): Likewise.
30544 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30546 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
30548 (rs6000_quadword_masked_address_p): Likewise.
30549 (quad_aligned_load_p): Likewise.
30550 (quad_aligned_store_p): Likewise.
30551 (const_load_sequence_p): Add comment to describe the outer-most loop.
30552 (mimic_memory_attributes_and_flags): New function.
30553 (rs6000_gen_stvx): Likewise.
30554 (replace_swapped_aligned_store): Likewise.
30555 (rs6000_gen_lvx): Likewise.
30556 (replace_swapped_aligned_load): Likewise.
30557 (replace_swapped_load_constant): Capitalize argument name in
30558 comment describing this function.
30559 (rs6000_analyze_swaps): Add a third pass to search for vector loads
30560 and stores that access quad-word aligned addresses and replace
30561 with stvx or lvx instructions when appropriate.
30562 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
30563 New function prototype.
30564 (rs6000_quadword_masked_address_p): Likewise.
30565 (rs6000_gen_lvx): Likewise.
30566 (rs6000_gen_stvx): Likewise.
30567 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
30568 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
30569 when memory address is aligned.
30570 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
30571 this split to select lvx instruction when memory address is aligned.
30572 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
30573 instruction when memory address is aligned.
30574 (*vsx_le_perm_load_v16qi): Likewise.
30575 (four unnamed splitters): Modify to select the stvx instruction
30576 when memory is aligned.
30578 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
30580 * predict.c (determine_unlikely_bbs): Handle correctly BBs
30581 which appears in the queue multiple times.
30583 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30584 Alan Hayward <alan.hayward@arm.com>
30585 David Sherwood <david.sherwood@arm.com>
30587 * tree-vectorizer.h (vec_lower_bound): New structure.
30588 (_loop_vec_info): Add check_nonzero and lower_bounds.
30589 (LOOP_VINFO_CHECK_NONZERO): New macro.
30590 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
30591 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
30592 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
30593 fields. Make seg_len the distance travelled, not including the
30595 (dr_direction_indicator): Declare.
30596 (dr_zero_step_indicator): Likewise.
30597 (dr_known_forward_stride_p): Likewise.
30598 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
30600 (runtime_alias_check_p): Allow runtime alias checks with
30602 (operator ==): Compare access_size and align.
30603 (prune_runtime_alias_test_list): Rework for new distinction between
30604 the access_size and seg_len.
30605 (create_intersect_range_checks_index): Likewise. Cope with polynomial
30607 (get_segment_min_max): New function.
30608 (create_intersect_range_checks): Use it.
30609 (dr_step_indicator): New function.
30610 (dr_direction_indicator): Likewise.
30611 (dr_zero_step_indicator): Likewise.
30612 (dr_known_forward_stride_p): Likewise.
30613 * tree-loop-distribution.c (data_ref_segment_size): Return
30614 DR_STEP * (niters - 1).
30615 (compute_alias_check_pairs): Update call to the dr_with_seg_len
30617 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
30618 (vect_preserves_scalar_order_p): New function, split out from...
30619 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
30620 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
30621 (vect_vfa_access_size): New function.
30622 (vect_vfa_align): Likewise.
30623 (vect_compile_time_alias): Take access_size_a and access_b arguments.
30624 (dump_lower_bound): New function.
30625 (vect_check_lower_bound): Likewise.
30626 (vect_small_gap_p): Likewise.
30627 (vectorizable_with_step_bound_p): Likewise.
30628 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
30629 depencies if the vectorization factor is 1. Convert the checks
30630 for nonzero steps into checks on the bounds of DR_STEP. Try using
30631 a bunds check for variable steps if the minimum required step is
30632 relatively small. Update calls to the dr_with_seg_len
30633 constructor and to vect_compile_time_alias.
30634 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
30636 (vect_loop_versioning): Call it.
30637 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
30639 (vect_estimate_min_profitable_iters): Account for any bounds checks.
30641 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30642 Alan Hayward <alan.hayward@arm.com>
30643 David Sherwood <david.sherwood@arm.com>
30645 * doc/sourcebuild.texi (vect_scatter_store): Document.
30646 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
30648 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
30650 * genopinit.c (main): Add supports_vec_scatter_store and
30651 supports_vec_scatter_store_cached to target_optabs.
30652 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
30653 IFN_MASK_SCATTER_STORE.
30654 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
30656 * internal-fn.h (internal_store_fn_p): Declare.
30657 (internal_fn_stored_value_index): Likewise.
30658 * internal-fn.c (scatter_store_direct): New macro.
30659 (expand_scatter_store_optab_fn): New function.
30660 (direct_scatter_store_optab_supported_p): New macro.
30661 (internal_store_fn_p): New function.
30662 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
30663 IFN_MASK_SCATTER_STORE.
30664 (internal_fn_mask_index): Likewise.
30665 (internal_fn_stored_value_index): New function.
30666 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
30667 for scatter stores.
30668 * optabs-query.h (supports_vec_scatter_store_p): Declare.
30669 * optabs-query.c (supports_vec_scatter_store_p): New function.
30670 * tree-vectorizer.h (vect_get_store_rhs): Declare.
30671 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
30672 true for scatter stores.
30673 (vect_gather_scatter_fn_p): Handle scatter stores too.
30674 (vect_check_gather_scatter): Consider using scatter stores if
30675 supports_vec_scatter_store_p.
30676 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
30677 scatter stores too.
30678 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30679 internal_fn_stored_value_index.
30680 (check_load_store_masking): Handle scatter stores too.
30681 (vect_get_store_rhs): Make public.
30682 (vectorizable_call): Use internal_store_fn_p.
30683 (vectorizable_store): Handle scatter store internal functions.
30684 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
30685 when deciding whether the end of the group has been reached.
30686 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
30687 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
30688 (mask_scatter_store<mode>): New insns.
30690 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30691 Alan Hayward <alan.hayward@arm.com>
30692 David Sherwood <david.sherwood@arm.com>
30694 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
30695 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
30696 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
30698 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
30699 Use vect_truncate_gather_scatter_offset if we can't treat the
30700 operation as a normal gather load or scatter store.
30701 (get_group_load_store_type): Take the gather_scatter_info
30702 as argument. Try using a gather load or scatter store for
30703 single-element groups.
30704 (get_load_store_type): Update calls to get_group_load_store_type
30705 and vect_use_strided_gather_scatters_p.
30707 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30708 Alan Hayward <alan.hayward@arm.com>
30709 David Sherwood <david.sherwood@arm.com>
30711 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
30712 optional tree argument.
30713 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
30715 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
30716 but continue to use the current value as a fallback.
30717 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
30718 to compare the updates.
30719 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
30720 (get_load_store_type): Use it when handling a strided access.
30721 (vect_get_strided_load_store_ops): New function.
30722 (vect_get_data_ptr_increment): Likewise.
30723 (vectorizable_load): Handle strided gather loads. Always pass
30724 a step to vect_create_data_ref_ptr and bump_vector_ptr.
30726 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30727 Alan Hayward <alan.hayward@arm.com>
30728 David Sherwood <david.sherwood@arm.com>
30730 * doc/md.texi (gather_load@var{m}): Document.
30731 (mask_gather_load@var{m}): Likewise.
30732 * genopinit.c (main): Add supports_vec_gather_load and
30733 supports_vec_gather_load_cached to target_optabs.
30734 * optabs-tree.c (init_tree_optimization_optabs): Use
30735 ggc_cleared_alloc to allocate target_optabs.
30736 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
30737 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
30739 * internal-fn.h (internal_load_fn_p): Declare.
30740 (internal_gather_scatter_fn_p): Likewise.
30741 (internal_fn_mask_index): Likewise.
30742 (internal_gather_scatter_fn_supported_p): Likewise.
30743 * internal-fn.c (gather_load_direct): New macro.
30744 (expand_gather_load_optab_fn): New function.
30745 (direct_gather_load_optab_supported_p): New macro.
30746 (direct_internal_fn_optab): New function.
30747 (internal_load_fn_p): Likewise.
30748 (internal_gather_scatter_fn_p): Likewise.
30749 (internal_fn_mask_index): Likewise.
30750 (internal_gather_scatter_fn_supported_p): Likewise.
30751 * optabs-query.c (supports_at_least_one_mode_p): New function.
30752 (supports_vec_gather_load_p): Likewise.
30753 * optabs-query.h (supports_vec_gather_load_p): Declare.
30754 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
30755 and memory_type field.
30756 (NUM_PATTERNS): Bump to 15.
30757 * tree-vect-data-refs.c: Include internal-fn.h.
30758 (vect_gather_scatter_fn_p): New function.
30759 (vect_describe_gather_scatter_call): Likewise.
30760 (vect_check_gather_scatter): Try using internal functions for
30761 gather loads. Recognize existing calls to a gather load function.
30762 (vect_analyze_data_refs): Consider using gather loads if
30763 supports_vec_gather_load_p.
30764 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
30765 (vect_get_gather_scatter_offset_type): Likewise.
30766 (vect_convert_mask_for_vectype): Likewise.
30767 (vect_add_conversion_to_patterm): Likewise.
30768 (vect_try_gather_scatter_pattern): Likewise.
30769 (vect_recog_gather_scatter_pattern): New pattern recognizer.
30770 (vect_vect_recog_func_ptrs): Add it.
30771 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30772 internal_fn_mask_index and internal_gather_scatter_fn_p.
30773 (check_load_store_masking): Take the gather_scatter_info as an
30774 argument and handle gather loads.
30775 (vect_get_gather_scatter_ops): New function.
30776 (vectorizable_call): Check internal_load_fn_p.
30777 (vectorizable_load): Likewise. Handle gather load internal
30779 (vectorizable_store): Update call to check_load_store_masking.
30780 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
30781 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
30782 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
30783 (aarch64_gather_scale_operand_d): New predicates.
30784 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
30785 (mask_gather_load<mode>): New insns.
30787 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30788 Alan Hayward <alan.hayward@arm.com>
30789 David Sherwood <david.sherwood@arm.com>
30791 * optabs.def (fold_left_plus_optab): New optab.
30792 * doc/md.texi (fold_left_plus_@var{m}): Document.
30793 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
30794 * internal-fn.c (fold_left_direct): Define.
30795 (expand_fold_left_optab_fn): Likewise.
30796 (direct_fold_left_optab_supported_p): Likewise.
30797 * fold-const-call.c (fold_const_fold_left): New function.
30798 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
30799 * tree-parloops.c (valid_reduction_p): New function.
30800 (gather_scalar_reductions): Use it.
30801 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
30802 (vect_finish_replace_stmt): Declare.
30803 * tree-vect-loop.c (fold_left_reduction_fn): New function.
30804 (needs_fold_left_reduction_p): New function, split out from...
30805 (vect_is_simple_reduction): ...here. Accept reductions that
30806 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
30807 (vect_force_simple_reduction): Also store the reduction type in
30808 the assignment's STMT_VINFO_REDUC_TYPE.
30809 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
30810 (merge_with_identity): New function.
30811 (vect_expand_fold_left): Likewise.
30812 (vectorize_fold_left_reduction): Likewise.
30813 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
30814 scalar phi in place for it. Check for target support and reject
30815 cases that would reassociate the operation. Defer the transform
30816 phase to vectorize_fold_left_reduction.
30817 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
30818 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
30819 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
30821 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30823 * tree-if-conv.c (predicate_mem_writes): Remove redundant
30824 call to ifc_temp_var.
30826 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30827 Alan Hayward <alan.hayward@arm.com>
30828 David Sherwood <david.sherwood@arm.com>
30830 * target.def (legitimize_address_displacement): Take the original
30831 offset as a poly_int.
30832 * targhooks.h (default_legitimize_address_displacement): Update
30834 * targhooks.c (default_legitimize_address_displacement): Likewise.
30835 * doc/tm.texi: Regenerate.
30836 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
30837 as an argument, moving assert of ad->disp == ad->disp_term to...
30838 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
30839 Try calling targetm.legitimize_address_displacement before expanding
30840 the address rather than afterwards, and adjust for the new interface.
30841 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
30842 Match the new hook interface. Handle SVE addresses.
30843 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
30844 new hook interface.
30846 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30848 * Makefile.in (OBJS): Add early-remat.o.
30849 * target.def (select_early_remat_modes): New hook.
30850 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
30851 * doc/tm.texi: Regenerate.
30852 * targhooks.h (default_select_early_remat_modes): Declare.
30853 * targhooks.c (default_select_early_remat_modes): New function.
30854 * timevar.def (TV_EARLY_REMAT): New timevar.
30855 * passes.def (pass_early_remat): New pass.
30856 * tree-pass.h (make_pass_early_remat): Declare.
30857 * early-remat.c: New file.
30858 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
30860 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
30862 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30863 Alan Hayward <alan.hayward@arm.com>
30864 David Sherwood <david.sherwood@arm.com>
30866 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
30867 vfm1 with a bound_epilog parameter.
30868 (vect_do_peeling): Update calls accordingly, and move the prologue
30869 call earlier in the function. Treat the base bound_epilog as 0 for
30870 fully-masked loops and retain vf - 1 for other loops. Add 1 to
30871 this base when peeling for gaps.
30872 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
30873 with fully-masked loops.
30874 (vect_estimate_min_profitable_iters): Handle the single peeled
30875 iteration in that case.
30877 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30878 Alan Hayward <alan.hayward@arm.com>
30879 David Sherwood <david.sherwood@arm.com>
30881 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
30882 single-element interleaving even if the size is not a power of 2.
30883 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
30884 accesses for single-element interleaving if the group size is
30887 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30888 Alan Hayward <alan.hayward@arm.com>
30889 David Sherwood <david.sherwood@arm.com>
30891 * doc/md.texi (fold_extract_last_@var{m}): Document.
30892 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
30893 * optabs.def (fold_extract_last_optab): New optab.
30894 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
30895 * internal-fn.c (fold_extract_direct): New macro.
30896 (expand_fold_extract_optab_fn): Likewise.
30897 (direct_fold_extract_optab_supported_p): Likewise.
30898 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
30899 * tree-vect-loop.c (vect_model_reduction_cost): Handle
30900 EXTRACT_LAST_REDUCTION.
30901 (get_initial_def_for_reduction): Do not create an initial vector
30902 for EXTRACT_LAST_REDUCTION reductions.
30903 (vectorizable_reduction): Leave the scalar phi in place for
30904 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
30905 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
30906 epilogue code for EXTRACT_LAST_REDUCTION and defer the
30907 transform phase to vectorizable_condition.
30908 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
30910 (vect_finish_stmt_generation): ...here.
30911 (vect_finish_replace_stmt): New function.
30912 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
30913 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
30915 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
30917 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30918 Alan Hayward <alan.hayward@arm.com>
30919 David Sherwood <david.sherwood@arm.com>
30921 * doc/md.texi (extract_last_@var{m}): Document.
30922 * optabs.def (extract_last_optab): New optab.
30923 * internal-fn.def (EXTRACT_LAST): New internal function.
30924 * internal-fn.c (cond_unary_direct): New macro.
30925 (expand_cond_unary_optab_fn): Likewise.
30926 (direct_cond_unary_optab_supported_p): Likewise.
30927 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
30928 loops using EXTRACT_LAST.
30929 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
30930 (extract_last_<mode>): ...this optab.
30931 (vec_extract<mode><Vel>): Update accordingly.
30933 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30934 Alan Hayward <alan.hayward@arm.com>
30935 David Sherwood <david.sherwood@arm.com>
30937 * target.def (empty_mask_is_expensive): New hook.
30938 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
30939 * doc/tm.texi: Regenerate.
30940 * targhooks.h (default_empty_mask_is_expensive): Declare.
30941 * targhooks.c (default_empty_mask_is_expensive): New function.
30942 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
30943 if the target says that empty masks are expensive.
30944 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
30946 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
30948 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30949 Alan Hayward <alan.hayward@arm.com>
30950 David Sherwood <david.sherwood@arm.com>
30952 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
30953 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
30954 (vect_use_loop_mask_for_alignment_p): New function.
30955 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
30956 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
30957 niters_skip argument. Make sure that the first niters_skip elements
30958 of the first iteration are inactive.
30959 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
30960 Update call to vect_set_loop_masks_directly.
30961 (get_misalign_in_elems): New function, split out from...
30962 (vect_gen_prolog_loop_niters): ...here.
30963 (vect_update_init_of_dr): Take a code argument that specifies whether
30964 the adjustment should be added or subtracted.
30965 (vect_update_init_of_drs): Likewise.
30966 (vect_prepare_for_masked_peels): New function.
30967 (vect_do_peeling): Skip prologue peeling if we're using a mask
30968 instead. Update call to vect_update_inits_of_drs.
30969 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
30971 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
30972 alignment. Do not include the number of peeled iterations in
30973 the minimum threshold in that case.
30974 (vectorizable_induction): Adjust the start value down by
30975 LOOP_VINFO_MASK_SKIP_NITERS iterations.
30976 (vect_transform_loop): Call vect_prepare_for_masked_peels.
30977 Take the number of skipped iterations into account when calculating
30979 * tree-vect-stmts.c (vect_gen_while_not): New function.
30981 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30982 Alan Hayward <alan.hayward@arm.com>
30983 David Sherwood <david.sherwood@arm.com>
30985 * doc/sourcebuild.texi (vect_fully_masked): Document.
30986 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
30987 default value to 0.
30988 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
30990 (vect_analyze_loop_2): ...here. Don't check the vectorization
30991 factor against the number of loop iterations if the loop is
30994 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30995 Alan Hayward <alan.hayward@arm.com>
30996 David Sherwood <david.sherwood@arm.com>
30998 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
30999 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
31000 (dump_groups): Update accordingly.
31001 (iv_use::mem_type): New member variable.
31002 (address_p): New function.
31003 (record_use): Add a mem_type argument and initialize the new
31005 (record_group_use): Add a mem_type argument. Use address_p.
31006 Remove obsolete null checks of base_object. Update call to record_use.
31007 (find_interesting_uses_op): Update call to record_group_use.
31008 (find_interesting_uses_cond): Likewise.
31009 (find_interesting_uses_address): Likewise.
31010 (get_mem_type_for_internal_fn): New function.
31011 (find_address_like_use): Likewise.
31012 (find_interesting_uses_stmt): Try find_address_like_use before
31013 calling find_interesting_uses_op.
31014 (addr_offset_valid_p): Use the iv mem_type field as the type
31015 of the addressed memory.
31016 (add_autoinc_candidates): Likewise.
31017 (get_address_cost): Likewise.
31018 (split_small_address_groups_p): Use address_p.
31019 (split_address_groups): Likewise.
31020 (add_iv_candidate_for_use): Likewise.
31021 (autoinc_possible_for_pair): Likewise.
31022 (rewrite_groups): Likewise.
31023 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
31024 (determine_group_iv_cost): Update after split of USE_ADDRESS.
31025 (get_alias_ptr_type_for_ptr_address): New function.
31026 (rewrite_use_address): Rewrite address uses in calls that were
31027 identified by find_address_like_use.
31029 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31030 Alan Hayward <alan.hayward@arm.com>
31031 David Sherwood <david.sherwood@arm.com>
31033 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
31035 * gimple-expr.h (is_gimple_addressable: Likewise.
31036 * gimple-expr.c (is_gimple_address): Likewise.
31037 * internal-fn.c (expand_call_mem_ref): New function.
31038 (expand_mask_load_optab_fn): Use it.
31039 (expand_mask_store_optab_fn): Likewise.
31041 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31042 Alan Hayward <alan.hayward@arm.com>
31043 David Sherwood <david.sherwood@arm.com>
31045 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
31046 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
31047 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
31048 (cond_umax@var{mode}): Document.
31049 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
31050 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
31051 (cond_umin_optab, cond_umax_optab): New optabs.
31052 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
31053 (COND_IOR, COND_XOR): New internal functions.
31054 * internal-fn.h (get_conditional_internal_fn): Declare.
31055 * internal-fn.c (cond_binary_direct): New macro.
31056 (expand_cond_binary_optab_fn): Likewise.
31057 (direct_cond_binary_optab_supported_p): Likewise.
31058 (get_conditional_internal_fn): New function.
31059 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
31060 Cope with reduction statements that are vectorized as calls rather
31062 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
31063 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
31064 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
31065 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
31066 (UNSPEC_COND_EOR): New unspecs.
31067 (optab): Add mappings for them.
31068 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
31069 (sve_int_op, sve_fp_op): New int attributes.
31071 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31072 Alan Hayward <alan.hayward@arm.com>
31073 David Sherwood <david.sherwood@arm.com>
31075 * optabs.def (while_ult_optab): New optab.
31076 * doc/md.texi (while_ult@var{m}@var{n}): Document.
31077 * internal-fn.def (WHILE_ULT): New internal function.
31078 * internal-fn.h (direct_internal_fn_supported_p): New override
31079 that takes two types as argument.
31080 * internal-fn.c (while_direct): New macro.
31081 (expand_while_optab_fn): New function.
31082 (convert_optab_supported_p): Likewise.
31083 (direct_while_optab_supported_p): New macro.
31084 * wide-int.h (wi::udiv_ceil): New function.
31085 * tree-vectorizer.h (rgroup_masks): New structure.
31086 (vec_loop_masks): New typedef.
31087 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
31088 and fully_masked_p.
31089 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
31090 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
31091 (vect_max_vf): New function.
31092 (slpeel_make_loop_iterate_ntimes): Delete.
31093 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
31094 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
31095 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
31096 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
31097 internal-fn.h, stor-layout.h and optabs-query.h.
31098 (vect_set_loop_mask): New function.
31099 (add_preheader_seq): Likewise.
31100 (add_header_seq): Likewise.
31101 (interleave_supported_p): Likewise.
31102 (vect_maybe_permute_loop_masks): Likewise.
31103 (vect_set_loop_masks_directly): Likewise.
31104 (vect_set_loop_condition_masked): Likewise.
31105 (vect_set_loop_condition_unmasked): New function, split out from
31106 slpeel_make_loop_iterate_ntimes.
31107 (slpeel_make_loop_iterate_ntimes): Rename to..
31108 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
31109 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
31110 (vect_do_peeling): Update call accordingly.
31111 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
31113 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
31114 mask_compare_type, can_fully_mask_p and fully_masked_p.
31115 (release_vec_loop_masks): New function.
31116 (_loop_vec_info): Use it to free the loop masks.
31117 (can_produce_all_loop_masks_p): New function.
31118 (vect_get_max_nscalars_per_iter): Likewise.
31119 (vect_verify_full_masking): Likewise.
31120 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
31121 retries, and free the mask rgroups before retrying. Check loop-wide
31122 reasons for disallowing fully-masked loops. Make the final decision
31123 about whether use a fully-masked loop or not.
31124 (vect_estimate_min_profitable_iters): Do not assume that peeling
31125 for the number of iterations will be needed for fully-masked loops.
31126 (vectorizable_reduction): Disable fully-masked loops.
31127 (vectorizable_live_operation): Likewise.
31128 (vect_halve_mask_nunits): New function.
31129 (vect_double_mask_nunits): Likewise.
31130 (vect_record_loop_mask): Likewise.
31131 (vect_get_loop_mask): Likewise.
31132 (vect_transform_loop): Handle the case in which the final loop
31133 iteration might handle a partial vector. Call vect_set_loop_condition
31134 instead of slpeel_make_loop_iterate_ntimes.
31135 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
31136 (check_load_store_masking): New function.
31137 (prepare_load_store_mask): Likewise.
31138 (vectorizable_store): Handle fully-masked loops.
31139 (vectorizable_load): Likewise.
31140 (supportable_widening_operation): Use vect_halve_mask_nunits for
31142 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
31143 (vect_gen_while): New function.
31144 * config/aarch64/aarch64.md (umax<mode>3): New expander.
31145 (aarch64_uqdec<mode>): New insn.
31147 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31148 Alan Hayward <alan.hayward@arm.com>
31149 David Sherwood <david.sherwood@arm.com>
31151 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
31152 (reduc_xor_scal_optab): New optabs.
31153 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
31154 (reduc_xor_scal_@var{m}): Document.
31155 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
31156 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
31157 internal functions.
31158 * fold-const-call.c (fold_const_call): Handle them.
31159 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
31160 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
31161 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
31162 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
31163 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
31164 (UNSPEC_XORV): New unspecs.
31165 (optab): Add entries for them.
31166 (BITWISEV): New int iterator.
31167 (bit_reduc_op): New int attributes.
31169 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31170 Alan Hayward <alan.hayward@arm.com>
31171 David Sherwood <david.sherwood@arm.com>
31173 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
31174 * internal-fn.def (VEC_SHL_INSERT): New internal function.
31175 * optabs.def (vec_shl_insert_optab): New optab.
31176 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
31177 (duplicate_and_interleave): Likewise.
31178 * tree-vect-loop.c: Include internal-fn.h.
31179 (neutral_op_for_slp_reduction): New function, split out from
31180 get_initial_defs_for_reduction.
31181 (get_initial_def_for_reduction): Handle option 2 for variable-length
31182 vectors by loading the neutral value into a vector and then shifting
31183 the initial value into element 0.
31184 (get_initial_defs_for_reduction): Replace the code argument with
31185 the neutral value calculated by neutral_op_for_slp_reduction.
31186 Use gimple_build_vector for constant-length vectors.
31187 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
31188 but the first group_size elements have a neutral value.
31189 Use duplicate_and_interleave otherwise.
31190 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
31191 Update call to get_initial_defs_for_reduction. Handle SLP
31192 reductions for variable-length vectors by creating one vector
31193 result for each scalar result, with the elements associated
31194 with other scalar results stubbed out with the neutral value.
31195 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
31196 Require IFN_VEC_SHL_INSERT for double reductions on
31197 variable-length vectors, or SLP reductions that have
31198 a neutral value. Require can_duplicate_and_interleave_p
31199 support for variable-length unchained SLP reductions if there
31200 is no neutral value, such as for MIN/MAX reductions. Also require
31201 the number of vector elements to be a multiple of the number of
31202 SLP statements when doing variable-length unchained SLP reductions.
31203 Update call to vect_create_epilog_for_reduction.
31204 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
31205 and remove initial values.
31206 (duplicate_and_interleave): Make public.
31207 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
31208 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
31210 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31211 Alan Hayward <alan.hayward@arm.com>
31212 David Sherwood <david.sherwood@arm.com>
31214 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
31215 (can_duplicate_and_interleave_p): New function.
31216 (vect_get_and_check_slp_defs): Take the vector of statements
31217 rather than just the current one. Remove excess parentheses.
31218 Restriction rejectinon of vect_constant_def and vect_external_def
31219 for variable-length vectors to boolean types, or types for which
31220 can_duplicate_and_interleave_p is false.
31221 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
31222 (duplicate_and_interleave): New function.
31223 (vect_get_constant_vectors): Use gimple_build_vector for
31224 constant-length vectors and suitable variable-length constant
31225 vectors. Use duplicate_and_interleave for other variable-length
31226 vectors. Don't defer the update when inserting new statements.
31228 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31229 Alan Hayward <alan.hayward@arm.com>
31230 David Sherwood <david.sherwood@arm.com>
31232 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
31233 min_profitable_iters doesn't go negative.
31235 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31236 Alan Hayward <alan.hayward@arm.com>
31237 David Sherwood <david.sherwood@arm.com>
31239 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
31240 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
31241 * optabs.def (vec_mask_load_lanes_optab): New optab.
31242 (vec_mask_store_lanes_optab): Likewise.
31243 * internal-fn.def (MASK_LOAD_LANES): New internal function.
31244 (MASK_STORE_LANES): Likewise.
31245 * internal-fn.c (mask_load_lanes_direct): New macro.
31246 (mask_store_lanes_direct): Likewise.
31247 (expand_mask_load_optab_fn): Handle masked operations.
31248 (expand_mask_load_lanes_optab_fn): New macro.
31249 (expand_mask_store_optab_fn): Handle masked operations.
31250 (expand_mask_store_lanes_optab_fn): New macro.
31251 (direct_mask_load_lanes_optab_supported_p): Likewise.
31252 (direct_mask_store_lanes_optab_supported_p): Likewise.
31253 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
31255 (vect_load_lanes_supported): Likewise.
31256 * tree-vect-data-refs.c (strip_conversion): New function.
31257 (can_group_stmts_p): Likewise.
31258 (vect_analyze_data_ref_accesses): Use it instead of checking
31259 for a pair of assignments.
31260 (vect_store_lanes_supported): Take a masked_p parameter.
31261 (vect_load_lanes_supported): Likewise.
31262 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
31263 vect_store_lanes_supported and vect_load_lanes_supported.
31264 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
31265 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
31266 parameter. Don't allow gaps for masked accesses.
31267 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
31268 and vect_load_lanes_supported.
31269 (get_load_store_type): Take a masked_p parameter and update
31270 call to get_group_load_store_type.
31271 (vectorizable_store): Update call to get_load_store_type.
31272 Handle IFN_MASK_STORE_LANES.
31273 (vectorizable_load): Update call to get_load_store_type.
31274 Handle IFN_MASK_LOAD_LANES.
31276 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31277 Alan Hayward <alan.hayward@arm.com>
31278 David Sherwood <david.sherwood@arm.com>
31280 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
31282 * config/aarch64/aarch64-protos.h
31283 (aarch64_sve_struct_memory_operand_p): Declare.
31284 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
31285 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
31286 (VPRED, vpred): Handle SVE structure modes.
31287 * config/aarch64/constraints.md (Utx): New constraint.
31288 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
31289 (aarch64_sve_struct_nonimmediate_operand): New predicates.
31290 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
31291 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
31292 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
31293 structure modes. Split into pieces after RA.
31294 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
31295 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
31297 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
31298 SVE structure modes.
31299 (aarch64_classify_address): Likewise.
31300 (sizetochar): Move earlier in file.
31301 (aarch64_print_operand): Handle SVE register lists.
31302 (aarch64_array_mode): New function.
31303 (aarch64_sve_struct_memory_operand_p): Likewise.
31304 (TARGET_ARRAY_MODE): Redefine.
31306 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31307 Alan Hayward <alan.hayward@arm.com>
31308 David Sherwood <david.sherwood@arm.com>
31310 * target.def (array_mode): New target hook.
31311 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
31312 * doc/tm.texi: Regenerate.
31313 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
31314 * hooks.c (hook_optmode_mode_uhwi_none): New function.
31315 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
31316 targetm.array_mode.
31317 * stor-layout.c (mode_for_array): Likewise. Support polynomial
31320 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31321 Alan Hayward <alan.hayward@arm.com>
31322 David Sherwood <david.sherwood@arm.com>
31324 * fold-const.c (fold_binary_loc): Check the argument types
31325 rather than the result type when testing for a vector operation.
31327 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31329 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
31330 * doc/tm.texi: Regenerate.
31332 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31333 Alan Hayward <alan.hayward@arm.com>
31334 David Sherwood <david.sherwood@arm.com>
31336 * doc/invoke.texi (-msve-vector-bits=): Document new option.
31337 (sve): Document new AArch64 extension.
31338 * doc/md.texi (w): Extend the description of the AArch64
31339 constraint to include SVE vectors.
31340 (Upl, Upa): Document new AArch64 predicate constraints.
31341 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
31343 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
31344 (msve-vector-bits=): New option.
31345 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
31346 SVE when these are disabled.
31347 (sve): New extension.
31348 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
31349 modes. Adjust their number of units based on aarch64_sve_vg.
31350 (MAX_BITSIZE_MODE_ANY_MODE): Define.
31351 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
31352 aarch64_addr_query_type.
31353 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
31354 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
31355 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
31356 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
31357 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
31358 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
31359 (aarch64_simd_imm_zero_p): Delete.
31360 (aarch64_check_zero_based_sve_index_immediate): Declare.
31361 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
31362 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
31363 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
31364 (aarch64_sve_float_mul_immediate_p): Likewise.
31365 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
31366 rather than an rtx.
31367 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
31368 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
31369 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
31370 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
31371 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
31372 (aarch64_regmode_natural_size): Likewise.
31373 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
31374 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
31376 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
31377 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
31378 for VG and the SVE predicate registers.
31379 (V_ALIASES): Add a "z"-prefixed alias.
31380 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
31381 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
31382 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
31383 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
31384 (REG_CLASS_NAMES): Add entries for them.
31385 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
31386 and the predicate registers.
31387 (aarch64_sve_vg): Declare.
31388 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
31389 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
31390 (REGMODE_NATURAL_SIZE): Define.
31391 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
31393 * config/aarch64/aarch64.c: Include cfgrtl.h.
31394 (simd_immediate_info): Add a constructor for series vectors,
31395 and an associated step field.
31396 (aarch64_sve_vg): New variable.
31397 (aarch64_dbx_register_number): Handle VG and the predicate registers.
31398 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
31399 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
31400 (VEC_ANY_DATA, VEC_STRUCT): New constants.
31401 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
31402 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
31403 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
31404 (aarch64_get_mask_mode): New functions.
31405 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
31406 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
31407 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
31408 predicate modes and predicate registers. Explicitly restrict
31409 GPRs to modes of 16 bytes or smaller. Only allow FP registers
31410 to store a vector mode if it is recognized by
31411 aarch64_classify_vector_mode.
31412 (aarch64_regmode_natural_size): New function.
31413 (aarch64_hard_regno_caller_save_mode): Return the original mode
31415 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
31416 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
31417 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
31418 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
31420 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
31421 does not overlap dest if the function is frame-related. Handle
31423 (aarch64_split_add_offset): New function.
31424 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
31425 them aarch64_add_offset.
31426 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
31427 and update call to aarch64_sub_sp.
31428 (aarch64_add_cfa_expression): New function.
31429 (aarch64_expand_prologue): Pass extra temporary registers to the
31430 functions above. Handle the case in which we need to emit new
31431 DW_CFA_expressions for registers that were originally saved
31432 relative to the stack pointer, but now have to be expressed
31433 relative to the frame pointer.
31434 (aarch64_output_mi_thunk): Pass extra temporary registers to the
31436 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
31437 IP0 and IP1 values for SVE frames.
31438 (aarch64_expand_vec_series): New function.
31439 (aarch64_expand_sve_widened_duplicate): Likewise.
31440 (aarch64_expand_sve_const_vector): Likewise.
31441 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
31442 Handle SVE constants. Use emit_move_insn to move a force_const_mem
31443 into the register, rather than emitting a SET directly.
31444 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
31445 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
31446 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
31447 (offset_9bit_signed_scaled_p): New functions.
31448 (aarch64_replicate_bitmask_imm): New function.
31449 (aarch64_bitmask_imm): Use it.
31450 (aarch64_cannot_force_const_mem): Reject expressions involving
31451 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
31452 (aarch64_classify_index): Handle SVE indices, by requiring
31453 a plain register index with a scale that matches the element size.
31454 (aarch64_classify_address): Handle SVE addresses. Assert that
31455 the mode of the address is VOIDmode or an integer mode.
31456 Update call to aarch64_classify_symbol.
31457 (aarch64_classify_symbolic_expression): Update call to
31458 aarch64_classify_symbol.
31459 (aarch64_const_vec_all_in_range_p): New function.
31460 (aarch64_print_vector_float_operand): Likewise.
31461 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
31462 "vN" for FP registers with SVE modes. Handle (const ...) vectors
31463 and the FP immediates 1.0 and 0.5.
31464 (aarch64_print_address_internal): Handle SVE addresses.
31465 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
31466 (aarch64_regno_regclass): Handle predicate registers.
31467 (aarch64_secondary_reload): Handle big-endian reloads of SVE
31469 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
31470 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
31471 (aarch64_convert_sve_vector_bits): New function.
31472 (aarch64_override_options): Use it to handle -msve-vector-bits=.
31473 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
31474 rather than an rtx.
31475 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
31476 Handle SVE vector and predicate modes. Accept VL-based constants
31477 that need only one temporary register, and VL offsets that require
31478 no temporary registers.
31479 (aarch64_conditional_register_usage): Mark the predicate registers
31480 as fixed if SVE isn't available.
31481 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
31482 Return true for SVE vector and predicate modes.
31483 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
31484 rather than an unsigned int. Handle SVE modes.
31485 (aarch64_preferred_simd_mode): Update call accordingly. Handle
31487 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
31489 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
31490 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
31491 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
31492 (aarch64_sve_float_mul_immediate_p): New functions.
31493 (aarch64_sve_valid_immediate): New function.
31494 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
31495 Explicitly reject structure modes. Check for INDEX constants.
31496 Handle PTRUE and PFALSE constants.
31497 (aarch64_check_zero_based_sve_index_immediate): New function.
31498 (aarch64_simd_imm_zero_p): Delete.
31499 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
31500 vector modes. Accept constants in the range of CNT[BHWD].
31501 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
31502 ask for an Advanced SIMD mode.
31503 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
31504 (aarch64_simd_vector_alignment): Handle SVE predicates.
31505 (aarch64_vectorize_preferred_vector_alignment): New function.
31506 (aarch64_simd_vector_alignment_reachable): Use it instead of
31508 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
31509 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
31511 (MAX_VECT_LEN): Delete.
31512 (expand_vec_perm_d): Add a vec_flags field.
31513 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
31514 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
31515 (aarch64_evpc_ext): Don't apply a big-endian lane correction
31517 (aarch64_evpc_rev): Rename to...
31518 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
31519 (aarch64_evpc_rev_global): New function.
31520 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
31521 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
31523 (aarch64_evpc_sve_tbl): New function.
31524 (aarch64_expand_vec_perm_const_1): Update after rename of
31525 aarch64_evpc_rev. Handle SVE permutes too, trying
31526 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
31527 than aarch64_evpc_tbl.
31528 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
31529 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
31530 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
31531 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
31532 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
31533 (aarch64_expand_sve_vcond): New functions.
31534 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
31535 of aarch64_vector_mode_p.
31536 (aarch64_dwarf_poly_indeterminate_value): New function.
31537 (aarch64_compute_pressure_classes): Likewise.
31538 (aarch64_can_change_mode_class): Likewise.
31539 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
31540 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
31541 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
31542 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
31543 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
31544 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
31545 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
31546 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
31548 (Dn, Dl, Dr): Accept const as well as const_vector.
31549 (Dz): Likewise. Compare against CONST0_RTX.
31550 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
31551 of "vector" where appropriate.
31552 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
31553 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
31554 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
31555 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
31556 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
31557 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
31558 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
31559 (v_int_equiv): Extend to SVE modes.
31560 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
31562 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
31563 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
31564 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
31565 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
31566 (SVE_COND_FP_CMP): New int iterators.
31567 (perm_hilo): Handle the new unpack unspecs.
31568 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
31570 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
31571 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
31572 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
31573 (aarch64_equality_operator, aarch64_constant_vector_operand)
31574 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
31575 (aarch64_sve_nonimmediate_operand): Likewise.
31576 (aarch64_sve_general_operand): Likewise.
31577 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
31578 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
31579 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
31580 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
31581 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
31582 (aarch64_sve_float_arith_immediate): Likewise.
31583 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
31584 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
31585 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
31586 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
31587 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
31588 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
31589 (aarch64_sve_float_arith_operand): Likewise.
31590 (aarch64_sve_float_arith_with_sub_operand): Likewise.
31591 (aarch64_sve_float_mul_operand): Likewise.
31592 (aarch64_sve_vec_perm_operand): Likewise.
31593 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
31594 (aarch64_mov_operand): Accept const_poly_int and const_vector.
31595 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
31596 as well as const_vector.
31597 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
31598 in file. Use CONST0_RTX and CONSTM1_RTX.
31599 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
31600 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
31601 Use aarch64_simd_imm_zero.
31602 * config/aarch64/aarch64-sve.md: New file.
31603 * config/aarch64/aarch64.md: Include it.
31604 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
31605 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
31606 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
31607 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
31608 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
31609 (sve): New attribute.
31610 (enabled): Disable instructions with the sve attribute unless
31612 (movqi, movhi): Pass CONST_POLY_INT operaneds through
31613 aarch64_expand_mov_immediate.
31614 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
31615 CNT[BHSD] immediates.
31616 (movti): Split CONST_POLY_INT moves into two halves.
31617 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
31618 Split additions that need a temporary here if the destination
31619 is the stack pointer.
31620 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
31621 (*add<mode>3_poly_1): New instruction.
31622 (set_clobber_cc): New expander.
31624 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31626 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
31627 parameter and use it instead of GET_MODE_SIZE (innermode). Use
31628 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
31629 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
31630 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
31631 Change innermode from fixed_mode_size to machine_mode.
31632 (simplify_subreg): Update call accordingly. Handle a constant-sized
31633 subreg of a variable-length CONST_VECTOR.
31635 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31636 Alan Hayward <alan.hayward@arm.com>
31637 David Sherwood <david.sherwood@arm.com>
31639 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
31640 (add_offset_to_base): New function, split out from...
31641 (create_mem_ref): ...here. When handling a scale other than 1,
31642 check first whether the address is valid without the offset.
31643 Add it into the base if so, leaving the index and scale as-is.
31645 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31648 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
31649 fold_for_warn before checking if arg2 is INTEGER_CST.
31651 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
31653 * config/rs6000/predicates.md (load_multiple_operation): Delete.
31654 (store_multiple_operation): Delete.
31655 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
31656 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
31657 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
31658 guarded by TARGET_STRING.
31659 (rs6000_output_load_multiple): Delete.
31660 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
31661 OPTION_MASK_STRING / TARGET_STRING handling.
31662 (print_operand) <'N', 'O'>: Add comment that these are unused now.
31663 (const rs6000_opt_masks) <"string">: Change mask to 0.
31664 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
31665 (MASK_STRING): Delete.
31666 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
31668 (load_multiple): Delete.
31675 (store_multiple): Delete.
31682 (movmemsi_8reg): Delete.
31683 (corresponding unnamed define_insn): Delete.
31684 (movmemsi_6reg): Delete.
31685 (corresponding unnamed define_insn): Delete.
31686 (movmemsi_4reg): Delete.
31687 (corresponding unnamed define_insn): Delete.
31688 (movmemsi_2reg): Delete.
31689 (corresponding unnamed define_insn): Delete.
31690 (movmemsi_1reg): Delete.
31691 (corresponding unnamed define_insn): Delete.
31692 * config/rs6000/rs6000.opt (mno-string): New.
31693 (mstring): Replace by deprecation warning stub.
31694 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
31696 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31698 * regrename.c (regrename_do_replace): If replacing the same
31699 reg multiple times, try to reuse last created gen_raw_REG.
31702 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
31703 main to workaround a bug in GDB.
31705 2018-01-12 Tom de Vries <tom@codesourcery.com>
31708 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
31710 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
31712 PR rtl-optimization/80481
31713 * ira-color.c (get_cap_member): New function.
31714 (allocnos_conflict_by_live_ranges_p): Use it.
31715 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
31716 (setup_slot_coalesced_allocno_live_ranges): Ditto.
31718 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
31721 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
31722 (*saddl_se_1): Ditto.
31723 (*ssubsi_1): Ditto.
31724 (*ssubl_se_1): Ditto.
31726 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31728 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
31729 rather than wi::to_widest for DR_INITs.
31730 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
31731 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
31732 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
31734 (vect_analyze_group_access_1): Note that here.
31736 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31738 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
31739 polynomial type sizes.
31741 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31743 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
31744 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
31745 (gimple_add_tmp_var): Likewise.
31747 2018-01-12 Martin Liska <mliska@suse.cz>
31749 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
31750 (gimple_alloc_sizes): Likewise.
31751 (dump_gimple_statistics): Use PRIu64 in printf format.
31752 * gimple.h: Change uint64_t to int.
31754 2018-01-12 Martin Liska <mliska@suse.cz>
31756 * tree-core.h: Use uint64_t instead of int.
31757 * tree.c (tree_node_counts): Likewise.
31758 (tree_node_sizes): Likewise.
31759 (dump_tree_statistics): Use PRIu64 in printf format.
31761 2018-01-12 Martin Liska <mliska@suse.cz>
31763 * Makefile.in: As qsort_chk is implemented in vec.c, add
31764 vec.o to linkage of gencfn-macros.
31765 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
31766 passing the info to record_node_allocation_statistics.
31767 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
31769 * ggc-common.c (struct ggc_usage): Add operator== and use
31770 it in operator< and compare function.
31771 * mem-stats.h (struct mem_usage): Likewise.
31772 * vec.c (struct vec_usage): Remove operator< and compare
31773 function. Can be simply inherited.
31775 2018-01-12 Martin Jambor <mjambor@suse.cz>
31778 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
31779 * tree-ssa-math-opts.c: Include domwalk.h.
31780 (convert_mult_to_fma_1): New function.
31781 (fma_transformation_info): New type.
31782 (fma_deferring_state): Likewise.
31783 (cancel_fma_deferring): New function.
31784 (result_of_phi): Likewise.
31785 (last_fma_candidate_feeds_initial_phi): Likewise.
31786 (convert_mult_to_fma): Added deferring logic, split actual
31787 transformation to convert_mult_to_fma_1.
31788 (math_opts_dom_walker): New type.
31789 (math_opts_dom_walker::after_dom_children): New method, body moved
31790 here from pass_optimize_widening_mul::execute, added deferring logic
31792 (pass_optimize_widening_mul::execute): Moved most of code to
31793 math_opts_dom_walker::after_dom_children.
31794 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
31795 * config/i386/i386.c (ix86_option_override_internal): Added
31796 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
31798 2018-01-12 Richard Biener <rguenther@suse.de>
31801 * dwarf2out.c (gen_variable_die): Do not reset old_die for
31802 inline instance vars.
31804 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
31807 * config/rx/rx.c (rx_is_restricted_memory_address):
31808 Handle SUBREG case.
31810 2018-01-12 Richard Biener <rguenther@suse.de>
31812 PR tree-optimization/80846
31813 * target.def (split_reduction): New target hook.
31814 * targhooks.c (default_split_reduction): New function.
31815 * targhooks.h (default_split_reduction): Declare.
31816 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
31817 target requests first reduce vectors by combining low and high
31819 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
31820 (get_vectype_for_scalar_type_and_size): Export.
31821 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
31822 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
31823 * doc/tm.texi: Regenerate.
31824 * config/i386/i386.c (ix86_split_reduction): Implement
31825 TARGET_VECTORIZE_SPLIT_REDUCTION.
31827 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31830 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
31831 in PIC mode except for TARGET_VXWORKS_RTP.
31832 * config/sparc/sparc.c: Include cfgrtl.h.
31833 (TARGET_INIT_PIC_REG): Define.
31834 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
31835 (sparc_pic_register_p): New predicate.
31836 (sparc_legitimate_address_p): Use it.
31837 (sparc_legitimize_pic_address): Likewise.
31838 (sparc_delegitimize_address): Likewise.
31839 (sparc_mode_dependent_address_p): Likewise.
31840 (gen_load_pcrel_sym): Remove 4th parameter.
31841 (load_got_register): Adjust call to above. Remove obsolete stuff.
31842 (sparc_expand_prologue): Do not call load_got_register here.
31843 (sparc_flat_expand_prologue): Likewise.
31844 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
31845 (sparc_use_pseudo_pic_reg): New function.
31846 (sparc_init_pic_reg): Likewise.
31847 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
31848 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
31850 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
31852 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31853 Add item for branch_cost.
31855 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31857 PR rtl-optimization/83565
31858 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
31859 not extend the result to a larger mode for rotate operations.
31860 (num_sign_bit_copies1): Likewise.
31862 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31865 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
31867 Use values-Xc.o for -pedantic.
31868 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
31870 2018-01-12 Martin Liska <mliska@suse.cz>
31873 * ipa-devirt.c (final_warning_record::grow_type_warnings):
31875 (possible_polymorphic_call_targets): Use it.
31876 (ipa_devirt): Likewise.
31878 2018-01-12 Martin Liska <mliska@suse.cz>
31880 * profile-count.h (enum profile_quality): Use 0 as invalid
31881 enum value of profile_quality.
31883 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
31885 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
31886 -mext-string options.
31888 2018-01-12 Richard Biener <rguenther@suse.de>
31890 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
31891 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
31892 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
31894 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
31896 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31898 * configure.ac (--with-long-double-format): Add support for the
31899 configuration option to change the default long double format on
31901 * config.gcc (powerpc*-linux*-*): Likewise.
31902 * configure: Regenerate.
31903 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
31904 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
31905 used without modification.
31907 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31909 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
31910 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
31911 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
31912 MISC_BUILTIN_SPEC_BARRIER.
31913 (rs6000_init_builtins): Likewise.
31914 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
31916 (speculation_barrier): New define_insn.
31917 * doc/extend.texi: Document __builtin_speculation_barrier.
31919 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31922 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
31923 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
31924 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
31926 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
31927 integral modes instead of "ss" and "sd".
31928 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
31929 vectors with 32-bit and 64-bit elements.
31930 (vecdupssescalarmodesuffix): New mode attribute.
31931 (vec_dup<mode>): Use it.
31933 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
31936 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
31937 frame if argument is passed on stack.
31939 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31942 * ree.c (combine_reaching_defs): Optimize also
31943 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
31944 reg2=any_extend(exp); reg1=reg2;, formatting fix.
31946 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31948 PR middle-end/83189
31949 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
31951 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31953 PR middle-end/83718
31954 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
31955 after they are computed.
31957 2018-01-11 Bin Cheng <bin.cheng@arm.com>
31959 PR tree-optimization/83695
31960 * gimple-loop-linterchange.cc
31961 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
31962 reset cached scev information after interchange.
31963 (pass_linterchange::execute): Remove call to scev_reset_htab.
31965 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31967 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
31968 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
31969 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
31970 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
31971 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
31972 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
31973 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
31974 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
31975 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
31976 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
31977 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
31978 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
31979 (V_lane_reg): Likewise.
31980 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
31982 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
31983 (vfmal_lane_low<mode>_intrinsic,
31984 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
31985 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
31986 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
31987 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
31988 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
31989 vfmsl_lane_high<mode>_intrinsic): New define_insns.
31991 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31993 * config/arm/arm-cpus.in (fp16fml): New feature.
31994 (ALL_SIMD): Add fp16fml.
31995 (armv8.2-a): Add fp16fml as an option.
31996 (armv8.3-a): Likewise.
31997 (armv8.4-a): Add fp16fml as part of fp16.
31998 * config/arm/arm.h (TARGET_FP16FML): Define.
31999 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
32001 * config/arm/arm-modes.def (V2HF): Define.
32002 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
32003 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
32004 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
32005 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
32006 vfmsl_low, vfmsl_high): New set of builtins.
32007 * config/arm/iterators.md (PLUSMINUS): New code iterator.
32008 (vfml_op): New code attribute.
32009 (VFMLHALVES): New int iterator.
32010 (VFML, VFMLSEL): New mode attributes.
32011 (V_reg): Define mapping for V2HF.
32012 (V_hi, V_lo): New mode attributes.
32013 (VF_constraint): Likewise.
32014 (vfml_half, vfml_half_selector): New int attributes.
32015 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
32017 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
32018 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
32020 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
32021 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
32022 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
32023 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
32025 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
32026 Document new effective target and option set.
32028 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32030 * config/arm/arm-cpus.in (armv8_4): New feature.
32031 (ARMv8_4a): New fgroup.
32032 (armv8.4-a): New arch.
32033 * config/arm/arm-tables.opt: Regenerate.
32034 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
32035 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
32036 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
32037 Add matching rules for -march=armv8.4-a and extensions.
32038 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
32040 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
32043 * config/rx/rx.md (BW): New mode attribute.
32044 (sync_lock_test_and_setsi): Add mode suffix to insn output.
32046 2018-01-11 Richard Biener <rguenther@suse.de>
32048 PR tree-optimization/83435
32049 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
32050 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
32051 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
32053 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32054 Alan Hayward <alan.hayward@arm.com>
32055 David Sherwood <david.sherwood@arm.com>
32057 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
32059 (aarch64_classify_address): Initialize it. Track polynomial offsets.
32060 (aarch64_print_address_internal): Use it to check for a zero offset.
32062 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32063 Alan Hayward <alan.hayward@arm.com>
32064 David Sherwood <david.sherwood@arm.com>
32066 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
32067 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
32068 Return a poly_int64 rather than a HOST_WIDE_INT.
32069 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
32070 rather than a HOST_WIDE_INT.
32071 * config/aarch64/aarch64.h (aarch64_frame): Protect with
32072 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
32073 hard_fp_offset, frame_size, initial_adjust, callee_offset and
32074 final_offset from HOST_WIDE_INT to poly_int64.
32075 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
32076 to_constant when getting the number of units in an Advanced SIMD
32078 (aarch64_builtin_vectorized_function): Check for a constant number
32080 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
32082 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
32083 attribute instead of GET_MODE_NUNITS.
32084 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
32085 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
32086 GET_MODE_SIZE for fixed-size registers.
32087 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
32088 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
32089 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
32090 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
32091 (aarch64_print_operand, aarch64_print_address_internal)
32092 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
32093 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
32094 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
32095 Handle polynomial GET_MODE_SIZE.
32096 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
32097 wider than SImode without modification.
32098 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
32099 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
32100 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
32101 passing and returning SVE modes.
32102 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
32103 rather than GEN_INT.
32104 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
32105 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
32106 (aarch64_allocate_and_probe_stack_space): Likewise.
32107 (aarch64_layout_frame): Cope with polynomial offsets.
32108 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
32109 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
32110 polynomial offsets.
32111 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
32112 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
32113 poly_int64 rather than a HOST_WIDE_INT.
32114 (aarch64_get_separate_components, aarch64_process_components)
32115 (aarch64_expand_prologue, aarch64_expand_epilogue)
32116 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
32117 (aarch64_anchor_offset): New function, split out from...
32118 (aarch64_legitimize_address): ...here.
32119 (aarch64_builtin_vectorization_cost): Handle polynomial
32120 TYPE_VECTOR_SUBPARTS.
32121 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
32123 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
32124 number of elements from the PARALLEL rather than the mode.
32125 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
32126 rather than GET_MODE_BITSIZE.
32127 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
32128 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
32129 (aarch64_expand_vec_perm_const_1): Handle polynomial
32130 d->perm.length () and d->perm elements.
32131 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
32132 Apply to_constant to d->perm elements.
32133 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
32134 polynomial CONST_VECTOR_NUNITS.
32135 (aarch64_move_pointer): Take amount as a poly_int64 rather
32137 (aarch64_progress_pointer): Avoid temporary variable.
32138 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
32139 the mode attribute instead of GET_MODE.
32141 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32142 Alan Hayward <alan.hayward@arm.com>
32143 David Sherwood <david.sherwood@arm.com>
32145 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
32146 x exists before using it.
32147 (aarch64_add_constant_internal): Rename to...
32148 (aarch64_add_offset_1): ...this. Replace regnum with separate
32149 src and dest rtxes. Handle the case in which they're different,
32150 including when the offset is zero. Replace scratchreg with an rtx.
32151 Use 2 additions if there is no spare register into which we can
32152 move a 16-bit constant.
32153 (aarch64_add_constant): Delete.
32154 (aarch64_add_offset): Replace reg with separate src and dest
32155 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
32156 Use aarch64_add_offset_1.
32157 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
32158 an rtx rather than an int. Take the delta as a poly_int64
32159 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
32160 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
32161 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
32162 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
32163 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
32164 and aarch64_add_sp.
32165 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
32166 aarch64_add_constant.
32168 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32170 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
32171 Use scalar_float_mode.
32173 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32175 * config/aarch64/aarch64-simd.md
32176 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
32177 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
32178 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
32179 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
32180 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
32181 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
32182 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
32183 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
32184 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
32185 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
32187 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32190 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
32191 targ_options->x_arm_arch_string is non NULL.
32193 2018-01-11 Tamar Christina <tamar.christina@arm.com>
32195 * config/aarch64/aarch64.h
32196 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
32198 2018-01-11 Sudakshina Das <sudi.das@arm.com>
32201 * expmed.c (emit_store_flag_force): Swap if const op0
32202 and change VOIDmode to mode of op0.
32204 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32206 PR rtl-optimization/83761
32207 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
32208 than bytes to mode_for_size.
32210 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32212 PR middle-end/83189
32213 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
32214 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
32217 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32219 PR middle-end/83575
32220 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
32221 when in layout mode.
32222 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
32223 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
32226 2018-01-10 Michael Collison <michael.collison@arm.com>
32228 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
32229 * config/aarch64/aarch64-option-extension.def: Add
32230 AARCH64_OPT_EXTENSION of 'fp16fml'.
32231 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32232 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
32233 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
32234 * config/aarch64/constraints.md (Ui7): New constraint.
32235 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
32236 (VFMLA_SEL_W): Ditto.
32239 (VFMLA16_LOW): New int iterator.
32240 (VFMLA16_HIGH): Ditto.
32241 (UNSPEC_FMLAL): New unspec.
32242 (UNSPEC_FMLSL): Ditto.
32243 (UNSPEC_FMLAL2): Ditto.
32244 (UNSPEC_FMLSL2): Ditto.
32245 (f16mac): New code attribute.
32246 * config/aarch64/aarch64-simd-builtins.def
32247 (aarch64_fmlal_lowv2sf): Ditto.
32248 (aarch64_fmlsl_lowv2sf): Ditto.
32249 (aarch64_fmlalq_lowv4sf): Ditto.
32250 (aarch64_fmlslq_lowv4sf): Ditto.
32251 (aarch64_fmlal_highv2sf): Ditto.
32252 (aarch64_fmlsl_highv2sf): Ditto.
32253 (aarch64_fmlalq_highv4sf): Ditto.
32254 (aarch64_fmlslq_highv4sf): Ditto.
32255 (aarch64_fmlal_lane_lowv2sf): Ditto.
32256 (aarch64_fmlsl_lane_lowv2sf): Ditto.
32257 (aarch64_fmlal_laneq_lowv2sf): Ditto.
32258 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
32259 (aarch64_fmlalq_lane_lowv4sf): Ditto.
32260 (aarch64_fmlsl_lane_lowv4sf): Ditto.
32261 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
32262 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
32263 (aarch64_fmlal_lane_highv2sf): Ditto.
32264 (aarch64_fmlsl_lane_highv2sf): Ditto.
32265 (aarch64_fmlal_laneq_highv2sf): Ditto.
32266 (aarch64_fmlsl_laneq_highv2sf): Ditto.
32267 (aarch64_fmlalq_lane_highv4sf): Ditto.
32268 (aarch64_fmlsl_lane_highv4sf): Ditto.
32269 (aarch64_fmlalq_laneq_highv4sf): Ditto.
32270 (aarch64_fmlsl_laneq_highv4sf): Ditto.
32271 * config/aarch64/aarch64-simd.md:
32272 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
32273 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
32274 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
32275 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
32276 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
32277 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
32278 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
32279 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
32280 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
32281 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
32282 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
32283 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
32284 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
32285 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
32286 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
32287 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
32288 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
32289 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
32290 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
32291 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
32292 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
32293 (vfmlsl_low_u32): Ditto.
32294 (vfmlalq_low_u32): Ditto.
32295 (vfmlslq_low_u32): Ditto.
32296 (vfmlal_high_u32): Ditto.
32297 (vfmlsl_high_u32): Ditto.
32298 (vfmlalq_high_u32): Ditto.
32299 (vfmlslq_high_u32): Ditto.
32300 (vfmlal_lane_low_u32): Ditto.
32301 (vfmlsl_lane_low_u32): Ditto.
32302 (vfmlal_laneq_low_u32): Ditto.
32303 (vfmlsl_laneq_low_u32): Ditto.
32304 (vfmlalq_lane_low_u32): Ditto.
32305 (vfmlslq_lane_low_u32): Ditto.
32306 (vfmlalq_laneq_low_u32): Ditto.
32307 (vfmlslq_laneq_low_u32): Ditto.
32308 (vfmlal_lane_high_u32): Ditto.
32309 (vfmlsl_lane_high_u32): Ditto.
32310 (vfmlal_laneq_high_u32): Ditto.
32311 (vfmlsl_laneq_high_u32): Ditto.
32312 (vfmlalq_lane_high_u32): Ditto.
32313 (vfmlslq_lane_high_u32): Ditto.
32314 (vfmlalq_laneq_high_u32): Ditto.
32315 (vfmlslq_laneq_high_u32): Ditto.
32316 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
32317 (AARCH64_FL_FOR_ARCH8_4): New.
32318 (AARCH64_ISA_F16FML): New ISA flag.
32319 (TARGET_F16FML): New feature flag for fp16fml.
32320 (doc/invoke.texi): Document new fp16fml option.
32322 2018-01-10 Michael Collison <michael.collison@arm.com>
32324 * config/aarch64/aarch64-builtins.c:
32325 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
32326 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32327 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
32328 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
32329 (AARCH64_ISA_SHA3): New ISA flag.
32330 (TARGET_SHA3): New feature flag for sha3.
32331 * config/aarch64/iterators.md (sha512_op): New int attribute.
32332 (CRYPTO_SHA512): New int iterator.
32333 (UNSPEC_SHA512H): New unspec.
32334 (UNSPEC_SHA512H2): Ditto.
32335 (UNSPEC_SHA512SU0): Ditto.
32336 (UNSPEC_SHA512SU1): Ditto.
32337 * config/aarch64/aarch64-simd-builtins.def
32338 (aarch64_crypto_sha512hqv2di): New builtin.
32339 (aarch64_crypto_sha512h2qv2di): Ditto.
32340 (aarch64_crypto_sha512su0qv2di): Ditto.
32341 (aarch64_crypto_sha512su1qv2di): Ditto.
32342 (aarch64_eor3qv8hi): Ditto.
32343 (aarch64_rax1qv2di): Ditto.
32344 (aarch64_xarqv2di): Ditto.
32345 (aarch64_bcaxqv8hi): Ditto.
32346 * config/aarch64/aarch64-simd.md:
32347 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
32348 (aarch64_crypto_sha512su0qv2di): Ditto.
32349 (aarch64_crypto_sha512su1qv2di): Ditto.
32350 (aarch64_eor3qv8hi): Ditto.
32351 (aarch64_rax1qv2di): Ditto.
32352 (aarch64_xarqv2di): Ditto.
32353 (aarch64_bcaxqv8hi): Ditto.
32354 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
32355 (vsha512h2q_u64): Ditto.
32356 (vsha512su0q_u64): Ditto.
32357 (vsha512su1q_u64): Ditto.
32358 (veor3q_u16): Ditto.
32359 (vrax1q_u64): Ditto.
32360 (vxarq_u64): Ditto.
32361 (vbcaxq_u16): Ditto.
32362 * config/arm/types.md (crypto_sha512): New type attribute.
32363 (crypto_sha3): Ditto.
32364 (doc/invoke.texi): Document new sha3 option.
32366 2018-01-10 Michael Collison <michael.collison@arm.com>
32368 * config/aarch64/aarch64-builtins.c:
32369 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
32370 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32371 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
32372 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
32373 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
32374 (AARCH64_ISA_SM4): New ISA flag.
32375 (TARGET_SM4): New feature flag for sm4.
32376 * config/aarch64/aarch64-simd-builtins.def
32377 (aarch64_sm3ss1qv4si): Ditto.
32378 (aarch64_sm3tt1aq4si): Ditto.
32379 (aarch64_sm3tt1bq4si): Ditto.
32380 (aarch64_sm3tt2aq4si): Ditto.
32381 (aarch64_sm3tt2bq4si): Ditto.
32382 (aarch64_sm3partw1qv4si): Ditto.
32383 (aarch64_sm3partw2qv4si): Ditto.
32384 (aarch64_sm4eqv4si): Ditto.
32385 (aarch64_sm4ekeyqv4si): Ditto.
32386 * config/aarch64/aarch64-simd.md:
32387 (aarch64_sm3ss1qv4si): Ditto.
32388 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
32389 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
32390 (aarch64_sm4eqv4si): Ditto.
32391 (aarch64_sm4ekeyqv4si): Ditto.
32392 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
32393 (sm3part_op): Ditto.
32394 (CRYPTO_SM3TT): Ditto.
32395 (CRYPTO_SM3PART): Ditto.
32396 (UNSPEC_SM3SS1): New unspec.
32397 (UNSPEC_SM3TT1A): Ditto.
32398 (UNSPEC_SM3TT1B): Ditto.
32399 (UNSPEC_SM3TT2A): Ditto.
32400 (UNSPEC_SM3TT2B): Ditto.
32401 (UNSPEC_SM3PARTW1): Ditto.
32402 (UNSPEC_SM3PARTW2): Ditto.
32403 (UNSPEC_SM4E): Ditto.
32404 (UNSPEC_SM4EKEY): Ditto.
32405 * config/aarch64/constraints.md (Ui2): New constraint.
32406 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
32407 * config/arm/types.md (crypto_sm3): New type attribute.
32408 (crypto_sm4): Ditto.
32409 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
32410 (vsm3tt1aq_u32): Ditto.
32411 (vsm3tt1bq_u32): Ditto.
32412 (vsm3tt2aq_u32): Ditto.
32413 (vsm3tt2bq_u32): Ditto.
32414 (vsm3partw1q_u32): Ditto.
32415 (vsm3partw2q_u32): Ditto.
32416 (vsm4eq_u32): Ditto.
32417 (vsm4ekeyq_u32): Ditto.
32418 (doc/invoke.texi): Document new sm4 option.
32420 2018-01-10 Michael Collison <michael.collison@arm.com>
32422 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
32423 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
32424 (AARCH64_FL_FOR_ARCH8_4): New.
32425 (AARCH64_FL_V8_4): New flag.
32426 (doc/invoke.texi): Document new armv8.4-a option.
32428 2018-01-10 Michael Collison <michael.collison@arm.com>
32430 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32431 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
32432 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
32433 * config/aarch64/aarch64-option-extension.def: Add
32434 AARCH64_OPT_EXTENSION of 'sha2'.
32435 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
32436 (crypto): Disable sha2 and aes if crypto disabled.
32437 (crypto): Enable aes and sha2 if enabled.
32438 (simd): Disable sha2 and aes if simd disabled.
32439 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
32441 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
32442 (TARGET_SHA2): New feature flag for sha2.
32443 (TARGET_AES): New feature flag for aes.
32444 * config/aarch64/aarch64-simd.md:
32445 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
32446 conditional on TARGET_AES.
32447 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
32448 (aarch64_crypto_sha1hsi): Make pattern conditional
32450 (aarch64_crypto_sha1hv4si): Ditto.
32451 (aarch64_be_crypto_sha1hv4si): Ditto.
32452 (aarch64_crypto_sha1su1v4si): Ditto.
32453 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
32454 (aarch64_crypto_sha1su0v4si): Ditto.
32455 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
32456 (aarch64_crypto_sha256su0v4si): Ditto.
32457 (aarch64_crypto_sha256su1v4si): Ditto.
32458 (doc/invoke.texi): Document new aes and sha2 options.
32460 2018-01-10 Martin Sebor <msebor@redhat.com>
32462 PR tree-optimization/83781
32463 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
32466 2018-01-11 Martin Sebor <msebor@gmail.com>
32467 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32469 PR tree-optimization/83501
32470 PR tree-optimization/81703
32472 * tree-ssa-strlen.c (get_string_cst): Rename...
32473 (get_string_len): ...to this. Handle global constants.
32474 (handle_char_store): Adjust.
32476 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
32477 Jim Wilson <jimw@sifive.com>
32479 * config/riscv/riscv-protos.h (riscv_output_return): New.
32480 * config/riscv/riscv.c (struct machine_function): New naked_p field.
32481 (riscv_attribute_table, riscv_output_return),
32482 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
32483 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
32484 (riscv_compute_frame_info): Only compute frame->mask if not a naked
32486 (riscv_expand_prologue): Add early return for naked function.
32487 (riscv_expand_epilogue): Likewise.
32488 (riscv_function_ok_for_sibcall): Return false for naked function.
32489 (riscv_set_current_function): New.
32490 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
32491 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
32492 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
32493 * doc/extend.texi (RISC-V Function Attributes): New.
32495 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
32497 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
32498 check for 128-bit long double before checking TCmode.
32499 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
32500 128-bit long doubles before checking TFmode or TCmode.
32501 (FLOAT128_IBM_P): Likewise.
32503 2018-01-10 Martin Sebor <msebor@redhat.com>
32505 PR tree-optimization/83671
32506 * builtins.c (c_strlen): Unconditionally return zero for the empty
32508 Use -Warray-bounds for warnings.
32509 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
32510 for non-constant array indices with COMPONENT_REF, arrays of
32511 arrays, and pointers to arrays.
32512 (gimple_fold_builtin_strlen): Determine and set length range for
32513 non-constant character arrays.
32515 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
32517 PR middle-end/81897
32518 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
32521 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
32523 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
32525 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
32528 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
32529 VECTOR_MEM_ALTIVEC_OR_VSX_P.
32530 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
32531 indexed_or_indirect_operand predicate.
32532 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
32533 (*vsx_le_perm_load_v8hi): Likewise.
32534 (*vsx_le_perm_load_v16qi): Likewise.
32535 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
32536 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
32537 (*vsx_le_perm_store_v8hi): Likewise.
32538 (*vsx_le_perm_store_v16qi): Likewise.
32539 (eight unnamed splitters): Likewise.
32541 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
32543 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
32544 * config/rs6000/emmintrin.h: Likewise.
32545 * config/rs6000/mmintrin.h: Likewise.
32546 * config/rs6000/xmmintrin.h: Likewise.
32548 2018-01-10 David Malcolm <dmalcolm@redhat.com>
32551 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
32553 * tree.c (tree_nop_conversion): Return true for location wrapper
32555 (maybe_wrap_with_location): New function.
32556 (selftest::check_strip_nops): New function.
32557 (selftest::test_location_wrappers): New function.
32558 (selftest::tree_c_tests): Call it.
32559 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
32560 (maybe_wrap_with_location): New decl.
32561 (EXPR_LOCATION_WRAPPER_P): New macro.
32562 (location_wrapper_p): New inline function.
32563 (tree_strip_any_location_wrapper): New inline function.
32565 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
32568 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
32569 stack_realign_offset for the largest alignment of stack slot
32571 (ix86_find_max_used_stack_alignment): New function.
32572 (ix86_finalize_stack_frame_flags): Use it. Set
32573 max_used_stack_alignment if we don't realign stack.
32574 * config/i386/i386.h (machine_function): Add
32575 max_used_stack_alignment.
32577 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
32579 * config/arm/arm.opt (-mbranch-cost): New option.
32580 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
32583 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
32586 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
32587 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
32589 2018-01-10 Richard Biener <rguenther@suse.de>
32592 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
32593 early out so it also covers the case where we have a non-NULL
32596 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
32598 PR tree-optimization/83753
32599 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
32600 for non-strided grouped accesses if the number of elements is 1.
32602 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32605 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
32606 * i386.h (TARGET_USE_GATHER): Define.
32607 * x86-tune.def (X86_TUNE_USE_GATHER): New.
32609 2018-01-10 Martin Liska <mliska@suse.cz>
32612 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
32613 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
32615 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
32616 CLEANUP_NO_PARTITIONING is not set.
32618 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
32620 * doc/rtl.texi: Remove documentation of (const ...) wrappers
32621 for vectors, as a partial revert of r254296.
32622 * rtl.h (const_vec_p): Delete.
32623 (const_vec_duplicate_p): Don't test for vector CONSTs.
32624 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
32625 * expmed.c (make_tree): Likewise.
32628 * common.md (E, F): Use CONSTANT_P instead of checking for
32630 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
32631 checking for CONST_VECTOR.
32633 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32635 PR middle-end/83575
32636 * predict.c (force_edge_cold): Handle in more sane way edges
32637 with no prediction.
32639 2018-01-09 Carl Love <cel@us.ibm.com>
32641 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
32643 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
32644 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
32645 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
32646 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
32647 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
32648 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
32649 * config/rs6000/rs6000-protos.h: Add extern defition for
32650 rs6000_generate_float2_double_code.
32651 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
32653 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
32654 (float2_v2df): Add define_expand.
32656 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
32659 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
32660 op_mode in the force_to_mode call.
32662 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32664 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
32665 instead of checking each element individually.
32666 (aarch64_evpc_uzp): Likewise.
32667 (aarch64_evpc_zip): Likewise.
32668 (aarch64_evpc_ext): Likewise.
32669 (aarch64_evpc_rev): Likewise.
32670 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
32671 instead of checking each element individually. Return true without
32673 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
32674 whether all selected elements come from the same input, instead of
32675 checking each element individually. Remove calls to gen_rtx_REG,
32676 start_sequence and end_sequence and instead assert that no rtl is
32679 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32681 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
32682 order of HIGH and CONST checks.
32684 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32686 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
32687 if the destination isn't an SSA_NAME.
32689 2018-01-09 Richard Biener <rguenther@suse.de>
32691 PR tree-optimization/83668
32692 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
32694 (canonicalize_loop_form): ... here, renamed from ...
32695 (canonicalize_loop_closed_ssa_form): ... this and amended to
32696 swap successor edges for loop exit blocks to make us use
32697 the RPO order we need for initial schedule generation.
32699 2018-01-09 Joseph Myers <joseph@codesourcery.com>
32701 PR tree-optimization/64811
32702 * match.pd: When optimizing comparisons with Inf, avoid
32703 introducing or losing exceptions from comparisons with NaN.
32705 2018-01-09 Martin Liska <mliska@suse.cz>
32708 * asan.c (shadow_mem_size): Add gcc_assert.
32710 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
32712 Don't save registers in main().
32715 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
32716 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
32717 * config/avr/avr.c (avr_set_current_function): Don't error if
32718 naked, OS_task or OS_main are specified at the same time.
32719 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
32721 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
32723 * common/config/avr/avr-common.c (avr_option_optimization_table):
32724 Switch on -mmain-is-OS_task for optimizing compilations.
32726 2018-01-09 Richard Biener <rguenther@suse.de>
32728 PR tree-optimization/83572
32729 * graphite.c: Include cfganal.h.
32730 (graphite_transform_loops): Connect infinite loops to exit
32731 and remove fake edges at the end.
32733 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32735 * ipa-inline.c (edge_badness): Revert accidental checkin.
32737 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32740 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
32741 symbols; not inline clones.
32743 2018-01-09 Jakub Jelinek <jakub@redhat.com>
32746 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
32747 hard registers. Formatting fixes.
32749 PR preprocessor/83722
32750 * gcc.c (try_generate_repro): Pass
32751 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
32752 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
32755 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
32756 Kito Cheng <kito.cheng@gmail.com>
32758 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
32759 (riscv_leaf_function_p): Delete.
32760 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
32762 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32764 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
32766 (do_ifelse): New function.
32767 (do_isel): New function.
32768 (do_sub3): New function.
32769 (do_add3): New function.
32770 (do_load_mask_compare): New function.
32771 (do_overlap_load_compare): New function.
32772 (expand_compare_loop): New function.
32773 (expand_block_compare): Call expand_compare_loop() when appropriate.
32774 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
32775 option description.
32776 (-mblock-compare-inline-loop-limit): New option.
32778 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32781 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
32782 Reverse order of second and third operands in first alternative.
32783 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
32784 of first and second elements in UNSPEC_VPERMR vector.
32785 (altivec_expand_vec_perm_le): Likewise.
32787 2018-01-08 Jeff Law <law@redhat.com>
32789 PR rtl-optimizatin/81308
32790 * tree-switch-conversion.c (cfg_altered): New file scoped static.
32791 (process_switch): If group_case_labels makes a change, then set
32793 (pass_convert_switch::execute): If a switch is converted, then
32794 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
32796 PR rtl-optimization/81308
32797 * recog.c (split_all_insns): Conditionally cleanup the CFG after
32800 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
32802 PR target/83663 - Revert r255946
32803 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
32804 generation for cases where splatting a value is not useful.
32805 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
32806 across a vec_duplicate and a paradoxical subreg forming a vector
32807 mode to a vec_concat.
32809 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32811 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
32812 -march=armv8.3-a variants.
32813 * config/arm/t-multilib: Likewise.
32814 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
32816 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32818 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
32820 (cceq_ior_compare_complement): Give it a name so I can use it, and
32821 change boolean_or_operator predicate to boolean_operator so it can
32822 be used to generate a crand.
32823 (eqne): New code iterator.
32824 (bd/bd_neg): New code_attrs.
32825 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
32826 a single define_insn.
32827 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
32828 decrement (bdnzt/bdnzf/bdzt/bdzf).
32829 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
32830 with the new names of the branch decrement patterns, and added the
32831 names of the branch decrement conditional patterns.
32833 2018-01-08 Richard Biener <rguenther@suse.de>
32835 PR tree-optimization/83563
32836 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
32839 2018-01-08 Richard Biener <rguenther@suse.de>
32841 PR middle-end/83713
32842 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
32844 2018-01-08 Richard Biener <rguenther@suse.de>
32846 PR tree-optimization/83685
32847 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
32848 references to abnormals.
32850 2018-01-08 Richard Biener <rguenther@suse.de>
32853 * dwarf2out.c (output_indirect_strings): Handle empty
32854 skeleton_debug_str_hash.
32855 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
32857 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32859 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
32860 (emit_store_direct): Likewise.
32861 (arc_trampoline_adjust_address): Likewise.
32862 (arc_asm_trampoline_template): New function.
32863 (arc_initialize_trampoline): Use asm_trampoline_template.
32864 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
32865 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
32866 * config/arc/arc.md (flush_icache): Delete pattern.
32868 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32870 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
32871 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
32874 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32877 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
32878 by not USED_FOR_TARGET.
32879 (make_pass_resolve_sw_modes): Likewise.
32881 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32883 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
32886 2018-01-08 Richard Biener <rguenther@suse.de>
32888 PR middle-end/83580
32889 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
32891 2018-01-08 Richard Biener <rguenther@suse.de>
32893 PR middle-end/83517
32894 * match.pd ((t * 2) / 2) -> t): Add missing :c.
32896 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
32898 PR middle-end/81897
32899 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
32900 basic blocks with a small number of successors.
32901 (convert_control_dep_chain_into_preds): Improve handling of
32903 (dump_predicates): Split apart into...
32904 (dump_pred_chain): ...here...
32905 (dump_pred_info): ...and here.
32906 (can_one_predicate_be_invalidated_p): Add debugging printfs.
32907 (can_chain_union_be_invalidated_p): Improve check for invalidation
32909 (uninit_uses_cannot_happen): Avoid unnecessary if
32910 convert_control_dep_chain_into_preds yielded nothing.
32912 2018-01-06 Martin Sebor <msebor@redhat.com>
32914 PR tree-optimization/83640
32915 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
32916 subtracting negative offset from size.
32917 (builtin_access::overlap): Adjust offset bounds of the access to fall
32918 within the size of the object if possible.
32920 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
32922 PR rtl-optimization/83699
32923 * expmed.c (extract_bit_field_1): Restrict the vector usage of
32924 extract_bit_field_as_subreg to cases in which the extracted
32925 value is also a vector.
32927 * lra-constraints.c (process_alt_operands): Test for the equivalence
32928 substitutions when detecting a possible reload cycle.
32930 2018-01-06 Jakub Jelinek <jakub@redhat.com>
32933 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
32934 by default if flag_selective_schedling{,2}. Formatting fixes.
32936 PR rtl-optimization/83682
32937 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
32938 if it has non-VECTOR_MODE element mode.
32939 (vec_duplicate_p): Likewise.
32941 PR middle-end/83694
32942 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
32943 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
32945 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32948 * config/i386/i386-builtin.def
32949 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
32950 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
32951 Require also OPTION_MASK_ISA_AVX512F in addition to
32952 OPTION_MASK_ISA_GFNI.
32953 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
32954 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
32955 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
32956 to OPTION_MASK_ISA_GFNI.
32957 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
32958 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
32959 OPTION_MASK_ISA_AVX512BW.
32960 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
32961 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
32962 addition to OPTION_MASK_ISA_GFNI.
32963 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
32964 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
32965 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
32966 to OPTION_MASK_ISA_GFNI.
32967 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
32968 a requirement for all ISAs rather than any of them with a few
32970 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
32972 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
32973 bitmasks to be enabled with 3 exceptions, instead of requiring any
32974 enabled ISA with lots of exceptions.
32975 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
32976 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
32977 Change avx512bw in isa attribute to avx512f.
32978 * config/i386/sgxintrin.h: Add license boilerplate.
32979 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
32980 to __AVX512F__ and __AVX512VL to __AVX512VL__.
32981 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
32982 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
32984 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
32985 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
32986 temporarily sse2 rather than sse if not enabled already.
32989 * config/i386/sse.md (VI248_VLBW): Rename to ...
32990 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
32991 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
32992 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
32993 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
32994 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
32995 mode iterator instead of VI248_VLBW.
32997 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
32999 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
33000 (record_modified): Skip clobbers; add debug output.
33001 (param_change_prob): Use sreal frequencies.
33003 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
33005 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
33006 punt for user-aligned variables.
33008 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
33010 * tree-chrec.c (chrec_contains_symbols): Return true for
33013 2018-01-05 Sudakshina Das <sudi.das@arm.com>
33016 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
33017 of (x|y) == x for BICS pattern.
33019 2018-01-05 Jakub Jelinek <jakub@redhat.com>
33021 PR tree-optimization/83605
33022 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
33023 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
33026 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
33028 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
33029 * config/epiphany/rtems.h: New file.
33031 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33032 Uros Bizjak <ubizjak@gmail.com>
33035 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
33036 QIreg_operand instead of register_operand predicate.
33037 * config/i386/i386.c (ix86_rop_should_change_byte_p,
33038 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
33039 comments instead of -fmitigate[-_]rop.
33041 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33044 * cgraphunit.c (symbol_table::compile): Switch to text_section
33045 before calling assembly_start debug hook.
33046 * run-rtl-passes.c (run_rtl_passes): Likewise.
33049 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33051 * tree-vrp.c (extract_range_from_binary_expr_1): Check
33052 range_int_cst_p rather than !symbolic_range_p before calling
33053 extract_range_from_multiplicative_op_1.
33055 2018-01-04 Jeff Law <law@redhat.com>
33057 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
33058 redundant test in assertion.
33060 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33062 * doc/rtl.texi: Document machine_mode wrapper classes.
33064 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33066 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
33067 using tree_to_uhwi.
33069 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33071 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
33072 the VEC_PERM_EXPR fold to fail.
33074 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33077 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
33078 to switched_sections.
33080 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33083 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
33084 test for d.testing.
33086 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
33089 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
33090 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
33092 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33095 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
33096 is BLKmode and bitpos not zero or mode change is needed.
33098 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33101 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
33104 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
33107 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
33108 instead of MULT rtx. Update all corresponding splitters.
33109 (*saddl_se): Ditto.
33110 (*ssub<modesuffix>): Ditto.
33111 (*ssubl_se): Ditto.
33112 (*cmp_sadd_di): Update split patterns.
33113 (*cmp_sadd_si): Ditto.
33114 (*cmp_sadd_sidi): Ditto.
33115 (*cmp_ssub_di): Ditto.
33116 (*cmp_ssub_si): Ditto.
33117 (*cmp_ssub_sidi): Ditto.
33118 * config/alpha/predicates.md (const23_operand): New predicate.
33119 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
33120 Look for ASHIFT, not MULT inner operand.
33121 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
33123 2018-01-04 Martin Liska <mliska@suse.cz>
33125 PR gcov-profile/83669
33126 * gcov.c (output_intermediate_file): Add version to intermediate
33128 * doc/gcov.texi: Document new field 'version' in intermediate
33129 file format. Fix location of '-k' option of gcov command.
33131 2018-01-04 Martin Liska <mliska@suse.cz>
33134 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
33136 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33138 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
33140 2018-01-03 Martin Sebor <msebor@redhat.com>
33142 PR tree-optimization/83655
33143 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
33144 checking calls with invalid arguments.
33146 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33148 * tree-vect-stmts.c (vect_get_store_rhs): New function.
33149 (vectorizable_mask_load_store): Delete.
33150 (vectorizable_call): Return false for masked loads and stores.
33151 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
33152 instead of gimple_assign_rhs1.
33153 (vectorizable_load): Handle IFN_MASK_LOAD.
33154 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
33156 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33158 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
33160 (vectorizable_mask_load_store): ...here.
33161 (vectorizable_load): ...and here.
33163 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33165 * tree-vect-stmts.c (vect_build_all_ones_mask)
33166 (vect_build_zero_merge_argument): New functions, split out from...
33167 (vectorizable_load): ...here.
33169 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33171 * tree-vect-stmts.c (vect_check_store_rhs): New function,
33173 (vectorizable_mask_load_store): ...here.
33174 (vectorizable_store): ...and here.
33176 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33178 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
33180 (vectorizable_mask_load_store): ...here.
33182 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33184 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
33185 (vect_model_store_cost): Take a vec_load_store_type instead of a
33187 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
33188 (vect_model_store_cost): Take a vec_load_store_type instead of a
33190 (vectorizable_mask_load_store): Update accordingly.
33191 (vectorizable_store): Likewise.
33192 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
33194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33196 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
33197 IFN_MASK_LOAD calls here rather than...
33198 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
33200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33201 Alan Hayward <alan.hayward@arm.com>
33202 David Sherwood <david.sherwood@arm.com>
33204 * expmed.c (extract_bit_field_1): For vector extracts,
33205 fall back to extract_bit_field_as_subreg if vec_extract
33208 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33209 Alan Hayward <alan.hayward@arm.com>
33210 David Sherwood <david.sherwood@arm.com>
33212 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
33213 they are variable or constant sized.
33214 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
33215 slots for constant-sized data.
33217 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33218 Alan Hayward <alan.hayward@arm.com>
33219 David Sherwood <david.sherwood@arm.com>
33221 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
33222 handling COND_EXPRs with boolean comparisons, try to find a better
33223 basis for the mask type than the boolean itself.
33225 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33227 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
33228 is calculated and how it can be overridden.
33229 * genmodes.c (max_bitsize_mode_any_mode): New variable.
33230 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
33232 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
33235 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33236 Alan Hayward <alan.hayward@arm.com>
33237 David Sherwood <david.sherwood@arm.com>
33239 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
33240 Remove the mode argument.
33241 (aarch64_simd_valid_immediate): Remove the mode and inverse
33243 * config/aarch64/iterators.md (bitsize): New iterator.
33244 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
33245 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
33246 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
33247 aarch64_simd_valid_immediate.
33248 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
33249 (aarch64_reg_or_bic_imm): Likewise.
33250 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
33251 with an insn_type enum and msl with a modifier_type enum.
33252 Replace element_width with a scalar_mode. Change the shift
33253 to unsigned int. Add constructors for scalar_float_mode and
33254 scalar_int_mode elements.
33255 (aarch64_vect_float_const_representable_p): Delete.
33256 (aarch64_can_const_movi_rtx_p)
33257 (aarch64_simd_scalar_immediate_valid_for_move)
33258 (aarch64_simd_make_constant): Update call to
33259 aarch64_simd_valid_immediate.
33260 (aarch64_advsimd_valid_immediate_hs): New function.
33261 (aarch64_advsimd_valid_immediate): Likewise.
33262 (aarch64_simd_valid_immediate): Remove mode and inverse
33263 arguments. Rewrite to use the above. Use const_vec_duplicate_p
33264 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
33265 and aarch64_float_const_representable_p on the result.
33266 (aarch64_output_simd_mov_immediate): Remove mode argument.
33267 Update call to aarch64_simd_valid_immediate and use of
33268 simd_immediate_info.
33269 (aarch64_output_scalar_simd_mov_immediate): Update call
33272 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33273 Alan Hayward <alan.hayward@arm.com>
33274 David Sherwood <david.sherwood@arm.com>
33276 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
33277 (mode_nunits): Likewise CONST_MODE_NUNITS.
33278 * machmode.def (ADJUST_NUNITS): Document.
33279 * genmodes.c (mode_data::need_nunits_adj): New field.
33280 (blank_mode): Update accordingly.
33281 (adj_nunits): New variable.
33282 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
33284 (emit_mode_size_inline): Set need_bytesize_adj for all modes
33285 listed in adj_nunits.
33286 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
33287 listed in adj_nunits. Don't emit case statements for such modes.
33288 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
33289 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
33290 nothing if adj_nunits is nonnull.
33291 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
33292 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
33293 (emit_mode_fbit): Update use of print_maybe_const_decl.
33294 (emit_move_size): Likewise. Treat the array as non-const
33296 (emit_mode_adjustments): Handle adj_nunits.
33298 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33300 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
33301 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
33302 (VECTOR_MODES): Use it.
33303 (make_vector_modes): Take the prefix as an argument.
33305 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33306 Alan Hayward <alan.hayward@arm.com>
33307 David Sherwood <david.sherwood@arm.com>
33309 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
33310 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
33311 for MODE_VECTOR_BOOL.
33312 * machmode.def (VECTOR_BOOL_MODE): Document.
33313 * genmodes.c (VECTOR_BOOL_MODE): New macro.
33314 (make_vector_bool_mode): New function.
33315 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
33317 * lto-streamer-in.c (lto_input_mode_table): Likewise.
33318 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
33320 * stor-layout.c (int_mode_for_mode): Likewise.
33321 * tree.c (build_vector_type_for_mode): Likewise.
33322 * varasm.c (output_constant_pool_2): Likewise.
33323 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
33324 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
33325 for MODE_VECTOR_BOOL.
33326 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
33327 of mode class checks.
33328 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
33329 instead of a list of mode class checks.
33330 (expand_vector_scalar_condition): Likewise.
33331 (type_for_widest_vector_mode): Handle BImode as an inner mode.
33333 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33334 Alan Hayward <alan.hayward@arm.com>
33335 David Sherwood <david.sherwood@arm.com>
33337 * machmode.h (mode_size): Change from unsigned short to
33339 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
33340 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
33341 or if measurement_type is not polynomial.
33342 (fixed_size_mode::includes_p): Check for constant-sized modes.
33343 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
33344 return a poly_uint16 rather than an unsigned short.
33345 (emit_mode_size): Change the type of mode_size from unsigned short
33346 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
33347 (emit_mode_adjustments): Cope with polynomial vector sizes.
33348 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33350 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33352 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
33353 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
33354 * caller-save.c (setup_save_areas): Likewise.
33355 (replace_reg_with_saved_mem): Likewise.
33356 * calls.c (emit_library_call_value_1): Likewise.
33357 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
33358 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
33359 (gen_lowpart_for_combine): Likewise.
33360 * convert.c (convert_to_integer_1): Likewise.
33361 * cse.c (equiv_constant, cse_insn): Likewise.
33362 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
33363 (cselib_subst_to_values): Likewise.
33364 * dce.c (word_dce_process_block): Likewise.
33365 * df-problems.c (df_word_lr_mark_ref): Likewise.
33366 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
33367 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
33368 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
33369 (rtl_for_decl_location): Likewise.
33370 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
33371 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
33372 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
33373 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
33374 (expand_expr_real_1): Likewise.
33375 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
33376 (pad_below): Likewise.
33377 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
33378 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
33379 * ira.c (get_subreg_tracking_sizes): Likewise.
33380 * ira-build.c (ira_create_allocno_objects): Likewise.
33381 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
33382 (ira_sort_regnos_for_alter_reg): Likewise.
33383 * ira-costs.c (record_operand_costs): Likewise.
33384 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
33385 (resolve_simple_move): Likewise.
33386 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
33387 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
33388 (lra_constraints): Likewise.
33389 (CONST_POOL_OK_P): Reject variable-sized modes.
33390 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
33391 (add_pseudo_to_slot, lra_spill): Likewise.
33392 * omp-low.c (omp_clause_aligned_alignment): Likewise.
33393 * optabs-query.c (get_best_extraction_insn): Likewise.
33394 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
33395 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
33396 (expand_mult_highpart, valid_multiword_target_p): Likewise.
33397 * recog.c (offsettable_address_addr_space_p): Likewise.
33398 * regcprop.c (maybe_mode_change): Likewise.
33399 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
33400 * regrename.c (build_def_use): Likewise.
33401 * regstat.c (dump_reg_info): Likewise.
33402 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
33403 (find_reloads, find_reloads_subreg_address): Likewise.
33404 * reload1.c (eliminate_regs_1): Likewise.
33405 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
33406 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
33407 (simplify_binary_operation_1, simplify_subreg): Likewise.
33408 * targhooks.c (default_function_arg_padding): Likewise.
33409 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
33410 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
33411 (verify_gimple_assign_ternary): Likewise.
33412 * tree-inline.c (estimate_move_cost): Likewise.
33413 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33414 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
33415 (get_address_cost_ainc): Likewise.
33416 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33417 (vect_supportable_dr_alignment): Likewise.
33418 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33419 (vectorizable_reduction): Likewise.
33420 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
33421 (vectorizable_operation, vectorizable_load): Likewise.
33422 * tree.c (build_same_sized_truth_vector_type): Likewise.
33423 * valtrack.c (cleanup_auto_inc_dec): Likewise.
33424 * var-tracking.c (emit_note_insn_var_location): Likewise.
33425 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
33426 (ADDR_VEC_ALIGN): Likewise.
33428 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33429 Alan Hayward <alan.hayward@arm.com>
33430 David Sherwood <david.sherwood@arm.com>
33432 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
33434 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
33435 or if measurement_type is polynomial.
33436 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
33437 * combine.c (make_extraction): Likewise.
33438 * dse.c (find_shift_sequence): Likewise.
33439 * dwarf2out.c (mem_loc_descriptor): Likewise.
33440 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
33441 (extract_bit_field, extract_low_bits): Likewise.
33442 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
33443 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
33444 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
33445 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
33446 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
33447 * reload.c (find_reloads): Likewise.
33448 * reload1.c (alter_reg): Likewise.
33449 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
33450 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
33451 * tree-if-conv.c (predicate_mem_writes): Likewise.
33452 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
33453 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
33454 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
33455 * valtrack.c (dead_debug_insert_temp): Likewise.
33456 * varasm.c (mergeable_constant_section): Likewise.
33457 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
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 * expr.c (expand_assignment): Cope with polynomial mode sizes
33464 when assigning to a CONCAT.
33466 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33467 Alan Hayward <alan.hayward@arm.com>
33468 David Sherwood <david.sherwood@arm.com>
33470 * machmode.h (mode_precision): Change from unsigned short to
33472 (mode_to_precision): Return a poly_uint16 rather than an unsigned
33474 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
33475 or if measurement_type is not polynomial.
33476 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
33477 in which the mode is already known to be a scalar_int_mode.
33478 * genmodes.c (emit_mode_precision): Change the type of mode_precision
33479 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
33481 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33482 for GET_MODE_PRECISION.
33483 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33484 for GET_MODE_PRECISION.
33485 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
33487 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
33488 (expand_field_assignment, make_extraction): Likewise.
33489 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
33490 (get_last_value): Likewise.
33491 * convert.c (convert_to_integer_1): Likewise.
33492 * cse.c (cse_insn): Likewise.
33493 * expr.c (expand_expr_real_1): Likewise.
33494 * lra-constraints.c (simplify_operand_subreg): Likewise.
33495 * optabs-query.c (can_atomic_load_p): Likewise.
33496 * optabs.c (expand_atomic_load): Likewise.
33497 (expand_atomic_store): Likewise.
33498 * ree.c (combine_reaching_defs): Likewise.
33499 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
33500 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
33501 * tree.h (type_has_mode_precision_p): Likewise.
33502 * ubsan.c (instrument_si_overflow): Likewise.
33504 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33505 Alan Hayward <alan.hayward@arm.com>
33506 David Sherwood <david.sherwood@arm.com>
33508 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
33509 polynomial numbers of units.
33510 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
33511 (valid_vector_subparts_p): New function.
33512 (build_vector_type): Remove temporary shim and take the number
33513 of units as a poly_uint64 rather than an int.
33514 (build_opaque_vector_type): Take the number of units as a
33515 poly_uint64 rather than an int.
33516 * tree.c (build_vector_from_ctor): Handle polynomial
33517 TYPE_VECTOR_SUBPARTS.
33518 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
33519 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
33520 (build_vector_from_val): If the number of units is variable,
33521 use build_vec_duplicate_cst for constant operands and
33522 VEC_DUPLICATE_EXPR otherwise.
33523 (make_vector_type): Remove temporary is_constant ().
33524 (build_vector_type, build_opaque_vector_type): Take the number of
33525 units as a poly_uint64 rather than an int.
33526 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
33528 * cfgexpand.c (expand_debug_expr): Likewise.
33529 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
33530 (store_constructor, expand_expr_real_1): Likewise.
33531 (const_scalar_mask_from_tree): Likewise.
33532 * fold-const-call.c (fold_const_reduction): Likewise.
33533 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
33534 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
33535 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
33536 (fold_relational_const): Likewise.
33537 (native_interpret_vector): Likewise. Change the size from an
33538 int to an unsigned int.
33539 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
33540 TYPE_VECTOR_SUBPARTS.
33541 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
33542 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
33543 duplicating a non-constant operand into a variable-length vector.
33544 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
33545 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
33546 * ipa-icf.c (sem_variable::equals): Likewise.
33547 * match.pd: Likewise.
33548 * omp-simd-clone.c (simd_clone_subparts): Likewise.
33549 * print-tree.c (print_node): Likewise.
33550 * stor-layout.c (layout_type): Likewise.
33551 * targhooks.c (default_builtin_vectorization_cost): Likewise.
33552 * tree-cfg.c (verify_gimple_comparison): Likewise.
33553 (verify_gimple_assign_binary): Likewise.
33554 (verify_gimple_assign_ternary): Likewise.
33555 (verify_gimple_assign_single): Likewise.
33556 * tree-pretty-print.c (dump_generic_node): Likewise.
33557 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33558 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
33559 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
33560 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
33561 (vect_shift_permute_load_chain): Likewise.
33562 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
33563 (expand_vector_condition, optimize_vector_constructor): Likewise.
33564 (lower_vec_perm, get_compute_type): Likewise.
33565 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33566 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
33567 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
33568 (vect_recog_mask_conversion_pattern): Likewise.
33569 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
33570 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
33571 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33572 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
33573 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
33574 (vectorizable_shift, vectorizable_operation, vectorizable_store)
33575 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
33576 (supportable_widening_operation): Likewise.
33577 (supportable_narrowing_operation): Likewise.
33578 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
33580 * varasm.c (output_constant): Likewise.
33582 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33583 Alan Hayward <alan.hayward@arm.com>
33584 David Sherwood <david.sherwood@arm.com>
33586 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
33587 so that both the length == 3 and length != 3 cases set up their
33588 own permute vectors. Add comments explaining why we know the
33589 number of elements is constant.
33590 (vect_permute_load_chain): Likewise.
33592 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33593 Alan Hayward <alan.hayward@arm.com>
33594 David Sherwood <david.sherwood@arm.com>
33596 * machmode.h (mode_nunits): Change from unsigned char to
33598 (ONLY_FIXED_SIZE_MODES): New macro.
33599 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
33600 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
33601 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
33603 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
33604 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
33605 or if measurement_type is not polynomial.
33606 * genmodes.c (ZERO_COEFFS): New macro.
33607 (emit_mode_nunits_inline): Make mode_nunits_inline return a
33609 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
33610 Use ZERO_COEFFS when emitting initializers.
33611 * data-streamer.h (bp_pack_poly_value): New function.
33612 (bp_unpack_poly_value): Likewise.
33613 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33614 for GET_MODE_NUNITS.
33615 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33616 for GET_MODE_NUNITS.
33617 * tree.c (make_vector_type): Remove temporary shim and make
33618 the real function take the number of units as a poly_uint64
33619 rather than an int.
33620 (build_vector_type_for_mode): Handle polynomial nunits.
33621 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
33622 * emit-rtl.c (const_vec_series_p_1): Likewise.
33623 (gen_rtx_CONST_VECTOR): Likewise.
33624 * fold-const.c (test_vec_duplicate_folding): Likewise.
33625 * genrecog.c (validate_pattern): Likewise.
33626 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
33627 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
33628 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
33629 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
33630 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
33631 * rtlanal.c (subreg_get_info): Likewise.
33632 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33633 (vect_grouped_load_supported): Likewise.
33634 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
33635 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
33636 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
33637 (simplify_const_unary_operation, simplify_binary_operation_1)
33638 (simplify_const_binary_operation, simplify_ternary_operation)
33639 (test_vector_ops_duplicate, test_vector_ops): Likewise.
33640 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
33641 instead of CONST_VECTOR_NUNITS.
33642 * varasm.c (output_constant_pool_2): Likewise.
33643 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
33644 explicit-encoded elements in the XVEC for variable-length vectors.
33646 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33648 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
33650 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33651 Alan Hayward <alan.hayward@arm.com>
33652 David Sherwood <david.sherwood@arm.com>
33654 * coretypes.h (fixed_size_mode): Declare.
33655 (fixed_size_mode_pod): New typedef.
33656 * builtins.h (target_builtins::x_apply_args_mode)
33657 (target_builtins::x_apply_result_mode): Change type to
33658 fixed_size_mode_pod.
33659 * builtins.c (apply_args_size, apply_result_size, result_vector)
33660 (expand_builtin_apply_args_1, expand_builtin_apply)
33661 (expand_builtin_return): Update accordingly.
33663 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33665 * cse.c (hash_rtx_cb): Hash only the encoded elements.
33666 * cselib.c (cselib_hash_rtx): Likewise.
33667 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
33668 CONST_VECTOR encoding.
33670 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33671 Jeff Law <law@redhat.com>
33674 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
33675 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
33676 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
33677 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
33680 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
33681 explicitly probe *sp in a noreturn function if there were any callee
33682 register saves or frame pointer is needed.
33684 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33687 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
33688 BLKmode for ternary, binary or unary expressions.
33691 * var-tracking.c (delete_vta_debug_insn): New inline function.
33692 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
33693 insns from get_insns () to NULL instead of each bb separately.
33694 Use delete_vta_debug_insn. No longer static.
33695 (vt_debug_insns_local, variable_tracking_main_1): Adjust
33696 delete_vta_debug_insns callers.
33697 * rtl.h (delete_vta_debug_insns): Declare.
33698 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
33699 instead of variable_tracking_main.
33701 2018-01-03 Martin Sebor <msebor@redhat.com>
33703 PR tree-optimization/83603
33704 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
33705 arguments past the endof the argument list in functions declared
33706 without a prototype.
33707 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
33708 Avoid checking when arguments are null.
33710 2018-01-03 Martin Sebor <msebor@redhat.com>
33713 * doc/extend.texi (attribute const): Fix a typo.
33714 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
33715 issuing -Wsuggest-attribute for void functions.
33717 2018-01-03 Martin Sebor <msebor@redhat.com>
33719 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
33720 offset_int::from instead of wide_int::to_shwi.
33721 (maybe_diag_overlap): Remove assertion.
33722 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
33723 * gimple-ssa-sprintf.c (format_directive): Same.
33724 (parse_directive): Same.
33725 (sprintf_dom_walker::compute_format_length): Same.
33726 (try_substitute_return_value): Same.
33728 2018-01-03 Jeff Law <law@redhat.com>
33730 PR middle-end/83654
33731 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
33732 non-constant residual for zero at runtime and avoid probing in
33733 that case. Reorganize code for trailing problem to mirror handling
33736 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33738 PR tree-optimization/83501
33739 * tree-ssa-strlen.c (get_string_cst): New.
33740 (handle_char_store): Call get_string_cst.
33742 2018-01-03 Martin Liska <mliska@suse.cz>
33744 PR tree-optimization/83593
33745 * tree-ssa-strlen.c: Include tree-cfg.h.
33746 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
33747 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
33748 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
33750 (strlen_dom_walker::before_dom_children): Call
33751 gimple_purge_dead_eh_edges. Dump tranformation with details
33753 (strlen_dom_walker::before_dom_children): Update call by adding
33754 new argument cleanup_eh.
33755 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
33757 2018-01-03 Martin Liska <mliska@suse.cz>
33760 * cif-code.def (VARIADIC_THUNK): New enum value.
33761 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
33764 2018-01-03 Jan Beulich <jbeulich@suse.com>
33766 * sse.md (mov<mode>_internal): Tighten condition for when to use
33767 vmovdqu<ssescalarsize> for TI and OI modes.
33769 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33771 Update copyright years.
33773 2018-01-03 Martin Liska <mliska@suse.cz>
33776 * ipa-visibility.c (function_and_variable_visibility): Skip
33777 functions with noipa attribure.
33779 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33781 * gcc.c (process_command): Update copyright notice dates.
33782 * gcov-dump.c (print_version): Ditto.
33783 * gcov.c (print_version): Ditto.
33784 * gcov-tool.c (print_version): Ditto.
33785 * gengtype.c (create_file): Ditto.
33786 * doc/cpp.texi: Bump @copying's copyright year.
33787 * doc/cppinternals.texi: Ditto.
33788 * doc/gcc.texi: Ditto.
33789 * doc/gccint.texi: Ditto.
33790 * doc/gcov.texi: Ditto.
33791 * doc/install.texi: Ditto.
33792 * doc/invoke.texi: Ditto.
33794 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33796 * vector-builder.h (vector_builder::m_full_nelts): Change from
33797 unsigned int to poly_uint64.
33798 (vector_builder::full_nelts): Update prototype accordingly.
33799 (vector_builder::new_vector): Likewise.
33800 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
33801 (vector_builder::operator ==): Likewise.
33802 (vector_builder::finalize): Likewise.
33803 * int-vector-builder.h (int_vector_builder::int_vector_builder):
33804 Take the number of elements as a poly_uint64 rather than an
33806 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
33807 from unsigned int to poly_uint64.
33808 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
33809 (vec_perm_indices::new_vector): Likewise.
33810 (vec_perm_indices::length): Likewise.
33811 (vec_perm_indices::nelts_per_input): Likewise.
33812 (vec_perm_indices::input_nelts): Likewise.
33813 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
33814 number of elements per input as a poly_uint64 rather than an
33815 unsigned int. Use the original encoding for variable-length
33816 vectors, rather than clamping each individual element.
33817 For the second and subsequent elements in each pattern,
33818 clamp the step and base before clamping their sum.
33819 (vec_perm_indices::series_p): Handle polynomial element counts.
33820 (vec_perm_indices::all_in_range_p): Likewise.
33821 (vec_perm_indices_to_tree): Likewise.
33822 (vec_perm_indices_to_rtx): Likewise.
33823 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
33824 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
33825 (tree_vector_builder::new_binary_operation): Handle polynomial
33826 element counts. Return false if we need to know the number
33827 of elements at compile time.
33828 * fold-const.c (fold_vec_perm): Punt if the number of elements
33829 isn't known at compile time.
33831 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33833 * vec-perm-indices.h (vec_perm_builder): Change element type
33834 from HOST_WIDE_INT to poly_int64.
33835 (vec_perm_indices::element_type): Update accordingly.
33836 (vec_perm_indices::clamp): Handle polynomial element_types.
33837 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33838 (vec_perm_indices::all_in_range_p): Likewise.
33839 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
33841 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
33842 polynomial vec_perm_indices element types.
33843 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
33844 * fold-const.c (fold_vec_perm): Likewise.
33845 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
33846 * tree-vect-generic.c (lower_vec_perm): Likewise.
33847 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33848 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
33849 element type to HOST_WIDE_INT.
33851 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33852 Alan Hayward <alan.hayward@arm.com>
33853 David Sherwood <david.sherwood@arm.com>
33855 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
33856 rather than an int. Use plus_constant.
33857 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
33858 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
33860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33861 Alan Hayward <alan.hayward@arm.com>
33862 David Sherwood <david.sherwood@arm.com>
33864 * calls.c (emit_call_1, expand_call): Change struct_value_size from
33865 a HOST_WIDE_INT to a poly_int64.
33867 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33868 Alan Hayward <alan.hayward@arm.com>
33869 David Sherwood <david.sherwood@arm.com>
33871 * calls.c (load_register_parameters): Cope with polynomial
33872 mode sizes. Require a constant size for BLKmode parameters
33873 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
33874 forces a parameter to be padded at the lsb end in order to
33875 fill a complete number of words, require the parameter size
33876 to be ordered wrt UNITS_PER_WORD.
33878 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33879 Alan Hayward <alan.hayward@arm.com>
33880 David Sherwood <david.sherwood@arm.com>
33882 * reload1.c (spill_stack_slot_width): Change element type
33883 from unsigned int to poly_uint64_pod.
33884 (alter_reg): Treat mode sizes as polynomial.
33886 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33887 Alan Hayward <alan.hayward@arm.com>
33888 David Sherwood <david.sherwood@arm.com>
33890 * reload.c (complex_word_subreg_p): New function.
33891 (reload_inner_reg_of_subreg, push_reload): Use it.
33893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33894 Alan Hayward <alan.hayward@arm.com>
33895 David Sherwood <david.sherwood@arm.com>
33897 * lra-constraints.c (process_alt_operands): Reject matched
33898 operands whose sizes aren't ordered.
33899 (match_reload): Refer to this check here.
33901 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33902 Alan Hayward <alan.hayward@arm.com>
33903 David Sherwood <david.sherwood@arm.com>
33905 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
33906 that the mode size is in the set {1, 2, 4, 8, 16}.
33908 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33909 Alan Hayward <alan.hayward@arm.com>
33910 David Sherwood <david.sherwood@arm.com>
33912 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
33913 Use plus_constant instead of gen_rtx_PLUS.
33915 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33916 Alan Hayward <alan.hayward@arm.com>
33917 David Sherwood <david.sherwood@arm.com>
33919 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
33920 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
33921 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
33922 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
33923 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
33924 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
33925 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
33926 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
33927 * config/i386/i386.c (ix86_push_rounding): ...this new function.
33928 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
33930 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
33931 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
33932 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
33933 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
33934 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
33935 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
33936 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
33937 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
33938 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
33939 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
33941 * expr.c (emit_move_resolve_push): Treat the input and result
33942 of PUSH_ROUNDING as a poly_int64.
33943 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
33944 (emit_push_insn): Likewise.
33945 * lra-eliminations.c (mark_not_eliminable): Likewise.
33946 * recog.c (push_operand): Likewise.
33947 * reload1.c (elimination_effects): Likewise.
33948 * rtlanal.c (nonzero_bits1): Likewise.
33949 * calls.c (store_one_arg): Likewise. Require the padding to be
33950 known at compile time.
33952 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33953 Alan Hayward <alan.hayward@arm.com>
33954 David Sherwood <david.sherwood@arm.com>
33956 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
33957 Use plus_constant instead of gen_rtx_PLUS.
33959 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33960 Alan Hayward <alan.hayward@arm.com>
33961 David Sherwood <david.sherwood@arm.com>
33963 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
33964 rather than an int.
33966 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33967 Alan Hayward <alan.hayward@arm.com>
33968 David Sherwood <david.sherwood@arm.com>
33970 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
33971 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
33972 via stack temporaries. Treat the mode size as polynomial too.
33974 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33975 Alan Hayward <alan.hayward@arm.com>
33976 David Sherwood <david.sherwood@arm.com>
33978 * expr.c (expand_expr_real_2): When handling conversions involving
33979 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
33980 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
33981 as a poly_uint64 too.
33983 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33984 Alan Hayward <alan.hayward@arm.com>
33985 David Sherwood <david.sherwood@arm.com>
33987 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
33989 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33990 Alan Hayward <alan.hayward@arm.com>
33991 David Sherwood <david.sherwood@arm.com>
33993 * combine.c (can_change_dest_mode): Handle polynomial
33994 REGMODE_NATURAL_SIZE.
33995 * expmed.c (store_bit_field_1): Likewise.
33996 * expr.c (store_constructor): Likewise.
33997 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
33998 and polynomial REGMODE_NATURAL_SIZE.
33999 (gen_lowpart_common): Likewise.
34000 * reginfo.c (record_subregs_of_mode): Likewise.
34001 * rtlanal.c (read_modify_subreg_p): Likewise.
34003 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34004 Alan Hayward <alan.hayward@arm.com>
34005 David Sherwood <david.sherwood@arm.com>
34007 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
34008 numbers of elements.
34010 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34011 Alan Hayward <alan.hayward@arm.com>
34012 David Sherwood <david.sherwood@arm.com>
34014 * match.pd: Cope with polynomial numbers of vector elements.
34016 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34017 Alan Hayward <alan.hayward@arm.com>
34018 David Sherwood <david.sherwood@arm.com>
34020 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
34021 in a POINTER_PLUS_EXPR.
34023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34024 Alan Hayward <alan.hayward@arm.com>
34025 David Sherwood <david.sherwood@arm.com>
34027 * omp-simd-clone.c (simd_clone_subparts): New function.
34028 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
34029 (ipa_simd_modify_function_body): Likewise.
34031 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34032 Alan Hayward <alan.hayward@arm.com>
34033 David Sherwood <david.sherwood@arm.com>
34035 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
34036 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
34037 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
34038 (expand_vector_condition, vector_element): Likewise.
34039 (subparts_gt): New function.
34040 (get_compute_type): Use subparts_gt.
34041 (count_type_subparts): Delete.
34042 (expand_vector_operations_1): Use subparts_gt instead of
34043 count_type_subparts.
34045 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34046 Alan Hayward <alan.hayward@arm.com>
34047 David Sherwood <david.sherwood@arm.com>
34049 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
34050 (vect_compile_time_alias): ...this new function. Do the calculation
34051 on poly_ints rather than trees.
34052 (vect_prune_runtime_alias_test_list): Update call accordingly.
34054 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34055 Alan Hayward <alan.hayward@arm.com>
34056 David Sherwood <david.sherwood@arm.com>
34058 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
34060 (vect_schedule_slp_instance): Likewise.
34062 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34063 Alan Hayward <alan.hayward@arm.com>
34064 David Sherwood <david.sherwood@arm.com>
34066 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
34067 constant and extern definitions for variable-length vectors.
34068 (vect_get_constant_vectors): Note that the number of units
34069 is known to be constant.
34071 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34072 Alan Hayward <alan.hayward@arm.com>
34073 David Sherwood <david.sherwood@arm.com>
34075 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
34076 of units as polynomial. Choose between WIDE and NARROW based
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-vect-stmts.c (simd_clone_subparts): New function.
34084 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
34086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34087 Alan Hayward <alan.hayward@arm.com>
34088 David Sherwood <david.sherwood@arm.com>
34090 * tree-vect-stmts.c (vectorizable_call): Treat the number of
34091 vectors as polynomial. Use build_index_vector for
34092 IFN_GOMP_SIMD_LANE.
34094 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34095 Alan Hayward <alan.hayward@arm.com>
34096 David Sherwood <david.sherwood@arm.com>
34098 * tree-vect-stmts.c (get_load_store_type): Treat the number of
34099 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
34100 for variable-length vectors.
34101 (vectorizable_mask_load_store): Treat the number of units as
34102 polynomial, asserting that it is constant if the condition has
34103 already been enforced.
34104 (vectorizable_store, vectorizable_load): Likewise.
34106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34107 Alan Hayward <alan.hayward@arm.com>
34108 David Sherwood <david.sherwood@arm.com>
34110 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
34111 of units as polynomial. Punt if we can't tell at compile time
34112 which vector contains the final result.
34114 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34115 Alan Hayward <alan.hayward@arm.com>
34116 David Sherwood <david.sherwood@arm.com>
34118 * tree-vect-loop.c (vectorizable_induction): Treat the number
34119 of units as polynomial. Punt on SLP inductions. Use an integer
34120 VEC_SERIES_EXPR for variable-length integer reductions. Use a
34121 cast of such a series for variable-length floating-point
34124 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34125 Alan Hayward <alan.hayward@arm.com>
34126 David Sherwood <david.sherwood@arm.com>
34128 * tree.h (build_index_vector): Declare.
34129 * tree.c (build_index_vector): New function.
34130 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
34131 of units as polynomial, forcibly converting it to a constant if
34132 vectorizable_reduction has already enforced the condition.
34133 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
34134 to create a {1,2,3,...} vector.
34135 (vectorizable_reduction): Treat the number of units as polynomial.
34136 Choose vectype_in based on the largest scalar element size rather
34137 than the smallest number of units. Enforce the restrictions
34140 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34141 Alan Hayward <alan.hayward@arm.com>
34142 David Sherwood <david.sherwood@arm.com>
34144 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
34145 number of units as polynomial.
34147 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34148 Alan Hayward <alan.hayward@arm.com>
34149 David Sherwood <david.sherwood@arm.com>
34151 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
34152 * target.def (autovectorize_vector_sizes): Return the vector sizes
34153 by pointer, using vector_sizes rather than a bitmask.
34154 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
34155 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
34156 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
34158 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
34159 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
34160 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
34161 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
34162 * omp-general.c (omp_max_vf): Likewise.
34163 * omp-low.c (omp_clause_aligned_alignment): Likewise.
34164 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
34165 * tree-vect-loop.c (vect_analyze_loop): Likewise.
34166 * tree-vect-slp.c (vect_slp_bb): Likewise.
34167 * doc/tm.texi: Regenerate.
34168 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
34170 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
34171 the vector size as a poly_uint64 rather than an unsigned int.
34172 (current_vector_size): Change from an unsigned int to a poly_uint64.
34173 (get_vectype_for_scalar_type): Update accordingly.
34174 * tree.h (build_truth_vector_type): Take the size and number of
34175 units as a poly_uint64 rather than an unsigned int.
34176 (build_vector_type): Add a temporary overload that takes
34177 the number of units as a poly_uint64 rather than an unsigned int.
34178 * tree.c (make_vector_type): Likewise.
34179 (build_truth_vector_type): Take the number of units as a poly_uint64
34180 rather than an unsigned int.
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 * target.def (get_mask_mode): Take the number of units and length
34187 as poly_uint64s rather than unsigned ints.
34188 * targhooks.h (default_get_mask_mode): Update accordingly.
34189 * targhooks.c (default_get_mask_mode): Likewise.
34190 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
34191 * doc/tm.texi: Regenerate.
34193 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34194 Alan Hayward <alan.hayward@arm.com>
34195 David Sherwood <david.sherwood@arm.com>
34197 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
34198 * omp-general.c (omp_max_vf): Likewise.
34199 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
34200 (expand_omp_simd): Handle polynomial safelen.
34201 * omp-low.c (omplow_simd_context): Add a default constructor.
34202 (omplow_simd_context::max_vf): Change from int to poly_uint64.
34203 (lower_rec_simd_input_clauses): Update accordingly.
34204 (lower_rec_input_clauses): Likewise.
34206 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34207 Alan Hayward <alan.hayward@arm.com>
34208 David Sherwood <david.sherwood@arm.com>
34210 * tree-vectorizer.h (vect_nunits_for_cost): New function.
34211 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
34212 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
34213 (vect_analyze_slp_cost): Likewise.
34214 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
34215 (vect_model_load_cost): Likewise.
34217 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34218 Alan Hayward <alan.hayward@arm.com>
34219 David Sherwood <david.sherwood@arm.com>
34221 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
34222 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
34223 from an unsigned int * to a poly_uint64_pod *.
34224 (calculate_unrolling_factor): New function.
34225 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
34227 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34228 Alan Hayward <alan.hayward@arm.com>
34229 David Sherwood <david.sherwood@arm.com>
34231 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
34232 from an unsigned int to a poly_uint64.
34233 (_loop_vec_info::slp_unrolling_factor): Likewise.
34234 (_loop_vec_info::vectorization_factor): Change from an int
34236 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
34237 (vect_get_num_vectors): New function.
34238 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
34239 (vect_get_num_copies): Use vect_get_num_vectors.
34240 (vect_analyze_data_ref_dependences): Change max_vf from an int *
34241 to an unsigned int *.
34242 (vect_analyze_data_refs): Change min_vf from an int * to a
34244 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
34245 than an unsigned HOST_WIDE_INT.
34246 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
34247 (vect_analyze_data_ref_dependence): Change max_vf from an int *
34248 to an unsigned int *.
34249 (vect_analyze_data_ref_dependences): Likewise.
34250 (vect_compute_data_ref_alignment): Handle polynomial vf.
34251 (vect_enhance_data_refs_alignment): Likewise.
34252 (vect_prune_runtime_alias_test_list): Likewise.
34253 (vect_shift_permute_load_chain): Likewise.
34254 (vect_supportable_dr_alignment): Likewise.
34255 (dependence_distance_ge_vf): Take the vectorization factor as a
34256 poly_uint64 rather than an unsigned HOST_WIDE_INT.
34257 (vect_analyze_data_refs): Change min_vf from an int * to a
34259 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
34260 vfm1 as a poly_uint64 rather than an int. Make the same change
34261 for the returned bound_scalar.
34262 (vect_gen_vector_loop_niters): Handle polynomial vf.
34263 (vect_do_peeling): Likewise. Update call to
34264 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
34265 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
34267 * tree-vect-loop.c (vect_determine_vectorization_factor)
34268 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
34269 (vect_get_known_peeling_cost): Likewise.
34270 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
34271 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
34272 (vect_transform_loop): Likewise. Use the lowest possible VF when
34273 updating the upper bounds of the loop.
34274 (vect_min_worthwhile_factor): Make static. Return an unsigned int
34275 rather than an int.
34276 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
34277 polynomial unroll factors.
34278 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
34279 (vect_make_slp_decision): Likewise.
34280 (vect_supported_load_permutation_p): Likewise, and polynomial
34282 (vect_analyze_slp_cost): Handle polynomial vf.
34283 (vect_slp_analyze_node_operations): Likewise.
34284 (vect_slp_analyze_bb_1): Likewise.
34285 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
34286 than an unsigned HOST_WIDE_INT.
34287 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
34288 (vectorizable_load): Handle polynomial vf.
34289 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
34291 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
34293 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34294 Alan Hayward <alan.hayward@arm.com>
34295 David Sherwood <david.sherwood@arm.com>
34297 * match.pd: Handle bit operations involving three constants
34298 and try to fold one pair.
34300 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34302 * tree-vect-loop-manip.c: Include gimple-fold.h.
34303 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
34304 niters_maybe_zero parameters. Handle other cases besides a step of 1.
34305 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
34306 Add a path that uses a step of VF instead of 1, but disable it
34308 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
34309 and niters_no_overflow parameters. Update calls to
34310 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
34311 Create a new SSA name if the latter choses to use a ste other
34312 than zero, and return it via niters_vector_mult_vf_var.
34313 * tree-vect-loop.c (vect_transform_loop): Update calls to
34314 vect_do_peeling, vect_gen_vector_loop_niters and
34315 slpeel_make_loop_iterate_ntimes.
34316 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
34317 (vect_gen_vector_loop_niters): Update declarations after above changes.
34319 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
34321 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
34322 128-bit round to integer instructions.
34323 (ceil<mode>2): Likewise.
34324 (btrunc<mode>2): Likewise.
34325 (round<mode>2): Likewise.
34327 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34329 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
34330 unaligned VSX load/store on P8/P9.
34331 (expand_block_clear): Allow the use of unaligned VSX
34332 load/store on P8/P9.
34334 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34336 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
34338 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
34339 swap associated with both a load and a store.
34341 2018-01-02 Andrew Waterman <andrew@sifive.com>
34343 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
34344 * config/riscv/riscv.md (clear_cache): Use it.
34346 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
34348 * web.c: Remove out-of-date comment.
34350 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34352 * expr.c (fixup_args_size_notes): Check that any existing
34353 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
34354 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
34355 (emit_single_push_insn): ...here.
34357 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34359 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
34360 (const_vector_encoded_nelts): New function.
34361 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
34362 (const_vector_int_elt, const_vector_elt): Declare.
34363 * emit-rtl.c (const_vector_int_elt_1): New function.
34364 (const_vector_elt): Likewise.
34365 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
34366 of CONST_VECTOR_ELT.
34368 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34370 * expr.c: Include rtx-vector-builder.h.
34371 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
34372 directly on the tree encoding.
34373 (const_vector_from_tree): Likewise.
34374 * optabs.c: Include rtx-vector-builder.h.
34375 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
34376 sequence of "u" values.
34377 * vec-perm-indices.c: Include rtx-vector-builder.h.
34378 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
34379 directly on the vec_perm_indices encoding.
34381 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34383 * doc/rtl.texi (const_vector): Describe new encoding scheme.
34384 * Makefile.in (OBJS): Add rtx-vector-builder.o.
34385 * rtx-vector-builder.h: New file.
34386 * rtx-vector-builder.c: Likewise.
34387 * rtl.h (rtx_def::u2): Add a const_vector field.
34388 (CONST_VECTOR_NPATTERNS): New macro.
34389 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
34390 (CONST_VECTOR_DUPLICATE_P): Likewise.
34391 (CONST_VECTOR_STEPPED_P): Likewise.
34392 (CONST_VECTOR_ENCODED_ELT): Likewise.
34393 (const_vec_duplicate_p): Check for a duplicated vector encoding.
34394 (unwrap_const_vec_duplicate): Likewise.
34395 (const_vec_series_p): Check for a non-duplicated vector encoding.
34396 Say that the function only returns true for integer vectors.
34397 * emit-rtl.c: Include rtx-vector-builder.h.
34398 (gen_const_vec_duplicate_1): Delete.
34399 (gen_const_vector): Call gen_const_vec_duplicate instead of
34400 gen_const_vec_duplicate_1.
34401 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
34402 (gen_const_vec_duplicate): Use rtx_vector_builder.
34403 (gen_const_vec_series): Likewise.
34404 (gen_rtx_CONST_VECTOR): Likewise.
34405 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
34406 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
34407 Build a new vector rather than modifying a CONST_VECTOR in-place.
34408 (handle_special_swappables): Update call accordingly.
34409 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
34410 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
34411 Build a new vector rather than modifying a CONST_VECTOR in-place.
34412 (handle_special_swappables): Update call accordingly.
34414 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34416 * simplify-rtx.c (simplify_const_binary_operation): Use
34417 CONST_VECTOR_ELT instead of XVECEXP.
34419 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34421 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
34422 the selector elements to be different from the data elements
34423 if the selector is a VECTOR_CST.
34424 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
34425 ssizetype for the selector.
34427 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34429 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
34430 before testing each element individually.
34431 * tree-vect-generic.c (lower_vec_perm): Likewise.
34433 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34435 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
34436 * selftest-run-tests.c (selftest::run_tests): Call it.
34437 * vector-builder.h (vector_builder::operator ==): New function.
34438 (vector_builder::operator !=): Likewise.
34439 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
34440 (vec_perm_indices::all_from_input_p): New function.
34441 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
34442 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
34443 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
34444 instead of reading the VECTOR_CST directly. Detect whether both
34445 vector inputs are the same before constructing the vec_perm_indices,
34446 and update the number of inputs argument accordingly. Use the
34447 utility functions added above. Only construct sel2 if we need to.
34449 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34451 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
34452 the broadcast of the low byte.
34453 (expand_mult_highpart): Use an explicit encoding for the permutes.
34454 * optabs-query.c (can_mult_highpart_p): Likewise.
34455 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
34456 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34457 (vectorizable_bswap): Likewise.
34458 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
34459 explicit encoding for the power-of-2 permutes.
34460 (vect_permute_store_chain): Likewise.
34461 (vect_grouped_load_supported): Likewise.
34462 (vect_permute_load_chain): Likewise.
34464 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34466 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
34467 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
34468 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
34469 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
34470 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
34471 (vect_gen_perm_mask_any): Likewise.
34473 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34475 * int-vector-builder.h: New file.
34476 * vec-perm-indices.h: Include int-vector-builder.h.
34477 (vec_perm_indices): Redefine as an int_vector_builder.
34478 (auto_vec_perm_indices): Delete.
34479 (vec_perm_builder): Redefine as a stand-alone class.
34480 (vec_perm_indices::vec_perm_indices): New function.
34481 (vec_perm_indices::clamp): Likewise.
34482 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
34483 (vec_perm_indices::new_vector): New function.
34484 (vec_perm_indices::new_expanded_vector): Update for new
34485 vec_perm_indices class.
34486 (vec_perm_indices::rotate_inputs): New function.
34487 (vec_perm_indices::all_in_range_p): Operate directly on the
34488 encoded form, without computing elided elements.
34489 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
34490 encoding. Update for new vec_perm_indices class.
34491 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
34492 the given vec_perm_builder.
34493 (expand_vec_perm_var): Update vec_perm_builder constructor.
34494 (expand_mult_highpart): Use vec_perm_builder instead of
34495 auto_vec_perm_indices.
34496 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
34497 vec_perm_indices instead of auto_vec_perm_indices. Use a single
34498 or double series encoding as appropriate.
34499 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
34500 vec_perm_indices instead of auto_vec_perm_indices.
34501 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34502 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34503 (vect_permute_store_chain): Likewise.
34504 (vect_grouped_load_supported): Likewise.
34505 (vect_permute_load_chain): Likewise.
34506 (vect_shift_permute_load_chain): Likewise.
34507 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34508 (vect_transform_slp_perm_load): Likewise.
34509 (vect_schedule_slp_instance): Likewise.
34510 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34511 (vectorizable_mask_load_store): Likewise.
34512 (vectorizable_bswap): Likewise.
34513 (vectorizable_store): Likewise.
34514 (vectorizable_load): Likewise.
34515 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
34516 vec_perm_indices instead of auto_vec_perm_indices. Use
34517 tree_to_vec_perm_builder to read the vector from a tree.
34518 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
34519 vec_perm_builder instead of a vec_perm_indices.
34520 (have_whole_vector_shift): Use vec_perm_builder and
34521 vec_perm_indices instead of auto_vec_perm_indices. Leave the
34522 truncation to calc_vec_perm_mask_for_shift.
34523 (vect_create_epilog_for_reduction): Likewise.
34524 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
34525 from auto_vec_perm_indices to vec_perm_indices.
34526 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
34527 instead of changing individual elements.
34528 (aarch64_vectorize_vec_perm_const): Use new_vector to install
34529 the vector in d.perm.
34530 * config/arm/arm.c (expand_vec_perm_d::perm): Change
34531 from auto_vec_perm_indices to vec_perm_indices.
34532 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
34533 instead of changing individual elements.
34534 (arm_vectorize_vec_perm_const): Use new_vector to install
34535 the vector in d.perm.
34536 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
34537 Update vec_perm_builder constructor.
34538 (rs6000_expand_interleave): Likewise.
34539 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
34540 (rs6000_expand_interleave): Likewise.
34542 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34544 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
34545 to qimode could truncate the indices.
34546 * optabs.c (expand_vec_perm_var): Likewise.
34548 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34550 * Makefile.in (OBJS): Add vec-perm-indices.o.
34551 * vec-perm-indices.h: New file.
34552 * vec-perm-indices.c: Likewise.
34553 * target.h (vec_perm_indices): Replace with a forward class
34555 (auto_vec_perm_indices): Move to vec-perm-indices.h.
34556 * optabs.h: Include vec-perm-indices.h.
34557 (expand_vec_perm): Delete.
34558 (selector_fits_mode_p, expand_vec_perm_var): Declare.
34559 (expand_vec_perm_const): Declare.
34560 * target.def (vec_perm_const_ok): Replace with...
34561 (vec_perm_const): ...this new hook.
34562 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
34563 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
34564 * doc/tm.texi: Regenerate.
34565 * optabs.def (vec_perm_const): Delete.
34566 * doc/md.texi (vec_perm_const): Likewise.
34567 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
34568 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
34569 expand_vec_perm for constant permutation vectors. Assert that
34570 the mode of variable permutation vectors is the integer equivalent
34571 of the mode that is being permuted.
34572 * optabs-query.h (selector_fits_mode_p): Declare.
34573 * optabs-query.c: Include vec-perm-indices.h.
34574 (selector_fits_mode_p): New function.
34575 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
34576 is defined, instead of checking whether the vec_perm_const_optab
34577 exists. Use targetm.vectorize.vec_perm_const instead of
34578 targetm.vectorize.vec_perm_const_ok. Check whether the indices
34579 fit in the vector mode before using a variable permute.
34580 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
34581 vec_perm_indices instead of an rtx.
34582 (expand_vec_perm): Replace with...
34583 (expand_vec_perm_const): ...this new function. Take the selector
34584 as a vec_perm_indices rather than an rtx. Also take the mode of
34585 the selector. Update call to shift_amt_for_vec_perm_mask.
34586 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
34587 Use vec_perm_indices::new_expanded_vector to expand the original
34588 selector into bytes. Check whether the indices fit in the vector
34589 mode before using a variable permute.
34590 (expand_vec_perm_var): Make global.
34591 (expand_mult_highpart): Use expand_vec_perm_const.
34592 * fold-const.c: Includes vec-perm-indices.h.
34593 * tree-ssa-forwprop.c: Likewise.
34594 * tree-vect-data-refs.c: Likewise.
34595 * tree-vect-generic.c: Likewise.
34596 * tree-vect-loop.c: Likewise.
34597 * tree-vect-slp.c: Likewise.
34598 * tree-vect-stmts.c: Likewise.
34599 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
34601 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
34602 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
34603 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
34604 (aarch64_vectorize_vec_perm_const): ...this new function.
34605 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34606 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34607 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
34608 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
34609 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34610 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34611 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
34613 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
34614 check for NEON modes.
34615 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
34616 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
34617 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
34618 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
34620 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
34621 the old VEC_PERM_CONST conditions.
34622 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
34623 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
34624 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
34625 (ia64_vectorize_vec_perm_const_ok): Merge into...
34626 (ia64_vectorize_vec_perm_const): ...this new function.
34627 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
34628 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
34629 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
34630 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
34631 * config/mips/mips.c (mips_expand_vec_perm_const)
34632 (mips_vectorize_vec_perm_const_ok): Merge into...
34633 (mips_vectorize_vec_perm_const): ...this new function.
34634 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
34635 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
34636 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
34637 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
34638 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
34639 (rs6000_expand_vec_perm_const): Delete.
34640 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
34642 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34643 (altivec_expand_vec_perm_const_le): Take each operand individually.
34644 Operate on constant selectors rather than rtxes.
34645 (altivec_expand_vec_perm_const): Likewise. Update call to
34646 altivec_expand_vec_perm_const_le.
34647 (rs6000_expand_vec_perm_const): Delete.
34648 (rs6000_vectorize_vec_perm_const_ok): Delete.
34649 (rs6000_vectorize_vec_perm_const): New function.
34650 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34651 an element count and rtx array.
34652 (rs6000_expand_extract_even): Update call accordingly.
34653 (rs6000_expand_interleave): Likewise.
34654 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
34655 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
34656 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
34657 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
34658 (rs6000_expand_vec_perm_const): Delete.
34659 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34660 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34661 (altivec_expand_vec_perm_const_le): Take each operand individually.
34662 Operate on constant selectors rather than rtxes.
34663 (altivec_expand_vec_perm_const): Likewise. Update call to
34664 altivec_expand_vec_perm_const_le.
34665 (rs6000_expand_vec_perm_const): Delete.
34666 (rs6000_vectorize_vec_perm_const_ok): Delete.
34667 (rs6000_vectorize_vec_perm_const): New function. Remove stray
34668 reference to the SPE evmerge intructions.
34669 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34670 an element count and rtx array.
34671 (rs6000_expand_extract_even): Update call accordingly.
34672 (rs6000_expand_interleave): Likewise.
34673 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
34674 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
34676 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34678 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34680 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
34681 vector mode and that that mode matches the mode of the data
34683 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
34684 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
34685 directly using expand_vec_perm_1 when forcing selectors into
34687 (expand_vec_perm_var): New function, split out from expand_vec_perm.
34689 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34691 * optabs-query.h (can_vec_perm_p): Delete.
34692 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
34693 * optabs-query.c (can_vec_perm_p): Split into...
34694 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
34695 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
34696 particular selector is valid.
34697 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34698 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34699 (vect_grouped_load_supported): Likewise.
34700 (vect_shift_permute_load_chain): Likewise.
34701 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34702 (vect_transform_slp_perm_load): Likewise.
34703 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34704 (vectorizable_bswap): Likewise.
34705 (vect_gen_perm_mask_checked): Likewise.
34706 * fold-const.c (fold_ternary_loc): Likewise. Don't take
34707 implementations of variable permutation vectors into account
34708 when deciding which selector to use.
34709 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
34710 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
34711 with a false third argument.
34712 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
34713 to test whether the constant selector is valid and can_vec_perm_var_p
34714 to test whether a variable selector is valid.
34716 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34718 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
34719 * optabs-query.c (can_vec_perm_p): Likewise.
34720 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
34721 instead of vec_perm_indices.
34722 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
34723 (vect_gen_perm_mask_checked): Likewise,
34724 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
34725 (vect_gen_perm_mask_checked): Likewise,
34727 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34729 * optabs-query.h (qimode_for_vec_perm): Declare.
34730 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
34731 (qimode_for_vec_perm): ...this new function.
34732 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
34734 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34736 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
34737 does not have a conditional at the top.
34739 2018-01-02 Richard Biener <rguenther@suse.de>
34741 * ipa-inline.c (big_speedup_p): Fix expression.
34743 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34746 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
34749 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34753 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
34754 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
34755 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
34756 cond_taken_branch_cost 3->4.
34758 2018-01-01 Jakub Jelinek <jakub@redhat.com>
34760 PR tree-optimization/83581
34761 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
34762 TODO_cleanup_cfg if any changes have been made.
34764 PR middle-end/83608
34765 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
34766 convert_modes if target mode has the right side, but different mode
34769 PR middle-end/83609
34770 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
34771 last argument when extracting from CONCAT. If either from_real or
34772 from_imag is NULL, use expansion through memory. If result is not
34773 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
34774 the parts directly to inner mode, if even that fails, use expansion
34777 PR middle-end/83623
34778 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
34779 check for bswap in mode rather than HImode and use that in expand_unop
34782 Copyright (C) 2018 Free Software Foundation, Inc.
34784 Copying and distribution of this file, with or without modification,
34785 are permitted in any medium without royalty provided the copyright
34786 notice and this notice are preserved.