1 2017-04-29 Jakub Jelinek <jakub@redhat.com>
3 PR rtl-optimization/80491
4 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
6 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
8 PR tree-optimization/80487
9 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
11 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13 PR tree-optimization/79697
14 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
18 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
19 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21 2017-04-28 Martin Sebor <msebor@redhat.com>
23 PR tree-optimization/80523
24 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25 (init_target_to_host_charmap, target_to_host, target_strtol10): New
27 (maybe_warn, format_directive, parse_directive): Use new functions.
28 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
32 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
34 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
36 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
37 target_header_dir): Set correctly.
38 * configure: Regenerated.
39 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
40 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
41 instead of SYSTEM_HEADER_DIR.
43 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
45 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
46 (estimate_local_effects): Likewise.
47 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
48 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
49 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
50 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
51 do_estimate_edge_time, estimate_edge_time): Likewise.
52 * ipa-inline-analysis.c (estimate_node_size_and_time,
53 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
54 (estimate_time_after_inlining): Remove.
56 2017-04-28 Martin Liska <mliska@suse.cz>
58 * doc/gcov.texi: Enhance documentation of gcov.
60 2017-04-28 Martin Liska <mliska@suse.cz>
62 * doc/gcov.texi: Sort options in alphabetic order.
63 * doc/gcov-dump.texi: Likewise.
64 * doc/gcov-tool.texi: Likewise.
65 * gcov.c (print_usage): Likewise.
66 * gcov-dump.c (print_usage): Likewise.
67 * gcov-tool.c (print_merge_usage_message): Likewise.
68 (print_rewrite_usage_message): Likewise.
69 (print_overlap_usage_message): Likewise.
71 2017-04-28 Martin Liska <mliska@suse.cz>
74 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
76 2017-04-28 Martin Liska <mliska@suse.cz>
79 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
80 is marked by compiler as living on a line.
81 (get_cycles_count): Remove usage of the union.
82 (output_intermediate_file): Likewise.
83 (find_source): Fix GNU coding style.
84 (accumulate_line_counts): Remove old non-all block mode.
85 (output_lines): Remove usage of the union.
86 * profile.c (output_location): Include all BBs, even if
87 belonging to a same line (and file) as a previous BB.
89 2017-04-28 Martin Liska <mliska@suse.cz>
91 * gcov.c (process_args): Handle new argument 'w'.
92 (read_graph_file): Assign ID to BBs.
93 (output_branch_count): Display BB # if verbose flag is set.
94 (output_lines): Likewise for arcs.
95 (print_usage): Add '--verbose' option help.
96 * doc/gcov.texi: Document --verbose (-w) option.
98 2017-04-28 Martin Liska <mliska@suse.cz>
100 * gcov.c (struct block_location_info): New struct.
101 (process_file): Fill up the new structure.
102 (read_graph_file): Replace usage of encoding by the newly added
104 (add_line_counts): Likewise.
105 (accumulate_line_counts): Remove usage of the union.
106 (function_info::function_info): New function.
107 (function_info::~function_info): Likewise.
108 (process_file): Call delete instead of release_function.
109 (release_function): Release the function.
110 (release_structures): Call delete instead of release_function.
111 (solve_flow_graph): Replace usage of num_blocks.
112 (find_exception_blocks): Likewise.
113 (output_lines): Fix GNU coding style.
115 2017-04-28 Martin Liska <mliska@suse.cz>
118 * coverage.c (coverage_remove_note_file): New function.
119 * coverage.h: Declare the function.
120 * toplev.c (finalize): Clean if an error has been seen.
122 2017-04-28 Martin Liska <mliska@suse.cz>
124 PR gcov-profile/80031
125 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
126 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
127 * gcov.c (read_graph_file): Read just number of blocks.
128 * profile.c (branch_prob): Do not stream 0 flags per a basic
131 2017-04-28 Martin Liska <mliska@suse.cz>
133 * gcov-dump.c (tag_*): Add new argument to declarations.
134 (dump_gcov_file): Likewise.
135 (tag_blocks): Add and use new argument depth.
136 (tag_arcs): Likewise.
137 (tag_lines): Likewise.
138 (tag_counters): Likewise.
139 (tag_summary): Likewise.
140 (dump_working_sets): Use depth to do a proper indentation.
142 2017-04-28 Jakub Jelinek <jakub@redhat.com>
145 * cgraph.h (symtab_node::debug_symtab): No longer inline.
146 * symtab.c (symtab_node::debug_symtab): Move definition here.
148 2017-04-28 Richard Biener <rguenther@suse.de>
150 * lto-streamer.h (LTO_major_version): Bump to 7.
152 2017-04-28 Richard Biener <rguenther@suse.de>
154 * tree-vrp.c (assert_info): New struct.
155 (add_assert_info): New helper.
156 (register_edge_assert_for_2): Refactor to add asserts to a vector
158 (register_edge_assert_for_1): Likewise.
159 (register_edge_assert_for): Likewise.
160 (finish_register_edge_assert_for): New helper actually registering
161 asserts where live on edge.
162 (find_conditional_asserts): Adjust.
163 (find_switch_asserts): Likewise.
164 (evrp_dom_walker::try_find_new_range): Generalize.
165 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
167 2017-04-27 Marek Polacek <polacek@redhat.com>
170 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
171 arg10 and arg11 to itype.
173 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
175 * doc/extend.texi (Object Size Checking): Improve grammar.
177 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
180 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
181 that the logic for permitting reciprocal estimates matches that
184 2017-04-27 Jakub Jelinek <jakub@redhat.com>
187 * tree.c (type_cache_hasher::equal): Only compare
188 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
189 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
190 non-aggregate element types.
191 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
192 about the flag on ARRAY_TYPEs in the comment, formatting fix.
194 2017-04-27 Richard Biener <rguenther@suse.de>
197 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
198 stripping ARRAY_REFs from MEM_EXPR make sure we're not
199 keeping a reference to a trailing array.
201 2017-04-27 Richard Biener <rguenther@suse.de>
204 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
205 being in loop-closed SSA form conservatively.
206 (chrec_fold_multiply_poly_poly): Likewise.
208 2017-04-27 Tamar Christina <tamar.christina@arm.com>
211 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
212 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
214 2017-04-27 Jakub Jelinek <jakub@redhat.com>
217 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
218 (aarch64_function_arg_alignment): Return unsigned int again, but still
219 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
220 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
221 Don't emit -Wpsabi note.
222 (aarch64_function_arg_boundary): Likewise.
223 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
226 2017-04-26 Nathan Sidwell <nathan@acm.org>
228 * tree.h (crc32_unsigned_n): Declare.
229 (crc32_unsigned, crc32_unsigned): Make inline.
230 * tree.c (crc32_unsigned_bits): Replace with ...
231 (crc32_unsigned_n): ... this.
232 (crc32_unsigned, crc32_byte): Remove.
233 (crc32_string): Remove unnecessary braces.
235 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
237 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
238 * ipa-inline-analysis.c (MAX_TIME): Remove.
239 (account_size_time): Use sreal for time.
240 (dump_inline_summary): Update.
241 (estimate_function_body_sizes): Update.
242 (estimate_edge_size_and_time): Update.
243 (estimate_calls_size_and_time): Update.
244 (estimate_node_size_and_time): Update.
245 (inline_merge_summary): Update.
246 (inline_update_overall_summary): Update.
247 (estimate_time_after_inlining): Update.
248 (inline_read_section): Update.
249 (inline_write_summary): Update.
250 * ipa-inline.c (compute_uninlined_call_time): Update.
251 (compute_inlined_call_time): Update.
252 (recursive_inlining): Update.
253 (inline_small_functions): Update.
254 (dump_overall_stats): Update.
255 * ipa-inline.h: Include sreal.h.
256 (size_time_entry): Turn time to sreal.
257 (inline_summary): Turn self_time nad time to sreal.
259 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
261 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
263 (sreal::stream_out, sreal::stream_in): New.
264 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
266 2017-04-25 Jakub Jelinek <jakub@redhat.com>
268 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
271 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
274 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
276 (dimode_scalar_chain::compute_convert_gain): Ditto.
277 (dimode_scalar_chain::make_vector_copies): Ditto.
278 (dimode_scalar_chain::convert_reg): Ditto.
279 (dimode_scalar_chain::convert_insn): Ditto.
280 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
281 (VI248_AVX512BW_1): New mode iterator.
282 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
283 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
286 2017-04-25 Martin Sebor <msebor@redhat.com>
288 PR tree-optimization/80497
289 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
290 constants are representable in HOST_WIDE_INT.
291 (parse_directive): Ditto.
293 2017-04-25 Martin Sebor <msebor@redhat.com>
296 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
297 (new_zero_array): Adjust signature.
298 (dom_info::dom_init): Used unsigned rather that size_t.
299 (dom_info::dom_info): Same.
301 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
302 Jakub Jelinek <jakub@redhat.com>
305 * config/arm/arm.c: Include gimple.h.
306 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
307 returns negative, increment ncrn only if it returned positive.
308 (arm_needs_doubleword_align): Return int instead of bool,
309 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
310 members, but if there is any such non-FIELD_DECL
311 > PARM_BOUNDARY aligned decl, return -1 instead of false.
312 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
313 returns negative, increment nregs only if it returned positive.
314 (arm_setup_incoming_varargs): Likewise.
315 (arm_function_arg_boundary): Emit -Wpsabi note if
316 arm_needs_doubleword_align returns negative, return
317 DOUBLEWORD_ALIGNMENT only if it returned positive.
319 2017-04-25 Marek Polacek <polacek@redhat.com>
322 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
323 first argument to type.
325 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
328 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
329 type checks to test for compatibility instead of equality.
331 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
332 Jakub Jelinek <jakub@redhat.com>
335 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
337 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
338 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
339 the alignment computation, but return their maximum in warn_alignment.
340 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
341 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
343 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
344 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
347 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
349 * config/arc/simdext.md (dmpyh): Fix typo.
351 2017-04-25 Richard Biener <rguenther@suse.de>
353 PR tree-optimization/80492
354 * alias.c (compare_base_decls): Handle registers with asm
355 specification conservatively.
356 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
357 compare_base_decls returning dont-know properly.
359 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
361 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
362 (legitimate_offset_address_p): New function.
363 (arc_legitimate_address_p): Use above function.
365 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
367 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
369 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
371 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
372 ACCH registers whenever they are available.
374 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
376 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
377 double regs fix when not used.
379 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
381 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
383 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
384 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
386 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
388 * config/arc/arc.c (arc_output_addsi): Check for h-register class
389 when emitting short ADD instructions.
391 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
393 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
395 (cmpsi_cc_c_insn): Likewise.
396 (cbranchsi4_scratch): Compute proper instruction length using
397 compact_hreg_operand.
398 * config/arc/predicates.md (compact_hreg_operand): New predicate.
400 2017-04-25 Richard Biener <rguenther@suse.de>
403 * passes.c (pass_manager::pass_manager): Initialize
406 2017-04-25 Richard Biener <rguenther@suse.de>
408 PR tree-optimization/79201
409 * tree-ssa-sink.c (statement_sink_location): Handle calls.
411 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
414 * config/s390/vector.md: Split MEM->GPR vector moves for
415 non-s_operand addresses.
417 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
420 * config/s390/predicates.md (reload_const_wide_int_operand): New
422 * config/s390/s390.md ("movti"): Remove d/P alternative.
423 ("movti_bigconst"): New pattern definition.
425 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
428 * s390-protos.h (s390_expand_cs_hqi): Removed.
429 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
430 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
431 modes as well as CCZ1mode and CCZmode.
432 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
433 signature of s390_emit_compare_and_swap.
434 (s390_expand_cs_hqi): Likewise, make static.
435 (s390_expand_cs_tdsi): Generate an explicit compare before trying
436 compare-and-swap, in some cases.
437 (s390_expand_cs): Wrapper function.
438 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
440 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
441 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
442 patterns for small and large integers. Forbid symref memory operands.
443 Move expander to s390.c. Require cc register.
444 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
445 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
446 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
447 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
448 symref memory operands. Remove CC mode and call s390_match_ccmode
450 ("atomic_exchange<mode>"): Allow and implement all integer modes.
452 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
454 * config/s390/s390.md (define_peephole2): New peephole to help
455 combining the load-and-test pattern with volatile memory.
457 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
459 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
460 with CCZmode for TARGET_Z196.
462 2017-04-25 Jakub Jelinek <jakub@redhat.com>
464 PR rtl-optimization/80501
465 * combine.c (make_compound_operation_int): Set subreg_code to SET
466 even for AND with mask of the sign bit of mode.
468 PR rtl-optimization/80500
469 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
472 2017-04-25 Julian Brown <julian@codesourcery.com>
473 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
475 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
477 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
479 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
481 2017-04-25 Julian Brown <julian@codesourcery.com>
482 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
484 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
485 (thunderx2t99_sha): New Reservation.
487 2017-04-25 Julian Brown <julian@codesourcery.com>
488 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
490 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
491 type for 1-element load.
493 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
495 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
497 2017-04-24 Martin Jambor <mjambor@suse.cz>
499 PR tree-optimization/80293
500 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
501 char arrays not totally scalarizable if it is false.
502 (analyze_all_variable_accesses): Pass correct value in the new
503 parameter. Add a statistics counter.
505 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
508 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
510 2017-04-24 Richard Biener <rguenther@suse.de>
512 PR tree-optimization/80494
513 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
514 out for complex types.
516 2017-04-24 Richard Biener <rguenther@suse.de>
518 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
519 * tree-ssa-sccvn.c (print_scc): Print SCC size.
520 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
521 (DFS): Adjust and never fail.
522 (sccvn_dom_walker::fail): Remove.
523 (sccvn_dom_walker::before_dom_children): Adjust.
524 (run_scc_vn): Likewise and never fail.
525 * tree-ssa-pre.c (pass_pre::execute): Adjust.
526 (pass_fre::execute): Likewise.
528 2017-04-24 Richard Biener <rguenther@suse.de>
530 PR tree-optimization/79725
531 * tree-ssa-sink.c (statement_sink_location): Return whether
532 failure reason was zero uses. Move that check later.
533 (sink_code_in_bb): Deal with zero uses by removing the stmt
536 2017-04-24 Richard Biener <rguenther@suse.de>
539 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
540 pointer-based references.
542 2017-04-24 Richard Biener <rguenther@suse.de>
545 * pass_manager.h (pass_manager::operator new): Remove.
546 (pass_manager::operator delete): Likewise.
547 * passes.c (pass_manager::operator new): Remove.
548 (pass_manager::operator delete): Likewise.
549 (pass_manager::pass_manager): Zero individual pass members.
551 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
554 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
555 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
556 Check "XEXP (src, 1)" operand here.
557 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
558 Check "XEXP (src, 1)" operand here.
559 (dimode_scalar_chain::make_vector_copies): Detect count register
560 of a shift instruction. Zero extend count register from QImode
561 to DImode to satisfy vector shift pattern count operand predicate.
562 Substitute vector shift count operand with a DImode copy.
563 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
566 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
568 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
569 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
570 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
571 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
572 (UNSPEC_NOREX_MEM): Remove definition.
574 2017-04-21 Richard Biener <rguenther@suse.de>
576 PR tree-optimization/79547
577 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
578 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
579 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
580 without any constraints.
582 2017-04-21 Richard Biener <rguenther@suse.de>
584 PR tree-optimization/78847
585 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
587 2017-04-21 Richard Biener <rguenther@suse.de>
589 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
590 (build_distinct_type_copy): Likewise.
591 (build_variant_type_copy): Likewise.
592 * tree.c (build_qualified_type): Pass down mem-stat info.
593 (build_distinct_type_copy): Likewise.
594 (build_variant_type_copy): Likewise.
596 2017-04-21 Richard Biener <rguenther@suse.de>
598 PR tree-optimization/80237
599 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
601 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
602 for a simplified result.
604 2016-04-21 Richard Biener <rguenther@suse.de>
606 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
607 sth as strict as a simple_iv but a chrec without symbols and an
608 operand defined in the loop we are peeling (and not some subloop).
609 (propagate_constants_for_unrolling): Propagate all constants.
611 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
614 * config/i386/i386.c (print_reg): Remove assert for disalowed
615 regno values, call output_operand_lossage instead.
617 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
620 * config/i386/constraints.md (Yc): New register constraint.
621 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
622 Use Yc constraint for alternative 2 of operand 0. Remove
623 preferred_for_speed attribute.
625 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
627 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
628 lastprivate clauses in SIMT case.
630 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
632 * doc/invoke.texi (-Wextra-semi): Document new warning option.
634 2017-04-20 Richard Biener <rguenther@suse.de>
636 PR tree-optimization/57796
637 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
639 (vect_model_load_cost): Cost gathers as N scalar loads.
641 2017-04-20 Richard Biener <rguenther@suse.de>
643 * ggc-page.c (ggc_allocated_p): Rename to ...
644 (safe_lookup_page_table_entry): ... this and return the lookup
646 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
648 2017-04-20 Richard Biener <rguenther@suse.de>
650 PR tree-optimization/80453
651 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
652 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
654 (vn_phi_eq): Pass them down.
655 (vn_phi_lookup): Record them.
656 (vn_phi_insert): Likewise.
658 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
660 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
661 uninitialized variable warning to avoid buffer overrun.
663 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
666 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
667 is suppressed for '{ 0 }' in C.
669 2017-04-20 Jakub Jelinek <jakub@redhat.com>
671 * BASE-VER: Set to 8.0.0.
673 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
675 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
676 priority .init_array and .fini_array section with SECTION_NOTYPE
679 2017-04-20 Jakub Jelinek <jakub@redhat.com>
682 * tree.h (build_array_type): Add typeless_storage default argument.
683 * tree.c (type_cache_hasher::equal): Also compare
684 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
685 (build_array_type): Add typeless_storage argument, set
686 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
688 (build_nonshared_array_type): Adjust build_array_type_1 caller.
689 (build_array_type): Likewise. Add typeless_storage argument.
691 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
692 Jakub Jelinek <jakub@redhat.com>
694 PR tree-optimization/80426
695 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
696 operation on symbolic operands, also compute the overflow for the
697 invariant part when the operation degenerates into a negation.
699 2017-04-19 Jakub Jelinek <jakub@redhat.com>
702 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
703 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
706 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
708 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
711 * config/avr/avr.c (tree.h): Include it.
712 (cgraph.h): Include it.
713 (avr_encode_section_info): Don't warn for uninitialized progmem
714 variable if it's just an alias.
716 2017-04-19 Richard Biener <rguenther@suse.de>
719 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
720 when needed by AutoPGO.
722 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
725 * doc/lto.texi: Remove an extra 'that'.
727 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
729 PR rtl-optimization/80429
730 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
731 are only used in debug insns.
733 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
734 Vladimir Makarov <vmakarov@redhat.com>
736 * config/sparc/predicates.md (input_operand): Add comment. Return
737 true for any memory operand when LRA is in progress.
738 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
740 2017-04-18 Jeff Law <law@redhat.com>
743 * mips.md ({return,simple_return}_internal): Do not overwrite
746 2017-04-18 Jakub Jelinek <jakub@redhat.com>
748 PR tree-optimization/80443
749 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
750 instead of adding 1, subtract -1 and similarly instead of subtracting
753 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
755 PR rtl-optimization/80357
756 * haifa-sched.c (tmp_bitmap): New variable.
757 (model_recompute): Handle duplicate use records.
758 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
759 (free_global_sched_pressure_data): Free it.
761 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
764 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
765 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
766 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
767 instead of SYSTEM_HEADER_DIR.
769 2017-04-18 Jeff Law <law@redhat.com>
772 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
773 predecessors after walking up the insn chain.
775 2017-04-18 Jakub Jelinek <jakub@redhat.com>
778 * dwarf2out.c (modified_type_die): Try harder not to emit internal
779 sizetype type into debug info.
781 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
784 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
785 unneeded test for TARGET_UPPER_REGS_SF.
786 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
788 2017-04-18 Jakub Jelinek <jakub@redhat.com>
791 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
792 instead of gsi_after_labels.
794 2017-04-18 Jeff Law <law@redhat.com>
796 * regcprop.c (maybe_mode_change): Avoid creating copies of the
800 2017-04-13 Jeff Law <law@redhat.com>
801 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
802 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
804 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
807 * config/avr/avr.c (intl.h): Include it.
808 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
810 2017-04-18 Martin Liska <mliska@suse.cz>
812 PR gcov-profile/78783
813 * gcov-tool.c (gcov_output_files): Validate that destination
814 file is either removed by the tool or by a user.
816 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
817 Guy Benyei <guybe@mellanox.com>
819 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
820 block, and do not negate it, the stored id is already negative.
822 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
824 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
826 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
829 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
830 masks of options that should be turned off if the VSX vector
831 options are turned off.
832 (OTHER_P8_VECTOR_MASKS): Likewise.
833 (OTHER_VSX_VECTOR_MASKS): Likewise.
834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
835 rs6000_disable_incompatible_switches to validate no type switches
837 (rs6000_incompatible_switch): New function to disallow turning on
838 other vector options if -mno-vsx, -mno-power8-vector, or
839 -mno-power9-vector are specified.
841 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
843 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
845 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
847 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
848 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
849 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
850 (ARG_POINTER_CFA_OFFSET): Likewise.
852 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
854 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
855 conditions to take advantage of various optimizations.
857 2017-04-13 Jeff Law <law@redhat.com>
859 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
860 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
861 (zero_extendsidi2_dext): Likewise.
863 2017-04-13 Jakub Jelinek <jakub@redhat.com>
866 * fold-const.c (fold_ternary_loc): Revert
867 use op0 instead of fold_convert_loc (loc, type, arg0) part of
870 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
872 PR rtl-optimization/80343
873 * lra-remat.c (update_scratch_ops): Assign original hard reg to
876 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
879 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
880 to ubsan_encode_value.
882 2017-04-13 Jeff Law <law@redhat.com>
884 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
885 appearing in DEBUG_INSNs.
887 2017-04-13 Martin Liska <mliska@suse.cz>
889 PR gcov-profile/80413
890 * gcov-io.c (gcov_write_string): Copy to buffer just when
891 allocated size is greater than zero.
893 2017-04-13 Jakub Jelinek <jakub@redhat.com>
896 * dwarf2out.c (decls_for_scope): Ignore declarations of
897 current_function_decl in BLOCK_NONLOCALIZED_VARS.
899 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
902 * ipa-visibility.c (non_local_p): Fix typos.
903 (localize_node): When localizing symbol in same comdat group,
904 dissolve the group only when we know external symbols are going
906 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
908 2017-04-12 Jakub Jelinek <jakub@redhat.com>
910 PR tree-optimization/79390
911 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
912 order does not result in usable sequence, retry with reversed operand
918 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
919 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
920 op0 instead of fold_convert_loc (loc, type, arg0).
922 2017-04-12 Jeff Law <law@redhat.com>
924 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
925 has a delay slot in the generated code.
927 * config/cris/cris.md (cris_preferred_reload_class): Return
928 GENNONACR_REGS rather than GENERAL_REGS.
930 2017-04-12 Jakub Jelinek <jakub@redhat.com>
933 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
934 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
935 signedness of the result type.
937 2017-04-12 Richard Biener <rguenther@suse.de>
938 Jeff Law <law@redhat.com>
940 PR tree-optimization/80359
941 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
942 trim stores to TARGET_MEM_REFs.
944 2017-04-12 Richard Biener <rguenther@suse.de>
946 PR tree-optimization/79390
947 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
948 threading case even more.
950 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
953 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
954 for quad_address_p for TImode, instead of just not indexed_address.
956 2017-04-12 Richard Biener <rguenther@suse.de>
957 Bernd Edlinger <bernd.edlinger@hotmail.de>
960 * alias.c (component_uses_parent_alias_set_from): Handle
961 TYPE_TYPELESS_STORAGE.
962 (get_alias_set): Likewise.
963 * tree-core.h (tree_type_common): Add typeless_storage flag.
964 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
965 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
966 for types containing members with TYPE_TYPELESS_STORAGE.
967 (place_field): Likewise.
968 (layout_type): Likewise for ARRAY_TYPE.
969 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
970 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
971 TYPE_TYPELESS_STORAGE.
972 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
974 2017-04-12 Jakub Jelinek <jakub@redhat.com>
977 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
978 first argument to type.
980 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
984 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
985 CONST0_RTX (mode) rather than const0_rtx where appropriate.
986 (rs6000_expand_binop_builtin): Likewise.
987 (rs6000_expand_ternop_builtin): Likewise; also add missing
988 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
990 * doc/extend.texi: Document that vec_xxpermdi's third argument
993 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
995 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
996 Use shift_const cost parameter when calculating gain of STV shifts.
998 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
1000 PR rtl-optimization/70478
1001 * lra-constraints.c (process_alt_operands): Check memory for
1002 disfavoring memory insn operand.
1004 2017-04-11 Jakub Jelinek <jakub@redhat.com>
1007 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
1008 left shift in unsigned HOST_WIDE_INT type.
1010 PR rtl-optimization/80385
1011 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
1012 (not (neg X)) into (plus X -1) for complex or non-integral modes.
1015 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
1016 if they have any depend clauses.
1018 2017-04-11 Martin Liska <mliska@suse.cz>
1021 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
1022 * ipa-split.c (split_function): Create a local comdat symbol
1023 if caller is in a comdat group.
1025 2017-04-11 Martin Liska <mliska@suse.cz>
1028 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
1031 2017-04-11 Martin Sebor <msebor@redhat.com>
1034 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
1035 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
1037 (directive::set_width, directive::set_precision, format_character):
1039 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
1042 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
1045 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
1046 conflict, set target->arch_name instead of target->cpu_name.
1048 2017-04-11 Richard Biener <rguenther@suse.de>
1050 PR tree-optimization/80374
1051 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
1052 build_zero_cst, remove fold_convertible_p check again.
1054 2017-04-11 Martin Liska <mliska@suse.cz>
1057 * ubsan.c (instrument_object_size): Do not instrument register
1060 2017-04-11 Jakub Jelinek <jakub@redhat.com>
1063 * config/i386/i386-builtin-types.def
1064 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
1065 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
1066 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
1067 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
1068 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
1069 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
1070 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
1071 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
1072 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
1073 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
1074 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
1075 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
1076 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
1077 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
1078 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
1079 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
1080 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
1081 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
1082 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
1083 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
1084 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
1085 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
1086 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
1087 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
1088 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
1089 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
1090 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
1091 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
1092 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
1093 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
1094 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
1095 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
1096 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
1097 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
1098 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
1099 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
1100 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
1101 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
1102 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
1103 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
1104 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
1105 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
1106 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
1107 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
1108 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
1109 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
1111 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
1112 flag to second_arg_count, handle 4 argument function type _COUNT
1113 aliases, handle second_arg_count on second argument rather than last.
1115 2017-04-10 Jeff Law <law@redhat.com>
1117 PR tree-optimization/80374
1118 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
1119 record anything if we can not convert integer_zero_node to the
1122 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
1125 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1126 Enhance special handling given to the TARGET_P9_MINMAX option in
1127 relation to certain other options.
1129 2017-04-10 Bin Cheng <bin.cheng@arm.com>
1131 PR tree-optimization/80153
1132 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
1133 remove POINTER_PLUS_EXPR's base part directly, rather than through
1136 2017-04-10 Richard Biener <rguenther@suse.de>
1137 Bin Cheng <bin.cheng@arm.com>
1139 PR tree-optimization/80153
1140 * tree-affine.c (aff_combination_to_tree): Get base pointer from
1141 the first element of pointer type aff_tree. Build result expr in
1143 (add_elt_to_tree): Convert to type unconditionally. Remove other
1145 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
1146 (rewrite_use_nonlinear_expr): Check invariant using iv information.
1148 2017-04-10 Richard Biener <rguenther@suse.de>
1150 * tree-ssa-structalias.c (find_func_aliases): Properly handle
1153 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
1155 PR rtl-optimization/70478
1156 * lra-constraints.c (curr_small_class_check): New.
1157 (update_and_check_small_class_inputs): New.
1158 (process_alt_operands): Update curr_small_class_check. Disfavor
1159 alternative insn memory operands. Check available regs for small
1162 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1165 * config/mips/mips.opt (-mvirt): Update description.
1166 * doc/invoke.texi (-mvirt): Likewise.
1168 2017-04-10 Richard Biener <rguenther@suse.de>
1171 * fold-const.c (fold_binary_loc): Look at unstripped ops when
1172 looking for NEGATE_EXPR in -A / -B to A / B folding.
1174 2017-04-10 Martin Liska <mliska@suse.cz>
1176 PR gcov-profile/80224
1177 * gcov.c (print_usage): Fix usage string.
1178 (get_gcov_intermediate_filename): Remove.
1179 (output_gcov_file): Use both for normal and intermediate format.
1180 (generate_results): Do not initialize special file for
1181 intermediate format.
1183 2017-04-10 Richard Biener <rguenther@suse.de>
1185 PR tree-optimization/80304
1186 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
1189 2017-04-10 Nathan Sidwell <nathan@acm.org>
1192 * config/rs6000/rs6000.c (rs6000_vector_type): New.
1193 (rs6000_init_builtins): Use it.
1195 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1197 * config/arm/arm.md (<mrc>): Add mode to SET source.
1200 2017-04-10 Richard Biener <rguenther@suse.de>
1203 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
1205 2017-04-10 Jakub Jelinek <jakub@redhat.com>
1208 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
1209 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
1210 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
1211 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
1212 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
1213 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
1214 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
1215 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
1216 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
1217 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
1218 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
1219 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
1220 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
1221 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
1222 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
1223 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
1224 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
1225 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
1226 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
1227 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
1228 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
1229 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
1230 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
1232 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
1234 PR rtl-optimization/70478
1235 * lra-constraints.c: Reverse the last patch.
1237 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
1239 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
1240 Add comment for WCHAR_T.
1242 2017-04-08 Martin Liska <mliska@suse.cz>
1245 2017-04-07 Martin Liska <mliska@suse.cz>
1248 * ipa-split.c (split_function): Add function part to a same comdat
1251 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1254 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
1256 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
1258 * rs6000/rs6000.c (vec_load_pendulum): Rename...
1259 (vec_pairing): ...to this.
1260 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
1261 (rs6000_sched_init): Adjust for name change.
1262 (struct rs6000_sched_context): Likewise.
1263 (rs6000_init_sched_context): Likewise.
1264 (rs6000_set_sched_context): Likewise.
1266 2017-04-07 Jakub Jelinek <jakub@redhat.com>
1272 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
1274 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
1275 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
1276 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
1278 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
1280 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
1282 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
1284 PR rtl-optimization/70703
1285 * ira-color.c (update_conflict_hard_regno_costs): Use
1286 int64_t instead of HOST_WIDE_INT.
1288 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
1290 PR rtl-optimization/70478
1291 * lra-constraints.c (process_alt_operands): Disfavor alternative
1292 insn memory operands.
1294 2017-04-07 Jeff Law <law@redhat.com>
1296 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
1297 CALL and NOTE_INSN_CALL_ARG_LOCATION.
1299 2017-04-07 Martin Liska <mliska@suse.cz>
1302 * config/aarch64/aarch64.c (aarch64_process_target_attr):
1303 Show error message instead of an ICE.
1305 2017-04-07 Martin Liska <mliska@suse.cz>
1308 * ipa-split.c (split_function): Add function part to a same comdat
1311 2017-04-07 Richard Biener <rguenther@suse.de>
1314 * tree.c (get_unwidened): Also handle ! for_type case for
1316 * convert.c (do_narrow): Split out from ...
1317 (convert_to_integer_1): ... here. Do not pass final truncation
1318 type to get_unwidened for TRUNC_DIV_EXPR.
1320 2017-04-07 Richard Biener <rguenther@suse.de>
1322 * tree-affine.c (wide_int_ext_for_comb): Take type rather
1324 (aff_combination_const): Adjust.
1325 (aff_combination_scale): Likewise.
1326 (aff_combination_add_elt): Likewise.
1327 (aff_combination_add_cst): Likewise.
1328 (aff_combination_convert): Likewise.
1329 (add_elt_to_tree): Likewise. Remove unused argument.
1330 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
1332 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
1334 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
1336 * config/arm/arm.c (arm_default_short_enums): Use
1337 ARM_DEFAULT_SHORT_ENUMS.
1338 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
1340 2017-04-06 Jakub Jelinek <jakub@redhat.com>
1343 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
1344 members with redundant out-of-class redeclaration.
1346 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
1349 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
1350 * config/i386/i386.md (*zero_extendsidi2):
1351 Add (?*x,*x) and (?*v,*v) alternatives.
1353 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
1356 * config/i386/i386.c (ix86_expand_builtin)
1357 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
1358 mode from insn data. Convert operands to insn operand mode.
1359 Copy operands that don't satisfy insn predicate to a register.
1361 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
1363 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
1366 2017-04-06 Richard Biener <rguenther@suse.de>
1368 PR tree-optimization/80334
1369 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
1370 preserve alignment of accesses.
1372 2017-04-06 Richard Biener <rguenther@suse.de>
1374 PR tree-optimization/80262
1375 * tree-sra.c (build_ref_for_offset): Preserve address-space
1377 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
1378 Drop useless address-space information on MEM_REF offsets.
1380 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
1382 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
1384 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
1386 PR rtl-optimization/70703
1387 * ira-color.c (update_conflict_hard_regno_costs): Use
1388 HOST_WIDE_INT instead of long.
1390 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
1393 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
1394 not defined for x86_64 target. Add -mmmx target option when __SSE2__
1396 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
1397 for x86_64 target. Handle -m3dnowa option.
1399 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
1401 PR rtl-optimization/70703
1402 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
1403 (update_conflict_hard_regno_costs): Use long instead of unsigned
1404 arithmetic for cost calculation.
1406 2017-04-05 Jakub Jelinek <jakub@redhat.com>
1407 Bernd Edlinger <bernd.edlinger@hotmail.de>
1410 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
1413 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
1416 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
1417 ptr_mode with Pmode throughout.
1418 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
1419 into probe_stack_range and use DImode.
1421 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
1424 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
1425 call_eh_return is true.
1427 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1429 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
1430 Initialize last_match_fntype_index.
1432 2017-04-05 Jakub Jelinek <jakub@redhat.com>
1435 * tree-nvr.c: Include internal-fn.h.
1436 (pass_return_slot::execute): Ignore internal calls without
1439 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1440 Richard Biener <rguenther@suse.de>
1443 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
1444 captures used multiple times, except for the last use.
1445 * generic-match-head.c: Include gimplify.h.
1447 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1449 PR tree-optimization/79390
1450 * target.h (struct noce_if_info): Declare.
1451 * targhooks.h (default_noce_conversion_profitable_p): Declare.
1452 * target.def (noce_conversion_profitable_p): New target hook.
1453 * ifcvt.h (struct noce_if_info): New type, moved from ...
1454 * ifcvt.c (struct noce_if_info): ... here.
1455 (noce_conversion_profitable_p): Renamed to ...
1456 (default_noce_conversion_profitable_p): ... this. No longer
1458 (noce_try_store_flag_constants, noce_try_addcc,
1459 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
1460 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
1461 instead of noce_conversion_profitable_p.
1462 * config/i386/i386.c: Include ifcvt.h.
1463 (ix86_option_override_internal): Don't override
1464 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
1465 (ix86_noce_conversion_profitable_p): New function.
1466 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
1467 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
1468 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
1469 * doc/tm.texi: Regenerated.
1471 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1473 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
1476 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
1479 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
1480 instructions for small multiply cores.
1482 2017-04-04 Jeff Law <law@redhat.com>
1484 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
1486 (mips_expand_vec_perm_const): Initialize elements in orig_perm
1487 that are not set by the loop over the elements.
1489 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1492 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
1493 int mode, convert_modes it to mode as unsigned, otherwise use
1494 lowpart_subreg to mode rather than SImode.
1495 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
1496 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
1497 Use DImode instead of SImode for the shift count operand.
1498 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
1501 2017-04-04 Richard Biener <rguenther@suse.de>
1504 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
1505 arithmetic done for the negate or the plus. Simplify.
1506 (A - (-B) -> A + B): Likewise.
1507 * fold-const.c (split_tree): Make sure to not negate pointers.
1509 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
1511 PR rtl-optimization/60818
1512 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
1513 a compare of comparisons with the thing compared if this results
1514 in a different machine mode.
1516 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
1518 * alias.c (base_alias_check): Fix typo in comment.
1519 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
1520 * cgraphunit.c (symbol_table::compile): Likewise.
1521 * collect2.c (maybe_run_lto_and_relink): Likewise.
1522 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
1523 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
1524 * config/avr/avr.c (avr_map_op_t): Likewise.
1525 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
1526 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
1527 * config/epiphany/epiphany.md (movcc): Likewise.
1528 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
1529 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
1531 * config/mips/mips.c (mips_save_restore_reg): Likewise.
1532 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
1533 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
1534 * config/sh/sh.c (sh_rtx_costs): Likewise.
1535 * fold-const.c (fold_truth_andor): Likewise.
1536 * genautomata.c (collapse_flag): Likewise.
1537 * gengtype.h (struct type::u::s): Likewise.
1538 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
1539 * input.c (FORMAT_AMOUNT): Likewise.
1540 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
1541 (known_aggs_to_agg_replacement_list): Likewise.
1542 * ipa-inline-analysis.c: Likewise.
1543 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
1544 * ipa-polymorphic-call.c
1545 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
1546 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
1547 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
1549 * modulo-sched.c (apply_reg_moves): Likewise.
1550 * omp-expand.c (build_omp_regions_1): Likewise.
1551 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
1552 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
1553 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
1554 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
1555 * value-prof.c: Likewise.
1556 * var-tracking.c (val_reset): Likewise.
1558 2017-04-03 Richard Biener <rguenther@suse.de>
1560 PR tree-optimization/80275
1561 * fold-const.c (split_address_to_core_and_offset): Handle
1564 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
1566 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
1567 descriptors is at least equal to that of functions.
1569 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1571 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
1573 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1576 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
1577 (mov<IMOD4:mode>): New expander.
1578 (*mov<IMOD4:mode>_internal): New insn and split pattern.
1580 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
1582 PR rtl-optimization/79405
1583 * fwprop.c (propagations_left): New variable.
1584 (forward_propagate_into): Decrement it.
1585 (fwprop_init): Initialize it.
1586 (fw_prop): If the variable has reached zero, stop propagating.
1587 (fwprop_addr): Ditto.
1589 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1592 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
1593 a FUNCTION_DECL, pass it as decl instead of origin to
1596 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
1598 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
1601 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
1604 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
1605 TARGET_VSX_SMALL_INTEGER.
1607 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1609 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1610 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
1612 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1614 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
1615 extraction from odd-numbered MSA register.
1617 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1620 * expmed.c (store_bit_field_1): Don't attempt to create
1621 a word subreg out of hard registers wider than word if they
1622 have HARD_REGNO_NREGS of 1 for their mode.
1625 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
1626 conversions to integer types wider than word and pointer.
1629 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
1630 (rtx_equal_for_cselib_p): Pass 0 to it.
1631 * cselib.c (cselib_hasher::equal): Likewise.
1632 (rtx_equal_for_cselib_1): Add depth argument. If depth
1633 is 128, don't look up VALUE locs and punt. Increment
1634 depth in recursive calls when walking VALUE locs.
1636 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
1638 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
1639 (make_gcov_file_name): Use the canonical path name for generating
1641 (read_line): Fix handling of files with ascii null bytes.
1643 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
1645 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
1646 to initialise a vector register instead
1647 of using a const_int.
1649 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1651 PR translation/80189
1652 * gimplify.c (omp_default_clause): Use %qs instead of %s in
1653 diagnostic messages.
1655 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
1658 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
1659 (dfp_diex_<mode>): Update mode of operand 1.
1660 * doc/extend.texi (dxex, dxexq): Document change to return type.
1661 (diex, diexq): Document change to argument type.
1663 2017-03-30 Martin Jambor <mjambor@suse.cz>
1666 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
1667 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
1668 it reflects the signature changes performed at the callee side.
1669 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
1670 to cgraph_build_function_type_skip_args.
1671 (build_function_decl_skip_args): Adjust call to the above function.
1673 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1676 * config/i386/sse.md
1677 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
1678 register as dest whenever it is a MEM not rtx_equal_p to the
1679 corresponding dup operand, and when forcing into reg move the
1680 reg into the memory afterwards.
1681 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
1682 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
1683 for the force_reg mode.
1684 (avx512vl_vextractf128<mode>): Use register as dest either
1685 always when a MEM, or when it is a MEM not rtx_equal_p to the
1686 corresponding dup operand, or even not when it is a CONST_VECTOR
1687 depending on the mode and lo vs. hi.
1688 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
1690 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
1691 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
1692 Likewise. Require that operands[2] is even.
1693 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
1694 Remove extraneous parens. Require that operands[2] is a multiple
1696 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
1697 operands[0] is a MEM if <mask_applied>, the predicates/constraints
1698 disallow memory then.
1700 2017-03-30 Richard Biener <rguenther@suse.de>
1702 PR tree-optimization/77498
1703 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
1704 to non-constants over backedges.
1706 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
1708 PR rtl-optimization/80233
1709 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
1710 as last_combined_insn. Do not test for BARRIER_P separately.
1712 2017-03-29 Andreas Schwab <schwab@suse.de>
1715 * calls.c (prepare_call_address): Convert funexp to Pmode before
1716 copying to temp reg.
1718 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1720 PR tree-optimization/80158
1721 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
1722 Handle possible future case of more than one alternate
1724 (replace_rhs_if_not_dup): Likewise.
1725 (replace_one_candidate): Likewise.
1727 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
1729 PR rtl-optimization/80193
1730 * ira.c (ira): Do not check allocation for LRA.
1732 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
1734 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
1735 (nvptx_output_simt_exit): Declare.
1736 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
1737 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
1738 (init_softstack_frame): Move initialization of crtl->is_leaf to...
1739 (nvptx_declare_function_name): ...here. Emit declaration of local
1740 memory space buffer for omp_simt_enter insn.
1741 (nvptx_output_unisimt_switch): New.
1742 (nvptx_output_softstack_switch): New.
1743 (nvptx_output_simt_enter): New.
1744 (nvptx_output_simt_exit): New.
1745 * config/nvptx/nvptx.h (struct machine_function): New fields
1746 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
1747 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
1748 (UNSPECV_SIMT_EXIT): Ditto.
1749 (omp_simt_enter_insn): New insn.
1750 (omp_simt_enter): New expansion.
1751 (omp_simt_exit): New insn.
1752 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
1754 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
1755 (expand_GOMP_SIMT_ENTER_ALLOC): New.
1756 (expand_GOMP_SIMT_EXIT): New.
1757 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
1758 (GOMP_SIMT_ENTER_ALLOC): Ditto.
1759 (GOMP_SIMT_EXIT): Ditto.
1760 * target-insns.def (omp_simt_enter): New insn.
1761 (omp_simt_exit): Ditto.
1762 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
1764 (lower_rec_simd_input_clauses): Implement SIMT privatization.
1765 (lower_rec_input_clauses): Likewise.
1766 (lower_lastprivate_clauses): Handle SIMT privatization.
1768 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
1769 (ompdevlow_adjust_simt_enter): New.
1770 (find_simtpriv_var_op): New.
1771 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
1772 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
1774 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
1775 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
1776 (copy_decl_for_dup_finish): Ditto.
1778 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
1780 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
1783 * config/i386/i386.c (ix86_option_override_internal): Always
1784 allow -mpreferred-stack-boundary=3 for 64-bit targets.
1786 2017-03-28 Bin Cheng <bin.cheng@arm.com>
1788 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
1790 2017-03-28 Bin Cheng <bin.cheng@arm.com>
1792 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
1793 mark new edge's irreducible flag accordign to it.
1794 (vect_do_peeling): Check loop preheader edge's irreducible flag
1795 and pass it to function slpeel_add_loop_guard.
1797 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
1799 PR tree-optimization/80218
1800 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
1801 Update block frequencies and counts.
1803 2017-03-28 Richard Biener <rguenther@suse.de>
1805 PR tree-optimization/78644
1806 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
1807 of a simplification result we may not use it at all.
1809 2017-03-28 Richard Biener <rguenther@suse.de>
1812 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
1813 without arguments, generate default definition of a SSA name.
1815 2017-03-28 Richard Biener <rguenther@suse.de>
1818 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
1819 TYPE_REF_CAN_ALIAS_ALL references.
1820 * fold-const.c (fold_indirect_ref_1): Likewise.
1822 2017-03-28 Martin Liska <mliska@suse.cz>
1825 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
1826 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
1828 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1829 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
1831 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
1832 (EXTRA_SPECS): Define.
1833 (SUBTARGET_EXTRA_SPECS): Likewise.
1834 (SUBTARGET_CPP_SPEC): Likewise.
1835 * config/arc/elf.h (EXTRA_SPECS): Renamed to
1836 SUBTARGET_EXTRA_SPECS.
1837 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
1839 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1841 * config/arc/simdext.md (vst64_insn): Update pattern.
1842 (vld32wh_insn): Likewise.
1843 (vld32wl_insn): Likewise.
1844 (vld64_insn): Likewise.
1845 (vld32_insn): Likewise.
1847 2017-03-28 Marek Polacek <polacek@redhat.com>
1850 * fold-const.c (fold_comparison): Use protected_set_expr_location
1851 instead of SET_EXPR_LOCATION.
1853 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
1855 * tree.c (add_expr): Avoid name lookup warning.
1857 2017-03-27 Jeff Law <law@redhat.com>
1859 PR tree-optimization/80216
1860 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
1861 function name. Limit recursion depth.
1862 (record_temporary_equivalences): Corresponding changes.
1864 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
1866 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
1869 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1872 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
1874 * cfgcleanup.c (reg_note_cfa_p): New array.
1875 (insns_have_identical_cfa_notes): New function.
1876 (old_insns_match_p): Don't cross-jump in between /f
1877 and non-/f instructions. If both i1 and i2 are frame related,
1878 verify all CFA notes, their order and content.
1880 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1883 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
1884 HImode and SImode with zero extend to DImode to one insn.
1885 (bswap<mode>2_extenddi): Likewise.
1886 (bswapsi2_extenddi): Likewise.
1887 (bswaphi2_extendsi): Likewise.
1888 (bswaphi2): Combine bswap HImode and SImode into one insn.
1889 Separate memory insns from swapping register.
1890 (bswapsi2): Likewise.
1891 (bswap<mode>2): Likewise.
1892 (bswaphi2_internal): Delete, no longer used.
1893 (bswapsi2_internal): Likewise.
1894 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
1895 store, and gpr<-gpr swap insns.
1896 (bswap<mode>2_store): Likewise.
1897 (bswaphi2_reg): Register only splitter, combine with the splitter.
1898 (bswaphi2 splitter): Likewise.
1899 (bswapsi2_reg): Likewise.
1900 (bswapsi2 splitter): Likewise.
1901 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
1902 the insns into load, store, and register/register insns.
1903 (bswapdi2_ldbrx): Likewise.
1904 (bswapdi2_load): Likewise.
1905 (bswapdi2_store): Likewise.
1906 (bswapdi2_reg): Likewise.
1908 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
1910 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
1911 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
1913 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1916 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
1918 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1919 special handling for target option conflicts between dform
1920 options (-mpower9-dform, -mpower9-dform-vector,
1921 -mpower9-dform-scalar) and -mno-direct-move.
1923 2017-03-27 Richard Biener <rguenther@suse.de>
1925 PR tree-optimization/80181
1926 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
1928 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1930 * config/arc/predicates.md (move_double_src_operand): Replace the
1931 call to move_double_src_operand with a call to address_operand.
1933 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1935 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
1936 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
1937 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
1939 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1941 * config/arc/predicates.md (long_immediate_loadstore_operand):
1942 Consider scaled addresses cases.
1944 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1946 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
1947 restored when in interrupt.
1948 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
1949 doesn't have delay slot.
1951 2017-03-27 Richard Biener <rguenther@suse.de>
1954 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
1955 inlined thunk clones.
1957 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1960 * asan.c (instrument_derefs): Copy over last operand from
1961 original COMPONENT_REF to the new COMPONENT_REF with
1962 DECL_BIT_FIELD_REPRESENTATIVE.
1963 * ubsan.c (instrument_object_size): Likewise.
1965 2017-03-27 Richard Biener <rguenther@suse.de>
1967 PR tree-optimization/80170
1968 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
1969 sure DR/SCEV didnt fold in constants we do not see when looking
1970 at the reference base alignment.
1972 2017-03-27 Richard Biener <rguenther@suse.de>
1975 * gimple-fold.c (fold_ctor_reference): Properly guard against
1976 NULL return value from canonicalize_constructor_val.
1978 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
1981 * config/i386/i386.c (ix86_expand_builtin)
1982 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
1983 flags reg setting and flags reg using instructions.
1984 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
1985 clobbering instructions to zero extend op2.
1987 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
1989 * doc/install.texi (Configuration) <--with-aix-soname>:
1990 Update link to AIX ld.
1992 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
1994 PR rtl-optimization/80160
1995 PR rtl-optimization/80159
1996 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
1997 reg_alternate_class into account.
1999 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
2002 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
2003 to consider in curr_insn_transform.
2005 2017-03-24 Jakub Jelinek <jakub@redhat.com>
2007 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
2008 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
2009 and emit_mode_inner.
2011 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2013 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
2014 argument to the overloaded builtin variants. Use the new flag to
2015 deprecate certain builtin variants.
2016 * config/s390/s390-builtin-types.def: Add new builtin types.
2017 * config/s390/s390-builtins.h: Support new flags field for
2018 overloaded builtins.
2019 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
2020 (s390_macro_to_expand): Enable vector float data type.
2021 (s390_cpu_cpp_builtins_internal): Indicate support of the new
2022 builtins by incrementing the __VEC__ version number.
2023 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
2025 (s390_resolve_overloaded_builtin): Emit error messages depending
2026 on the builtin flags.
2027 * config/s390/s390.c (s390_expand_builtin): Support additional
2028 flags argument. Change error message to match the messages
2029 emitted in s390-c.c.
2030 * config/s390/s390.md: New UNSPEC_* constants.
2031 (op_type): Add new instruction types.
2032 * config/s390/vecintrin.h: Add new builtins and test data class
2034 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
2035 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
2036 (VEC_INEXACT, VEC_NOINEXACT): New constants.
2037 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
2038 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
2039 ("vec_mergel<mode>"): V_HW -> VEC_HW.
2041 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
2042 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
2043 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
2044 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
2046 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
2047 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
2048 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
2049 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
2051 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
2052 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
2053 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
2054 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
2055 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
2056 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
2057 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
2059 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
2060 ("vec_scatter_element<V_HW_4:mode>_DI")
2061 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
2062 ("vec_fpint<mode>", "vflls")
2063 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
2064 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
2065 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
2066 ("*vec_cmphe<mode>_cc"): ... these.
2068 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
2069 mode constant instead of magic value.
2071 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2073 * config/s390/s390.c (s390_expand_vec_compare): Support other
2074 vector floating point modes than just V2DF.
2075 (s390_expand_vcond): Likewise.
2076 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
2077 (s390_cannot_change_mode_class): Prevent mode changes between TF
2078 and V1TF in vector registers.
2079 * config/s390/s390.md (DF, SF): New mode attributes.
2080 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
2081 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
2082 SFmode support for VRs.
2083 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
2085 (VFT, VF_HW): New mode iterators.
2086 (vw, sdx): New mode attributes.
2087 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
2088 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
2089 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
2090 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
2091 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
2092 also the new vector floating point modes. Renaming to ...
2094 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
2095 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
2096 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
2097 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
2098 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
2099 ("vec_unordered<mode>"): ... these.
2101 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
2102 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
2103 ("*vec_extendv2df"): New insn definitions.
2105 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2107 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
2108 ("mulditi3_2", "*muldi3_sign"): New patterns.
2109 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
2110 rename the pattern definition.
2112 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2114 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
2116 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
2118 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2120 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
2121 instruction if possible.
2122 * config/s390/vector.md (vec_halfnumelts): New mode
2124 ("*vec_vllezlf<mode>"): New pattern.
2126 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2128 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
2129 ("popcountv4si2", "popcountv2di2"): Rename to ...
2130 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
2131 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
2133 ("popcount<mode>2_vxe"): New pattern.
2135 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2137 * common/config/s390/s390-common.c (processor_flags_table): Add
2139 * config.gcc: Add arch12.
2140 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2141 Default to arch12 for unknown CPU model numbers.
2142 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
2143 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
2144 PROCESSOR_max sanity check.
2145 * config/s390/s390-opts.h (enum processor_type): Add
2147 * config/s390/s390.c (processor_table): Add arch12.
2148 (s390_expand_builtin): Add check for B_VXE flag.
2149 (s390_issue_rate): Add PROCESSOR_ARCH12.
2150 (s390_get_sched_attrmask): Likewise.
2151 (s390_get_unit_mask): Likewise.
2152 (s390_sched_score): Enable z13 scheduling for arch12.
2153 (s390_sched_reorder): Likewise.
2154 (s390_sched_variable_issue): Likewise.
2155 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
2157 (s390_tune_attr): Use z13 scheduling also for arch12.
2158 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
2159 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
2160 (TARGET_VXE_P): New macros.
2161 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
2162 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
2163 * config/s390/s390.opt: Add arch12 as processor_type.
2165 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2167 * config/s390/s390.md
2168 ("fixuns_truncdddi2", "fixuns_trunctddi2")
2169 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
2170 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
2172 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
2173 Rename expanders to ...
2175 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
2176 ("fixuns_truncdddi2_emu"): ... these.
2178 ("fixuns_trunc<mode>si2_emu"): New expander.
2180 ("*fixuns_truncdfdi2_z13"): Rename to ...
2181 ("*fixuns_truncdfdi2_vx"): ... this.
2183 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2185 * config/s390/2964.md: Remove the single element vector compare
2186 instructions which are no longer used.
2187 * config/s390/s390.c (s390_select_ccmode): Remove handling of
2189 (s390_canonicalize_comparison): Remove handling of DFmode
2191 (s390_expand_vec_compare_scalar): Remove function.
2192 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
2193 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
2195 ("*cmp<mode>_ccs"): Add wfcdb instruction.
2197 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2199 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
2201 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
2202 will anyway by matched by mov<mode>_64dfp.
2204 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2206 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
2207 vlef/vstef. Add missing operand to vleif.
2209 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2211 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
2212 pair for all vector types with 64 bit elements.
2213 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
2214 * config/s390/vector.md (V_HW_64): ... here.
2215 (V_128_NOSINGLE): New mode iterator.
2216 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
2217 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
2218 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
2219 ("*vec_load_pairv2di"): Change to ...
2220 ("*vec_load_pair<mode>"): ... this one.
2222 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2224 * config/s390/constraints.md: Add comments.
2225 (jKK): Reject element sizes > 8 bytes.
2226 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
2228 * config/s390/s390.md: Add the s_operand checks formerly in
2229 s390_split_ok_p to various splitters where they are still
2231 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
2232 for 128 bit vectors. Plus two splitters.
2234 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2236 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
2239 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2242 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
2243 error if the boundary argument is not constant.
2245 2017-03-24 Jakub Jelinek <jakub@redhat.com>
2247 PR rtl-optimization/80112
2248 * loop-doloop.c (doloop_condition_get): Don't check condition
2249 if cmp isn't SET with IF_THEN_ELSE src.
2251 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2253 PR tree-optimization/80158
2254 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
2255 replacing a candidate statement, also replace it for the
2256 candidate's alternate interpretation.
2257 (replace_rhs_if_not_dup): Likewise.
2258 (replace_one_candidate): Likewise.
2260 2017-03-24 Richard Biener <rguenther@suse.de>
2262 PR tree-optimization/80167
2263 * graphite-isl-ast-to-gimple.c
2264 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
2266 (translate_isl_ast_to_gimple::get_rename): Likewise.
2268 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
2270 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
2271 handling of certain combinations of target options, including the
2272 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
2273 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
2275 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2278 * config/arm/arm.md (*load_multiple): Add reload_completed to
2281 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2282 Richard Biener <rguenth@suse.de>
2284 PR tree-optimization/79908
2285 PR tree-optimization/80136
2286 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2287 been cast away, gimplify_and_add suffices.
2289 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
2291 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
2293 2017-03-23 Richard Biener <rguenther@suse.de>
2295 PR tree-optimization/80032
2296 * gimplify.c (gimple_push_cleanup): Forced unconditional
2297 cleanups still have to go to the conditional_cleanups
2300 2017-03-22 Jakub Jelinek <jakub@redhat.com>
2302 PR tree-optimization/80072
2303 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
2305 (next_operand_entry_id): Change type to unsigned int.
2306 (sort_by_operand_rank): Make sure to return the right return value
2307 even if unsigned fields are bigger than INT_MAX.
2308 (struct oecount): Change cnt and id type to unsigned int.
2309 (oecount_hasher::equal): Formatting fix.
2310 (oecount_cmp): Make sure to return the right return value
2311 even if unsigned fields are bigger than INT_MAX.
2312 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
2315 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
2316 TREE_READONLY on result if writing it more than once.
2319 * doc/invoke.texi (-fsanitize=thread): Document that with
2320 -fnon-call-exceptions atomics are not able to throw
2324 * tsan.c: Include tree-eh.h.
2325 (instrument_builtin_call): Call maybe_clean_eh_stmt or
2326 maybe_clean_or_replace_eh_stmt where needed.
2327 (instrument_memory_accesses): Add cfg_changed argument.
2328 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
2329 if it returned true.
2330 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
2332 PR rtl-optimization/63191
2333 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
2334 wrapper function, moved the whole old content into ...
2335 (ix86_delegitimize_address_1): ... this. New inline function.
2336 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
2337 true as last argument instead of ix86_delegitimize_address.
2339 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
2341 * config/aarch64/aarch64.c (generic_branch_cost): Copy
2342 cortexa57_branch_cost.
2344 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
2346 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
2348 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2351 * doc/md.texi (Constraints): Document wA constraint.
2352 * config/rs6000/constraints.md (wA): New.
2353 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
2354 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
2355 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
2356 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
2358 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
2361 * gimplify.c (is_oacc_declared): New function.
2362 (oacc_default_clause): Use it to set default flags for acc declared
2363 variables inside parallel regions.
2364 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
2366 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
2367 declare attribute to any decl as necessary.
2369 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2372 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
2373 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
2374 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
2375 (arm_arch_lpae): This.
2376 * config/arm/arm.c (arm_arch7ve): Rename into ...
2377 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
2378 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
2381 2017-03-22 Martin Liska <mliska@suse.cz>
2384 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
2385 error message instead of an ICE.
2387 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2389 * doc/extend.texi (6.11 Additional Floating Types): Revise.
2391 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
2393 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2395 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2398 2017-03-21 Martin Sebor <msebor@redhat.com>
2400 * doc/extend.texi: Use "cannot" instead of "can't."
2401 * doc/hostconfig.texi: Same.
2402 * doc/install.texi: Same.
2403 * doc/invoke.texi: Same.
2404 * doc/loop.texi: Same.
2405 * doc/md.texi: Same.
2406 * doc/objc.texi: Same.
2407 * doc/rtl.texi: Same.
2408 * doc/tm.texi: Same.
2409 * doc/tm.texi.in: Same.
2410 * doc/trouble.texi: Same.
2412 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
2415 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
2416 (collect_checksum_attributes): Set it.
2417 (die_checksum_ordered): Use it.
2419 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2421 PR tree-optimization/79908
2422 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
2423 change: For a VA_ARG whose LHS has been cast away, use
2424 force_gimple_operand to construct the side effects.
2426 2017-03-21 David Malcolm <dmalcolm@redhat.com>
2428 PR translation/80001
2429 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
2430 more amenable to translation.
2431 (oacc_loop_auto_partitions): Likewise.
2433 2017-03-21 Marek Polacek <polacek@redhat.com>
2434 Martin Sebor <msebor@redhat.com>
2436 PR tree-optimization/80109
2437 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
2440 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2441 Segher Boessenkool <segher@kernel.crashing.org>
2444 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
2445 check reg_used_between_p between insn and one of succ or succ2
2446 depending on if succ is artificial insn not inserted into insn
2449 2017-03-21 Martin Liska <mliska@suse.cz>
2451 PR gcov-profile/80081
2452 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
2453 * doc/gcc.texi: Include gcov-dump stuff.
2454 * doc/gcov-dump.texi: New file.
2456 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
2458 PR rtl-optimization/79150
2459 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
2460 conditional jump, if the jump is the last insn of the loop.
2462 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2463 Richard Biener <rguenth@suse.de>
2465 PR tree-optimization/79908
2466 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2467 been cast away, use force_gimple_operand to construct the side
2470 2017-03-21 Martin Liska <mliska@suse.cz>
2473 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
2476 2017-03-21 Brad Spengler <spender@grsecurity.net>
2479 * plugin.c (htab_hash_plugin): New function.
2480 (add_new_plugin): Use it and adjust.
2481 (parse_plugin_arg_opt): Adjust.
2482 (init_one_plugin): Likewise.
2484 2017-03-21 Richard Biener <rguenther@suse.de>
2486 PR tree-optimization/80032
2487 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
2488 if set force the cleanup to happen unconditionally.
2489 (gimplify_target_expr): Push inserted clobbers with force_uncond
2490 to avoid them being removed by control-dependent DCE.
2492 2017-03-21 Richard Biener <rguenther@suse.de>
2494 PR tree-optimization/80122
2495 * tree-inline.c (copy_bb): Do not expans va-arg packs or
2496 va_arg_pack_len when the inlined call stmt requires pack
2498 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
2500 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2503 * tsan.c (instrument_builtin_call): If the memory model argument
2504 is not a constant, assume it is valid.
2507 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
2510 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2512 PR rtl-optimization/79910
2513 * combine.c (can_combine_p): Do not allow combining an I0 or I1
2514 if its dest is used by an insn before I2 (other than the combined
2515 insns themselves, which are properly handled already).
2517 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2520 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2522 * combine.c (record_used_regs): New static function.
2523 (try_combine): Handle situations where there is an additional
2524 instruction between I2 and I3 which needs to have a LOG_LINK
2528 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2530 * combine.c (try_combine): Delete redundant i1 test. Call
2531 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2533 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2536 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
2539 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2541 PR tree-optimization/80054
2542 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
2543 the optimization if a PHI or any of its arguments is not dominated
2544 by the candidate's basis. Use gphi* rather than gimple* as
2546 (replace_profitable_candidates): Clean up a gimple* variable that
2547 should be a gphi* variable.
2549 2017-03-20 Martin Sebor <msebor@redhat.com>
2552 * doc/extend.texi (attribute constructor): Document present limitation.
2554 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
2557 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
2558 __POWER9_VECTOR__ #ifdef control, change template definition to
2559 use Power9-specific built-in function.
2560 (vec_any_eq): Likewise.
2561 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
2562 to control outcomes from this test.
2563 (vector_ae_<mode>p): For VEC_F modes, likewise.
2565 2017-03-20 Ian Lance Taylor <iant@google.com>
2567 * config/i386/i386.c (ix86_function_regparm): Save an extra
2568 register for -fsplit-stack with DECL_STATIC_CHAIN.
2570 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2573 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
2574 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
2576 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2578 * config/riscv/riscv.c (riscv_print_operand): Use "fence
2580 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
2583 2017-03-20 Marek Polacek <polacek@redhat.com>
2586 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
2588 2017-03-20 Richard Biener <rguenther@suse.de>
2590 PR tree-optimization/80113
2591 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
2592 allocate extra SSA name for PHI def.
2593 (add_close_phis_to_outer_loops): Likewise.
2594 (add_close_phis_to_merge_points): Likewise.
2595 (copy_loop_close_phi_args): Likewise.
2596 (copy_cond_phi_nodes): Likewise.
2598 2017-03-20 Martin Liska <mliska@suse.cz>
2601 * tree-chkp.c (chkp_build_returned_bound): Do not build
2602 returned bounds for a LHS that's not a BOUNDED_P type.
2604 2017-03-20 Martin Liska <mliska@suse.cz>
2608 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
2609 COMPLEX_CST and VECTOR_CST.
2611 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2614 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
2615 target operand. A new splitter adds the clobber statement in case
2616 the target operand is dead anyway.
2618 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
2620 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
2621 to age-old versions of binutils and glibc.
2623 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
2625 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
2627 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2629 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
2631 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2633 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
2634 requirement for binutils 2.13.
2636 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2638 * combine.c (try_combine): Delete redundant i1 test. Call
2639 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2641 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
2643 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
2644 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
2646 <riscv64-*-elf>: Re-arrange section
2647 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
2648 <riscv32-*-linux>: Likewise.
2649 <riscv64-*-elf>: Likewise
2650 <riscv64-*-linux>: Likewise.
2652 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
2655 * aarch64.opt(verbose-cost-dump): Fix typo.
2657 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
2660 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
2661 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
2663 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2665 * reload.c (find_reloads): When reloading a nonoffsettable address,
2666 use RELOAD_OTHER for it and its address reloads.
2668 PR rtl-optimization/79910
2669 * combine.c (record_used_regs): New static function.
2670 (try_combine): Handle situations where there is an additional
2671 instruction between I2 and I3 which needs to have a LOG_LINK
2674 2017-03-17 Jeff Law <law@redhat.com>
2676 PR tree-optimization/71437
2677 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
2678 conditional in the hash table first.
2679 (vrp_dom_walker::before_dom_children): Extract condition from
2680 ASSERT_EXPR. Record condition, its inverion and any implied
2683 2017-03-17 Marek Polacek <polacek@redhat.com>
2684 Markus Trippelsdorf <markus@trippelsdorf.de>
2686 PR tree-optimization/80079
2687 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
2690 2017-03-17 Richard Biener <rguenther@suse.de>
2693 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
2694 Properly verify the LHS before the RHS possibly claims to be
2696 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
2699 2017-03-17 Martin Jambor <mjambor@suse.cz>
2701 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
2702 (List of -O2 options): Likewise.
2703 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
2706 2017-03-17 Tom de Vries <tom@codesourcery.com>
2708 * gcov-dump.c (print_usage): Print bug_report_url.
2710 2017-03-17 Richard Biener <rguenther@suse.de>
2713 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
2714 (parser::peek): Likewise.
2716 2017-03-17 Richard Biener <rguenther@suse.de>
2718 PR tree-optimization/80048
2719 * sese.c (free_sese_info): Properly release rename_map and
2720 copied_bb_map elements.
2722 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
2724 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
2725 Add linked-list forward and backlinks. Insert on
2726 construction, remove on destruction.
2727 (class pass_store_merging): Add m_stores_head field.
2728 (pass_store_merging::terminate_and_process_all_chains):
2729 Iterate over m_stores_head list.
2730 (pass_store_merging::terminate_all_aliasing_chains):
2732 (pass_store_merging::execute): Check for debug stmts first.
2733 Push new chains onto the m_stores_head stack.
2735 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2738 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
2739 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
2740 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
2742 2017-03-16 Jeff Law <law@redhat.com>
2744 PR tree-optimization/71437
2745 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
2746 member function. Implementation moved into after_dom_children
2747 member function and into the threader's thread_outgoing_edges
2749 (dom_opt_dom_walker::after_dom_children): Simplify by moving
2750 some code into new thread_outgoing_edges.
2751 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
2752 definition. Simplify marker handling (do it here). Assume we always
2753 have the available expression and the const/copies tables.
2754 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
2756 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
2757 * tree-vrp.c (equiv_stack): No longer file scoped.
2758 (vrp_dom_walker): New class.
2759 (vrp_dom_walker::before_dom_children): New member function.
2760 (vrp_dom_walker::after_dom_children): Likewise.
2761 (identify_jump_threads): Setup domwalker. Use it rather than
2762 walking edges in a random order by hand. Simplify setup/finalization.
2763 (finalize_jump_threads): Remove.
2764 (vrp_finalize): Do not call identify_jump_threads here.
2765 (execute_vrp): Do it here instead and call thread_through_all_blocks
2768 PR tree-optimization/71437
2769 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
2771 (simplify_stmt_for_jump_threading): Add basic_block argument. All
2773 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
2774 (dom_opt_dom_walker::thread_across_edge): Remove
2775 handle_dominating_asserts argument. All callers changed.
2776 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
2777 changes. Remove calls to lhs_of_dominating_assert. Other
2778 uses of handle_dominating_asserts turn into unconditional code
2779 (simplify_control_stmt_condition_1): Likewise.
2780 (simplify_control_stmt_condition): Likewise.
2781 (thread_through_normal_block, thread_across_edge): Likewise.
2782 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
2783 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
2784 object if it is not an SSA_NAME.
2785 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
2786 before calling into the VRP specific simplifiers.
2787 (identify_jump_threads): Remove handle_dominating_asserts
2790 2017-03-16 Jakub Jelinek <jakub@redhat.com>
2793 * tree-diagnostic.c (default_tree_printer): No longer static.
2794 * tree-diagnostic.h (default_tree_printer): New prototype.
2796 2017-03-16 Tamar Christina <tamar.christina@arm.com>
2798 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
2799 Change ins into fmov.
2801 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2803 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
2804 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
2805 Use h_con constraint for operand 1.
2806 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
2807 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
2809 2017-03-15 Jeff Law <law@redhat.com>
2811 PR tree-optimization/71437
2812 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
2813 (record_temporary_equivalences): Use it.
2815 PR tree-optimization/71437
2816 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
2817 tree-ssa-scopedtables.
2818 (lookup_avail_expr, build_and_record_new_cond): Likewise.
2819 (record_conditions, record_cond, vuse_eq): Likewise.
2820 (record_edge_info): Adjust to API tweak of record_conditions.
2821 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
2822 (record_temporary_equivalences, optimize_stmt): Likewise.
2823 (eliminate_redundant_computations): Likewise.
2824 (record_equivalences_from_stmt): Likewise.
2825 * tree-ssa-scopedtables.c: Include options.h and params.h.
2826 (vuse_eq): New function, moved from tree-ssa-dom.c
2827 (build_and_record_new_cond): Likewise.
2828 (record_conditions): Likewise. Accept vector of conditions rather
2829 than edge_equivalence structure for first argument.
2830 for the first argument.
2831 (avail_exprs_stack::lookup_avail_expr): New member function, moved
2832 from tree-ssa-dom.c.
2833 (avail_exprs_stack::record_cond): Likewise.
2834 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
2835 from tree-ssa-dom.c.
2836 (avail_exprs_stack): Add new member functions lookup_avail_expr
2838 (record_conditions): Declare.
2840 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
2843 * lra-constraints.c (process_alt_operands): Increase reject for
2844 reloading an input/output operand.
2846 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2849 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
2850 insns to convert from signed/unsigned char/short to IEEE 128-bit
2852 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
2854 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
2857 * config/i386/i386.c (ix86_vector_duplicate_value): Create
2858 subreg of inner mode for values already in registers.
2860 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
2862 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
2863 iteration reg is used after the loop.
2865 2017-03-14 Martin Sebor <msebor@redhat.com>
2867 PR tree-optimization/79800
2868 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
2869 precision in negative-positive range.
2870 (format_floating): Call non-const overload with adjusted precision.
2872 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2875 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
2878 2017-03-14 Martin Sebor <msebor@redhat.com>
2881 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
2882 * builtins.def (aligned_alloc): Use it.
2885 * Makefile.in (GTFILES): Add calls.c.
2886 * calls.c: Include "gt-calls.h".
2888 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
2890 PR rtl-optimization/79728
2891 * regs.h (struct target_regs): New field
2892 x_contains_allocatable_regs_of_mode.
2893 (contains_allocatable_regs_of_mode): New macro.
2894 * reginfo.c (init_reg_sets_1): Initialize it, and change
2895 contains_reg_of_mode so it includes global regs as well.
2896 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
2897 rather than contains_regs_of_mode.
2899 2017-03-14 Martin Liska <mliska@suse.cz>
2901 * doc/invoke.texi: Document options that can't be combined with
2902 -fcheck-pointer-bounds.
2904 2017-03-14 Martin Liska <mliska@suse.cz>
2907 * doc/invoke.texi (-Wchkp): Document the option.
2909 2017-03-14 Martin Liska <mliska@suse.cz>
2911 * Makefile.in: Install gcov-dump.
2913 2017-03-14 Martin Liska <mliska@suse.cz>
2915 * multiple_target.c (expand_target_clones): Bail out for
2916 an invalid attribute.
2918 2017-03-14 Richard Biener <rguenther@suse.de>
2920 * alias.c (struct alias_set_entry): Pack properly.
2921 * cfgloop.h (struct loop): Likewise.
2922 * cse.c (struct set): Likewise.
2923 * ipa-utils.c (struct searchc_env): Likewise.
2924 * loop-invariant.c (struct invariant): Likewise.
2925 * lra-remat.c (struct cand): Likewise.
2926 * recog.c (struct change_t): Likewise.
2927 * rtl.h (struct address_info): Likewise.
2928 * symbol-summary.h (function_summary): Likewise.
2929 * tree-loop-distribution.c (struct partition): Likewise.
2930 * tree-object-size.c (struct object_size_info): Likewise.
2931 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
2932 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
2933 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
2934 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
2935 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
2936 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
2937 (struct _stmt_vec_info): Likewise.
2939 2017-03-14 Martin Liska <mliska@suse.cz>
2942 * multiple_target.c (create_dispatcher_calls): Check that
2943 a target can create a function dispatcher.
2945 2017-03-14 Martin Liska <mliska@suse.cz>
2948 * multiple_target.c (expand_target_clones): Drop local.local
2949 flag for default implementation.
2951 2017-03-14 Richard Biener <rguenther@suse.de>
2953 PR tree-optimization/80030
2954 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
2956 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
2958 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
2959 gcc_fallthrough() instead of __attribute__((fallthrough));
2961 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2963 * doc/gcc.texi: Remove "up" link to (DIR).
2964 * doc/gccint.texi: Ditto.
2966 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2968 * doc/install.texi (Specific) <avr>: Remove reference to
2971 2017-03-13 Jeff Law <law@redhat.com>
2973 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
2974 attribute rather than comments.
2976 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
2977 match_scratch operand is highest.
2979 2017-03-13 Martin Liska <mliska@suse.cz>
2982 * ipa-pure-const.c (warn_function_noreturn): If the declarations
2983 is a CHKP clone, use original declaration.
2985 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2987 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
2988 (arc_conditional_register_usage): Use a different allocation order
2989 when optimizing for size.
2990 * common/config/arc/arc-common.c (arc_option_optimization_table):
2991 Section anchors default on when optimizing for size.
2993 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2995 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
2997 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2999 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
3000 * config/arc/arc.md (cpu_facility): Add cd variant.
3001 (*movqi_insn): Add code density variant.
3002 (*movhi_insn): Likewise.
3003 (*movqi_insn): Likewise.
3004 (*addsi3_mixed): Likewise.
3005 (subsi3_insn): Likewise.
3007 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3009 * config/arc/arc.md (movsi_cond_exec): Update constraint.
3011 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
3013 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
3014 expressions with MINUS and UNARY ops.
3016 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3019 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
3021 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
3022 between vec_select and vector argument.
3023 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
3024 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
3025 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
3026 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
3027 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
3028 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
3030 2017-03-13 Richard Biener <rguenther@suse.de>
3033 * params.def (vect-max-peeling-for-alignment): Fix typo.
3035 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
3037 * doc/install.texi (Specific) <mips-*-*>: Remove description of
3038 issue that only occurred with binutils below 2.18.
3040 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
3042 * doc/install.texi (Specific) <cris-axis-elf>: No longer
3043 refer to binutils 2.11/2.12 minimum.
3045 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
3047 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
3048 ftp.kernel.org and simplify binutils requirement.
3050 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
3052 * doc/invoke.texi (Warning Options): Fix spelling of link-time
3054 (Optimize Options): Ditto. Also remove redundancy.
3056 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3058 PR translation/79848
3059 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
3061 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
3062 to G_ to avoid double translation.
3064 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3066 PR translation/79923
3067 * auto-profile.c (get_combined_location): Convert leading
3068 character of diagnostics to lower case and remove trailing period.
3069 (read_profile): Likewise for various diagnostics.
3070 * config/arm/arm.c (arm_option_override): Remove trailing period
3071 from various diagnostics.
3072 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
3073 (msp430_expand_delay_cycles): Likewise.
3075 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3078 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
3079 full command-line argument, rather than just "str".
3080 (aarch64_validate_march): Likewise.
3081 (aarch64_validate_mtune): Likewise.
3083 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
3085 PR rtl-optimization/78911
3086 * lra-assigns.c (must_not_spill_p): New function.
3087 (spill_for): Use it.
3089 2017-03-10 Jakub Jelinek <jakub@redhat.com>
3091 PR tree-optimization/79981
3092 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
3093 ATOMIC_COMPARE_EXCHANGE ifn result.
3094 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
3095 IFN_ATOMIC_COMPARE_EXCHANGE.
3097 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3100 * opts.c (parse_sanitizer_options): Add missing question mark to
3101 "did you mean" message.
3103 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3105 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
3107 (VMULEUH_UNS): Likewise.
3108 (VMULOUB_UNS): Likewise.
3109 (VMULOUH_UNS): Likewise.
3110 * config/rs6000/rs6000.c (builtin_function_type): Remove
3111 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
3113 2017-03-10 David Malcolm <dmalcolm@redhat.com>
3116 * read-rtl-function.c (function_reader::read_rtx_operand): Update
3117 x with result of extra_parsing_for_operand_code_0.
3118 (function_reader::extra_parsing_for_operand_code_0): Convert
3119 return type from void to rtx, returning x. When reading
3120 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
3121 larger size containing struct block_symbol.
3123 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
3125 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
3126 -mfloat128-hardware without -m64.
3128 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
3131 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
3132 entries to the case statement that marks unsigned arguments to
3133 overloaded functions.
3135 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
3137 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
3138 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
3140 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
3143 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
3144 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
3146 2017-03-10 Martin Liska <mliska@suse.cz>
3150 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
3151 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
3154 2017-03-10 Olivier Hainque <hainque@adacore.com>
3156 * tree-switch-conversion (array_value_type): Start by resetting
3157 candidate type to it's main variant.
3159 2017-03-10 Jakub Jelinek <jakub@redhat.com>
3161 PR rtl-optimization/79909
3162 * combine.c (try_combine): Use simplify_replace_rtx on individual
3163 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
3164 of the whole CALL_INSN_FUNCTION_USAGE.
3166 PR tree-optimization/79972
3167 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
3168 get_range_info on SSA_NAMEs. Formatting fixes.
3170 2017-03-10 Richard Biener <rguenther@suse.de>
3171 Jakub Jelinek <jakub@redhat.com>
3173 PR tree-optimization/77975
3174 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
3175 edge to be constant.
3176 (get_val_for): For constant x return it. Formatting fix.
3177 (loop_niter_by_eval): Avoid pointless looping if the next iteration
3178 would use the same bases as the current one.
3180 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3182 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
3183 instead of vec_select for V1TImode.
3184 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
3186 (VSX_LE_128): Add V1TI to this mode iterator.
3187 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
3188 (*vsx_le_perm_store_<mode>): Likewise.
3189 (pre-reload splitter for VSX stores): Likewise.
3190 (post-reload splitter for VSX stores): Likewise.
3191 (*vsx_xxpermdi2_le_<mode>): Likewise.
3192 (*vsx_lxvd2x2_le_<mode>): Likewise.
3193 (*vsx_stxvd2x2_le_<mode>): Likewise.
3195 2017-03-09 Michael Eager <eager@eagercon.com>
3197 Correct failures with --enable-checking=yes,rtl.
3199 * config/microblaze/microblaze.c (microblaze_expand_shift):
3200 Replace GET_CODE test with CONST_INT_P and INTVAL test with
3201 test for const0_rtx.
3202 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
3203 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
3205 2017-03-09 Richard Biener <rguenther@suse.de>
3207 PR tree-optimization/79977
3208 * graphite-scop-detection.c (scop_detection::merge_sese):
3209 Handle the case of extra exits to blocks dominating the entry.
3211 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
3213 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3216 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
3218 PR rtl-optimization/79949
3219 * lra-constraints.c (process_alt_operands): Check memory when
3220 trying to predict a cycle. Print about the overall increase.
3222 2017-03-09 Richard Biener <rguenther@suse.de>
3225 * gimple-expr.c (useless_type_conversion_p): Preserve
3226 TYPE_SATURATING for fixed-point types.
3228 2017-03-09 Richard Biener <rguenther@suse.de>
3231 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
3232 alignment of BLKmode params.
3234 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3237 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
3238 (VALL_NO_V2Q): Likewise.
3240 * config/aarch64/aarch64-simd.md
3241 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
3243 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
3244 VALL_NO_V2Q mode iterator.
3245 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
3247 2017-03-09 Martin Liska <mliska@suse.cz>
3249 PR tree-optimization/79631
3250 * tree-chkp-opt.c (chkp_is_constant_addr): Call
3251 tree_int_cst_sign_bit just for INTEGER constants.
3253 2017-03-09 Martin Liska <mliska@suse.cz>
3257 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
3260 2017-03-09 Marek Polacek <polacek@redhat.com>
3263 * tree.c (inchash::add_expr): Handle TREE_VEC.
3265 2017-03-09 Martin Liska <mliska@suse.cz>
3268 (chkp_narrow_size_and_offset): New function.
3269 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
3270 (void chkp_parse_bit_field_ref): New function.
3271 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
3272 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
3274 2017-03-09 Martin Liska <mliska@suse.cz>
3277 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
3278 (chkp_find_bounds_1): Remove gcc_unreachable.
3280 2017-03-09 Jakub Jelinek <jakub@redhat.com>
3283 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
3284 BUILT_IN_SYNC*, determine the access type from the size suffix and
3285 always build a MEM_REF with that type. Handle forgotten
3286 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
3289 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
3290 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
3291 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
3292 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
3293 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
3294 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
3295 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
3296 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
3297 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
3298 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
3299 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
3300 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
3301 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
3302 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
3303 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
3304 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
3305 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
3306 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
3307 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
3308 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
3309 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
3310 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
3311 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
3312 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
3313 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
3314 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
3315 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
3316 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
3317 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
3318 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
3319 definitions outside of __OPTIMIZE__ guarded section.
3322 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
3323 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
3324 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
3325 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
3328 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3330 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
3331 ("vfenez<mode>"): Add missing constraints.
3333 2017-03-08 Martin Sebor <msebor@redhat.com>
3336 * config/nds32/nds32.c (nds32_option_override):
3337 Fix misspelled diagnostic.
3339 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3342 * gimplify.c (gimplify_omp_for): Replace index var in outer
3343 taskloop statement with an artificial variable and add
3344 OMP_CLAUSE_PRIVATE clause for it.
3346 2017-03-08 Richard Biener <rguenther@suse.de>
3348 PR tree-optimization/79955
3349 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
3350 for accesses that are completely outside of the variable.
3352 2017-03-08 Andrew Haley <aph@redhat.com>
3354 PR tree-optimization/79943
3355 * tree-ssa-loop-split.c (compute_new_first_bound): When
3356 calculating the new upper bound, (END-BEG) should be added, not
3359 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3361 * config/avr/avr.md (setmemhi): Make sure match_dup
3362 operand number comes before match_scratch.
3364 2017-03-08 Richard Biener <rguenther@suse.de>
3366 PR tree-optimization/79920
3367 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
3368 with ncopies == 1 to ...
3369 (vect_transform_slp_perm_load): ... here. Properly compute
3370 all element loads by iterating VF times over the group. Do
3371 not handle ncopies (computed in a broken way) in
3372 vect_create_mask_and_perm.
3374 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3377 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
3378 is a uniform vector, use uniform_vector_p return value instead of
3379 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
3381 2017-03-07 Marek Polacek <polacek@redhat.com>
3384 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
3385 (alloca_call_type): Likewise.
3387 2017-03-07 Martin Liska <mliska@suse.cz>
3389 * gcov.c (process_args): Put comment to correct location.
3391 2017-03-07 Martin Liska <mliska@suse.cz>
3394 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
3395 Use array_at_struct_end_p instead of DECL_CHAIN (field).
3396 (chkp_narrow_bounds_for_field): Likewise.
3397 (chkp_parse_array_and_component_ref): Pass one more argument to
3400 2017-03-07 Richard Biener <rguenther@suse.de>
3402 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
3405 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
3407 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
3408 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
3410 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3413 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
3414 to end of description.
3415 (PARAM_MAX_STORES_TO_MERGE): Likewise.
3417 2017-03-07 Jakub Jelinek <jakub@redhat.com>
3419 PR rtl-optimization/79901
3420 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
3422 (*avx512f_<code><mode>3<mask_name>): ... this.
3423 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
3424 iterator instead of VI8_AVX2_AVX512BW.
3426 PR rtl-optimization/79901
3427 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
3428 min/max expander, expand it using expand_vec_cond_expr.
3431 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
3434 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3437 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
3438 to void * for PCH reasons.
3439 * dwarf2out.c (output_loc_operands, output_die): Cast
3440 v.val_vec.array to unsigned char *.
3442 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
3445 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
3448 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
3450 PR rtl-optimization/79571
3451 * lra-constraints.c (process_alt_operands): Calculate static
3452 reject and subtract it from overall when only addresses will be
3455 2017-03-06 Julia Koval <julia.koval@intel.com>
3458 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
3459 incoming stack boundary to 128 for 64-bit targets.
3461 2017-03-06 Richard Biener <rguenther@suse.de>
3463 PR tree-optimization/79894
3464 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
3465 to NULL after folding it.
3467 2017-03-06 Richard Biener <rguenther@suse.de>
3469 PR tree-optimization/79824
3470 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
3471 check disabling peeling for gaps.
3473 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
3475 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
3476 attributes): Document gettimeofday.
3478 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3480 * config/s390/s390.c (s390_option_override_internal): Set
3481 PARAM_MIN_VECT_LOOP_BOUND
3483 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3485 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
3486 * config/s390/s390.md: Likewise.
3488 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3491 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
3492 (<avx2_avx512>_perm<mode>): Rename to ...
3493 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
3495 (<avx512>_perm<mode>_mask): Rename to ...
3496 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
3498 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
3499 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
3500 instead of VI8F_256_512.
3501 (avx512f_perm<mode>): New define_expand.
3502 (avx512f_perm<mode>_mask): Likewise.
3503 (avx512f_perm<mode>_1<mask_name>): New define_insn.
3504 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
3506 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3508 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
3509 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
3511 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
3513 2017-03-06 Martin Liska <mliska@suse.cz>
3516 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
3517 when having a SSA NAME w/o VAR_DECL assigned to it.
3519 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3521 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
3522 msa_dpsub_<su>_d): Fix MODE for vec_select.
3524 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3526 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
3528 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
3530 2017-03-06 Richard Biener <rguenther@suse.de>
3532 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
3533 * plugin.c (register_plugin_info): Likewise.
3534 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
3536 2017-03-05 Jakub Jelinek <jakub@redhat.com>
3538 * config/i386/sse.md (sse_storehps, sse_storelps,
3539 avx_<castmode><avxsizesuffix>_<castmode>,
3540 avx512f_<castmode><avxsizesuffix>_<castmode>,
3541 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
3542 in condition that at least one operand is not a MEM.
3544 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3547 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
3548 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
3550 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
3551 gimple_call_nothrow_p flag based on whether original builtin can throw.
3552 If it can, emit following stmts on the fallthrough edge.
3553 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
3554 don't create new bb if inserting just debug stmts on the edge, try to
3555 insert them on the fallthru bb or just reset debug stmts.
3557 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
3560 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
3561 restore recog_data (including the operand rtxes inside it) around
3562 the call to get_insn_template.
3564 2017-03-03 Martin Sebor <msebor@redhat.com>
3566 PR tree-optimization/79699
3567 * context.c (context::~context): Free MPFR caches to avoid
3568 a memory leak on program exit.
3570 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3572 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3573 Use wide_int::ulow () instead of .elt (0).
3575 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3577 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
3578 (*pushxf): Limit oF constraint to 32bit targets and add oC
3579 constraint for 64bit targets.
3580 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
3581 (*pushdf): Change rmF constraint to rmC.
3583 2017-03-03 Martin Liska <mliska@suse.cz>
3585 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
3586 Remove unused variable.
3588 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3591 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
3592 is a memory operand, increase num_memory.
3593 (ix86_expand_args_builtin): Likewise.
3595 2017-03-03 Jan Hubicka <jh@suse.cz>
3598 * ipa-devirt.c (maybe_record_node): Properly handle
3599 __cxa_pure_virtual visibility.
3601 2017-03-03 Martin Liska <mliska@suse.cz>
3603 PR tree-optimization/79803
3604 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
3606 (pass_loop_prefetch::execute): Disabled optimization if an
3607 assumption about L1 cache size is not met.
3609 2017-03-03 Martin Liska <mliska@suse.cz>
3611 PR rtl-optimization/79574
3612 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
3613 (hash_scan_set): Likewise.
3614 (dump_hash_table): Likewise.
3615 (hoist_code): Likewise.
3617 2017-03-03 Richard Biener <rguenther@suse.de>
3619 * fixed-value.c (fixed_from_string): Restore use of elt (1)
3620 in place of uhigh ().
3621 (fixed_convert_from_real): Likewise.
3623 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3626 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
3628 2017-03-03 Richard Biener <rguenther@suse.de>
3631 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
3632 TYPE_OVERFLOW_UNDEFINED check.
3634 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3636 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
3638 (vector_ae_<mode>_p): Likewise.
3639 (vector_nez_<mode>_p): Likewise.
3640 (vector_ne_v2di_p): Likewise.
3641 (vector_ae_v2di_p): Likewise.
3642 (vector_ne_<mode>_p): Likewise.
3643 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
3645 (vsx_tsqrt<mode>2_fe): Likewise.
3647 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
3650 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
3652 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3654 PR rtl-optimization/79780
3655 * cprop.c (one_cprop_pass): When second and further conditional trap
3656 in a single basic block is turned into an unconditional trap, turn it
3657 into a deleted note to avoid RTL verification failures.
3659 2017-03-02 Richard Biener <rguenther@suse.de>
3661 * fold-const.c (const_binop): Use ulow () instead of elt (0).
3663 2017-03-02 Richard Biener <rguenther@suse.de>
3665 PR tree-optimization/79345
3667 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
3668 param and abort the walk, returning -1 if it is hit.
3669 (walk_aliased_vdefs): Take a limit param and pass it on.
3670 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
3671 defaulting to 0 and return a signed int.
3672 * tree-ssa-uninit.c (struct check_defs_data): New struct.
3673 (check_defs): New helper.
3674 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
3675 about uninitialized memory.
3676 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
3677 bogus uninitialized warning.
3678 (fixed_convert_from_real): Likewise.
3680 2017-03-02 Bin Cheng <bin.cheng@arm.com>
3682 PR tree-optimization/66768
3683 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
3684 iv_use if base object can't be determined.
3686 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3688 PR tree-optimization/79345
3689 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
3690 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
3691 (get_pattern_stats): Initialize it.
3692 * genemit.c (gen_expand): Verify match_scratch numbers come after
3693 match_operand/match_dup numbers.
3694 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
3695 match_scratch numbers.
3696 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
3698 * config/s390/s390.md (trunctdsd2): Likewise.
3700 2017-03-02 Richard Biener <rguenther@suse.de>
3702 * wide-int.h (wide_int_storage::operator=): Implement in terms
3705 2017-03-02 Richard Biener <rguenther@suse.de>
3707 PR tree-optimization/79777
3708 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
3709 the to insert expression to sth existing.
3711 2017-03-01 Martin Sebor <msebor@redhat.com>
3714 * gimple-ssa-sprintf.c
3715 (directive::known_width_and_precision): New function.
3716 (format_integer): Use it.
3717 (get_mpfr_format_length): Consider the full range of precision
3718 when computing %g output with the # flag. Set the likely byte
3719 count to 3 rather than 1 when precision is indeterminate.
3720 (format_floating): Correct the lower bound of precision.
3722 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3724 * doc/invoke.texi: Document default code model for 64-bit Linux.
3726 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3729 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
3730 udiv rather than div since input pattern is unsigned.
3732 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
3734 * config/i386/i386.c (print_reg): Warn for values of
3735 unsupported size in integer register.
3737 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3740 * config/rs6000/predicates.md (current_file_function_operand): Do
3741 not allow self calls to be local if the function is replaceable.
3743 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
3746 * config/rs6000/altivec.h (vec_ctz and others): Change the
3747 preprocessor macro that controls conditional compilation from
3748 _ARCH_PWR9 to __POWER9_VECTOR__.
3749 (vec_all_ne): Change parameterization of __altivec_scalar_pred
3750 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
3751 control (instead of _ARCH_PWR9 control) so that template
3752 definition uses power9-specific function.
3753 (vec_any_eq): Likewise.
3754 (vec_all_ne): Change macro definition to use a power9-specific
3755 expansion under #ifdef __POWER9_VECTOR__ control (instead of
3756 _ARCH_PWR9 control).
3757 (vec_any_eq) Likewise.
3758 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
3759 expansion for CMPNEF to remove support for xvcmpnesp instruction.
3760 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
3761 support for xvcmpnedp instruction.
3762 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
3763 macro expansion so that Power9 implementation of vec_all_ne does
3764 not use the AltiVec predicate framework.
3765 (VCMPNEH_P): Likewise.
3766 (VCMPNEW_P): Likewise.
3767 (VCMPNED_P): Likewise.
3768 (VCMPNEFP_P): Likewise.
3769 (VCMPNEDP_P): Likewise.
3770 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
3771 implementation of vec_any_eq to not use AltiVec predicate
3773 (VCMPAEH_P): Likewise.
3774 (VCMPAEW_P): Likewise.
3775 (VCMPAED_P): Likewise.
3776 (VCMPAEFP_P): Likewise.
3777 (VCMPAEDP_P): Likewise.
3778 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
3779 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
3780 not use the AltiVec predicate framework.
3781 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
3782 of vec_any_eq to not use AltiVec predicate framework.
3783 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
3784 support for predefined __POWER9_VECTOR__ macro to indicate that
3785 Power9 instruction selection is enabled.
3786 (altivec_overloaded_builtins): Remove extraneous
3787 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
3788 function argument types RS6000_BTI_bool_V16QI and
3789 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
3790 entry for overloaded function argument types RS6000_BTI_bool_V4SI
3791 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
3792 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
3793 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
3794 Power9 for implementations of vec_cmpne. Change the signature for
3795 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
3796 (representing vec_all_ne) to remove the previously described first
3797 argument of type RS6000_BTI_INTSI, as this was an artifact of
3798 reliance on the AltiVec predicate framework, which is no longer
3799 used in the implementation of these functions. Add
3800 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
3801 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
3802 since, unlike the AltiVec predicate framework implementation, we
3803 do not share function descriptors between vec_alle and vec_anyeq.
3804 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
3805 set of modes that receive special treatment even when
3806 TARGET_P9_VECTOR is true. The special treatment emits code that
3807 does not depend on Power9 instructions.
3808 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
3809 define_expand to not rely on AltiVec predicate framework.
3810 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3812 (vector_ne_v2di_p): Change this define_expand to not rely on
3813 AltiVec predicate framework.
3814 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
3816 (vector_ne_<mode>_p): Change this define_expand to not rely on
3817 AltiVec predicate framework.
3818 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3820 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
3821 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
3822 define_insn pattern.
3823 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
3824 define_insn pattern because the xvcmpne<VSs>. instruction is not
3826 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
3827 instruction is not supported.
3829 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3831 * config/nvptx/nvptx.c: Include intl.h.
3833 2017-03-01 Martin Jambor <mjambor@suse.cz>
3836 * ipa-prop.h (ipa_bits): Removed field known.
3837 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
3838 to pointers. Adjusted their comments to warn about their sharing.
3839 (ipcp_transformation_summary): Change bits to a vector of pointers.
3840 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
3841 (ipa_get_ipa_bits_for_value): Declare.
3842 * tree-vrp.h (value_range): Mark as GTY((for_user)).
3843 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
3844 (ipa_bits_hash_table): Likewise.
3845 (ipa_vr_ggc_hash_traits): Likewise.
3846 (ipa_vr_hash_table): Likewise.
3847 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
3848 being pointers and vr_known being removed.
3849 (ipa_set_jf_unknown): Likewise.
3850 (ipa_get_ipa_bits_for_value): New function.
3851 (ipa_set_jfunc_bits): Likewise.
3852 (ipa_get_value_range): New overloaded functions.
3853 (ipa_set_jfunc_vr): Likewise.
3854 (ipa_compute_jump_functions_for_edge): Use the above functions to
3855 construct bits and vr parts of jump functions.
3856 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
3857 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3859 (ipcp_grow_transformations_if_necessary): Also allocate
3860 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3862 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
3863 them. Fix too long lines.
3864 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
3865 vr_known being removed.
3866 (ipa_read_jump_function): Use new setter functions to construct bits
3867 and vr parts of jump functions or set them to NULL.
3868 (write_ipcp_transformation_info): Adjust for bits being pointers.
3869 (read_ipcp_transformation_info): Likewise.
3870 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
3872 Include gt-ipa-prop.h.
3873 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
3875 (ipcp_store_bits_results): Likewise.
3876 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
3877 Do not write to existing jump functions but use a temporary instead.
3879 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3882 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
3883 attempt to use its first operand as BIT_FIELD_REF base.
3885 2017-03-01 Richard Biener <rguenther@suse.de>
3888 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
3889 interpolating formula in wrapping arithmetic.
3890 (chrec_apply): Convert chrec_evaluate return value to wanted type.
3892 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3894 PR tree-optimization/79734
3895 * tree-vect-generic.c (expand_vector_condition): Optimize
3896 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
3897 Handle VEC_COND_EXPR where comparison has different inner width from
3900 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
3902 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
3903 markup, and similar issues. Remove @opindex entries for things
3904 that aren't options. Add missing -mmpy-option entries.
3906 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3908 PR tree-optimization/79737
3909 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
3910 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
3911 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
3912 instead of byte_size. Formatting fix.
3913 (shift_bytes_in_array_right): Formatting fix.
3915 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
3918 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
3919 condition on optimize for the leaf function test.
3921 2017-02-28 Martin Liska <mliska@suse.cz>
3924 * read-rtl-function.c (function_reader::handle_unknown_directive):
3925 Bail out when one uses -flto.
3927 2017-02-28 Martin Liska <mliska@suse.cz>
3929 * common.opt: Replace space with tabular for options of <number>
3931 * config/i386/i386.opt: Show <number> value for
3932 -mlarge-data-threshold.
3933 * opts.c (print_filtered_help): Do not display number in hexadecimal
3936 2017-02-28 Martin Liska <mliska@suse.cz>
3938 * common.opt: Fix --help=option -Q for options which are of
3941 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
3943 * config/i386/i386.c (print_reg): Error out for values
3944 of 8-bit size in invalid integer register.
3946 2017-02-28 Martin Sebor <msebor@redhat.com>
3948 PR tree-optimization/79691
3949 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
3951 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3954 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
3955 gcc_unreachable with output_operand_lossage.
3957 2017-02-28 Richard Biener <rguenther@suse.de>
3959 PR tree-optimization/79740
3960 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
3962 (visit_nary_op): Insert the nary into the hashtable if we
3963 pattern-matched sth.
3964 * tree-ssa-pre.c (eliminate_insert): Robustify.
3966 2017-02-28 Richard Biener <rguenther@suse.de>
3969 * fold-const.c (decode_field_reference): Reject out-of-bound
3972 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3974 * config/i386/i386.c: Include intl.h.
3975 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
3976 instead of just cond ? "..." : "...".
3977 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
3978 * coverage.c (read_counts_file): Likewise.
3979 * omp-offload.c: Include intl.h.
3980 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
3981 of just cond ? "..." : "...".
3982 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
3983 of just cond ? "..." : "...".
3985 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
3988 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
3990 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
3991 'tune for' CPU name.
3992 * config/arm/arm-cpu-data.h: Regenerated.
3994 2017-02-28 Richard Biener <rguenther@suse.de>
3996 PR tree-optimization/79732
3997 * tree-inline.c (expand_call_inline): Do not shadow var.
3999 2017-02-28 Richard Biener <rguenther@suse.de>
4001 PR tree-optimization/79723
4002 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
4003 address-space properly.
4005 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
4007 * doc/optinfo.texi (Optimization groups): Fix option used for
4009 * doc/invoke.texi (-fopt-info): Document "omp".
4010 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
4011 (OPTGROUP_ALL): Add OPTGROUP_OMP.
4012 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
4013 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
4014 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
4016 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
4018 * dumpfile.c (optgroup_options): Instead of "openmp", associate
4019 OPTGROUP_OMP with "omp".
4021 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
4024 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
4025 for arithmetic shift of unsigned V2DI.
4027 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
4029 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
4030 arc/linux.h headers.
4031 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
4032 (LINK_SPEC): Likewise.
4033 (ARC_TLS_EXTRA_START_SPEC): Likewise.
4034 (EXTRA_SPECS): Likewise.
4035 (STARTFILE_SPEC): Likewise.
4036 (ENDFILE_SPEC): Likewise.
4037 (LIB_SPEC): Likewise.
4038 (TARGET_SDATA_DEFAULT): Likewise.
4039 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
4040 (MULTILIB_DEFAULTS): Likewise.
4041 (DWARF2_UNWIND_INFO): Likewise.
4042 * config/arc/big.h: New file.
4043 * config/arc/elf.h: Likewise.
4044 * config/arc/linux.h: Likewise.
4045 * config/arc/t-uClibc: Remove.
4047 2017-02-27 Bin Cheng <bin.cheng@arm.com>
4049 PR tree-optimization/77536
4050 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
4051 (tree_transform_and_unroll_loop): Use above function to compute the
4052 estimated niter of unrolled loop and use it when scaling profile.
4053 Also use count info rather than frequency if it's non-zero.
4054 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
4055 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
4056 (vect_transform_loop): Call above function.
4058 2017-02-27 Richard Biener <rguenther@suse.de>
4060 PR tree-optimization/45397
4061 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
4062 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
4063 (visit_nary_op): Add pattern matching for CSEing sign-changed
4064 or truncated operations with wider ones.
4066 2017-02-27 Richard Biener <rguenther@suse.de>
4068 PR tree-optimization/79690
4069 * tree-vect-stmts.c (vectorizable_store): Use vector type
4070 built from the DR with address-space.
4072 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
4074 * doc/invoke.texi (Optimize Options): Refine the description
4075 of asan-use-after-return.
4077 2017-02-25 Alan Modra <amodra@gmail.com>
4079 PR rtl-optimization/79584
4080 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
4081 base, not ad->base_term, the reg within base. Remove assertion
4082 that ad->base == ad->base_term. Replace gen_int_mode using
4083 bogus mode with const0_rtx.
4085 2017-02-25 Jakub Jelinek <jakub@redhat.com>
4088 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
4089 FMA_EXPR like tcc_binary or tcc_unary.
4091 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
4094 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
4096 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
4097 (output_loc_operands): Handle DW_OP_call_ref and
4098 DW_OP_GNU_variable_value.
4099 (struct variable_value_struct): New type.
4100 (struct variable_value_hasher): Likewise.
4101 (variable_value_hash): New variable.
4102 (string_types): Remove.
4103 (copy_loc_descr): New function.
4104 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
4105 (prepend_loc_descr_to_each): New function.
4106 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
4107 instead of add_loc_descr_to_each if the first argument is single
4108 location list and the second has multiple.
4109 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
4110 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
4111 when looking for variable value which doesn't have other location info.
4112 (loc_list_from_tree): Formatting fix.
4113 (gen_array_type_die): Simplify DW_AT_string_length handling.
4114 (adjust_string_types): Remove.
4115 (gen_subprogram_die): Don't call adjust_string_types nor test/set
4116 string_types. Call resolve_variable_values.
4117 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
4118 (resolve_addr_in_expr): Likewise. Add A argument.
4119 (copy_deref_exprloc): Remove deref argument. Adjust for the
4120 original expression being DW_OP_GNU_variable_value with optionally
4121 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
4122 optionally after it.
4123 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
4124 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
4125 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
4126 (variable_value_hasher::hash, variable_value_hasher::equal): New
4128 (resolve_variable_value_in_expr, resolve_variable_value,
4129 resolve_variable_values, note_variable_value_in_expr,
4130 note_variable_value): New functions.
4131 (dwarf2out_early_finish): Call note_variable_value on all toplevel
4134 2017-02-24 Jakub Jelinek <jakub@redhat.com>
4137 * opts.h (handle_generated_option): Add GENERATED_P argument.
4138 * opts-common.c (handle_option): Adjust function comment.
4139 (handle_generated_option): Add GENERATED_P argument, pass it to
4141 (control_warning_option): Pass false to handle_generated_option
4143 * opts.c (maybe_default_option): Pass true to handle_generated_option
4145 * optc-gen.awk: Likewise.
4147 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4149 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
4150 a REG, look at the REG it is a SUBREG of.
4151 (splitter for cmpeqsi_t): Ditto.
4153 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4155 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
4156 the special USEs with the pattern of the insn, not the insn itself.
4158 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
4161 * doc/invoke.texi: Document -mload-store-pairs.
4163 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4164 Sandra Loosemore <sandra@codesourcery.com>
4166 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
4167 argument isn't a CONST_INT.
4168 (nios2_alternate_compare_const): Assert op is a CONST_INT.
4169 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
4170 (nios2_validate_compare): Bypass alternate compare logic if *op2
4172 (ldstwm_operation_p): Return false if first_base is not a REG or
4173 if first_offset is not a CONST_INT.
4175 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4177 * config/cris/cris.md: Use correct operand in a define_peephole2.
4179 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4181 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
4183 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4185 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
4186 this_insn if it is an INSN or JUMP_INSN.
4187 (force_offsettable): Look at base, not at addr.
4188 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
4189 on things that aren't necessarily CONST_INTs.
4191 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
4193 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
4194 -mfpmath=sse is the default also for x86-32 targets with SSE2
4195 instruction set when @option{-ffast-math} is enabled
4197 2017-02-24 Jeff Law <law@redhat.com>
4199 PR rtl-optimizatoin/79286
4200 * ira.c (update_equiv_regs): Drop may_trap_p exception to
4203 2017-02-24 Richard Biener <rguenther@suse.de>
4205 PR tree-optimization/79389
4206 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
4209 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
4211 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
4212 function comment to reflect reality.
4213 (loop_exits_before_overflow): Fix typo in function description.
4215 2017-02-24 Richard Biener <rguenther@suse.de>
4217 PR tree-optimization/79389
4218 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
4219 properly that a threading opportunity exists. Detect conditional
4220 copy/constant propagation opportunities.
4222 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
4224 * config/visium/visium.md (type): Add trap.
4225 (b): New mode attribute.
4226 (*btst): Rename into...
4227 (*btst<mode>): ...this and adjust.
4228 (*cbranchsi4_btst_insn): Rename into...
4229 (*cbranch<mode>4_btst_insn): ...this and adjust.
4230 (trap): New define_insn.
4232 2017-02-23 Jakub Jelinek <jakub@redhat.com>
4234 PR tree-optimization/79389
4235 * ifcvt.c (struct noce_if_info): Add rev_cond field.
4236 (noce_reversed_cond_code): New function.
4237 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
4238 reversed_comparison_code. Formatting fix.
4239 (noce_try_store_flag): Test rev_cond != NULL in addition to
4240 reversed_comparison_code.
4241 (noce_try_store_flag_constants): Likewise.
4242 (noce_try_store_flag_mask): Likewise.
4243 (noce_try_addcc): Use rev_cond if non-NULL instead of
4244 reversed_comparison_code.
4245 (noce_try_cmove_arith): Likewise. Formatting fixes.
4246 (noce_try_minmax, noce_try_abs): Clear rev_cond.
4247 (noce_find_if_block): Initialize rev_cond.
4248 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
4249 instead of false as last argument never attempt to reverse it
4252 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4254 PR tree-optimization/79663
4255 * tree-predcom.c (combine_chains): Process refs in reverse order
4256 only for ZERO length chains, and add explaining comment.
4258 2017-02-23 Jeff Law <law@redhat.com>
4260 PR tree-optimization/79578
4261 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
4262 in call to operand_equal_p.
4264 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
4267 * config/i386/cpuid.h: Fix another undefined behavior.
4269 2017-02-23 Richard Biener <rguenther@suse.de>
4271 PR tree-optimization/79683
4272 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
4273 vector types for data-refs.
4275 2017-02-23 Martin Liska <mliska@suse.cz>
4277 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
4279 2017-02-23 Jakub Jelinek <jakub@redhat.com>
4282 * internal-fn.c (get_range_pos_neg): Moved to ...
4283 * tree.c (get_range_pos_neg): ... here. No longer static.
4284 * tree.h (get_range_pos_neg): New prototype.
4285 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
4286 are known to be in between 0 and signed maximum inclusive, try to
4287 expand both unsigned and signed divmod and use the cheaper one from
4290 2017-02-22 Jeff Law <law@redhat.com>
4292 PR tree-optimization/79578
4293 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
4294 to compare base operands.
4296 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
4299 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
4300 gpc_reg_operand instead of fpr_reg_operand.
4302 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
4304 * config/mips/mips.c (mips_return_in_memory): Force FP
4305 vector types to be returned in memory for o32 ABI.
4307 2017-02-22 Jakub Jelinek <jakub@redhat.com>
4309 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
4310 instead of DW_TAG_member for static data member declarations and don't
4311 set no_linkage_name for static inline data members.
4312 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
4315 2017-02-22 Martin Liska <mliska@suse.cz>
4317 * doc/invoke.texi: Replace inequality signs with square brackets
4320 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4322 PR tree-optimization/68644
4323 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
4325 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
4328 * lra-constraints.c (simplify_operand_subreg): Handle
4329 WORD_REGISTER_OPERATIONS targets.
4331 2017-02-22 Jakub Jelinek <jakub@redhat.com>
4334 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
4335 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
4336 elimination by swapping fld*.
4338 2017-02-22 Richard Biener <rguenther@suse.de>
4340 PR tree-optimization/79673
4341 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
4342 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
4343 irrelevant address-space qualifiers and avoiding a
4344 ADDR_SPACE_CONVERT_EXPR from fold_convert.
4346 2017-02-22 Richard Biener <rguenther@suse.de>
4348 PR tree-optimization/79666
4349 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
4350 to not symbolically negate if that may introduce undefined
4353 2017-02-22 Martin Liska <mliska@suse.cz>
4356 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
4357 * data-streamer-out.c (streamer_write_gcov_count_stream):
4359 * value-prof.c (stream_out_histogram_value): Make assert more
4360 precise based on type of counter.
4362 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
4365 * config/i386/i386.md (standard_x87sse_constant_load splitter):
4366 Use nonimmediate_operand instead of memory_operand for operand 1.
4367 (float-extend standard_x87sse_constant_load splitter): Ditto.
4369 2017-02-21 Jeff Law <law@redhat.com>
4371 PR tree-optimization/79621
4372 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
4373 blocks with edges to themselves.
4375 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4378 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
4379 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
4380 Use gimple_call_builtin_p.
4383 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
4384 on temporarily removed DEBUG_INSNs.
4386 PR tree-optimization/79649
4387 * tree-loop-distribution.c (classify_partition): Give up on
4388 non-generic address space loads/stores.
4390 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
4392 * doc/loop.texi (Loop manipulation): Remove nonexistent
4393 tree_ssa_loop_version from the documentation.
4394 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
4396 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4399 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
4400 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
4401 * config/rs6000/rs6000.c: Include except.h.
4402 (rs6000_expand_split_stack_prologue): Call
4403 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
4405 2017-02-21 Martin Jambor <mjambor@suse.cz>
4408 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
4411 2017-02-21 Martin Jambor <mjambor@suse.cz>
4413 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
4414 for backward compatibility only.
4415 * doc/invoke.texi (Option Summary): Remove all references to
4418 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
4422 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4424 * lra-constraints.c (curr_insn_transform): Handle
4425 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4427 2017-02-21 Martin Liska <mliska@suse.cz>
4429 * config/i386/i386.opt: Replace -masm-dialect with -masm.
4431 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
4433 PR translation/79638
4434 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
4436 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
4439 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
4440 (arm_function_ok_for_sibcall): Return false for an indirect call by
4441 descriptor if all the argument registers are used.
4442 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
4443 alignment of the function.
4445 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4447 PR tree-optimization/61441
4448 * simplify-rtx.c (simplify_const_unary_operation): For
4449 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
4450 the sNaN unmodified.
4452 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4454 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
4455 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
4456 instead of SYSTEM_HEADER_DIR.
4458 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
4459 Martin Liška <mliska@suse.cz>
4461 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
4462 Fix typos and grammar, use active voice, and clarify.
4464 2017-02-20 Marek Polacek <polacek@redhat.com>
4467 * gimplify.c (gimplify_expr): Handle unused *&&L;.
4470 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
4472 2017-02-20 Jakub Jelinek <jakub@redhat.com>
4475 * config/i386/i386.c (ix86_expand_builtin): Handle
4476 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
4477 ix86_builtins_isa[fcode].isa as a requirement of those
4478 flags and any other flag in the bitmask.
4479 (ix86_init_mmx_sse_builtins): Use 0 instead of
4480 ~OPTION_MASK_ISA_64BIT as mask.
4481 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
4482 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
4483 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
4484 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
4486 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4489 * lra-constraints.c (split_reg): Check requested split mode
4490 is supported by the register.
4492 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4494 * lra-constraints.c (simplify_operand_subreg): Remove early
4497 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4500 * lra-constraints.c (curr_insn_transform): Tighten condition
4501 for converting SUBREG reloads from OP_OUT to OP_INOUT.
4503 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4506 * lra-constraints.c (curr_insn_transform): Handle
4507 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4509 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
4512 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
4514 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
4516 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
4519 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
4522 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4524 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
4527 2017-02-18 Jakub Jelinek <jakub@redhat.com>
4529 * final.c (last_columnnum, override_columnnum): New variables.
4530 (final_start_function): Set last_columnnum, pass it to begin_prologue
4531 hook and pass 0 to dwarf2out_begin_prologue.
4532 (final_scan_insn): Update override_columnnum. Pass last_columnnum
4533 to source_line debug hook.
4534 (notice_source_line): Compute last_columnnum and for debug_column_info
4535 return true on column changes.
4536 * debug.h (struct gcc_debug_hooks): Add column argument to
4537 source_line and begin_prologue hooks.
4538 (debug_nothing_int_charstar_int_bool): Remove prototype.
4539 (debug_nothing_int_int_charstar,
4540 debug_nothing_int_int_charstar_int_bool): New prototypes.
4541 (dwarf2out_begin_prologue): Add column argument.
4542 * debug.c (do_nothing_debug_hooks): Adjust source_line and
4543 begin_prologue hooks.
4544 (debug_nothing_int_charstar_int_bool): Remove.
4545 (debug_nothing_int_int_charstar,
4546 debug_nothing_int_int_charstar_int_bool): New functions.
4547 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
4548 through to dwarf2out_source_line.
4549 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
4550 (dwarf2out_source_line): Add column argument, emit it if requested.
4551 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
4553 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4554 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4555 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
4556 through to dwarf2out_begin_prologue.
4557 (vmsdbgout_source_line): Add column argument, pass it through to
4558 dwarf2out_source_line.
4559 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
4560 dbxout_source_line caller.
4561 (dbxout_source_line): Add column argument.
4563 * common.opt (gno-column-info, gcolumn-info): New options.
4564 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
4565 (check_die): Also test for multiple DW_AT_decl_column attributes.
4566 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
4567 DW_AT_decl_column if requested.
4568 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
4570 (gen_variable_die): Likewise.
4571 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
4572 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
4575 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
4576 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
4577 (ix86_handle_option): Handle OPT_m3dnowa.
4578 * doc/invoke.texi (-m3dnowa): Document.
4579 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
4580 -m3dnowa instead of -m3dnow -march=athlon.
4583 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
4584 instead of gcc_assert for K, r and R code checks. Formatting fixes.
4586 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4589 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
4590 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
4591 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
4592 generator for vsx_xxpermdi_<mode>_be.
4593 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
4594 force big-endian semantics.
4595 (vsx_xxpermdi_<mode>_be): New define_expand with same
4596 implementation as previous version of vsx_xxpermdi_<mode>.
4598 2017-02-17 Jakub Jelinek <jakub@redhat.com>
4600 PR tree-optimization/79327
4601 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
4602 variable, its initialization and use.
4604 2017-02-17 Julia Koval <julia.koval@intel.com>
4606 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
4607 (OPTION_MASK_ISA_PKU_UNSET): New.
4608 (ix86_handle_option): Handle -mrdpid.
4609 * config/i386/cpuid.h (bit_RDPID): New.
4610 * config/i386/driver-i386.c (host_detect_local_cpu):
4611 Detect RDPID feature.
4612 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
4613 * config/i386/i386-c.c (ix86_target_macros_internal):
4615 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
4616 (ix86_valid_target_attribute_inner_p): Add "rdpid".
4617 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
4618 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
4619 * config/i386/i386.md (define_insn "rdpid"): New.
4620 * config/i386/i386.opt Add -mrdpid.
4621 * config/i386/immintrin.h (_rdpid_u32): New.
4623 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
4625 PR rtl-optimization/79541
4626 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
4627 instead of transforming it into USE.
4629 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
4631 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
4632 If HONOR_SNANS (SFmode) force the input to a register.
4633 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
4634 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
4635 an frsp or similar insn.
4637 2017-02-17 Martin Liska <mliska@suse.cz>
4639 PR rtl-optimization/79577
4640 * params.def (selsched-max-sched-times): Increase minimum to 1.
4642 2017-02-17 Martin Liska <mliska@suse.cz>
4644 PR rtl-optimization/79574
4645 * gcse.c (want_to_gcse_p): Prevent integer overflow.
4647 2017-02-17 Martin Liska <mliska@suse.cz>
4649 PR tree-optimization/79529
4650 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
4651 ssa_defined_default_def_p to handle cases which are implicitly
4653 * tree-ssa.c (ssa_defined_default_def_p): New function.
4654 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
4655 which are implicitly defined.
4656 * tree-ssa.h (ssa_defined_default_def_p): Declare.
4658 2017-02-17 Richard Biener <rguenther@suse.de>
4661 * params.def (max-ssa-name-query-depth): Limit to 10.
4663 2017-02-17 Richard Biener <rguenther@suse.de>
4665 PR tree-optimization/79552
4666 * tree-ssa-structalias.c (visit_loadstore): Properly verify
4669 2017-02-17 Richard Biener <rguenther@suse.de>
4672 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
4674 2017-02-17 Marek Polacek <polacek@redhat.com>
4677 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
4678 (fold_negate_expr): New wrapper.
4680 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
4682 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
4683 Correct terminology and de-emphasize pre-standard behavior.
4685 2017-02-16 Alan Modra <amodra@gmail.com>
4687 PR rtl-optimization/79286
4688 * ira.c (def_dominates_uses): New function.
4689 (update_equiv_regs): Don't create an equivalence for insns that
4690 may trap where the register def does not dominate the use.
4692 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
4694 PR rtl-optimization/78127
4695 * lra.c (lra): Call lra_eliminate before finish the loop after
4698 2017-02-16 Richard Biener <rguenther@suse.de>
4700 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
4702 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
4703 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
4704 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
4705 (isl_val_int_from_wi): New function.
4706 (extract_affine_gmp): Rename to ...
4707 (extract_affine_wi): ... this, take a widest_int.
4708 (extract_affine_int): Just wrap extract_affine_wi.
4709 (add_param_constraints): Use isl_val_int_from_wi.
4710 (add_loop_constraints): Likewise, and extract_affine_wi.
4712 2017-02-15 Jeff Law <law@redhat.com>
4715 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
4716 ira_init_register_move_cost_if_necessary.
4718 2017-02-15 Martin Sebor <msebor@redhat.com>
4721 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
4722 removed in a prior commit.
4724 2017-02-15 Bin Cheng <bin.cheng@arm.com>
4726 PR tree-optimization/79347
4727 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
4728 counters during peeling.
4730 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
4732 * Makefile.in (site.exp): Remove "set ISLVER".
4734 2017-02-15 Jakub Jelinek <jakub@redhat.com>
4737 * real.c (real_from_integer): Call real_convert even for decimal.
4739 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
4742 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
4744 2017-02-14 Andrew Pinski <apinski@cavium.com>
4746 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
4747 cores and change the partno/implementer to be correct.
4748 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
4749 the 'B" as the implementer.
4750 * config/aarch64/aarch64-tune.md: Regenerate.
4752 2017-02-14 Carl Love <cel@us.ibm.com>
4754 * config/rs6000/rs6000.c: Add case statement entry to make the
4755 xvcvuxdsp built-in argument unsigned.
4756 * config/rs6000/vsx.md: Fix the source and return operand types so they
4757 match the instruction definitions from the ISA document. Fix typo
4758 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
4761 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
4764 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
4765 member early_clobber_alts.
4766 * lra-lives.c (reg_early_clobber_p): New.
4767 (process_bb_lives): Use it.
4768 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
4769 (debug_operand_data): Initialize early_clobber_alts.
4770 (setup_operand_alternative): Set up early_clobber_alts.
4771 (collect_non_operand_hard_regs): Ditto. Pass early clobber
4772 alternatives to new_insn_reg.
4773 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
4775 (lra_update_insn_regno_info): Pass the new arg.
4777 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4780 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
4781 (new_oacc_loop_raw): Don't clear already cleared fields.
4784 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
4785 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
4786 _mm512_prefetch_i64gather_ps): New inline functions and macros.
4788 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
4791 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
4793 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
4796 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
4797 the extra instruction to the right place to store 128-bit constant
4800 2017-02-14 Martin Sebor <msebor@redhat.com>
4803 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
4804 warning for strings of unknown length.
4806 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
4808 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
4810 2017-02-14 Jeff Law <law@redhat.com>
4813 * ira-costs.c (scan_one_insn): Initialize register move costs
4814 for pseudos seen in USE/CLOBBER insns.
4816 PR tree-optimization/79095
4817 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
4818 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
4819 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
4820 if the operands are known to be not equal, then the resulting range
4822 (intersect_ranges): If the new range is ~[0,0] and the old range is
4823 wide, then prefer ~[0,0].
4824 * tree-vrp.c (overflow_comparison_p_1): New function.
4825 (overflow_comparison_p): New function.
4826 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
4827 if NAME is used in an overflow test.
4828 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
4829 overflow check that can be expressed as an equality test, then adjust
4830 ops to be that equality test.
4832 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4834 * config/s390/s390-builtin-types.def: Remove flags argument.
4835 * config/s390/s390.c (s390_init_builtins): Likewise.
4837 2017-02-14 Martin Liska <mliska@suse.cz>
4839 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
4840 vector. Fix trailing white spaces.
4842 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
4844 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
4847 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4849 PR rtl-optimization/68664
4850 * config/arm/arm.c (arm_sched_can_speculate_insn):
4851 New function. Declare prototype.
4852 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4854 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4856 PR rtl-optimization/68664
4857 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
4859 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4861 2017-02-14 Amit Pawar <amit.pawar@amd.com>
4863 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
4864 max skip bytes for function, loop and jump.
4866 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4868 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
4869 ABS_EXPR for gimple dump.
4871 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4874 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
4876 PR tree-optimization/79408
4877 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
4878 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
4879 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
4880 also if rhs1 is INTEGER_CST.
4882 2017-02-14 Richard Biener <rguenther@suse.de>
4885 * tree-into-ssa.c (insert_phi_nodes): When the function can
4886 have abnormal edges rewrite SSA names with broken use-def
4887 dominance out of SSA and register them for PHI insertion.
4889 2017-02-13 Martin Sebor <msebor@redhat.com>
4892 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
4893 clearing info.nowrite flag when snprintf size argument is a range.
4895 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4897 * cprop.c (cprop_jump): Add missing space in string literal.
4898 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
4899 (get_constraint_for_component_ref): Likewise.
4900 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
4901 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
4902 * lra-constraints.c (process_alt_operands): Likewise.
4903 * ipa-inline.c (inline_small_functions): Likewise.
4904 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
4905 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4906 * trans-mem.c (diagnose_tm_1_op): Likewise.
4907 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
4908 (grid_parallel_clauses_gridifiable): Likewise.
4910 * config/nvptx/mkoffload.c (process): Add space in between
4913 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
4914 "MOD4_SSE_REGS" and "ALL_REGS".
4916 * spellcheck.c (test_data): Add , in between "foo" and "food".
4918 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4921 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
4922 boundary crossing check and subsequent code generation agree.
4924 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4926 * config/aarch64/aarch64.c (has_memory_op): Delete.
4927 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
4930 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4932 PR rtl-optimization/79388
4933 PR rtl-optimization/79450
4934 * combine.c (distribute_notes): When removing TEM_INSN for which
4935 corresponding dest has last value recorded, invalidate that last
4938 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4940 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
4941 of explicit '@'. Add missing assembly comment marker on branch costs
4944 2017-02-13 Nathan Sidwell <nathan@acm.org>
4946 * gengtype-lex.l (<in_struct>): Add '/'.
4948 2017-02-13 Martin Liska <mliska@suse.cz>
4951 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
4953 2017-02-13 Richard Biener <rguenther@suse.de>
4955 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
4957 * configure: Re-generate.
4958 * config.in: Likewise.
4959 * graphite-dependences.c: Simplify as if
4960 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
4961 * graphite-isl-ast-to-gimple.c: Likewise.
4962 * graphite-optimize-isl.c: Likewise.
4963 * graphite-poly.c: Likewise.
4964 * graphite-sese-to-poly.c: Likewise.
4965 * graphite.h: Likewise.
4966 * toplev.c: Include isl/version.h and use isl_version () for
4967 printing the ISL version.
4968 * doc/install.texi: Update ISL requirement.
4970 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4972 * doc/standards.texi (Standards): Update reference to
4975 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4977 * doc/extend.texi (Named Address Spaces): sourceware.org now
4979 * doc/install.texi (Binaries): Ditto.
4982 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4984 * doc/cpp.texi: Replace "stringify"/"stringification" with C
4985 standard terminology "stringize"/"stringizing" throughout.
4986 * doc/cppinternals.texi: Likewise.
4988 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4990 * doc/extend.texi: Fix some spelling mistakes and typos.
4991 * doc/invoke.texi: Likewise.
4993 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4996 * params.def (inline-min-speedup) Change from 10 to 8.
4998 2017-02-11 Jakub Jelinek <jakub@redhat.com>
5000 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
5003 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
5006 * ipa-inline-analysis.c (get_minimal_bb): New function.
5007 (record_modified): Use it.
5008 (remap_edge_change_prob): Handle also ancestor functions.
5010 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
5012 * doc/contrib.texi (Contributors): Remove broken link into
5013 the Mauve CVS repository.
5015 2017-02-11 Jakub Jelinek <jakub@redhat.com>
5018 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
5019 result computation whenever lhs doesn't have vector mode, not
5020 just when it has BLKmode.
5022 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
5024 * doc/makefile.texi (profiledbootstrap): Refer to the
5025 installation instructions only in textual form.
5027 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5030 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
5032 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
5034 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
5035 (Specific): Update mingw-w64 reference.
5037 (Specific): Remove broken link to Renesas RX processor.
5039 2017-02-10 Richard Biener <rguenther@suse.de>
5041 * toplev.c (process_options): Do not mention obsolete graphite
5042 options when printing sorry message about missing graphite support.
5043 Mention -floop-nest-optimize.
5045 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
5047 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
5048 (vtst_p16): Likewise.
5049 (vtstq_p8): Likewise.
5050 (vtstq_p16): Likewise.
5052 (vtstq_p64): Likewise.
5053 * config/arm/arm_neon.h (vgetq_lane_p64): New.
5054 (vset_lane_p64): New.
5055 (vsetq_lane_p64): New.
5057 2017-02-10 Jakub Jelinek <jakub@redhat.com>
5059 PR tree-optimization/79411
5060 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
5061 stmt operands are SSA_NAMEs used in abnormal phis.
5062 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
5065 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
5068 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
5071 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
5073 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
5075 2017-02-09 Jakub Jelinek <jakub@redhat.com>
5077 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
5081 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
5082 not arbitrary TREE_CONSTANT.
5085 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
5086 "omp declare target link" attribute unless is_global_var.
5087 * omp-offload.c (find_link_var_op): Likewise.
5089 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
5090 Chung-Lin Tang <cltang@codesourcery.com>
5092 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
5094 (gimplify_adjust_omp_clauses): Don't delete TILE.
5095 (gimplify_omp_for): Deal with TILE.
5096 * internal-fn.c (expand_GOACC_TILE): New function.
5097 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
5099 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
5100 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
5102 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
5103 avoid DIV for outermost collapse var.
5104 (expand_oacc_for): Insert tile element loop as needed. Adjust.
5105 Remove out of date comments, fix whitespace.
5106 * omp-general.c (omp_extract_for_data): Deal with tiling.
5107 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
5108 adjust OLF_DIM_BASE value.
5109 (struct omp_for_data): Add tiling field.
5110 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
5111 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
5112 for auto loops. Remove default auto determining, moved to
5113 oacc_loop_fixed_partitions.
5114 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
5115 stmts, add e_mask field.
5116 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
5117 (oacc_thread_numbers): Use oacc_dim_call.
5118 (oacc_xform_tile): New.
5119 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
5120 (finish_oacc_loop): Adjust for ifns vector.
5121 (oacc_loop_discover_walk): Append loop abstraction sites to list,
5122 add case for GOACC_TILE fns.
5123 (oacc_loop_xform_loop): Delete.
5124 (oacc_loop_process): Iterate over call list directly, and add
5125 handling for GOACC_TILE fns.
5126 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
5128 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
5129 vector partitioning to outer loops. Assign 2 partitions to loops
5130 when available. Add TILE handling.
5131 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
5132 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
5133 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
5134 * tree.c (omp_clause_num_ops): Adjust TILE ops.
5135 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
5137 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
5139 * configure.ac (ACX_BUGURL): Update.
5140 * configure: Regenerate.
5142 2017-02-09 Richard Biener <rguenther@suse.de>
5144 PR tree-optimization/69823
5145 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
5146 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
5148 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
5150 * config/arc/arc-c.def: Add __NPS400__ definition.
5151 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
5152 (TARGET_NPS400): Define.
5154 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
5156 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
5158 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
5160 (arc_cpu_types): Fill the arch_info field with a pointer into the
5161 arc_arch_types table.
5162 (arc_selected_cpu): Declare.
5163 * config/arc/arc.c (arc_selected_cpu): Make global.
5164 (arc_selected_arch): Delete.
5165 (arc_base_cpu): Delete.
5166 (arc_override_options): Remove references to deleted variables,
5167 update access to arch information.
5168 (ARC_OPT): Update access to arch information.
5169 (ARC_OPTX): Likewise.
5170 * config/arc/arc.h (arc_base_cpu): Remove declaration.
5171 (TARGET_ARC600): Update access to arch information.
5172 (TARGET_ARC601): Likewise.
5173 (TARGET_ARC700): Likewise.
5174 (TARGET_EM): Likewise.
5175 (TARGET_HS): Likewise.
5176 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
5179 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
5182 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
5183 condition/operands for integer GE/LE/GEU/LEU operations.
5185 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
5187 PR translation/79397
5188 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
5191 2017-02-08 Martin Jambor <mjambor@suse.cz>
5194 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
5195 whether allocation happened.
5196 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
5197 nothing was allocated.
5199 2017-02-08 Jakub Jelinek <jakub@redhat.com>
5201 PR tree-optimization/79408
5202 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
5203 constant, but SSA_NAME with a known integer range, use the minimum
5204 of that range instead of op1 to determine if modulo can be replaced
5205 with its first operand.
5207 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5209 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
5211 2017-02-08 Richard Biener <rguenther@suse.de>
5213 PR tree-optimization/71824
5214 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
5215 Check all loops contained in the merged region.
5217 2017-02-07 Andrew Pinski <apinski@cavium.com>
5219 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
5221 2017-02-07 Andrew Pinski <apinski@cavium.com>
5223 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
5224 (thunderxt88): Likewise.
5225 (thunderxt81): Disable LSE and change v8.1 to v8.
5226 (thunderxt83): Likewise.
5228 2017-02-07 Jakub Jelinek <jakub@redhat.com>
5229 Richard Biener <rguenther@suse.de>
5232 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
5233 type from int to size_t.
5234 * ira-costs.c (struct_costs_size): Change type from int to size_t.
5236 2017-02-07 Jakub Jelinek <jakub@redhat.com>
5238 PR rtl-optimization/79386
5239 * cprop.c (bypass_conditional_jumps): Initialize
5240 bypass_last_basic_block already before splitting bbs after
5241 unconditional traps...
5242 (bypass_conditional_jumps): ... rather than here.
5245 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
5246 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
5247 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
5248 fix -masm=intel patterns.
5250 2017-02-07 Richard Biener <rguenther@suse.de>
5252 PR tree-optimization/79256
5254 * builtins.c (get_object_alignment_2): Use min_align_of_type
5255 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
5256 and ADJUST_FIELD_ALIGN.
5258 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
5260 * doc/tm.texi: Regenerate.
5261 * stor-layout.c (layout_decl): Adjust.
5262 (update_alignment_for_field): Likewise.
5263 (place_field): Likewise.
5264 (min_align_of_type): Likewise.
5265 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
5266 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
5267 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
5268 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
5269 * config/frv/frv.c (frv_adjust_field_align): Likewise.
5270 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
5271 * config/i386/i386.c (x86_field_alignment): Likewise.
5272 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
5273 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
5274 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
5275 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
5276 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
5277 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
5281 2017-01-30 Richard Biener <rguenther@suse.de>
5283 PR tree-optimization/79256
5284 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5285 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5288 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
5290 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
5291 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
5292 builtins to SImode and emit a zero-extend, if necessary.
5294 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5296 * docs/invoke.texi (RISC-V Options): Alphabetize.
5298 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5300 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
5303 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5305 * config/riscv/riscv.c: New file.
5306 * gcc/common/config/riscv/riscv-common.c: Likewise.
5307 * config.gcc: Likewise.
5308 * config/riscv/constraints.md: Likewise.
5309 * config/riscv/elf.h: Likewise.
5310 * config/riscv/generic.md: Likewise.
5311 * config/riscv/linux.h: Likewise.
5312 * config/riscv/multilib-generator: Likewise.
5313 * config/riscv/peephole.md: Likewise.
5314 * config/riscv/pic.md: Likewise.
5315 * config/riscv/predicates.md: Likewise.
5316 * config/riscv/riscv-builtins.c: Likewise.
5317 * config/riscv/riscv-c.c: Likewise.
5318 * config/riscv/riscv-ftypes.def: Likewise.
5319 * config/riscv/riscv-modes.def: Likewise.
5320 * config/riscv/riscv-opts.h: Likewise.
5321 * config/riscv/riscv-protos.h: Likewise.
5322 * config/riscv/riscv.h: Likewise.
5323 * config/riscv/riscv.md: Likewise.
5324 * config/riscv/riscv.opt: Likewise.
5325 * config/riscv/sync.md: Likewise.
5326 * config/riscv/t-elf-multilib: Likewise.
5327 * config/riscv/t-linux: Likewise.
5328 * config/riscv/t-linux-multilib: Likewise.
5329 * config/riscv/t-riscv: Likewise.
5330 * configure.ac: Likewise.
5331 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
5332 Waterman as RISC-V maintainers.
5333 * doc/install.texi: Add RISC-V entries.
5334 * doc/invoke.texi: Add RISC-V options section.
5335 * doc/md.texi: Add RISC-V constraints section.
5336 * configure: Regenerated.
5338 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5341 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
5342 false values to be constant vectors with all 0 or all 1 bits set.
5343 (vcondu<mode><mode>): Likewise.
5344 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
5346 (fpmask_comparison_operator): Update comment.
5347 (vecint_comparison_operator): New predicate.
5348 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
5349 vector conditionals when the true and false values are constant
5350 vectors with all 0 bits or all 1 bits set.
5352 2017-02-06 Martin Sebor <msebor@redhat.com>
5354 PR tree-optimization/79376
5355 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
5357 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
5359 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
5360 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
5361 to simplify split condition.
5363 2017-02-06 Jakub Jelinek <jakub@redhat.com>
5365 * omp-expand.c (oxpand_omp_atomic_fetch_op,
5366 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
5369 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
5371 PR rtl-optimization/68664
5372 * target.def (can_speculate_insn): New hook.
5373 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
5374 * doc/tm.texi: Regenerate.
5375 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
5376 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
5377 (rs6000_sched_can_speculate_insn): New function.
5379 2017-02-06 Jakub Jelinek <jakub@redhat.com>
5381 PR tree-optimization/79284
5382 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
5383 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
5384 vectorizable_mask_load_store, vectorizable_operation,
5385 vect_is_simple_cond, get_same_sized_vectype): Use it instead
5386 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
5387 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
5388 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
5389 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5390 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
5391 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
5392 is_gimple_assign (stmt). Replace another such test with
5393 is_gimple_assign (stmt).
5395 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
5398 * config/avr/avr.c (rtl-iter.h): Include it.
5399 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
5400 (avr_legitimate_combined_insn): ...and implementation.
5402 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
5404 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
5405 * config/s390/s390.c (s390_const_operand_ok)
5406 (s390_canonicalize_comparison, s390_extract_part)
5407 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
5408 (s390_contiguous_bitmask_p, s390_rtx_costs)
5409 (legitimize_pic_address): Likewise.
5410 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
5411 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
5412 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
5413 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
5414 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
5416 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
5418 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
5419 REGNO($0) == REGNO($1).
5421 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5423 * config/s390/linux.h(SIZE_TYPE): Add comment.
5425 2017-02-06 Julian Brown <julian@codesourcery.com>
5426 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5427 Virendra Pathak <virendra.pathak@broadcom.com>
5429 * config/aarch64/aarch64-cores.def: Change the scheduler
5431 * config/aarch64/aarch64.md: Include thunderx2t99.md.
5432 * config/aarch64/thunderx2t99.md: New file.
5434 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5436 * doc/standards.texi (Go Language): Update link to language
5439 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
5441 * tree-eh.c (lower_resx): Sanitize profile.
5442 (cleanup_empty_eh_move_lp): Likewise.
5444 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
5447 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
5449 * cfgloopmanip.h (loop_version): Update prototype.
5450 * modulo-sched.c (sms_schedule): Update call of loop_version.
5451 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
5452 * tree-parloops.c (gen_parallel_loop): Likewise.
5453 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
5454 * tree-ssa-loop-split.c (split_loop): Likewise.
5455 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
5456 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
5458 2017-02-05 Martin Liska <mliska@suse.cz>
5461 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
5463 (print_operand_address): Initialize a struct to zero.
5465 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5467 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
5468 garbage collector only in textual form.
5470 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5472 * doc/extend.texi (x86 specific memory model extensions for
5473 transactional memory): Simplify a phrase.
5475 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
5478 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
5479 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
5480 (atomic_storedi_1): Likewise.
5482 2017-02-04 Jakub Jelinek <jakub@redhat.com>
5484 PR tree-optimization/79338
5485 * tree-parloops.c (gather_scalar_reductions): Don't call
5486 vect_analyze_loop_form for loop->inner before destroying loop's
5489 2017-02-03 Martin Sebor <msebor@redhat.com>
5491 PR tree-optimization/79327
5492 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
5493 when precision has resulted in leading zeros.
5494 (format_integer): Adjust the likely counter to assume an unknown
5495 argument that may be zero is non-zero.
5497 2017-02-03 Jason Merrill <jason@redhat.com>
5500 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
5501 avoid copying non-taken branch.
5503 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5505 PR tree-optimization/79340
5506 * tree-vect-loop.c (vectorizable_reduction): Release
5507 vec_defs elements after safe_splicing them into other vectors.
5510 PR tree-optimization/79327
5511 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
5512 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
5514 (format_integer): Use wide_int_to_tree instead of build_int_cst
5515 + to_?hwi. If argmin is NULL, just set argmin and argmax to
5516 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
5517 of shortest and longest sequence.
5519 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
5521 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
5522 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
5524 2017-02-03 Walter Lee <walt@tilera.com>
5527 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
5528 after initial stackframe link reg save.
5529 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
5531 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5534 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
5535 wu for stxssp alternative.
5537 2017-02-03 Martin Sebor <msebor@redhat.com>
5539 PR tree-optimization/79352
5540 * gimple-fold.c (get_range_strlen): Add argument.
5541 (get_range_strlen): Change return type to bool.
5542 (get_maxval_strlen): Pass in a dummy argument.
5543 * gimple-fold.h (get_range_strlen): Change return type to bool.
5544 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
5545 * tree.h (array_at_struct_end_p): Add argument.
5546 * tree.c (array_at_struct_end_p): Handle it.
5548 2017-02-03 Martin Liska <mliska@suse.cz>
5551 * multiple_target.c (create_dispatcher_calls): Redirect edge
5552 from a caller of a dispatcher.
5553 (expand_target_clones): Make the clones local.
5554 (ipa_target_clone): Do both target clones and resolvers.
5555 (ipa_dispatcher_calls): Remove the pass.
5556 (pass_dispatcher_calls::gate): Likewise.
5557 (make_pass_dispatcher_calls): Likewise.
5558 * passes.def (pass_target_clone): Put as very first IPA early
5561 2017-02-03 Martin Liska <mliska@suse.cz>
5563 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
5564 in case of a function with ifunc attribute.
5566 2017-02-03 Martin Liska <mliska@suse.cz>
5568 * cgraph.c (cgraph_node::dump): Dump function version info.
5569 * symtab.c (symtab_node::dump_base): Add missing new line.
5571 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5573 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
5574 (ifcombine_ifandif): Use it.
5576 2017-02-03 Martin Liska <mliska@suse.cz>
5578 * doc/invoke.texi: Document default value for
5579 use-after-scope-direct-emission-threshold.
5581 2017-02-03 Martin Liska <mliska@suse.cz>
5583 PR tree-optimization/79339
5584 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
5585 (format_floating): Likewise.
5587 2017-02-03 Martin Liska <mliska@suse.cz>
5590 * ipa-prop.c (ipa_node_params_t::insert): Remove current
5592 (ipa_node_params_t::remove): Likewise.
5593 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
5594 initialization from removed ipa_node_params_t::insert.
5595 (ipa_node_params::~ipa_node_params): Move from removed
5596 ipa_node_params_t::release.
5597 * symbol-summary.h (symbol_summary::m_released): New member.
5598 Do not release a summary twice. Do not allow to call finalizer
5599 for types of a summary that live in GGC memory.
5601 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5603 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
5606 2017-02-02 Martin Sebor <msebor@redhat.com>
5609 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
5610 (format_string): Tighten up the range of output for non-constant
5611 strings and correct the expected range for wide non-constant strings.
5613 2017-02-02 Martin Sebor <msebor@redhat.com>
5615 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
5618 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
5620 (-fdump-tree-@var): Add to index and document how to come up
5621 with pass-specific option and dump file names.
5622 (-fdump-passes): Clarify where to look for output.
5624 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5627 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
5628 statistics of the analyzed path; allow threading for speed when
5629 any of BBs along the path are optimized for speed.
5631 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
5634 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
5635 settings of the virtual registers.
5638 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
5640 * explow.c (get_dynamic_stack_size): Take known alignment of stack
5641 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
5644 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5646 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
5647 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
5649 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5651 * config/s390/s390.md: Add missing comments with the expanded
5653 * config/s390/vector.md: Likewise.
5654 * config/s390/vx-builtins.md: Likewise.
5656 2017-02-02 Jakub Jelinek <jakub@redhat.com>
5659 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
5660 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
5661 conditions on a single line.
5663 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5665 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5666 __S390_VX__ to __VX__.
5668 2017-02-01 Andrew Pinski <apinski@cavium.com>
5670 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
5671 stmt_info to record_stmt_cost.
5672 (vect_get_known_peeling_cost): Pass stmt_info if known to
5674 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
5675 cpu_vector_cost field into
5676 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
5677 field into vec_int_stmt_cost and vec_fp_stmt_cost.
5678 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
5679 splitting of scalar_stmt_cost and vec_stmt_cost.
5680 (thunderx_vector_cost): Likewise.
5681 (cortexa57_vector_cost): LIkewise.
5682 (exynosm1_vector_cost): Likewise.
5683 (xgene1_vector_cost): Likewise.
5684 (thunderx2t99_vector_cost): Improve after the splitting of the two
5686 (aarch64_builtin_vectorization_cost): Update for the splitting of
5687 scalar_stmt_cost and vec_stmt_cost.
5689 2017-02-01 Torvald Riegel <triegel@redhat.com>
5690 Richard Henderson <rth@redhat.com>
5692 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
5693 conditional on existance of a fast atomic load.
5694 * optabs-query.c (can_atomic_load_p): New function.
5695 * optabs-query.h (can_atomic_load_p): Declare it.
5696 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
5697 no fast atomic load is available for the particular size of access.
5698 (expand_atomic_compare_and_swap): Likewise.
5699 (expand_atomic_load): Likewise.
5700 (expand_atomic_store): Likewise.
5701 (expand_atomic_fetch_op): Likewise.
5702 * testsuite/lib/target-supports.exp
5703 (check_effective_target_sync_int_128): Remove x86 because it provides
5704 no fast atomic load.
5705 (check_effective_target_sync_int_128_runtime): Likewise.
5707 2017-02-01 Richard Biener <rguenther@suse.de>
5709 * graphite.c: Include tree-vectorizer.h for find_loop_location.
5710 (graphite_transform_loops): Provide opt-info for optimized nests.
5711 * tree-parloop.c (parallelize_loops): Provide opt-info for
5714 2017-02-01 Richard Biener <rguenther@suse.de>
5717 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
5720 2017-02-01 Richard Biener <rguenther@suse.de>
5722 PR tree-optimization/71824
5723 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
5724 Verify the loops are valid in the merged SESE region.
5725 (scop_detection::can_represent_loop_1): Check analyzing the
5726 evolution of the number of iterations in the region succeeds.
5728 2017-01-31 Ian Lance Taylor <iant@golang.org>
5730 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
5731 REG_ARGS_SIZE note to 32-bit push insns and call insn.
5733 2017-01-31 David Malcolm <dmalcolm@redhat.com>
5735 PR preprocessor/79210
5736 * input.c (get_substring_ranges_for_loc): Replace line_width
5737 assertion with error-handling.
5739 2017-01-31 Richard Biener <rguenther@suse.de>
5741 PR tree-optimization/77318
5742 * graphite-sese-to-poly.c (extract_affine): Fix assert.
5743 (create_pw_aff_from_tree): Take loop parameter.
5744 (add_condition_to_pbb): Pass loop of the condition to
5745 create_pw_aff_from_tree.
5747 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5749 * config/s390/s390.c (s390_asan_shadow_offset): New function.
5750 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
5752 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5756 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
5758 (convert_int_to_float128): Likewise.
5759 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
5760 (convert_int_to_float128): Likewise.
5761 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
5762 (UNSPEC_IEEE128_CONVERT): Likewise.
5763 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
5764 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
5765 Use local variables for IBM extended format.
5766 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
5767 (fix_trunc<mode>si2_fprs): Likewise.
5768 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
5769 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
5770 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
5771 to know that we can now have integers of all sizes in vector
5773 (fix<uns>_<mode>di2_hw): Likewise.
5774 (float<uns>_<mode>si2_hw): Likewise.
5775 (fix_<mode>si2_hw): Likewise.
5776 (fixuns_<mode>si2_hw): Likewise.
5777 (float<uns>_<mode>di2_hw): Likewise.
5778 (float_<mode>di2_hw): Likewise.
5779 (float_<mode>si2_hw): Likewise.
5780 (floatuns_<mode>di2_hw): Likewise.
5781 (floatuns_<mode>si2_hw): Likewise.
5782 (xscvqp<su>wz_<mode>): Delete, no longer used.
5783 (xscvqp<su>dz_<mode>): Likewise.
5784 (xscv<su>dqp_<mode>): Likewise.
5785 (ieee128_mfvsrd_64bit): Likewise.
5786 (ieee128_mfvsrd_32bit): Likewise.
5787 (ieee128_mfvsrwz): Likewise.
5788 (ieee128_mtvsrw): Likewise.
5789 (ieee128_mtvsrd_64bit): Likewise.
5790 (ieee128_mtvsrd_32bit): Likewise.
5792 2017-01-31 Martin Liska <mliska@suse.cz>
5795 * ipa-prop.c (ipa_free_all_node_params): Call release method
5796 instead of ~sumbol_summary to not to trigger double times
5799 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
5801 PR tree-optimization/71691
5802 * bitmap.h (class auto_bitmap): New.
5803 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
5804 is_maybe_undefined instead of ssa_undefined_value_p.
5806 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5808 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5809 __S390_ARCH_LEVEL__ to __ARCH__.
5811 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5813 PR tree-optimization/79267
5814 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
5815 if should_remove_lhs_p is true.
5817 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
5820 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
5821 (add_alignment_attribute): New.
5822 (base_type_die): Add alignment attribute.
5823 (subrange_type_die): Likewise.
5824 (modified_type_die): Likewise.
5825 (gen_array_type_die): Likewise.
5826 (gen_descr_array_type_die: Likewise.
5827 (gen_enumeration_type_die): Likewise.
5828 (gen_subprogram_die): Likewise.
5829 (gen_variable_die): Likewise.
5830 (gen_field_die): Likewise.
5831 (gen_ptr_to_mbr_type_die): Likewise.
5832 (gen_struct_or_union_type_die): Likewise.
5833 (gen_subroutine_type_die): Likewise.
5834 (gen_typedef_die): Likewise.
5835 (base_type_cmp): Compare alignment attribute.
5837 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5840 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
5841 (setb_unsigned) New pattern for setb with CCUNS.
5842 * config/rs6000/rs6000.c (expand_block_compare): Use a different
5843 subfc./subfe sequence to avoid overflow problems. Generate a
5844 shorter sequence with cmpld/setb for power9.
5845 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
5846 for generating subfc. instruction.
5847 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
5848 now uses this instruction.
5850 2017-01-30 Ian Lance Taylor <iant@google.com>
5853 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
5854 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
5856 2017-01-30 Martin Sebor <msebor@redhat.com>
5858 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
5859 Move constant to the right of a relational operator.
5860 (get_mpfr_format_length, format_character, format_string): Ditto.
5861 (should_warn_p, maybe_warn): Same.
5863 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
5865 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5868 * asan.c (get_translation_unit_decl): Remove function.
5869 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
5871 2017-01-30 Martin Liska <mliska@suse.cz>
5873 PR gcov-profile/79259
5874 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
5877 2017-01-30 Martin Liska <mliska@suse.cz>
5880 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
5881 Initialize variables with NULL value.
5883 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
5886 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
5888 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
5890 2017-01-30 Richard Biener <rguenther@suse.de>
5892 PR tree-optimization/79276
5893 * tree-vrp.c (process_assert_insertions): Properly adjust common
5894 when removing a duplicate.
5896 * gcc.dg/torture/pr79276.c: New testcase.
5898 2017-01-30 Richard Biener <rguenther@suse.de>
5900 PR tree-optimization/79256
5901 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5902 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5904 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
5906 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
5909 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
5910 ("*r<noxa>sbg_<mode>_sll_bitmask")
5911 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
5912 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
5913 Use contiguous_bitmask_nowrap_operand.
5915 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5918 * config/rs6000/altivec.h (vec_xl): Revise #define.
5919 (vec_xst): Likewise.
5921 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
5923 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
5925 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
5927 PR rtl-optimization/79194
5928 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
5929 traps before call to bypass_conditional_jumps.
5931 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5933 PR tree-optimization/71374
5934 * lra-constraints.c (check_conflict_input_operands): New.
5935 (match_reload): Use it.
5937 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5940 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
5941 account to calculate conflict_set.
5943 2017-01-27 Bin Cheng <bin.cheng@arm.com>
5945 PR rtl-optimization/78559
5946 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
5947 other_insn in combine.
5949 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
5951 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
5952 uint16_type_node for BT_UINT16.
5954 2017-01-27 David Malcolm <dmalcolm@redhat.com>
5956 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
5957 "RTL Tests" to menu.
5958 (GIMPLE Tests): New node.
5959 (RTL Tests): New node.
5961 2017-01-27 Richard Biener <rguenther@suse.de>
5963 PR tree-optimization/79245
5964 * tree-loop-distribution.c (distribute_loop): Apply cost
5965 modeling also to detected patterns.
5967 2017-01-27 Richard Biener <rguenther@suse.de>
5969 PR tree-optimization/71433
5970 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
5971 (compare_assert_loc): New function.
5972 (process_assert_insertions): Sort and optimize assert locations
5973 to remove duplicates and push down identical assertions on
5974 edges to their destination block.
5976 2017-01-27 Richard Biener <rguenther@suse.de>
5978 PR tree-optimization/79244
5979 * tree-vrp.c (remove_range_assertions): Forcefully propagate
5980 out SSA names even if abnormal.
5982 2017-01-27 Jakub Jelinek <jakub@redhat.com>
5984 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
5985 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
5986 instead of MPFR_RNDN.
5988 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
5991 * arm.c (arm_option_override): Don't call build_target_option_node
5992 until after doing all option overrides.
5993 (arm_valid_target_attribute_tree): Likewise.
5995 2017-01-27 Martin Liska <mliska@suse.cz>
5997 * doc/invoke.texi (-fprofile-arcs): Document profiling support
5998 for {cd}tors and C++ {cd}tors.
6000 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
6002 * config/s390/s390.md ("*setmem_long_and")
6003 ("*setmem_long_and_31z"): Use zero_extend instead of and.
6005 2017-01-26 Martin Sebor <msebor@redhat.com>
6007 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
6010 2017-01-26 Martin Sebor <msebor@redhat.com>
6012 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
6013 HAVE_DFmode before using XFmode or DFmode.
6014 (parse_directive): Avoid using the z length modifier to avoid
6015 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
6018 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
6019 to accept adjustment as an array.
6020 (get_int_range): New function.
6021 (struct directive): Make width and prec arrays.
6022 (directive::set_width, directive::set_precision): Call get_int_range.
6023 (format_integer, format_floating): Handle width and precision ranges.
6024 (format_string, parse_directive): Same.
6026 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6029 * dwarf2out.c (generate_skeleton_bottom_up): For children with
6030 comdat_type_p set, just clone them, but keep the children in the
6034 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
6035 which have direct callers with -fvar-tracking-assignments enabled
6037 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
6038 inside of type units.
6040 2017-01-26 Martin Sebor <msebor@redhat.com>
6043 * gimple-ssa-sprintf.c (struct result_range): Add likely and
6045 (struct format_result): Replace number_chars, number_chars_min,
6046 and number_chars_max with a single member of struct result_range.
6048 (format_result::operator+=): Adjust.
6049 (struct fmtresult): Remove bounded. Handle likely and unlikely
6051 (fmtresult::adjust_for_width_or_precision): New function.
6052 (fmtresult:type_max_digits): New function.
6053 (bytes_remaining): Handle likely and unlikely counters.
6054 (min_bytes_remaining): Remove.
6055 (format_percent): Simplify.
6056 (format_integer, format_floating): Set likely and unlikely counters.
6057 (get_string_length, format_character, format_string): Same.
6058 (format_plain, should_warn_p): New function.
6059 (maybe_warn): Call should_warn_p. Update diagnostic messages
6060 and handle those for all directives, including plain strings.
6061 (format_directive): Handle likely and unlikely counters.
6062 Remove unnecessary quoting from diagnostics. Add an informational
6064 (add_bytes): Remove.
6065 (pass_sprintf_length::compute_format_length): Simplify.
6066 (try_substitute_return_value): Handle likely and unlikely counters.
6068 2017-01-26 Carl Love <cel@us.ibm.com>
6070 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
6071 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
6073 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
6076 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
6077 endianess for subregs into account.
6078 * lra-constraints.c (lra_constraints): Do risky transformations
6079 always on the first iteration.
6080 * lra-lives.c (check_pseudos_live_through_calls): Add arg
6081 last_call_used_reg_set.
6082 (process_bb_lives): Define and use last_call_used_reg_set.
6083 * lra.c (lra): Always continue after lra_constraints on the first
6086 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
6088 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
6090 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
6092 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6094 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
6095 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
6096 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
6097 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
6098 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
6099 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
6100 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
6101 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
6102 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
6104 2017-01-26 Marek Polacek <polacek@redhat.com>
6107 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
6108 for the third operand.
6110 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6113 * omp-low.c (struct omp_context): Add simt_stmt field.
6114 (scan_omp_for): Return omp_context *.
6115 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
6116 context to the _simt_ SIMD stmt.
6117 (lower_omp_for): For combined SIMD with sibling _simt_
6118 SIMD, make sure to use the same decls in _looptemp_
6119 clauses as in the sibling.
6121 2017-01-26 David Sherwood <david.sherwood@arm.com>
6124 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
6127 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6130 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
6131 emit fld b; fld a; if possible.
6133 * brig-builtins.def: Update copyright years.
6134 * config/arm/arm_acle_builtins.def: Update copyright years.
6136 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6139 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
6140 constraint instead of o for the stxsd instruction.
6142 2017-01-25 Carl Love <cel@us.ibm.com>
6144 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
6145 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
6147 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
6149 * doc/invoke.texi (C++ Dialect Options): Fix typo.
6151 2017-01-25 Richard Biener <rguenther@suse.de>
6153 PR tree-optimization/69264
6154 * target.def (vector_alignment_reachable): Improve documentation.
6155 * doc/tm.texi: Regenerate.
6156 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
6158 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
6159 earlier changes with respect to TYPE_USER_ALIGN.
6160 (vector_alignment_reachable_p): Likewise. Improve dumping.
6162 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6165 * config/arm/arm.md (xordi3): Force constant operand into a register
6168 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6170 * doc/invoke.texi (-fstore-merging): Correct default optimization
6171 levels at which it is enabled.
6172 (-O): Move -fstore-merging from list to...
6175 2017-01-25 Richard Biener <rguenther@suse.de>
6178 * omp-expand.c: Include debug.h.
6179 (expand_omp_taskreg): Make sure to generate early debug before
6180 outlining anything from a function.
6181 (expand_omp_target): Likewise.
6182 (grid_expand_target_grid_body): Likewise.
6184 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
6187 * asan.c (get_translation_unit_decl): New function.
6188 (asan_add_global): Extract modules file name from globals
6189 TRANSLATION_UNIT_DECL name.
6191 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
6194 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
6195 for long calls with APCS frame and VFP.
6197 2017-01-24 David Malcolm <dmalcolm@redhat.com>
6199 * cfg.c (original_copy_tables_initialized_p): New function.
6200 * cfg.h (original_copy_tables_initialized_p): New decl.
6201 * cfgrtl.c (relink_block_chain): Guard the call to
6202 free_original_copy_tables with a call to
6203 original_copy_tables_initialized_p.
6204 * cgraph.h (symtab_node::native_rtl_p): New decl.
6205 * cgraphunit.c (symtab_node::native_rtl_p): New function.
6206 (symtab_node::needed_p): Don't assert for early assembly output
6207 for __RTL functions.
6208 (cgraph_node::finalize_function): Set "force_output" for __RTL
6210 (cgraph_node::analyze): Bail out early for __RTL functions.
6211 (analyze_functions): Update assertion to support __RTL functions.
6212 (cgraph_node::expand): Bail out early for __RTL functions.
6213 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
6215 * function.h (struct function): Update comment for field
6217 * gimple-expr.c: Include "tree-pass.h".
6218 (gimple_has_body_p): Return false for __RTL functions.
6219 * Makefile.in (OBJS): Add run-rtl-passes.o.
6220 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
6222 (gcc::pass_manager::get_clean_slate): New accessor.
6223 * passes.c: Include "insn-addr.h".
6224 (should_skip_pass_p): Add logging. Update logic for running
6225 "expand" to be compatible with both __GIMPLE and __RTL. Guard
6226 property-provider override so it is only done for gimple passes.
6228 (skip_pass): New function.
6229 (execute_one_pass): Call skip_pass when skipping passes.
6230 * read-md.c (md_reader::read_char): Support filtering
6231 the input to a subset of line numbers.
6232 (md_reader::md_reader): Initialize fields
6233 m_first_line and m_last_line.
6234 (md_reader::read_file_fragment): New function.
6235 * read-md.h (md_reader::read_file_fragment): New decl.
6236 (md_reader::m_first_line): New field.
6237 (md_reader::m_last_line): New field.
6238 * read-rtl-function.c (function_reader::create_function): Only
6239 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
6240 curr_properties. Set DECL_INITIAL to a dummy block.
6241 (read_rtl_function_body_from_file_range): New function.
6242 * read-rtl-function.h (read_rtl_function_body_from_file_range):
6244 * run-rtl-passes.c: New file.
6245 * run-rtl-passes.h: New file.
6247 2017-01-24 Jeff Law <law@redhat.com>
6249 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
6252 2017-01-24 Bin Cheng <bin.cheng@arm.com>
6254 PR tree-optimization/79159
6255 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
6256 (record_nonwrapping_iv): Improve boundary using above function if no
6257 value range information.
6259 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
6260 Martin Jambor <mjambor@suse.cz>
6262 * brig-builtins.def: New file.
6263 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
6264 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
6265 (DEF_HSAIL_SAT_BUILTIN): Likewise.
6266 (DEF_HSAIL_INTR_BUILTIN): Likewise.
6267 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
6268 * builtin-types.def (BT_INT8): New.
6269 (BT_INT16): Likewise.
6270 (BT_UINT8): Likewise.
6271 (BT_UINT16): Likewise.
6272 (BT_FN_ULONG): Likewise.
6273 (BT_FN_UINT_INT): Likewise.
6274 (BT_FN_UINT_ULONG): Likewise.
6275 (BT_FN_UINT_LONG): Likewise.
6276 (BT_FN_UINT_PTR): Likewise.
6277 (BT_FN_ULONG_PTR): Likewise.
6278 (BT_FN_INT8_FLOAT): Likewise.
6279 (BT_FN_INT16_FLOAT): Likewise.
6280 (BT_FN_UINT32_FLOAT): Likewise.
6281 (BT_FN_UINT16_FLOAT): Likewise.
6282 (BT_FN_UINT8_FLOAT): Likewise.
6283 (BT_FN_UINT64_FLOAT): Likewise.
6284 (BT_FN_UINT16_UINT32): Likewise.
6285 (BT_FN_UINT32_UINT16): Likewise.
6286 (BT_FN_UINT16_UINT16_UINT16): Likewise.
6287 (BT_FN_INT_PTR_INT): Likewise.
6288 (BT_FN_UINT_PTR_UINT): Likewise.
6289 (BT_FN_LONG_PTR_LONG): Likewise.
6290 (BT_FN_ULONG_PTR_ULONG): Likewise.
6291 (BT_FN_VOID_UINT64_UINT64): Likewise.
6292 (BT_FN_UINT8_UINT8_UINT8): Likewise.
6293 (BT_FN_INT8_INT8_INT8): Likewise.
6294 (BT_FN_INT16_INT16_INT16): Likewise.
6295 (BT_FN_INT_INT_INT): Likewise.
6296 (BT_FN_UINT_FLOAT_UINT): Likewise.
6297 (BT_FN_FLOAT_UINT_UINT): Likewise.
6298 (BT_FN_ULONG_UINT_UINT): Likewise.
6299 (BT_FN_ULONG_UINT_PTR): Likewise.
6300 (BT_FN_ULONG_ULONG_ULONG): Likewise.
6301 (BT_FN_UINT_UINT_UINT): Likewise.
6302 (BT_FN_VOID_UINT_PTR): Likewise.
6303 (BT_FN_UINT_UINT_PTR: Likewise.
6304 (BT_FN_UINT32_UINT64_PTR): Likewise.
6305 (BT_FN_INT_INT_UINT_UINT): Likewise.
6306 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
6307 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
6308 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
6309 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
6310 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
6311 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
6312 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
6313 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
6314 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
6315 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
6316 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
6317 * doc/frontends.texi: List BRIG FE.
6318 * doc/install.texi (Testing): Add BRIG tesring requirements.
6319 * doc/invoke.texi (Overall Options): Mention BRIG.
6320 * doc/standards.texi (Standards): Doucment BRIG HSA version.
6322 2017-01-24 Richard Biener <rguenther@suse.de>
6324 PR translation/79208
6325 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
6327 2017-01-24 Martin Jambor <mjambor@suse.cz>
6330 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
6331 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
6334 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
6337 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
6338 casts from signed to unsigned really don't have a range.
6340 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
6342 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
6343 GMP_RNDx for compatiblity.
6345 2017-01-24 Martin Liska <mliska@suse.cz>
6348 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
6349 that would prevent us to call alloca with -1 as argument.
6351 2017-01-24 Jakub Jelinek <jakub@redhat.com>
6353 * dwarf2out.c (output_compilation_unit_header, output_file_names):
6354 Avoid -Wformat-security warning.
6356 2017-01-23 Andrew Pinski <apinski@cavium.com>
6358 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
6361 2017-01-23 Martin Sebor <msebor@redhat.com>
6364 * gimple-ssa-sprintf.c (warn_level): New global.
6365 (format_integer): Use it here and throughout the rest of the file.
6366 Use the same switch to compute sign as base.
6367 (maybe_warn): New function.
6368 (format_directive): Factor out warnings into maybe_warn.
6369 Add debugging output. Use warn_level.
6370 (add_bytes): Use warn_level.
6371 (pass_sprintf_length::compute_format_length): Add debugging output.
6372 (try_substitute_return_value): Same.
6373 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
6376 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
6377 (struct fmtresult, format_integer, format_floating): Adjust.
6378 (fmtresult::fmtresult): Set max correctly in two argument ctor.
6379 (get_string_length, format_string,format_directive): Same.
6380 (pass_sprintf_length::compute_format_length): Same.
6381 (try_substitute_return_value): Simplify slightly.
6384 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
6385 (fmtresult::operator+=): Outlined.
6386 (struct fmtresult): Add ctors.
6387 (struct conversion_spec): Rename...
6388 (struct directive): ...to this. Add and remove data members.
6389 (directive::set_width, directive::set_precision): New functions.
6390 (format_percent): Use fmtresult ctor.
6391 (get_width_and_precision): Remove.
6392 (format_integer): Make naming changes. Avoid computing width and
6394 (format_floating): Same. Adjust indentation.
6395 (format_character, format_none): New functions.
6396 (format_string): Moved character handling to format_character.
6397 (format_directive): Remove arguments, change return type.
6398 (parse_directive): New function.
6399 (pass_sprintf_length::compute_format_length): Move directive
6400 parsing to parse_directive.
6402 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6404 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
6405 (assign_assembler_name_if_needed): ... this.
6406 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
6407 (assign_assembler_name_if_needed): ... this.
6408 (free_lang_data_in_cgraph): Adjust callers.
6409 * cgraphunit.c (cgraph_node::analyze): Likewise.
6410 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
6413 2017-01-23 Richard Biener <rguenther@suse.de>
6415 PR tree-optimization/79088
6416 PR tree-optimization/79188
6417 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
6418 resetting loop bounds after last path deletion. Reset loop
6419 bounds of the target loop, make code match the comments.
6420 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
6421 Make sure loops need no fixups.
6423 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
6425 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
6426 exponent support with double type for first argument.
6427 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
6428 type returned by __builtin_vec_extract_sig,
6429 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
6430 functions from "vector int" to "vector unsigned int" or from
6431 "vector long long int" to "vector unsigned long long int".
6432 Changed type returned by __builtin_vec_extract_exp,
6433 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
6434 functions from "vector int" to "vector unsigned int" or from
6435 "vector long long int" to "vector unsigned long long int".
6436 Changed return type of __builtin_vec_test_data_class,
6437 __builtin_vec_test_data_class_sp, and
6438 __builtin_vec_test_data_class_dp from "vector int" to
6439 "vector bool int" or from "vector long long int" to "vector bool
6440 long long int" and changed second argument type from "unsigned
6441 int" to "int". Added new overloaded function forms "vector float
6442 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
6443 "vector float __builtin_vec_insert_exp_sp (vector float, vector
6444 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
6445 double, vector unsigned long long int)" and "vector double
6446 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
6447 long int)". Changed return type of
6448 __builtin_scalar_test_data_class and
6449 __builtin_scalar_test_data_class_sp and
6450 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
6451 int" and changed second argument from "unsigned int" to "int".
6452 Changed type returned by __builtin_scalar_test_neg,
6453 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
6454 from "int" to "bool int". Added new overloaded function form
6455 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
6456 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
6457 exponent double-precision with floating point first argument.
6458 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
6459 documentation of scalar_test_data_class, scalar_test_neg,
6460 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
6461 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
6462 vec_test_data_class built-in functions to reflect refinements in
6463 their type signatures.
6465 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
6467 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
6469 (aarch64_elf_asm_destructor): Likewise.
6471 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
6473 PR rtl-optimization/78634
6474 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
6475 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
6476 * ifcvt.c (noce_try_cmove): Add missing cost check.
6478 PR rtl-optimization/71724
6479 * combine.c (if_then_else_cond): Look for situations where it is
6480 beneficial to undo the work of one of the recursive calls.
6482 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6484 PR tree-optimization/70754
6485 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6486 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6487 combined stmt before it if not NULL.
6488 (combine_chains): Process refs reversely and compute dominance point
6491 2017-01-23 Martin Liska <mliska@suse.cz>
6493 PR tree-optimization/79196
6494 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
6495 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
6497 (strlen_optimize_stmt): Call the renamed function.
6499 2017-01-23 Michael Matz <matz@suse.de>
6501 PR tree-optimization/78384
6502 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
6504 2017-01-23 Richard Biener <rguenther@suse.de>
6506 PR tree-optimization/79186
6507 * tree-vrp.c (register_new_assert_for): Make sure we've seen
6508 both incoming edges before moving an assert.
6510 2017-01-23 Martin Jambor <mjambor@suse.cz>
6512 * ipa-prop.c (load_from_param_1): Removed.
6513 (load_from_unmodified_param): Bits from load_from_param_1 put back
6515 (load_from_param): Removed.
6516 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
6517 with stmt. Reverted back to use of load_from_unmodified_param.
6519 2017-01-23 Martin Jambor <mjambor@suse.cz>
6522 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
6523 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
6524 field a pointer to garbage collected vector, mark lattices and
6525 ipcp_orig_node with GTY((skip)).
6526 (ipa_get_param_count): Adjust to descriptors being a pointer.
6527 (ipa_get_param): Likewise.
6528 (ipa_get_type): Likewise.
6529 (ipa_get_param_move_cost): Likewise.
6530 (ipa_set_param_used): Likewise.
6531 (ipa_get_controlled_uses): Likewise.
6532 (ipa_set_controlled_uses): Likewise.
6533 (ipa_is_param_used): Likewise.
6534 (ipa_node_params_t): Move into garbage collector. New methods insert
6536 (ipa_node_params_sum): Annotate wth GTY(()).
6537 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
6539 (ipa_load_from_parm_agg): Adjust declaration.
6540 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
6541 * ipa-profile.c (ipa_profile): Likewise.
6542 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
6543 (ipa_populate_param_decls): Make descriptors parameter garbage
6545 (ipa_dump_param): Adjust to descriptors being a pointer.
6546 (ipa_alloc_node_params): Likewise.
6547 (ipa_initialize_node_params): Likewise.
6548 (load_from_param_1): Make descriptors parameter garbage collected.
6549 (load_from_unmodified_param): Likewise.
6550 (load_from_param): Likewise.
6551 (ipa_load_from_parm_agg): Likewise.
6552 (ipa_node_params::~ipa_node_params): Removed.
6553 (ipa_free_all_node_params): Remove call to delete operator.
6554 (ipa_node_params_t::insert): New.
6555 (ipa_node_params_t::remove): Likewise.
6556 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
6557 copy known_csts and known_contexts vectors.
6558 (ipa_read_node_info): Adjust to descriptors being a pointer.
6559 (ipcp_modif_dom_walker): Make m_descriptors field garbage
6561 (ipcp_transform_function): Make descriptors variable garbage
6564 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
6566 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
6567 * config/i386/avx512dqintrin.h: Ditto.
6568 * config/i386/avx512fintrin.h: Ditto.
6569 * gcc/config/i386/i386.c: Handle new builtins.
6570 * config/i386/i386-builtin.def: Add new builtins.
6571 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
6572 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
6574 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6575 Martin Liska <mliska@suse.cz>
6577 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
6578 * asan.c (asan_expand_poison_ifn): Support stores and use
6579 appropriate ASAN report function.
6580 * internal-fn.c (expand_ASAN_POISON_USE): New function.
6581 * internal-fn.def (ASAN_POISON_USE): Declare.
6582 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
6583 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
6584 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
6585 ASAN_POISON calls w/o LHS.
6586 * tree-ssa.c (execute_update_addresses_taken): Create clobber
6587 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
6588 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
6589 * gimplify.c (asan_poison_variables): Add attribute
6590 use_after_scope_memory to variables that really needs to live
6592 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
6593 having the attribute.
6595 2017-01-23 Martin Liska <mliska@suse.cz>
6597 * asan.c (create_asan_shadow_var): New function.
6598 (asan_expand_poison_ifn): Likewise.
6599 * asan.h (asan_expand_poison_ifn): New declaration.
6600 * internal-fn.c (expand_ASAN_POISON): Likewise.
6601 * internal-fn.def (ASAN_POISON): New builtin.
6602 * sanopt.c (pass_sanopt::execute): Expand
6603 asan_expand_poison_ifn.
6604 * tree-inline.c (copy_decl_for_dup_finish): Make function
6606 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
6607 * tree-ssa.c (is_asan_mark_p): New function.
6608 (execute_update_addresses_taken): Rewrite local variables
6609 (identified just by use-after-scope as addressable) into SSA.
6611 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6613 * doc/install.texi (Specific): opensource.apple.com uses https
6614 now. Remove trailing slash.
6616 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6618 * README.Portability: Remove note on an Irix compatibility issue.
6620 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
6622 * gcov.c (INCLUDE_ALGORITHM): Define.
6623 (INCLUDE_VECTOR): Define.
6624 No longer include <vector> and <algorithm> directly.
6626 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
6628 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
6630 * doc/invoke.texi (Code Gen Options): Ditto.
6632 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
6635 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
6637 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
6639 rtl-optimization/79125
6640 * cprop.c (local_cprop_pass): Handle cases where we make an
6643 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
6647 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
6648 read from, for big endian.
6650 2017-01-20 Jiong Wang <jiong.wang@arm.com>
6652 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
6653 register pauth builtins for LP64 only.
6655 2017-01-20 Marek Polacek <polacek@redhat.com>
6658 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
6661 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6663 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
6665 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
6666 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
6667 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
6669 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6672 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
6673 in big-endian mode when they are not a single duplicated value.
6675 2017-01-20 Richard Biener <rguenther@suse.de>
6677 * BASE-VER: Bump to 7.0.1.
6679 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6681 * omp-low.c (omplow_simd_context): New struct. Use it...
6682 (lower_rec_simd_input_clauses): ...here and...
6683 (lower_rec_input_clauses): ...here to hold common data. Adjust all
6684 references to idx, lane, max_vf, is_simt.
6686 2017-01-20 Graham Markall <graham.markall@embecosm.com>
6688 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
6691 2017-01-20 Martin Jambor <mjambor@suse.cz>
6693 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
6694 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
6696 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
6697 (GTFILES): Rename hsa.c to hsa-common.c.
6698 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
6699 * hsa-dump.c: Likewise.
6700 * hsa-gen.c: Likewise.
6701 * hsa-regalloc.c: Likewise.
6702 * ipa-hsa.c: Likewise.
6703 * omp-expand.c: Likewise.
6704 * omp-low.c: Likewise.
6705 * toplev.c: Likewise.
6707 2017-01-20 Marek Polacek <polacek@redhat.com>
6710 * doc/invoke.texi: Document -Wduplicated-branches.
6711 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
6712 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
6713 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
6714 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
6715 return 0 only when not OEP_LEXICOGRAPHIC.
6716 (fold_build_cleanup_point_expr): Use the expression
6717 location when building CLEANUP_POINT_EXPR.
6718 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
6719 * tree.c (add_expr): Handle error_mark_node.
6721 2017-01-20 Martin Liska <mliska@suse.cz>
6724 * tree-profile.c (init_ic_make_global_vars): Do not call
6726 (gimple_init_gcov_profiler): Likewise.
6728 2017-01-20 Martin Liska <mliska@suse.cz>
6731 * cgraph.h (maybe_create_reference): Remove argument and
6733 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
6735 * ipa-cp.c (create_specialized_node): Likewise.
6736 * symtab.c (symtab_node::maybe_create_reference): Handle
6737 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
6739 2017-01-20 Martin Liska <mliska@suse.cz>
6741 * read-rtl-function.c (function_reader::create_function): Use
6742 build_decl instread of build_decl_stat.
6744 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
6746 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
6747 * config/i386/avx512dqintrin.h: Ditto.
6748 * config/i386/avx512fintrin.h: Ditto.
6749 * config/i386/i386-builtin-types.def: Add new types.
6750 * gcc/config/i386/i386.c: Handle new types.
6751 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
6752 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
6753 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
6754 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
6755 (__builtin_ia32_kshiftridi): New.
6756 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
6758 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
6762 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
6763 define to rs6000_init_stack_protect_guard.
6764 (rs6000_init_stack_protect_guard): New function.
6766 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6767 Yunqiang Su <yunqiang.su@imgtec.com>
6769 * config.gcc (supported_defaults): Add madd4.
6770 (with_madd4): Add validation.
6771 (all_defaults): Add madd4.
6772 * config/mips/mips.opt (mmadd4): New option.
6773 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6775 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6777 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
6778 (ISA_HAS_FUSED_MADD4): Likewise.
6779 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
6780 * gcc/doc/install.texi (--with-madd4): Document the new option.
6782 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6784 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
6785 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
6786 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
6787 (aarch64_init_pauth_hint_builtins): New.
6788 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
6789 (aarch64_expand_builtin): Expand new builtins.
6791 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6793 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
6794 * combine-stack-adj.c (no_unhandled_cfa): Handle
6795 REG_CFA_TOGGLE_RA_MANGLE.
6796 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
6797 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
6798 info for return address signing.
6799 (aarch64_expand_epilogue): Likewise.
6801 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6803 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
6804 * config/aarch64/aarch64-protos.h
6805 (aarch64_return_address_signing_enabled): New declaration.
6806 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
6808 (aarch64_expand_prologue): Sign return address before it's pushed onto
6810 (aarch64_expand_epilogue): Authenticate return address fetched from
6812 (aarch64_override_options): Sanity check for ILP32 and ISA level.
6813 (aarch64_attributes): New function attributes for "sign-return-address".
6814 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
6815 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
6816 ("*do_return"): Generate combined instructions according to key index.
6817 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
6818 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
6820 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
6821 * config/aarch64/aarch64.opt (msign-return-address=): New.
6822 * doc/extend.texi (AArch64 Function Attributes): Documents
6823 "sign-return-address=".
6824 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
6826 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6828 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
6829 overall option summary.
6831 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6833 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
6834 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
6835 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
6836 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
6838 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
6840 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
6841 -mpower9-minmax by default for -mcpu=power9.
6842 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
6843 128-bit floating point.
6845 2017-01-20 Alan Modra <amodra@gmail.com>
6847 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
6848 optimizing for size.
6850 2017-01-20 Alan Modra <amodra@gmail.com>
6853 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
6854 for strcmp and strncmp from corresponding builtin decl.
6856 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6858 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
6859 instead of i386/rtems-64.h.
6860 * config/i386/rtems-64.h: Remove.
6862 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6866 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
6868 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
6870 2017-01-19 Tamar Christina <tamar.christina@arm.com>
6872 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6873 Change int to HOST_WIDE_INT.
6874 * config/aarch64/aarch64-protos.h
6875 (aarch64_simd_gen_const_vector_dup): Likewise.
6876 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
6878 2017-01-19 David Malcolm <dmalcolm@redhat.com>
6880 * langhooks-def.h (lhd_type_for_size): New decl.
6881 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
6882 * langhooks.c (lhd_type_for_size): New function, taken from
6885 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
6887 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
6888 define_bypass for CR latency.
6889 (power9-cracked-alu): Update bypass latency and remove power9-branch.
6890 (power9-alu2): Add define_bypass for CR latency.
6892 (power9-mul): Update insn latency.
6893 (power9-mul-compare): Update insn latency, bypass latency and remove
6896 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6898 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
6900 * config/aarch64/aarch64.md
6901 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
6902 aarch64_nopcrelative_literal_loads.
6903 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
6905 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
6907 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
6909 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
6911 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
6914 * config.gcc (supported_defaults): Add lxc1-sxc1.
6915 (with_lxc1_sxc1): Add validation.
6916 (all_defaults): Add lxc1-sxc1.
6917 * config/mips/mips.opt (mlxc1-sxc1): New option.
6918 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6920 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6921 __mips_no_lxc1_sxc1.
6922 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
6923 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
6924 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
6926 2017-01-19 Richard Biener <rguenther@suse.de>
6928 PR tree-optimization/72488
6929 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
6930 sure to restore SSA info.
6931 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
6933 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
6935 PR rtl-optimization/79121
6936 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
6937 of the inner type when shifting an extended value.
6939 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6942 * symtab.c (symtab_node::equal_address_to): Fix comparing of
6943 interposable aliases.
6945 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
6948 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
6949 Use the evmergelohi instruction.
6950 (mov_si<mode>_e500_subreg4_2_le): Likewise.
6951 (mov_sitf_e500_subreg8_2_be): Likewise.
6952 (mov_sitf_e500_subreg12_2_le): Likewise.
6953 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
6954 (mov_si<mode>_e500_subreg4_2_be): Likewise.
6955 (mov_sitf_e500_subreg8_2_le): Likewise.
6956 (mov_sitf_e500_subreg12_2_be): Likewise.
6958 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6960 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
6961 attribute from vecsimple to vecperm.
6962 (altivec_vbpermq2): Likewise.
6964 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6967 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
6969 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6970 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
6971 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
6972 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
6973 case where N arg is SIZE_MAX.
6974 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
6975 (cmpstrsi): Add pattern.
6977 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6979 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6980 __builtin_vec_revb builtins.
6981 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
6982 built-in functions to support generation of the ISA 3.0 XXBR<x>
6983 vector byte reverse instructions.
6984 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
6985 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
6986 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
6987 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
6988 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
6989 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
6990 (P9V_BUILTIN_VEC_REVB): Likewise.
6991 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
6992 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
6993 (p9_xxbrq_v16qi): Likewise.
6994 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
6995 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
6996 (p9_xxbrh_v8hi): Likewise.
6997 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
6998 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
6999 vec_revb built-in functions.
7001 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
7003 PR rtl-optimization/78952
7004 * config/i386/i386.md (any_extract): New code iterator.
7005 (*insvqi_2): Use any_extract for source operand.
7006 (*insvqi_3): Use any_shiftrt for source operand.
7008 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
7010 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
7012 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
7014 2017-01-18 Matthias Klose <doko@ubuntu.com>
7016 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
7018 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7020 * config/rs6000/altivec.h (vec_bperm): Change #define.
7021 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
7022 (altivec_vbpermq2): New define_insn.
7023 (altivec_vbpermd): Likewise.
7024 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
7026 (VBPERMD): Likewise.
7027 (VBPERM): New polymorphic function interface.
7028 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
7029 Add entries for P9V_BUILTIN_VEC_VBPERM.
7030 * doc/extend.texi: Add interfaces for vec_bperm.
7032 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7034 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
7035 first letter of error messages.
7036 (s390_resolve_overloaded_builtin): Likewise.
7037 * config/s390/s390.c (s390_expand_builtin): Likewise.
7038 (s390_invalid_arg_for_unprototyped_fn): Likewise.
7039 (s390_valid_target_attribute_inner_p): Likewise.
7040 * config/s390/s390.md ("tabort"): Likewise.
7042 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
7044 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
7045 (ISA_AVOID_DIV_HILO): New macro.
7046 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
7047 (ISA_HAS_DDIV): Likewise.
7049 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7051 * doc/invoke.texi (fabi-version): Correct number of occurrences.
7053 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7055 * doc/invoke.texi (fabi-version): Spelling fix.
7057 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7060 * doc/invoke.texi (fabi-version): Mention mangling fix for
7063 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
7066 * doc/invoke.texi (fabi-version): Document discriminator mangling.
7068 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
7071 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
7072 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
7074 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
7076 (stack_protect_test): Ditto.
7077 * config/rs6000/rs6000.opt (mstack-protector-guard=,
7078 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
7080 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
7081 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
7082 -mstack-protector-guard-offset=.
7083 (RS/6000 and PowerPC Options): Ditto.
7085 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
7087 * config/i386/i386.h (MASK_CLASS_P): New define.
7088 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
7089 there are no registers from different register sets also when
7090 mask registers are used. Update function comment.
7091 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
7092 to (*k/*r) and (*k/*km) alternatives.
7094 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
7096 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
7097 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
7098 (EH_RETURN_HANDLER_RTX): New define.
7099 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
7100 Force frame pointer in EH return functions.
7101 (aarch64_expand_epilogue): Add barrier for eh_return.
7102 (aarch64_final_eh_return_addr): Remove.
7103 (aarch64_eh_return_handler_rtx): New function.
7104 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
7106 (aarch64_eh_return_handler_rtx): New prototype.
7108 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7110 * config/rs6000/altivec.h (vec_rlmi): New #define.
7111 (vec_vrlnm): Likewise.
7112 (vec_rlnm): Likewise.
7113 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
7114 (UNSPEC_VRLNM): Likewise.
7115 (VIlong): New mode iterator.
7116 (altivec_vrl<VI_char>mi): New define_insn.
7117 (altivec_vrl<VI_char>nm): Likewise.
7118 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
7121 (RLNM): New polymorphic function entry.
7122 (VRLWMI): New monomorphic function entry.
7124 (RLMI): New polymorphic function entry.
7125 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
7126 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
7127 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
7130 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7133 * dwarf2out.c (field_byte_offset): Restore the
7134 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
7135 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
7136 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
7139 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
7142 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
7144 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7147 * dwarf2out.c (add_data_member_location_attribute): For constant
7148 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
7149 instead of DW_AT_data_member_location, DW_AT_bit_offset and
7150 DW_AT_byte_size attributes.
7152 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
7154 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
7155 after forcing to constant memory when the code model is medium.
7157 2017-01-17 Julia Koval <julia.koval@intel.com>
7160 * config/i386/avx512fintrin.h
7161 (_mm512_i32gather_ps): Change __addr type to void const*.
7162 (_mm512_mask_i32gather_ps): Ditto.
7163 (_mm512_i32gather_pd): Ditto.
7164 (_mm512_mask_i32gather_pd): Ditto.
7165 (_mm512_i64gather_ps): Ditto.
7166 (_mm512_mask_i64gather_ps): Ditto.
7167 (_mm512_i64gather_pd): Ditto.
7168 (_mm512_mask_i64gather_pd): Ditto.
7169 (_mm512_i32gather_epi32): Ditto.
7170 (_mm512_mask_i32gather_epi32): Ditto.
7171 (_mm512_i32gather_epi64): Ditto.
7172 (_mm512_mask_i32gather_epi64): Ditto.
7173 (_mm512_i64gather_epi32): Ditto.
7174 (_mm512_mask_i64gather_epi32): Ditto.
7175 (_mm512_i64gather_epi64): Ditto.
7176 (_mm512_mask_i64gather_epi64): Ditto.
7177 (_mm512_i32scatter_ps): Change __addr type to void*.
7178 (_mm512_mask_i32scatter_ps): Ditto.
7179 (_mm512_i32scatter_pd): Ditto.
7180 (_mm512_mask_i32scatter_pd): Ditto.
7181 (_mm512_i64scatter_ps): Ditto.
7182 (_mm512_mask_i64scatter_ps): Ditto.
7183 (_mm512_i64scatter_pd): Ditto.
7184 (_mm512_mask_i64scatter_pd): Ditto.
7185 (_mm512_i32scatter_epi32): Ditto.
7186 (_mm512_mask_i32scatter_epi32): Ditto.
7187 (_mm512_i32scatter_epi64): Ditto.
7188 (_mm512_mask_i32scatter_epi64): Ditto.
7189 (_mm512_i64scatter_epi32): Ditto.
7190 (_mm512_mask_i64scatter_epi32): Ditto.
7191 (_mm512_i64scatter_epi64): Ditto.
7192 (_mm512_mask_i64scatter_epi64): Ditto.
7193 * config/i386/avx512pfintrin.h
7194 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
7195 (_mm512_mask_prefetch_i32gather_ps): Ditto.
7196 (_mm512_mask_prefetch_i64gather_pd): Ditto.
7197 (_mm512_mask_prefetch_i64gather_ps): Ditto.
7198 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
7199 (_mm512_prefetch_i32scatter_ps): Ditto.
7200 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
7201 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
7202 (_mm512_prefetch_i64scatter_pd): Ditto.
7203 (_mm512_prefetch_i64scatter_ps): Ditto.
7204 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
7205 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
7206 * config/i386/avx512vlintrin.h
7207 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
7208 (_mm_mmask_i32gather_ps): Ditto.
7209 (_mm256_mmask_i32gather_pd): Ditto.
7210 (_mm_mmask_i32gather_pd): Ditto.
7211 (_mm256_mmask_i64gather_ps): Ditto.
7212 (_mm_mmask_i64gather_ps): Ditto.
7213 (_mm256_mmask_i64gather_pd): Ditto.
7214 (_mm_mmask_i64gather_pd): Ditto.
7215 (_mm256_mmask_i32gather_epi32): Ditto.
7216 (_mm_mmask_i32gather_epi32): Ditto.
7217 (_mm256_mmask_i32gather_epi64): Ditto.
7218 (_mm_mmask_i32gather_epi64): Ditto.
7219 (_mm256_mmask_i64gather_epi32): Ditto.
7220 (_mm_mmask_i64gather_epi32): Ditto.
7221 (_mm256_mmask_i64gather_epi64): Ditto.
7222 (_mm_mmask_i64gather_epi64): Ditto.
7223 (_mm256_i32scatter_ps): Change __addr type to void*.
7224 (_mm256_mask_i32scatter_ps): Ditto.
7225 (_mm_i32scatter_ps): Ditto.
7226 (_mm_mask_i32scatter_ps): Ditto.
7227 (_mm256_i32scatter_pd): Ditto.
7228 (_mm256_mask_i32scatter_pd): Ditto.
7229 (_mm_i32scatter_pd): Ditto.
7230 (_mm_mask_i32scatter_pd): Ditto.
7231 (_mm256_i64scatter_ps): Ditto.
7232 (_mm256_mask_i64scatter_ps): Ditto.
7233 (_mm_i64scatter_ps): Ditto.
7234 (_mm_mask_i64scatter_ps): Ditto.
7235 (_mm256_i64scatter_pd): Ditto.
7236 (_mm256_mask_i64scatter_pd): Ditto.
7237 (_mm_i64scatter_pd): Ditto.
7238 (_mm_mask_i64scatter_pd): Ditto.
7239 (_mm256_i32scatter_epi32): Ditto.
7240 (_mm256_mask_i32scatter_epi32): Ditto.
7241 (_mm_i32scatter_epi32): Ditto.
7242 (_mm_mask_i32scatter_epi32): Ditto.
7243 (_mm256_i32scatter_epi64): Ditto.
7244 (_mm256_mask_i32scatter_epi64): Ditto.
7245 (_mm_i32scatter_epi64): Ditto.
7246 (_mm_mask_i32scatter_epi64): Ditto.
7247 (_mm256_i64scatter_epi32): Ditto.
7248 (_mm256_mask_i64scatter_epi32): Ditto.
7249 (_mm_i64scatter_epi32): Ditto.
7250 (_mm_mask_i64scatter_epi32): Ditto.
7251 (_mm256_i64scatter_epi64): Ditto.
7252 (_mm256_mask_i64scatter_epi64): Ditto.
7253 (_mm_i64scatter_epi64): Ditto.
7254 (_mm_mask_i64scatter_epi64): Ditto.
7255 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
7256 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
7257 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
7258 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
7259 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
7260 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
7261 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
7262 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
7263 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
7264 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
7265 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
7266 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
7267 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
7268 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
7269 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
7270 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
7271 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
7272 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
7273 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
7274 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
7275 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
7276 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
7277 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
7278 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
7279 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
7280 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
7281 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
7282 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
7283 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
7284 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
7285 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
7286 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
7287 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
7288 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
7289 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
7290 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
7291 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
7292 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
7293 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
7294 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
7295 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
7296 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
7297 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
7298 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
7299 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
7300 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
7301 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
7302 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
7303 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
7304 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
7305 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
7306 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
7307 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
7308 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
7309 definitions accordingly.
7311 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
7312 Kuan-Lin Chen <kuanlinchentw@gmail.com>
7315 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
7318 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
7321 * ira-conflicts.c (ira_build_conflicts): Update total conflict
7322 hard regs for inner regno.
7324 2017-01-17 Martin Liska <mliska@suse.cz>
7327 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
7328 assumption and add comment.
7330 2017-01-17 Nathan Sidwell <nathan@acm.org>
7332 * ipa-visibility.c (localize_node): New function, broken out of ...
7333 (function_and_variable_visibility): ... here. Call it.
7335 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
7338 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
7339 correctly set frequency of oudgoing edge.
7340 (duplicate_thread_path): Fix profile updating.
7342 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7345 * configure.ac: Add GCC_BASE_VER.
7346 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
7347 version from BASE-VER file.
7348 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
7349 (gcc.o): Depend on $(BASEVER).
7350 * common.opt (dumpfullversion): New option.
7351 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
7352 * doc/invoke.texi: Document -dumpfullversion.
7353 * doc/install.texi: Document --with-gcc-major-version-only.
7354 * configure: Regenerated.
7356 2017-01-17 Richard Biener <rguenther@suse.de>
7358 PR tree-optimization/71433
7359 * tree-vrp.c (register_new_assert_for): Merge same asserts
7360 on all incoming edges.
7361 (process_assert_insertions_for): Handle insertions at the
7364 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
7366 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
7367 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
7369 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
7372 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
7375 2017-01-17 Alan Modra <amodra@gmail.com>
7378 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
7379 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
7380 symbolic stack limit when pic.
7382 2017-01-16 Martin Sebor <msebor@redhat.com>
7384 PR tree-optimization/78608
7385 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
7387 2017-01-16 Jeff Law <law@redhat.com>
7390 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
7391 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
7392 for several include directories that may be relative to sysroot.
7393 * config/i386/x-mingw32 (gplus_includedir): Define.
7394 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
7395 (native_system_includedir): Likewise.
7396 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
7397 override if TARGET_SYSTEM_ROOT is defined.
7398 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
7400 PR tree-optimization/79090
7401 PR tree-optimization/33562
7402 PR tree-optimization/61912
7403 PR tree-optimization/77485
7404 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
7405 and computed trims into the dump file.
7407 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
7409 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
7411 2017-01-16 Jakub Jelinek <jakub@redhat.com>
7414 * gimplify.c (gimplify_init_constructor): If want_value and
7415 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
7419 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
7420 sequence. Formatting fixes.
7421 (doloop_optimize): Formatting fixes.
7424 * gcc.c (debug_level_greater_than_spec_func): New function.
7425 (static_spec_functions): Add debug-level-gt spec function.
7426 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
7428 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
7429 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
7430 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
7431 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
7432 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
7433 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
7435 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
7437 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
7438 QImode fixups to general and mask registers only.
7440 2017-01-16 Carl Love <cel@us.ibm.com>
7442 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
7443 for built-in functions
7444 vector signed char vec_nabs (vector signed char)
7445 vector signed short vec_nabs (vector signed short)
7446 vector signed int vec_nabs (vector signed int)
7447 vector signed long long vec_nabs (vector signed long long)
7448 vector float vec_nabs (vector float)
7449 vector double vec_nabs (vector double)
7450 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
7452 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
7453 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
7454 * doc/extend.texi: Update the documentation file for the new built-in
7457 2017-01-16 Martin Sebor <msebor@redhat.com>
7459 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
7462 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7464 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
7465 UNSPEC_VSX__XXSPLTD to require special splat handling.
7467 2017-01-16 David Malcolm <dmalcolm@redhat.com>
7470 * system.h: Poison strndup.
7472 2017-01-16 Alan Modra <amodra@gmail.com>
7475 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
7478 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
7480 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
7482 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
7484 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
7485 call recog here. Assert that INSN_CODE (insn) is non-negative.
7487 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7490 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
7492 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
7493 in the currently scheduled RTL fragment.
7495 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7497 PR rtl-optimization/78751
7498 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
7501 2017-01-14 Jeff Law <law@redhat.com>
7503 PR tree-optimization/79090
7504 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
7505 variable length stores.
7506 (compute_trims): Delete dead assignment to *trim_tail.
7507 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
7510 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
7512 PR rtl-optimization/78626
7513 PR rtl-optimization/78727
7514 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
7515 of a block, and split such blocks after everything else is finished.
7517 2017-01-14 Alan Modra <amodra@gmail.com>
7520 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
7521 target legitimate_combined_insn.
7522 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
7523 (rs6000_legitimate_combined_insn): New function.
7524 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
7526 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
7527 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
7528 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
7530 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
7532 * doc/frontends.texi (G++ and GCC): Remove references to Java.
7534 2017-01-13 Jeff Law <law@redhat.com>
7536 PR tree-optimization/33562
7537 PR tree-optimization/61912
7538 PR tree-optimization/77485
7539 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
7541 (delete_dead_assignment): Likewise.
7542 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
7543 statement to delete_dead_call and delete_dead_assignment.
7545 2017-01-13 David Malcolm <dmalcolm@redhat.com>
7548 * substring-locations.c (format_warning_va): Strengthen case 1 so
7549 that both endpoints of the substring must be within the format
7550 range for just the substring to be printed.
7552 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
7554 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
7555 * config/i386/i386.c (ix86_target_string): Add missing options
7556 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
7557 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
7558 flags_other and ix86_target_other to flags2_other. Display unknown
7560 (ix86_valid_target_attribute_inner_p): Add missing options and
7561 reorder options by implied ISAs, as in ix86_target_string.
7563 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7565 * hash-table.h (hash_table::too_empty_p): New function.
7566 (hash_table::expand): Use it.
7567 (hash_table::traverse): Likewise.
7568 (hash_table::empty_slot): Use sizeof (value_type) instead of
7569 sizeof (PTR) to convert bytes to elements. Shrink the table
7570 if the current size is excessive for the current number of
7573 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7575 * ira-costs.c (record_reg_classes): Break from the inner loop
7576 early once alt_fail is known to be true. Update outer loop
7577 handling accordingly.
7579 2017-01-13 Jeff Law <law@redhat.com>
7581 * tree-ssa-dse.c (decrement_count): New function.
7582 (increment_start_addr, maybe_trim_memstar_call): Likewise.
7583 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
7584 when we know the partially dead statement is a mem* function.
7586 PR tree-optimization/61912
7587 PR tree-optimization/77485
7588 * tree-ssa-dse.c: Include expr.h.
7589 (maybe_trim_constructor_store): New function.
7590 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
7592 PR tree-optimization/33562
7593 PR tree-optimization/61912
7594 PR tree-optimization/77485
7595 * doc/invoke.texi: Document new dse-max-object-size param.
7596 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
7597 * tree-ssa-dse.c: Include params.h.
7598 (dse_store_status): New enum.
7599 (initialize_ao_ref_for_dse): New, partially extracted from
7601 (valid_ao_ref_for_dse, normalize_ref): New.
7602 (setup_live_bytes_from_ref, compute_trims): Likewise.
7603 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
7604 (maybe_trim_partially_dead_store): Likewise.
7605 (maybe_trim_complex_store): Likewise.
7606 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
7607 Track what bytes live from the original store. Return tri-state
7608 for dead, partially dead or live.
7609 (dse_dom_walker): Add constructor, destructor and new private members.
7610 (delete_dead_call, delete_dead_assignment): New extracted from
7612 (dse_optimize_stmt): Make a member of dse_dom_walker.
7613 Use initialize_ao_ref_for_dse.
7615 PR tree-optimization/33562
7616 PR tree-optimization/61912
7617 PR tree-optimization/77485
7618 * sbitmap.h (bitmap_count_bits): Prototype.
7619 (bitmap_clear_range, bitmap_set_range): Likewise.
7620 * sbitmap.c (bitmap_clear_range): New function.
7621 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
7623 2017-01-13 Martin Liska <mliska@suse.cz>
7626 * function.c (set_cfun): Add new argument force.
7627 * function.h (set_cfun): Likewise.
7628 * ipa-inline-transform.c (inline_call): Use the function when
7629 strict alising from is dropped for function we inline to.
7631 2017-01-13 Richard Biener <rguenther@suse.de>
7633 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
7634 for dumping GIMPLE INTEGER_CSTs.
7636 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7638 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
7639 to 201112L since C++17.
7641 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7644 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
7645 if -fsanitize=kernel-address is present.
7647 2017-01-13 Richard Biener <rguenther@suse.de>
7649 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
7650 as _Literal ( type ) number in case usual suffixes do not
7651 preserve all information.
7653 2017-01-13 Richard Biener <rguenther@suse.de>
7655 PR tree-optimization/77283
7656 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
7657 and ssa-iterators.h.
7658 (is_feasible_trace): Implement a cost model based on joiner
7661 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7664 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
7665 char or short to __float128/_Float128 directly.
7667 2017-01-12 Martin Sebor <msebor@redhat.com>
7669 to -Wformat-overflow.
7670 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
7671 (min_bytes_remaining): Same.
7672 (get_string_length): Same.
7673 (format_string): Same.
7674 (format_directive): Same.
7676 (pass_sprintf_length::handle_gimple_call): Same.
7678 2017-01-12 Jakub Jelinek <jakub@redhat.com>
7680 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
7681 info.nowrite calls with no lhs that can't throw. Return bool
7682 whether gsi_remove has been called or not.
7683 (pass_sprintf_length::handle_gimple_call): Return bool whether
7684 try_substitute_return_value called gsi_remove. Formatting fix.
7685 (pass_sprintf_length::execute): Don't use gsi_remove if
7686 handle_gimple_call returned true.
7689 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
7690 be removed due to side-effects, don't remove following barrier nor
7691 turn the successor edge into fallthru edge.
7693 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7696 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
7697 element-reversing loads and stores as not swappable.
7699 2017-01-12 Nathan Sidwell <nathan@acm.org>
7700 Nicolai Stange <nicstange@gmail.com>
7702 * combine.c (try_combine): Don't ignore result of overlap checking
7703 loop. Combine overlap & asm check into single loop.
7705 2017-01-12 Richard Biener <rguenther@suse.de>
7707 * tree-pretty-print.c (dump_generic_node): Provide -gimple
7708 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
7710 2017-01-12 Richard Biener <rguenther@suse.de>
7712 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
7713 and TS_TARGET_OPTION directly derive from TS_BASE.
7714 * tree-core.h (tree_optimization_option): Derive from tree_base.
7715 (tree_target_option): Likewise.
7717 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
7719 * config/i386/i386.c (memory_address_length): Increase len
7720 only when rip_relative_addr_p returns false.
7722 2017-01-11 Julia Koval <julia.koval@intel.com>
7724 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
7725 (OPTION_MASK_ISA_SGX_SET): New.
7726 (ix86_handle_option): Handle OPT_msgx.
7727 * config.gcc: Added sgxintrin.h.
7728 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
7729 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
7730 * config/i386/i386.c (ix86_target_string): Add -msgx.
7732 (ix86_option_override_internal): Handle new options.
7733 (ix86_valid_target_attribute_inner_p): Add sgx.
7734 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
7735 * config/i386/i386.opt: Add msgx.
7736 * config/i386/sgxintrin.h: New file.
7737 * config/i386/x86intrin.h: Add sgxintrin.h.
7739 2017-01-11 Jakub Jelinek <jakub@redhat.com>
7742 * fold-const.c (maybe_nonzero_address): Return 1 for function
7744 (tree_single_nonzero_warnv_p): Don't handle function local objects
7748 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
7751 2017-01-11 David Malcolm <dmalcolm@redhat.com>
7754 * opts.c: Include "spellcheck.h"
7755 (struct string_fragment): New struct.
7756 (struct edit_distance_traits<const string_fragment &>): New
7758 (get_closest_sanitizer_option): New function.
7759 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
7761 2017-01-11 Jakub Jelinek <jakub@redhat.com>
7763 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
7765 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
7766 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
7767 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
7768 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
7769 for initial die_offset if dwarf_split_debug_info.
7770 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
7771 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
7773 (output_skeleton_debug_sections): Formatting fix. Use
7774 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
7775 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
7777 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
7779 * config/arm/cortex-a53.md: Add bypasses for
7781 (cortex_a53_r2f): Only use for transfers.
7782 (cortex_a53_f2r): Likewise.
7783 (cortex_a53_r2f_cvt): Add reservation for conversions.
7784 (cortex_a53_f2r_cvt): Likewise.
7786 2017-01-11 Tamar Christina <tamar.christina@arm.com>
7788 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
7789 to all inlined functions, change static to extern.
7791 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
7794 * config/arm/arm.c (legitimize_pic_address): Handle reference to
7796 (arm_assemble_integer): Likewise.
7798 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
7800 * config.gcc: Use new awk script to check CPU, FPU and architecture
7801 parameters for --with-... options.
7802 * config/arm/parsecpu.awk: New file
7803 * config/arm/arm-cpus.in: New file.
7804 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
7806 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
7808 * config/arm/t-arm: Update dependency rules.
7809 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
7810 of processing .def files.
7811 * config/arm/genopt.sh: Deleted.
7812 * config/arm/gentune.sh: Deleted.
7813 * config/arm/arm-cores.def: Deleted.
7814 * config/arm/arm-arches.def: Deleted.
7815 * config/arm/arm-fpus.def: Deleted.
7816 * config/arm/arm-tune.md: Regenerated.
7817 * config/arm/arm-tables.opt: Regenerated.
7818 * config/arm/arm-cpu.h: New generated file.
7819 * config/arm/arm-cpu-data.h: New generated file.
7820 * config/arm/arm-cpu-cdata.h: New generated file.
7822 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
7825 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
7827 (input_varpool_node): Unpack dynamically_initialized bit.
7829 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
7831 PR rtl-optimization/79032
7832 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
7833 the alignment of the adjusted memory reference against that of MODE,
7834 instead of the alignment of the original memory reference.
7836 2017-01-11 Martin Jambor <mjambor@suse.cz>
7838 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
7840 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
7841 decorated functions.
7843 2017-01-11 Richard Biener <rguenther@suse.de>
7845 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
7846 set range/nonnull info for PHI results. Do not set it on
7847 stmts marked for removal.
7849 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
7851 * expr.c (store_field): In the bitfield case, fetch the return value
7852 from the registers before applying a single big-endian adjustment.
7853 Always do a final load for a BLKmode value not larger than a word.
7855 2017-01-10 David Malcolm <dmalcolm@redhat.com>
7858 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7859 that we correctly handle column numbers greater than
7860 LINE_MAP_MAX_COLUMN_NUMBER.
7862 2017-01-10 Martin Sebor <msebor@redhat.com>
7865 * gimple-ssa-sprintf.c (get_destination_size): Call
7866 {init,fini}object_sizes.
7867 * tree-object-size.c (addr_object_size): Adjust.
7868 (pass_through_call): Adjust.
7869 (pass_object_sizes::execute): Adjust.
7870 * tree-object-size.h (fini_object_sizes): Declare.
7872 2017-01-10 Martin Sebor <msebor@redhat.com>
7874 PR tree-optimization/78775
7875 * builtins.c (get_size_range): Move...
7876 * calls.c: ...to here.
7877 (alloc_max_size): Accept zero argument.
7878 (operand_signed_p): Remove.
7879 (maybe_warn_alloc_args_overflow): Call get_size_range.
7880 * calls.h (get_size_range): Declare.
7882 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
7884 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
7885 from TI's devices.csv file as of September 2016.
7886 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
7888 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
7890 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
7891 * doc/invoke.texi: Likewise.
7892 * doc/md.texi: Likewise.
7893 * doc/objc.texi: Likewise.
7895 2017-01-10 Joshua Conner <joshconner@google.com>
7897 * config/arm/fuchsia-elf.h: New file.
7898 * config/fuchsia.h: New file.
7899 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
7900 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
7902 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
7904 2016-01-10 Richard Biener <rguenther@suse.de>
7906 PR tree-optimization/79034
7907 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
7908 Propagate out degenerate PHIs in the joiner.
7910 2017-01-10 Martin Liska <mliska@suse.cz>
7912 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
7913 (sort_congruence_classes_by_decl_uid): Likewise.
7914 (sort_congruence_class_groups_by_decl_uid): Likewise.
7915 (sem_item_optimizer::merge_classes): Sort class, groups in these
7916 classes and members in the groups by DECL_UID of declarations.
7917 This would make merge operations stable.
7919 2017-01-10 Martin Liska <mliska@suse.cz>
7921 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
7922 usage of m_classes_vec.
7923 (sem_item_optimizer::~sem_item_optimizer): Likewise.
7924 (sem_item_optimizer::get_group_by_hash): Likewise.
7925 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
7926 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
7927 (sem_item_optimizer::verify_classes): Likewise.
7928 (sem_item_optimizer::process_cong_reduction): Likewise.
7929 (sem_item_optimizer::dump_cong_classes): Likewise.
7930 (sem_item_optimizer::merge_classes): Likewise.
7931 * ipa-icf.h (congruence_class_hash): Rename from
7932 congruence_class_group_hash. Remove declaration of m_classes_vec.
7934 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
7936 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
7937 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
7938 * config.gcc: Add avx512vpopcntdqintrin.h.
7939 * config/i386/avx512vpopcntdqintrin.h: New.
7940 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
7941 * config/i386/i386-builtin-types.def: Add new types.
7942 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
7943 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
7944 __builtin_ia32_vpopcountq_v8di_mask): New.
7945 * config/i386/i386-c.c (ix86_target_macros_internal): Define
7946 __AVX512VPOPCNTDQ__.
7947 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
7948 (PTA_AVX512VPOPCNTDQ): Define.
7949 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
7950 TARGET_AVX512VPOPCNTDQ_P): Define.
7951 * config/i386/i386.opt: Add mavx512vpopcntdq.
7952 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
7953 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
7955 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7958 * predict.def (PRED_CALL): Set to 67.
7960 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
7962 * expr.c (store_field): In the bitfield case, if the value comes from
7963 a function call and is of an aggregate type returned in registers, do
7964 not modify the field mode; extract the value in all cases if the mode
7965 is BLKmode and the size is not larger than a word.
7967 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
7970 * config/i386/cpuid.h: Fix undefined behavior.
7972 2017-01-04 Jeff Law <law@redhat.com>
7974 PR tree-optimization/79007
7975 PR tree-optimization/67955
7976 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
7977 conservative for pt.null when flag_non_call_exceptions is on.
7979 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7981 PR translation/79019
7982 PR translation/79020
7983 * params.def (PARAM_INLINE_MIN_SPEEDUP,
7984 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
7985 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
7987 * config/avr/avr.opt (maccumulate-args): Likewise.
7988 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
7989 * common.opt (freport-bug): Likewise.
7990 * cif-code.def (CIF_FINAL_ERROR): Likewise.
7991 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
7992 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
7993 translatable string.
7994 * config/i386/i386.c (function_value_32): Likewise.
7995 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
7996 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
7998 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
7999 * common/config/msp430/msp430-common.c (msp430_handle_option):
8001 * symtab.c (symtab_node::verify_base): Likewise.
8002 * opts.c (set_debug_level): Likewise.
8003 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
8004 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
8005 missing whitespace to translatable strings.
8006 * config/avr/avr.md (bswapsi2): Fix typo in comment.
8007 * config/sh/superh.h: Likewise.
8008 * config/i386/xopintrin.h: Likewise.
8009 * config/i386/znver1.md: Likewise.
8010 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
8011 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
8012 * double-int.h (struct double_int): Likewise.
8013 * double-int.c (div_and_round_double): Likewise.
8014 * wide-int.cc: Likewise.
8015 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
8016 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
8017 * cfgcleanup.c (crossjumps_occured): Renamed to ...
8018 (crossjumps_occurred): ... this.
8019 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
8022 PR tree-optimization/78899
8023 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
8024 returning bool return struct loop *, NULL for failure and the new
8026 (versionable_outer_loop_p): Don't version outer loop if it has
8027 dont_vectorized bit set.
8028 (tree_if_conversion): When versioning outer loop, ensure
8029 tree_if_conversion is performed also on the inner loop of the
8030 non-vectorizable outer loop copy.
8031 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
8032 LOOP_VECTORIZED in inner loop of the scalar outer loop and
8033 prevent vectorization of it.
8034 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
8035 the outer loop vectorization of the non-scalar version is attempted
8036 before vectorization of the inner loop in scalar version. If
8037 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
8038 vectorization of its inner loop.
8039 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
8040 has 2 inner loops, rename also on edges from bb whose single pred
8041 is outer_loop->header. Fix typo in function comment.
8043 2017-01-09 Martin Sebor <msebor@redhat.com>
8046 * asan.c (asan_emit_stack_protection): Increase local buffer size
8047 to avoid snprintf truncation warning.
8049 2017-01-09 Andrew Pinski <apinski@cavium.com>
8051 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
8052 to reference thunderx2t99 for the tuning structure
8053 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
8055 (thunderx2t99_extra_costs): This.
8056 * config/aarch64/aarch64-tune.md: Regenerate.
8057 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
8058 (vulcan_addrcost_table): This.
8059 (vulcan_regmove_cost): Rename to ...
8060 (thunderx2t99_regmove_cost): This.
8061 (vulcan_vector_cost): Rename to ...
8062 (thunderx2t99_vector_cost): this.
8063 (vulcan_branch_cost): Rename to ...
8064 (thunderx2t99_branch_cost): This.
8065 (vulcan_tunings): Rename to ...
8066 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
8067 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
8069 2017-01-09 Martin Jambor <mjambor@suse.cz>
8073 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
8074 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
8075 (propagate_vr_accross_jump_function): Use the above function for all
8076 value range computations for pass-through jump functions and type
8077 converasion from explicit value range values.
8078 (ipcp_propagate_stage): Do not attempt to deduce types of formal
8079 parameters from TYPE_ARG_TYPES.
8080 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
8081 (ipa_write_node_info): Stream type of the actual argument.
8082 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
8084 2017-01-09 Martin Liska <mliska@suse.cz>
8087 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
8088 (lookup_compiler): Do not show error message with have_E.
8090 2017-01-09 Jakub Jelinek <jakub@redhat.com>
8092 PR tree-optimization/78938
8093 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
8094 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
8095 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
8096 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
8099 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8101 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
8104 2017-01-09 Richard Biener <rguenther@suse.de>
8106 PR tree-optimization/78997
8107 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
8108 name condition properly.
8110 2017-01-09 Richard Biener <rguenther@suse.de>
8113 * dwarf2out.c (is_cxx): New overload with context.
8114 (is_naming_typedef_decl): Use it.
8116 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8118 * invoke.texi (Option Summary): Correct spacing in option lists
8119 and add line breaks to fix over-long lines.
8121 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8125 * extend.texi (Common Variable Attributes): Add xref to GCC
8126 Internals manual to explain mode attribute keywords.
8128 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8131 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
8132 and Preprocessor Options.
8133 (Options for Linking): Document -pthread here....
8134 (RS/6000 and PowerPC Options): ...not here.
8135 (Solaris 2 Options): ...or here.
8136 * doc/cppopts.texi: Document -pthread.
8138 2017-01-08 Martin Sebor <msebor@redhat.com>
8141 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
8142 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
8143 New member functions.
8144 (format_directive): Used them.
8146 (pass_sprintf_length::handle_gimple_call): Same.
8147 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
8148 to avoid truncation for any argument.
8149 (extract_affine_mul): Same.
8150 * tree.c (get_file_function_name): Same.
8152 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8155 * predict.def (PRED_INDIR_CALL): Set to 86.
8157 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8159 PR preprocessor/54124
8160 * doc/cppopts.texi: Reformat -d subtable to list the full name
8161 of the options. Add cross-reference to the docs for the general
8162 compiler -d options.
8163 * doc/invoke.texi (Developer Options): Add cross-reference to the
8164 preprocessor-specific -d option documentation.
8166 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8168 PR preprocessor/13498
8169 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
8170 redudant material, and reflect new command-line options.
8171 (System Headers): Likewise.
8173 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8175 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
8176 -isystem, and -idirafter. Copy-edit.
8177 * doc/cppopts.texi: Copy-edit. Remove contradiction about
8178 default for -ftrack-macro-expansion. Delete obsolete and
8179 badly-formatted implementation details about -fdebug-cpp output.
8180 * doc/cppwarnopts.texi: Copy-edit.
8182 2017-01-07 David Malcolm <dmalcolm@redhat.com>
8185 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
8186 that the transition from a max line width >= 1<<10 to narrower
8187 lines works correctly.
8189 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
8191 * doc/options.texi (PerFunction): New.
8192 * opt-functions.awk (switch_flags): Map both Optimization and
8193 PerFunction to CL_OPTIMIZATION.
8194 * opth-gen.awk: Test for PerFunction flag along with
8196 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
8197 it only when the latter is present. Skip those that don't in
8198 the hash function generator.
8199 * common.opt (fvar-tracking): Mark as PerFunction instead of
8201 (fvar-tracking-assignments): Likewise.
8202 (fvar-tracking-assignments-toggle): Likewise.
8203 (fvar-tracking-uninit): Likewise.
8205 2017-01-07 Jakub Jelinek <jakub@redhat.com>
8207 PR translation/79018
8208 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
8211 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
8214 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
8215 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
8216 TARGET_LONG_JUMP_TABLE_OFFSETS.
8217 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
8218 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
8219 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
8220 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
8221 * config/m68k/m68k.md (tablejump expander): Likewise.
8222 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
8223 TARGET_LONG_JUMP_TABLE_OFFSETS.
8224 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
8225 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
8227 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
8228 David Holsgrove <david.holsgrove@xilinx.com>
8230 * common/config/microblaze/microblaze-common.c
8231 (TARGET_EXCEPT_UNWIND_INFO): Remove.
8232 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
8234 * config/microblaze/microblaze.c (microblaze_must_save_register)
8235 (microblaze_expand_epilogue, microblaze_return_addr): Handle
8237 (microblaze_eh_return): New function.
8238 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
8239 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
8240 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
8241 * config/microblaze/microblaze.md (eh_return): New pattern.
8243 2017-01-06 Jakub Jelinek <jakub@redhat.com>
8245 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
8246 GCC_DIAGNOSTIC_STRINGIFY): Define.
8248 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
8250 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8252 * config/arm/arm.md (<mcrr>): New.
8254 * config/arm/arm.c (arm_arch5te): New.
8255 (arm_option_override): Set arm_arch5te.
8256 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
8258 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
8259 (arm_mcrr_qualifiers): ... this. New.
8260 (MRRC_QUALIFIERS): Define to...
8261 (arm_mrrc_qualifiers): ... this. New.
8262 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
8264 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
8265 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
8266 (MRRCI, mrrc, MRRC): New.
8267 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
8268 VUNSPEC_MRRC2): New.
8270 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8272 * config/arm/arm.md (<mcr>): New.
8274 * config/arm/arm.c (arm_coproc_builtin_available): Add
8275 support for mcr, mrc, mcr2 and mrc2.
8276 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
8277 (arm_mcr_qualifiers): ... this. New.
8278 (MRC_QUALIFIERS): Define to ...
8279 (arm_mrc_qualifiers): ... this. New.
8280 (MCR_QUALIFIERS): Define to ...
8281 (arm_mcr_qualifiers): ... this. New.
8282 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
8284 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
8285 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
8286 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
8289 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8291 * config/arm/arm.md (*ldc): New.
8295 * config/arm/arm.c (arm_coproc_builtin_available): Add
8296 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
8297 (arm_coproc_ldc_stc_legitimate_address): New.
8298 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
8299 'qualifier_const_pointer'.
8300 (LDC_QUALIFIERS): Define to...
8301 (arm_ldc_qualifiers): ... this. New.
8302 (STC_QUALIFIERS): Define to...
8303 (arm_stc_qualifiers): ... this. New.
8304 * config/arm/arm-protos.h
8305 (arm_coproc_ldc_stc_legitimate_address): New.
8306 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
8307 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
8308 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
8309 stc2, stcl, stc2l): New.
8310 * config/arm/constraints.md (Uz): New.
8311 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
8312 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
8313 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
8314 VUNSPEC_STC2L): New.
8316 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8318 * config/arm/arm.md (<cdp>): New.
8319 * config/arm/arm.c (neon_const_bounds): Rename this ...
8320 (arm_const_bounds): ... this.
8321 (arm_coproc_builtin_available): New.
8322 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
8323 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
8324 (CDP_QUALIFIERS): Define to...
8325 (arm_cdp_qualifiers): ... this. New.
8327 (arm_expand_builtin_args): Add case for 6 arguments.
8328 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
8329 (arm_const_bounds): ... this.
8330 (arm_coproc_builtin_available): New.
8331 * config/arm/arm_acle.h (__arm_cdp): New.
8333 * config/arm/arm_acle_builtins.def (cdp): New.
8335 * config/arm/iterators.md (CDPI,CDP,cdp): New.
8336 * config/arm/neon.md: Rename all 'neon_const_bounds' to
8338 * config/arm/types.md (coproc): New.
8339 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
8340 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
8341 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
8342 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
8344 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8346 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
8347 (UBINOP_QUALIFIERS): New.
8349 (acle_builtin_data): New. Change comment.
8350 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
8351 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
8352 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
8353 arm_acle_builtins.def.
8354 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
8355 (arm_init_acle_builtins): New.
8356 (CRC32_BUILTIN): Remove.
8357 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
8358 crc32cb, crc32ch and crc32cw.
8359 (arm_init_crc32_builtins): Remove.
8360 (arm_init_builtins): Use arm_init_acle_builtins rather
8361 than arm_init_crc32_builtins.
8362 (arm_expand_acle_builtin): New.
8363 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
8364 * config/arm/arm_acle_builtins.def: New.
8366 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8368 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
8369 (arm_builtin_datum): ... this.
8370 (arm_init_neon_builtin): Rename to ...
8371 (arm_init_builtin): ... this. Add a new parameters PREFIX
8372 and USE_SIG_IN_NAME.
8373 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
8374 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
8375 'arm_builtin_datum'.
8376 (arm_init_vfp_builtins): Likewise.
8377 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
8378 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
8379 (arm_expand_neon_args): Rename to ...
8380 (arm_expand_builtin_args): ... this. Rename builtin_arg
8381 enum values and differentiate between ARG_BUILTIN_MEMORY
8382 and ARG_BUILTIN_NEON_MEMORY.
8383 (arm_expand_neon_builtin_1): Rename to ...
8384 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
8385 values, arm_expand_builtin_args and add bool parameter NEON.
8386 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
8387 (arm_expand_vfp_builtin): Likewise.
8388 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
8390 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8393 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
8394 * predict.c (tree_estimate_probability_bb): Reverse direction of
8395 polymorphic call predictor.
8397 2017-01-06 David Malcolm <dmalcolm@redhat.com>
8399 * passes.c (execute_one_pass): Split out pass-skipping logic into...
8400 (determine_pass_name_match): ...this new function and...
8401 (should_skip_pass_p): ...this new function.
8403 2017-01-06 Nathan Sidwell <nathan@acm.org>
8405 * ipa-visibility.c (function_and_variable_visibility): Reformat
8406 comments and long lines. Remove extrneous if.
8407 * symtab.c (symtab_node::make_decl_local): Fix code format.
8408 (symtab_node::set_section_for_node): Fix comment typo.
8410 2017-01-06 Martin Liska <mliska@suse.cz>
8413 * lra-constraints.c: Rename invariant to lra_invariant.
8414 * predict.c (set_even_probabilities): Initialize e to NULL.
8416 2017-01-05 Martin Sebor <msebor@redhat.com>
8418 PR tree-optimization/78910
8419 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
8420 (format_integer): Correct off-by-one error in the handling
8421 of precision with negative numbers in signed conversions..
8423 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
8425 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
8427 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8429 PR tree-optimization/71016
8430 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
8431 factor_out_conditional_conversion. Formatting fix.
8432 (factor_out_conditional_conversion): Add cond_stmt argument.
8433 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
8434 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
8437 2017-01-05 David Malcolm <dmalcolm@redhat.com>
8439 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
8440 read-rtl-function.o, and selftest-rtl.o.
8441 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
8442 (selftest::aarch64_test_loading_full_dump): New function.
8443 (selftest::aarch64_run_selftests): New function.
8444 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
8445 selftest::aarch64_run_selftests.
8446 * config/i386/i386.c
8447 (selftest::ix86_test_loading_dump_fragment_1): New function.
8448 (selftest::ix86_test_loading_call_insn): New function.
8449 (selftest::ix86_test_loading_full_dump): New function.
8450 (selftest::ix86_test_loading_unspec): New function.
8451 (selftest::ix86_run_selftests): Call the new functions.
8452 * emit-rtl.c (maybe_set_max_label_num): New function.
8453 * emit-rtl.h (maybe_set_max_label_num): New decl.
8454 * function.c (instantiate_decls): Guard call to
8455 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
8456 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
8458 * gensupport.c (gen_reader::gen_reader): Pass "false"
8459 for new "compact" param of rtx_reader.
8460 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
8461 rather than an empty string for NULL strings.
8462 * read-md.c: Potentially include config.h rather than bconfig.h.
8463 Wrap include of errors.h with #ifdef GENERATOR_FILE.
8464 (have_error): New global, copied from errors.c.
8465 (md_reader::read_name): Rename to...
8466 (md_reader::read_name_1): ...this, adding "out_loc" param,
8467 and converting "missing name or number" to returning false, rather
8469 (md_reader::read_name): Reimplement in terms of read_name_1.
8470 (md_reader::read_name_or_nil): New function.
8471 (md_reader::read_string): Handle "(nil)" by returning NULL.
8472 (md_reader::md_reader): Add new param "compact".
8473 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
8474 (md_reader::read_file): New method.
8475 * read-md.h (md_reader::md_reader): Add new param "compact".
8476 (md_reader::read_file): New method.
8477 (md_reader::is_compact): New accessor.
8478 (md_reader::read_name): Convert return type from void to file_location.
8479 (md_reader::read_name_or_nil): New decl.
8480 (md_reader::read_name_1): New decl.
8481 (md_reader::m_compact): New field.
8482 (noop_reader::noop_reader): Pass "false" for new "compact" param
8484 (rtx_reader::rtx_reader): Add new "compact" param.
8485 (rtx_reader::read_rtx_operand): Make virtual and convert return
8486 type from void to rtx.
8487 (rtx_reader::read_until): New decl.
8488 (rtx_reader::handle_any_trailing_information): New virtual function.
8489 (rtx_reader::postprocess): New virtual function.
8490 (rtx_reader::finalize_string): New virtual function.
8491 (rtx_reader::m_in_call_function_usage): New field.
8492 (rtx_reader::m_reuse_rtx_by_id): New field.
8493 * read-rtl-function.c: New file.
8494 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
8495 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
8496 (selftest::verify_three_block_rtl_cfg): New decl.
8497 * read-rtl-function.h: New file.
8498 * read-rtl.c: Potentially include config.h rather than bconfig.h.
8499 For host, include function.h, memmodel.h, and emit-rtl.h.
8500 (one_time_initialization): New function.
8501 (struct compact_insn_name): New struct.
8502 (compact_insn_names): New array.
8503 (find_code): Handle insn codes in compact dumps.
8504 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
8505 (bind_subst_iter_and_attr): Likewise.
8506 (add_condition_to_string): Likewise.
8507 (add_condition_to_rtx): Likewise.
8508 (apply_attribute_uses): Likewise.
8509 (add_current_iterators): Likewise.
8510 (apply_iterators): Likewise.
8511 (initialize_iterators): Guard usage of apply_subst_iterator with
8512 #ifdef GENERATOR_FILE.
8513 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
8514 (md_reader::read_mapping): Likewise.
8515 (add_define_attr_for_define_subst): Likewise.
8516 (add_define_subst_attr): Likewise.
8517 (read_subst_mapping): Likewise.
8518 (check_code_iterator): Likewise.
8519 (rtx_reader::read_rtx): Likewise. Move one-time initialization
8521 (one_time_initialization): New function.
8522 (rtx_reader::read_until): New method.
8523 (read_flags): New function.
8524 (parse_reg_note_name): New function.
8525 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
8526 Handle reuse_rtx ids.
8527 Wrap iterator lookup within #ifdef GENERATOR_FILE.
8528 Add parsing support for RTL dumps, mirroring the special-cases in
8529 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
8530 values, and calling handle_any_trailing_information.
8531 (rtx_reader::read_rtx_operand): Convert return type from void
8532 to rtx, returning return_rtx. Handle case 'e'. Call
8533 finalize_string on XSTR and XTMPL fields.
8534 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
8535 "(nil)" values were omitted. Call the postprocess vfunc on the
8537 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
8538 class ctor. Initialize m_in_call_function_usage. Call
8539 one_time_initialization.
8540 * rtl-tests.c (selftest::test_uncond_jump): Call
8541 set_new_first_and_last_insn.
8542 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
8543 * selftest-rtl.c: New file.
8544 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
8545 (selftest::get_insn_by_uid): New decl.
8546 * selftest-run-tests.c (selftest::run_tests): Call
8547 read_rtl_function_c_tests.
8548 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
8549 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
8552 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
8554 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
8555 operands in a special way. Assert that pos+len <= mode precision.
8557 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8559 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
8560 3 argument Alias with unlimited for the negative form.
8561 (fno-vect-cost-model): Removed.
8563 2017-01-05 Martin Liska <mliska@suse.cz>
8565 * hsa-gen.c (gen_hsa_divmod): New function.
8566 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
8568 2017-01-05 Martin Liska <mliska@suse.cz>
8571 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
8574 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8576 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
8577 small constant length operands.
8579 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8581 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
8582 between loop iterations.
8584 2017-01-05 Martin Liska <mliska@suse.cz>
8587 * gimplify.c (gimplify_decl_expr): Compare to
8588 asan_poisoned_variables instread of checking flags.
8589 (gimplify_target_expr): Likewise.
8590 (gimplify_expr): Likewise.
8591 (gimplify_function_tree): Conditionally initialize
8592 asan_poisoned_variables.
8594 2017-01-04 Jeff Law <law@redhat.com>
8596 PR tree-optimizatin/78812
8597 * rtl.h (contains_mem_rtx_p): Prototype.
8598 * ifcvt.c (containts_mem_rtx_p): Move from here to...
8599 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
8600 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
8601 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
8602 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
8604 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8606 * input.c (assert_char_at_range): Default-initialize actual_range.
8608 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8610 * df-scan.c (df_ref_create_structure): Make regno unsigned,
8611 to match the caller.
8613 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8615 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
8616 insns after final jump in test to emit dummy move.
8618 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8620 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
8621 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
8623 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8625 * multiple_target.c (create_dispatcher_calls): Init e_next.
8626 * tree-ssa-loop-split.c (split_loop): Init border.
8627 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
8630 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
8635 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
8636 (altivec_register_operand): Do not return true if the operand
8637 contains a SUBREG mixing SImode and SFmode.
8638 (vsx_register_operand): Likewise.
8639 (vsx_reg_sfsubreg_ok): New predicate.
8640 (vfloat_operand): Do not return true if the operand contains a
8641 SUBREG mixing SImode and SFmode.
8642 (vint_operand): Likewise.
8643 (vlogical_operand): Likewise.
8644 (gpc_reg_operand): Likewise.
8645 (int_reg_operand): Likewise.
8646 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
8647 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
8648 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
8650 (rs6000_emit_move_si_sf_subreg): New helper function.
8651 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
8652 fixup SUBREGs involving SImode and SFmode.
8653 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
8654 numbers for the new peephole2 optimization.
8655 (peephole2 for SFmode unions): New peephole2 to optimize cases in
8656 the GLIBC math library that do AND/IOR/XOR operations on single
8657 precision floating point.
8658 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
8659 target macros to say whether we need to avoid SUBREGs mixing
8661 (TARGET_ALLOW_SF_SUBREG): Likewise.
8662 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
8663 (UNSPEC_SI_FROM_SF): Likewise.
8664 (iorxor): Change spacing.
8665 (and_ior_xor): New iterator for AND, IOR, and XOR.
8666 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
8667 (movdi_from_sf_zero_ext): Likewise.
8668 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
8669 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
8670 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
8671 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
8672 (fms<mode>4): Likewise.
8673 (fnma<mode>4): Likewise.
8674 (fnms<mode>4): Likewise.
8675 (nfma<mode>4): Likewise.
8676 (nfms<mode>4): Likewise.
8678 2017-01-04 Marek Polacek <polacek@redhat.com>
8681 * doc/invoke.texi: Document -Wpointer-compare.
8683 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8685 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
8688 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
8689 descriptions for -gdwarf-5 and emit them as uleb128 instead of
8692 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
8695 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
8696 documentation of the powerpc_popcntb_ok attribute.
8697 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
8698 code to issue warning messages if a requested CPU configuration is
8699 not supported by the binary (assembler and loader) toolchain.
8700 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
8701 made to define a built-in function that has been disabled.
8702 (paired_init_builtins): Add assertion to prevent ICE if attempt is
8703 made to define a built-in function that has been disabled.
8704 (altivec_init_builtins): Add comment explaining why definition
8705 of the DST built-in functions is not preceded by an assertion
8706 check. Add assertions to prevent ICE if attempts are made to
8707 define an altivec predicate or an abs* built-in function that has
8709 (htm_init_builtins): Add comment explaining why definition of the
8710 htm built-in functions is not preceded by an assertion check.
8712 2017-01-04 Jeff Law <law@redhat.com>
8714 PR tree-optimizatin/67955
8715 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
8716 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
8717 the points-to solution does not include pt_null. Use DECL_PT_UID
8720 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
8722 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
8723 Use gen_int_mode instead of gen_lopwart for const_int operands.
8725 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8727 PR tree-optimization/71563
8728 * match.pd: Simplify X << Y into X if Y is known to be 0 or
8729 out of range value - has low bits known to be zero.
8731 2017-01-04 Alan Modra <amodra@gmail.com>
8733 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
8734 * configure: Regenerate.
8735 * config.in: Regenerate.
8737 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8740 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
8741 a substring of the message, but strcmp with the whole message. Ifdef
8742 ENABLE_NLS, translate the message first using dgettext.
8744 2017-01-03 Jeff Law <law@redhat.com>
8746 PR tree-optimizatin/78856
8747 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
8748 (mark_threaded_blocks): Remove code to truncate thread paths that
8749 cross multiple loop headers. Instead invalidate the cached loop
8750 iteration information and handle case of a thread path walking
8751 into an irreducible region.
8753 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8756 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
8757 assertions. Add support for doing the signbit if the IEEE 128-bit
8758 floating point value is in a GPR.
8759 * config/rs6000/rs6000.md (Fsignbit): Delete.
8760 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
8761 Update the length attribute if the value is in a GPR.
8762 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
8763 the sign or zero extension instruction, since the value is always 0/1.
8764 (signbit<mode>2_dm2): Delete using <Fsignbit>.
8767 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
8768 extracting SImode to a GPR register so that we can generate a
8769 store, limit the vector to be in a traditional Altivec register
8770 for the vextuwrx instruction.
8772 2017-01-03 Ian Lance Taylor <iant@google.com>
8774 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
8776 2017-01-03 Martin Sebor <msebor@redhat.com>
8778 PR tree-optimization/78696
8779 * gimple-ssa-sprintf.c (format_floating): Correct handling of
8780 precision. Use MPFR for %f for greater fidelity. Correct handling
8782 (pass_sprintf_length::compute_format_length): Set width and precision
8783 specified by asrerisk to void_node for vararg functions.
8784 (try_substitute_return_value): Adjust dump output.
8786 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
8788 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
8790 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
8792 * doc/invoke.texi (SPARC options): Document -mlra as the default.
8793 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
8794 -mlra/-mno-lra was passed to the compiler.
8796 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
8798 PR rtl-optimization/65618
8799 * emit-rtl.c (try_split): Move initialization of "before" and
8800 "after" to just before the call to emit_insn_after_setloc.
8802 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
8804 * doc/md.texi (Standard Names): Remove reference to Java frontend.
8806 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
8808 * dwarf2out.c (gen_enumeration_type_die): When
8809 -gno-strict-dwarf, add a DW_AT_encoding attribute.
8811 2017-01-03 Jakub Jelinek <jakub@redhat.com>
8813 PR tree-optimization/78965
8814 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
8815 Change first argument from const call_info & to call_info &. For %n
8816 set info.nowrite to false.
8819 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
8820 possibly throwing calls.
8822 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
8823 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
8824 and fns handling, rather than in a separate case SSA_NAME.
8826 2017-01-02 Jeff Law <law@redhat.com>
8828 * config/darwin-driver.c (darwin_driver_init): Const-correctness
8829 fixes for first_period and second_period variables.
8831 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
8834 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
8835 (*insvqi_1): New insn pattern.
8836 (*insvqi_1_mem_rex64): Ditto.
8838 (*insvqi_3): Rename from *insvqi.
8840 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
8842 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
8844 * doc/cfg.texi (Edges): Remove reference to Java.
8845 (Maintaining the CFG): Ditto.
8847 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8850 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
8851 transparent aliases.
8853 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8856 * predict.def (PRED_CALL): Update hitrate.
8857 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
8858 * predict.c (tree_estimate_probability_bb): Split CALL predictor
8859 into direct/indirect/polymorphic variants.
8861 2017-01-01 Jakub Jelinek <jakub@redhat.com>
8863 Update copyright years.
8865 * gcc.c (process_command): Update copyright notice dates.
8866 * gcov-dump.c (print_version): Ditto.
8867 * gcov.c (print_version): Ditto.
8868 * gcov-tool.c (print_version): Ditto.
8869 * gengtype.c (create_file): Ditto.
8870 * doc/cpp.texi: Bump @copying's copyright year.
8871 * doc/cppinternals.texi: Ditto.
8872 * doc/gcc.texi: Ditto.
8873 * doc/gccint.texi: Ditto.
8874 * doc/gcov.texi: Ditto.
8875 * doc/install.texi: Ditto.
8876 * doc/invoke.texi: Ditto.
8878 Copyright (C) 2017 Free Software Foundation, Inc.
8880 Copying and distribution of this file, with or without modification,
8881 are permitted in any medium without royalty provided the copyright
8882 notice and this notice are preserved.