1 2017-04-28 Martin Liska <mliska@suse.cz>
4 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
5 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
6 * gcov.c (read_graph_file): Read just number of blocks.
7 * profile.c (branch_prob): Do not stream 0 flags per a basic
10 2017-04-28 Martin Liska <mliska@suse.cz>
12 * gcov-dump.c (tag_*): Add new argument to declarations.
13 (dump_gcov_file): Likewise.
14 (tag_blocks): Add and use new argument depth.
16 (tag_lines): Likewise.
17 (tag_counters): Likewise.
18 (tag_summary): Likewise.
19 (dump_working_sets): Use depth to do a proper indentation.
21 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25 * symtab.c (symtab_node::debug_symtab): Move definition here.
27 2017-04-28 Richard Biener <rguenther@suse.de>
29 * lto-streamer.h (LTO_major_version): Bump to 7.
31 2017-04-28 Richard Biener <rguenther@suse.de>
33 * tree-vrp.c (assert_info): New struct.
34 (add_assert_info): New helper.
35 (register_edge_assert_for_2): Refactor to add asserts to a vector
37 (register_edge_assert_for_1): Likewise.
38 (register_edge_assert_for): Likewise.
39 (finish_register_edge_assert_for): New helper actually registering
40 asserts where live on edge.
41 (find_conditional_asserts): Adjust.
42 (find_switch_asserts): Likewise.
43 (evrp_dom_walker::try_find_new_range): Generalize.
44 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
46 2017-04-27 Marek Polacek <polacek@redhat.com>
49 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
50 arg10 and arg11 to itype.
52 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
54 * doc/extend.texi (Object Size Checking): Improve grammar.
56 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
59 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
60 that the logic for permitting reciprocal estimates matches that
63 2017-04-27 Jakub Jelinek <jakub@redhat.com>
66 * tree.c (type_cache_hasher::equal): Only compare
67 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
68 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
69 non-aggregate element types.
70 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
71 about the flag on ARRAY_TYPEs in the comment, formatting fix.
73 2017-04-27 Richard Biener <rguenther@suse.de>
76 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
77 stripping ARRAY_REFs from MEM_EXPR make sure we're not
78 keeping a reference to a trailing array.
80 2017-04-27 Richard Biener <rguenther@suse.de>
83 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
84 being in loop-closed SSA form conservatively.
85 (chrec_fold_multiply_poly_poly): Likewise.
87 2017-04-27 Tamar Christina <tamar.christina@arm.com>
90 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
91 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
93 2017-04-27 Jakub Jelinek <jakub@redhat.com>
96 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
97 (aarch64_function_arg_alignment): Return unsigned int again, but still
98 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
99 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
100 Don't emit -Wpsabi note.
101 (aarch64_function_arg_boundary): Likewise.
102 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
105 2017-04-26 Nathan Sidwell <nathan@acm.org>
107 * tree.h (crc32_unsigned_n): Declare.
108 (crc32_unsigned, crc32_unsigned): Make inline.
109 * tree.c (crc32_unsigned_bits): Replace with ...
110 (crc32_unsigned_n): ... this.
111 (crc32_unsigned, crc32_byte): Remove.
112 (crc32_string): Remove unnecessary braces.
114 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
116 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
117 * ipa-inline-analysis.c (MAX_TIME): Remove.
118 (account_size_time): Use sreal for time.
119 (dump_inline_summary): Update.
120 (estimate_function_body_sizes): Update.
121 (estimate_edge_size_and_time): Update.
122 (estimate_calls_size_and_time): Update.
123 (estimate_node_size_and_time): Update.
124 (inline_merge_summary): Update.
125 (inline_update_overall_summary): Update.
126 (estimate_time_after_inlining): Update.
127 (inline_read_section): Update.
128 (inline_write_summary): Update.
129 * ipa-inline.c (compute_uninlined_call_time): Update.
130 (compute_inlined_call_time): Update.
131 (recursive_inlining): Update.
132 (inline_small_functions): Update.
133 (dump_overall_stats): Update.
134 * ipa-inline.h: Include sreal.h.
135 (size_time_entry): Turn time to sreal.
136 (inline_summary): Turn self_time nad time to sreal.
138 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
140 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
142 (sreal::stream_out, sreal::stream_in): New.
143 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
145 2017-04-25 Jakub Jelinek <jakub@redhat.com>
147 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
150 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
153 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
155 (dimode_scalar_chain::compute_convert_gain): Ditto.
156 (dimode_scalar_chain::make_vector_copies): Ditto.
157 (dimode_scalar_chain::convert_reg): Ditto.
158 (dimode_scalar_chain::convert_insn): Ditto.
159 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
160 (VI248_AVX512BW_1): New mode iterator.
161 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
162 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
165 2017-04-25 Martin Sebor <msebor@redhat.com>
167 PR tree-optimization/80497
168 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
169 constants are representable in HOST_WIDE_INT.
170 (parse_directive): Ditto.
172 2017-04-25 Martin Sebor <msebor@redhat.com>
175 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
176 (new_zero_array): Adjust signature.
177 (dom_info::dom_init): Used unsigned rather that size_t.
178 (dom_info::dom_info): Same.
180 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
181 Jakub Jelinek <jakub@redhat.com>
184 * config/arm/arm.c: Include gimple.h.
185 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
186 returns negative, increment ncrn only if it returned positive.
187 (arm_needs_doubleword_align): Return int instead of bool,
188 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
189 members, but if there is any such non-FIELD_DECL
190 > PARM_BOUNDARY aligned decl, return -1 instead of false.
191 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
192 returns negative, increment nregs only if it returned positive.
193 (arm_setup_incoming_varargs): Likewise.
194 (arm_function_arg_boundary): Emit -Wpsabi note if
195 arm_needs_doubleword_align returns negative, return
196 DOUBLEWORD_ALIGNMENT only if it returned positive.
198 2017-04-25 Marek Polacek <polacek@redhat.com>
201 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
202 first argument to type.
204 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
207 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
208 type checks to test for compatibility instead of equality.
210 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
211 Jakub Jelinek <jakub@redhat.com>
214 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
216 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
217 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
218 the alignment computation, but return their maximum in warn_alignment.
219 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
220 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
222 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
223 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
226 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
228 * config/arc/simdext.md (dmpyh): Fix typo.
230 2017-04-25 Richard Biener <rguenther@suse.de>
232 PR tree-optimization/80492
233 * alias.c (compare_base_decls): Handle registers with asm
234 specification conservatively.
235 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
236 compare_base_decls returning dont-know properly.
238 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
240 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
241 (legitimate_offset_address_p): New function.
242 (arc_legitimate_address_p): Use above function.
244 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
246 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
248 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
250 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
251 ACCH registers whenever they are available.
253 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
255 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
256 double regs fix when not used.
258 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
260 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
262 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
263 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
265 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
267 * config/arc/arc.c (arc_output_addsi): Check for h-register class
268 when emitting short ADD instructions.
270 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
272 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
274 (cmpsi_cc_c_insn): Likewise.
275 (cbranchsi4_scratch): Compute proper instruction length using
276 compact_hreg_operand.
277 * config/arc/predicates.md (compact_hreg_operand): New predicate.
279 2017-04-25 Richard Biener <rguenther@suse.de>
282 * passes.c (pass_manager::pass_manager): Initialize
285 2017-04-25 Richard Biener <rguenther@suse.de>
287 PR tree-optimization/79201
288 * tree-ssa-sink.c (statement_sink_location): Handle calls.
290 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
293 * config/s390/vector.md: Split MEM->GPR vector moves for
294 non-s_operand addresses.
296 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
299 * config/s390/predicates.md (reload_const_wide_int_operand): New
301 * config/s390/s390.md ("movti"): Remove d/P alternative.
302 ("movti_bigconst"): New pattern definition.
304 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
307 * s390-protos.h (s390_expand_cs_hqi): Removed.
308 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
309 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
310 modes as well as CCZ1mode and CCZmode.
311 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
312 signature of s390_emit_compare_and_swap.
313 (s390_expand_cs_hqi): Likewise, make static.
314 (s390_expand_cs_tdsi): Generate an explicit compare before trying
315 compare-and-swap, in some cases.
316 (s390_expand_cs): Wrapper function.
317 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
319 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
320 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
321 patterns for small and large integers. Forbid symref memory operands.
322 Move expander to s390.c. Require cc register.
323 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
324 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
325 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
326 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
327 symref memory operands. Remove CC mode and call s390_match_ccmode
329 ("atomic_exchange<mode>"): Allow and implement all integer modes.
331 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
333 * config/s390/s390.md (define_peephole2): New peephole to help
334 combining the load-and-test pattern with volatile memory.
336 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
338 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
339 with CCZmode for TARGET_Z196.
341 2017-04-25 Jakub Jelinek <jakub@redhat.com>
343 PR rtl-optimization/80501
344 * combine.c (make_compound_operation_int): Set subreg_code to SET
345 even for AND with mask of the sign bit of mode.
347 PR rtl-optimization/80500
348 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
351 2017-04-25 Julian Brown <julian@codesourcery.com>
352 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
354 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
356 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
358 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
360 2017-04-25 Julian Brown <julian@codesourcery.com>
361 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
363 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
364 (thunderx2t99_sha): New Reservation.
366 2017-04-25 Julian Brown <julian@codesourcery.com>
367 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
369 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
370 type for 1-element load.
372 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
374 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
376 2017-04-24 Martin Jambor <mjambor@suse.cz>
378 PR tree-optimization/80293
379 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
380 char arrays not totally scalarizable if it is false.
381 (analyze_all_variable_accesses): Pass correct value in the new
382 parameter. Add a statistics counter.
384 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
387 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
389 2017-04-24 Richard Biener <rguenther@suse.de>
391 PR tree-optimization/80494
392 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
393 out for complex types.
395 2017-04-24 Richard Biener <rguenther@suse.de>
397 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
398 * tree-ssa-sccvn.c (print_scc): Print SCC size.
399 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
400 (DFS): Adjust and never fail.
401 (sccvn_dom_walker::fail): Remove.
402 (sccvn_dom_walker::before_dom_children): Adjust.
403 (run_scc_vn): Likewise and never fail.
404 * tree-ssa-pre.c (pass_pre::execute): Adjust.
405 (pass_fre::execute): Likewise.
407 2017-04-24 Richard Biener <rguenther@suse.de>
409 PR tree-optimization/79725
410 * tree-ssa-sink.c (statement_sink_location): Return whether
411 failure reason was zero uses. Move that check later.
412 (sink_code_in_bb): Deal with zero uses by removing the stmt
415 2017-04-24 Richard Biener <rguenther@suse.de>
418 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
419 pointer-based references.
421 2017-04-24 Richard Biener <rguenther@suse.de>
424 * pass_manager.h (pass_manager::operator new): Remove.
425 (pass_manager::operator delete): Likewise.
426 * passes.c (pass_manager::operator new): Remove.
427 (pass_manager::operator delete): Likewise.
428 (pass_manager::pass_manager): Zero individual pass members.
430 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
433 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
434 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
435 Check "XEXP (src, 1)" operand here.
436 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
437 Check "XEXP (src, 1)" operand here.
438 (dimode_scalar_chain::make_vector_copies): Detect count register
439 of a shift instruction. Zero extend count register from QImode
440 to DImode to satisfy vector shift pattern count operand predicate.
441 Substitute vector shift count operand with a DImode copy.
442 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
445 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
447 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
448 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
449 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
450 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
451 (UNSPEC_NOREX_MEM): Remove definition.
453 2017-04-21 Richard Biener <rguenther@suse.de>
455 PR tree-optimization/79547
456 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
457 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
458 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
459 without any constraints.
461 2017-04-21 Richard Biener <rguenther@suse.de>
463 PR tree-optimization/78847
464 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
466 2017-04-21 Richard Biener <rguenther@suse.de>
468 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
469 (build_distinct_type_copy): Likewise.
470 (build_variant_type_copy): Likewise.
471 * tree.c (build_qualified_type): Pass down mem-stat info.
472 (build_distinct_type_copy): Likewise.
473 (build_variant_type_copy): Likewise.
475 2017-04-21 Richard Biener <rguenther@suse.de>
477 PR tree-optimization/80237
478 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
480 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
481 for a simplified result.
483 2016-04-21 Richard Biener <rguenther@suse.de>
485 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
486 sth as strict as a simple_iv but a chrec without symbols and an
487 operand defined in the loop we are peeling (and not some subloop).
488 (propagate_constants_for_unrolling): Propagate all constants.
490 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
493 * config/i386/i386.c (print_reg): Remove assert for disalowed
494 regno values, call output_operand_lossage instead.
496 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
499 * config/i386/constraints.md (Yc): New register constraint.
500 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
501 Use Yc constraint for alternative 2 of operand 0. Remove
502 preferred_for_speed attribute.
504 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
506 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
507 lastprivate clauses in SIMT case.
509 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
511 * doc/invoke.texi (-Wextra-semi): Document new warning option.
513 2017-04-20 Richard Biener <rguenther@suse.de>
515 PR tree-optimization/57796
516 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
518 (vect_model_load_cost): Cost gathers as N scalar loads.
520 2017-04-20 Richard Biener <rguenther@suse.de>
522 * ggc-page.c (ggc_allocated_p): Rename to ...
523 (safe_lookup_page_table_entry): ... this and return the lookup
525 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
527 2017-04-20 Richard Biener <rguenther@suse.de>
529 PR tree-optimization/80453
530 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
531 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
533 (vn_phi_eq): Pass them down.
534 (vn_phi_lookup): Record them.
535 (vn_phi_insert): Likewise.
537 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
539 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
540 uninitialized variable warning to avoid buffer overrun.
542 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
545 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
546 is suppressed for '{ 0 }' in C.
548 2017-04-20 Jakub Jelinek <jakub@redhat.com>
550 * BASE-VER: Set to 8.0.0.
552 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
554 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
555 priority .init_array and .fini_array section with SECTION_NOTYPE
558 2017-04-20 Jakub Jelinek <jakub@redhat.com>
561 * tree.h (build_array_type): Add typeless_storage default argument.
562 * tree.c (type_cache_hasher::equal): Also compare
563 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
564 (build_array_type): Add typeless_storage argument, set
565 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
567 (build_nonshared_array_type): Adjust build_array_type_1 caller.
568 (build_array_type): Likewise. Add typeless_storage argument.
570 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
571 Jakub Jelinek <jakub@redhat.com>
573 PR tree-optimization/80426
574 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
575 operation on symbolic operands, also compute the overflow for the
576 invariant part when the operation degenerates into a negation.
578 2017-04-19 Jakub Jelinek <jakub@redhat.com>
581 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
582 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
585 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
587 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
590 * config/avr/avr.c (tree.h): Include it.
591 (cgraph.h): Include it.
592 (avr_encode_section_info): Don't warn for uninitialized progmem
593 variable if it's just an alias.
595 2017-04-19 Richard Biener <rguenther@suse.de>
598 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
599 when needed by AutoPGO.
601 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
604 * doc/lto.texi: Remove an extra 'that'.
606 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
608 PR rtl-optimization/80429
609 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
610 are only used in debug insns.
612 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
613 Vladimir Makarov <vmakarov@redhat.com>
615 * config/sparc/predicates.md (input_operand): Add comment. Return
616 true for any memory operand when LRA is in progress.
617 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
619 2017-04-18 Jeff Law <law@redhat.com>
622 * mips.md ({return,simple_return}_internal): Do not overwrite
625 2017-04-18 Jakub Jelinek <jakub@redhat.com>
627 PR tree-optimization/80443
628 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
629 instead of adding 1, subtract -1 and similarly instead of subtracting
632 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
634 PR rtl-optimization/80357
635 * haifa-sched.c (tmp_bitmap): New variable.
636 (model_recompute): Handle duplicate use records.
637 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
638 (free_global_sched_pressure_data): Free it.
640 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
643 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
644 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
645 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
646 instead of SYSTEM_HEADER_DIR.
648 2017-04-18 Jeff Law <law@redhat.com>
651 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
652 predecessors after walking up the insn chain.
654 2017-04-18 Jakub Jelinek <jakub@redhat.com>
657 * dwarf2out.c (modified_type_die): Try harder not to emit internal
658 sizetype type into debug info.
660 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
663 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
664 unneeded test for TARGET_UPPER_REGS_SF.
665 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
667 2017-04-18 Jakub Jelinek <jakub@redhat.com>
670 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
671 instead of gsi_after_labels.
673 2017-04-18 Jeff Law <law@redhat.com>
675 * regcprop.c (maybe_mode_change): Avoid creating copies of the
679 2017-04-13 Jeff Law <law@redhat.com>
680 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
681 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
683 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
686 * config/avr/avr.c (intl.h): Include it.
687 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
689 2017-04-18 Martin Liska <mliska@suse.cz>
691 PR gcov-profile/78783
692 * gcov-tool.c (gcov_output_files): Validate that destination
693 file is either removed by the tool or by a user.
695 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
696 Guy Benyei <guybe@mellanox.com>
698 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
699 block, and do not negate it, the stored id is already negative.
701 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
703 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
705 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
708 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
709 masks of options that should be turned off if the VSX vector
710 options are turned off.
711 (OTHER_P8_VECTOR_MASKS): Likewise.
712 (OTHER_VSX_VECTOR_MASKS): Likewise.
713 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
714 rs6000_disable_incompatible_switches to validate no type switches
716 (rs6000_incompatible_switch): New function to disallow turning on
717 other vector options if -mno-vsx, -mno-power8-vector, or
718 -mno-power9-vector are specified.
720 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
722 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
724 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
726 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
727 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
728 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
729 (ARG_POINTER_CFA_OFFSET): Likewise.
731 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
733 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
734 conditions to take advantage of various optimizations.
736 2017-04-13 Jeff Law <law@redhat.com>
738 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
739 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
740 (zero_extendsidi2_dext): Likewise.
742 2017-04-13 Jakub Jelinek <jakub@redhat.com>
745 * fold-const.c (fold_ternary_loc): Revert
746 use op0 instead of fold_convert_loc (loc, type, arg0) part of
749 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
751 PR rtl-optimization/80343
752 * lra-remat.c (update_scratch_ops): Assign original hard reg to
755 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
758 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
759 to ubsan_encode_value.
761 2017-04-13 Jeff Law <law@redhat.com>
763 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
764 appearing in DEBUG_INSNs.
766 2017-04-13 Martin Liska <mliska@suse.cz>
768 PR gcov-profile/80413
769 * gcov-io.c (gcov_write_string): Copy to buffer just when
770 allocated size is greater than zero.
772 2017-04-13 Jakub Jelinek <jakub@redhat.com>
775 * dwarf2out.c (decls_for_scope): Ignore declarations of
776 current_function_decl in BLOCK_NONLOCALIZED_VARS.
778 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
781 * ipa-visibility.c (non_local_p): Fix typos.
782 (localize_node): When localizing symbol in same comdat group,
783 dissolve the group only when we know external symbols are going
785 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
787 2017-04-12 Jakub Jelinek <jakub@redhat.com>
789 PR tree-optimization/79390
790 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
791 order does not result in usable sequence, retry with reversed operand
797 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
798 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
799 op0 instead of fold_convert_loc (loc, type, arg0).
801 2017-04-12 Jeff Law <law@redhat.com>
803 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
804 has a delay slot in the generated code.
806 * config/cris/cris.md (cris_preferred_reload_class): Return
807 GENNONACR_REGS rather than GENERAL_REGS.
809 2017-04-12 Jakub Jelinek <jakub@redhat.com>
812 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
813 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
814 signedness of the result type.
816 2017-04-12 Richard Biener <rguenther@suse.de>
817 Jeff Law <law@redhat.com>
819 PR tree-optimization/80359
820 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
821 trim stores to TARGET_MEM_REFs.
823 2017-04-12 Richard Biener <rguenther@suse.de>
825 PR tree-optimization/79390
826 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
827 threading case even more.
829 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
832 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
833 for quad_address_p for TImode, instead of just not indexed_address.
835 2017-04-12 Richard Biener <rguenther@suse.de>
836 Bernd Edlinger <bernd.edlinger@hotmail.de>
839 * alias.c (component_uses_parent_alias_set_from): Handle
840 TYPE_TYPELESS_STORAGE.
841 (get_alias_set): Likewise.
842 * tree-core.h (tree_type_common): Add typeless_storage flag.
843 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
844 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
845 for types containing members with TYPE_TYPELESS_STORAGE.
846 (place_field): Likewise.
847 (layout_type): Likewise for ARRAY_TYPE.
848 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
849 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
850 TYPE_TYPELESS_STORAGE.
851 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
853 2017-04-12 Jakub Jelinek <jakub@redhat.com>
856 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
857 first argument to type.
859 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
863 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
864 CONST0_RTX (mode) rather than const0_rtx where appropriate.
865 (rs6000_expand_binop_builtin): Likewise.
866 (rs6000_expand_ternop_builtin): Likewise; also add missing
867 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
869 * doc/extend.texi: Document that vec_xxpermdi's third argument
872 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
874 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
875 Use shift_const cost parameter when calculating gain of STV shifts.
877 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
879 PR rtl-optimization/70478
880 * lra-constraints.c (process_alt_operands): Check memory for
881 disfavoring memory insn operand.
883 2017-04-11 Jakub Jelinek <jakub@redhat.com>
886 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
887 left shift in unsigned HOST_WIDE_INT type.
889 PR rtl-optimization/80385
890 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
891 (not (neg X)) into (plus X -1) for complex or non-integral modes.
894 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
895 if they have any depend clauses.
897 2017-04-11 Martin Liska <mliska@suse.cz>
900 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
901 * ipa-split.c (split_function): Create a local comdat symbol
902 if caller is in a comdat group.
904 2017-04-11 Martin Liska <mliska@suse.cz>
907 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
910 2017-04-11 Martin Sebor <msebor@redhat.com>
913 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
914 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
916 (directive::set_width, directive::set_precision, format_character):
918 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
921 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
924 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
925 conflict, set target->arch_name instead of target->cpu_name.
927 2017-04-11 Richard Biener <rguenther@suse.de>
929 PR tree-optimization/80374
930 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
931 build_zero_cst, remove fold_convertible_p check again.
933 2017-04-11 Martin Liska <mliska@suse.cz>
936 * ubsan.c (instrument_object_size): Do not instrument register
939 2017-04-11 Jakub Jelinek <jakub@redhat.com>
942 * config/i386/i386-builtin-types.def
943 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
944 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
945 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
946 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
947 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
948 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
949 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
950 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
951 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
952 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
953 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
954 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
955 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
956 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
957 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
958 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
959 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
960 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
961 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
962 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
963 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
964 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
965 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
966 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
967 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
968 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
969 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
970 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
971 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
972 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
973 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
974 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
975 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
976 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
977 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
978 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
979 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
980 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
981 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
982 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
983 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
984 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
985 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
986 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
987 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
988 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
990 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
991 flag to second_arg_count, handle 4 argument function type _COUNT
992 aliases, handle second_arg_count on second argument rather than last.
994 2017-04-10 Jeff Law <law@redhat.com>
996 PR tree-optimization/80374
997 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
998 record anything if we can not convert integer_zero_node to the
1001 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
1004 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1005 Enhance special handling given to the TARGET_P9_MINMAX option in
1006 relation to certain other options.
1008 2017-04-10 Bin Cheng <bin.cheng@arm.com>
1010 PR tree-optimization/80153
1011 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
1012 remove POINTER_PLUS_EXPR's base part directly, rather than through
1015 2017-04-10 Richard Biener <rguenther@suse.de>
1016 Bin Cheng <bin.cheng@arm.com>
1018 PR tree-optimization/80153
1019 * tree-affine.c (aff_combination_to_tree): Get base pointer from
1020 the first element of pointer type aff_tree. Build result expr in
1022 (add_elt_to_tree): Convert to type unconditionally. Remove other
1024 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
1025 (rewrite_use_nonlinear_expr): Check invariant using iv information.
1027 2017-04-10 Richard Biener <rguenther@suse.de>
1029 * tree-ssa-structalias.c (find_func_aliases): Properly handle
1032 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
1034 PR rtl-optimization/70478
1035 * lra-constraints.c (curr_small_class_check): New.
1036 (update_and_check_small_class_inputs): New.
1037 (process_alt_operands): Update curr_small_class_check. Disfavor
1038 alternative insn memory operands. Check available regs for small
1041 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1044 * config/mips/mips.opt (-mvirt): Update description.
1045 * doc/invoke.texi (-mvirt): Likewise.
1047 2017-04-10 Richard Biener <rguenther@suse.de>
1050 * fold-const.c (fold_binary_loc): Look at unstripped ops when
1051 looking for NEGATE_EXPR in -A / -B to A / B folding.
1053 2017-04-10 Martin Liska <mliska@suse.cz>
1055 PR gcov-profile/80224
1056 * gcov.c (print_usage): Fix usage string.
1057 (get_gcov_intermediate_filename): Remove.
1058 (output_gcov_file): Use both for normal and intermediate format.
1059 (generate_results): Do not initialize special file for
1060 intermediate format.
1062 2017-04-10 Richard Biener <rguenther@suse.de>
1064 PR tree-optimization/80304
1065 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
1068 2017-04-10 Nathan Sidwell <nathan@acm.org>
1071 * config/rs6000/rs6000.c (rs6000_vector_type): New.
1072 (rs6000_init_builtins): Use it.
1074 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1076 * config/arm/arm.md (<mrc>): Add mode to SET source.
1079 2017-04-10 Richard Biener <rguenther@suse.de>
1082 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
1084 2017-04-10 Jakub Jelinek <jakub@redhat.com>
1087 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
1088 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
1089 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
1090 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
1091 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
1092 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
1093 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
1094 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
1095 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
1096 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
1097 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
1098 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
1099 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
1100 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
1101 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
1102 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
1103 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
1104 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
1105 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
1106 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
1107 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
1108 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
1109 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
1111 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
1113 PR rtl-optimization/70478
1114 * lra-constraints.c: Reverse the last patch.
1116 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
1118 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
1119 Add comment for WCHAR_T.
1121 2017-04-08 Martin Liska <mliska@suse.cz>
1124 2017-04-07 Martin Liska <mliska@suse.cz>
1127 * ipa-split.c (split_function): Add function part to a same comdat
1130 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1133 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
1135 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
1137 * rs6000/rs6000.c (vec_load_pendulum): Rename...
1138 (vec_pairing): ...to this.
1139 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
1140 (rs6000_sched_init): Adjust for name change.
1141 (struct rs6000_sched_context): Likewise.
1142 (rs6000_init_sched_context): Likewise.
1143 (rs6000_set_sched_context): Likewise.
1145 2017-04-07 Jakub Jelinek <jakub@redhat.com>
1151 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
1153 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
1154 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
1155 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
1157 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
1159 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
1161 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
1163 PR rtl-optimization/70703
1164 * ira-color.c (update_conflict_hard_regno_costs): Use
1165 int64_t instead of HOST_WIDE_INT.
1167 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
1169 PR rtl-optimization/70478
1170 * lra-constraints.c (process_alt_operands): Disfavor alternative
1171 insn memory operands.
1173 2017-04-07 Jeff Law <law@redhat.com>
1175 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
1176 CALL and NOTE_INSN_CALL_ARG_LOCATION.
1178 2017-04-07 Martin Liska <mliska@suse.cz>
1181 * config/aarch64/aarch64.c (aarch64_process_target_attr):
1182 Show error message instead of an ICE.
1184 2017-04-07 Martin Liska <mliska@suse.cz>
1187 * ipa-split.c (split_function): Add function part to a same comdat
1190 2017-04-07 Richard Biener <rguenther@suse.de>
1193 * tree.c (get_unwidened): Also handle ! for_type case for
1195 * convert.c (do_narrow): Split out from ...
1196 (convert_to_integer_1): ... here. Do not pass final truncation
1197 type to get_unwidened for TRUNC_DIV_EXPR.
1199 2017-04-07 Richard Biener <rguenther@suse.de>
1201 * tree-affine.c (wide_int_ext_for_comb): Take type rather
1203 (aff_combination_const): Adjust.
1204 (aff_combination_scale): Likewise.
1205 (aff_combination_add_elt): Likewise.
1206 (aff_combination_add_cst): Likewise.
1207 (aff_combination_convert): Likewise.
1208 (add_elt_to_tree): Likewise. Remove unused argument.
1209 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
1211 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
1213 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
1215 * config/arm/arm.c (arm_default_short_enums): Use
1216 ARM_DEFAULT_SHORT_ENUMS.
1217 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
1219 2017-04-06 Jakub Jelinek <jakub@redhat.com>
1222 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
1223 members with redundant out-of-class redeclaration.
1225 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
1228 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
1229 * config/i386/i386.md (*zero_extendsidi2):
1230 Add (?*x,*x) and (?*v,*v) alternatives.
1232 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
1235 * config/i386/i386.c (ix86_expand_builtin)
1236 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
1237 mode from insn data. Convert operands to insn operand mode.
1238 Copy operands that don't satisfy insn predicate to a register.
1240 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
1242 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
1245 2017-04-06 Richard Biener <rguenther@suse.de>
1247 PR tree-optimization/80334
1248 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
1249 preserve alignment of accesses.
1251 2017-04-06 Richard Biener <rguenther@suse.de>
1253 PR tree-optimization/80262
1254 * tree-sra.c (build_ref_for_offset): Preserve address-space
1256 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
1257 Drop useless address-space information on MEM_REF offsets.
1259 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
1261 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
1263 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
1265 PR rtl-optimization/70703
1266 * ira-color.c (update_conflict_hard_regno_costs): Use
1267 HOST_WIDE_INT instead of long.
1269 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
1272 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
1273 not defined for x86_64 target. Add -mmmx target option when __SSE2__
1275 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
1276 for x86_64 target. Handle -m3dnowa option.
1278 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
1280 PR rtl-optimization/70703
1281 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
1282 (update_conflict_hard_regno_costs): Use long instead of unsigned
1283 arithmetic for cost calculation.
1285 2017-04-05 Jakub Jelinek <jakub@redhat.com>
1286 Bernd Edlinger <bernd.edlinger@hotmail.de>
1289 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
1292 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
1295 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
1296 ptr_mode with Pmode throughout.
1297 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
1298 into probe_stack_range and use DImode.
1300 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
1303 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
1304 call_eh_return is true.
1306 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1308 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
1309 Initialize last_match_fntype_index.
1311 2017-04-05 Jakub Jelinek <jakub@redhat.com>
1314 * tree-nvr.c: Include internal-fn.h.
1315 (pass_return_slot::execute): Ignore internal calls without
1318 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1319 Richard Biener <rguenther@suse.de>
1322 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
1323 captures used multiple times, except for the last use.
1324 * generic-match-head.c: Include gimplify.h.
1326 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1328 PR tree-optimization/79390
1329 * target.h (struct noce_if_info): Declare.
1330 * targhooks.h (default_noce_conversion_profitable_p): Declare.
1331 * target.def (noce_conversion_profitable_p): New target hook.
1332 * ifcvt.h (struct noce_if_info): New type, moved from ...
1333 * ifcvt.c (struct noce_if_info): ... here.
1334 (noce_conversion_profitable_p): Renamed to ...
1335 (default_noce_conversion_profitable_p): ... this. No longer
1337 (noce_try_store_flag_constants, noce_try_addcc,
1338 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
1339 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
1340 instead of noce_conversion_profitable_p.
1341 * config/i386/i386.c: Include ifcvt.h.
1342 (ix86_option_override_internal): Don't override
1343 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
1344 (ix86_noce_conversion_profitable_p): New function.
1345 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
1346 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
1347 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
1348 * doc/tm.texi: Regenerated.
1350 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1352 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
1355 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
1358 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
1359 instructions for small multiply cores.
1361 2017-04-04 Jeff Law <law@redhat.com>
1363 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
1365 (mips_expand_vec_perm_const): Initialize elements in orig_perm
1366 that are not set by the loop over the elements.
1368 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1371 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
1372 int mode, convert_modes it to mode as unsigned, otherwise use
1373 lowpart_subreg to mode rather than SImode.
1374 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
1375 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
1376 Use DImode instead of SImode for the shift count operand.
1377 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
1380 2017-04-04 Richard Biener <rguenther@suse.de>
1383 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
1384 arithmetic done for the negate or the plus. Simplify.
1385 (A - (-B) -> A + B): Likewise.
1386 * fold-const.c (split_tree): Make sure to not negate pointers.
1388 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
1390 PR rtl-optimization/60818
1391 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
1392 a compare of comparisons with the thing compared if this results
1393 in a different machine mode.
1395 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
1397 * alias.c (base_alias_check): Fix typo in comment.
1398 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
1399 * cgraphunit.c (symbol_table::compile): Likewise.
1400 * collect2.c (maybe_run_lto_and_relink): Likewise.
1401 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
1402 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
1403 * config/avr/avr.c (avr_map_op_t): Likewise.
1404 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
1405 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
1406 * config/epiphany/epiphany.md (movcc): Likewise.
1407 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
1408 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
1410 * config/mips/mips.c (mips_save_restore_reg): Likewise.
1411 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
1412 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
1413 * config/sh/sh.c (sh_rtx_costs): Likewise.
1414 * fold-const.c (fold_truth_andor): Likewise.
1415 * genautomata.c (collapse_flag): Likewise.
1416 * gengtype.h (struct type::u::s): Likewise.
1417 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
1418 * input.c (FORMAT_AMOUNT): Likewise.
1419 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
1420 (known_aggs_to_agg_replacement_list): Likewise.
1421 * ipa-inline-analysis.c: Likewise.
1422 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
1423 * ipa-polymorphic-call.c
1424 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
1425 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
1426 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
1428 * modulo-sched.c (apply_reg_moves): Likewise.
1429 * omp-expand.c (build_omp_regions_1): Likewise.
1430 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
1431 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
1432 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
1433 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
1434 * value-prof.c: Likewise.
1435 * var-tracking.c (val_reset): Likewise.
1437 2017-04-03 Richard Biener <rguenther@suse.de>
1439 PR tree-optimization/80275
1440 * fold-const.c (split_address_to_core_and_offset): Handle
1443 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
1445 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
1446 descriptors is at least equal to that of functions.
1448 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1450 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
1452 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1455 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
1456 (mov<IMOD4:mode>): New expander.
1457 (*mov<IMOD4:mode>_internal): New insn and split pattern.
1459 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
1461 PR rtl-optimization/79405
1462 * fwprop.c (propagations_left): New variable.
1463 (forward_propagate_into): Decrement it.
1464 (fwprop_init): Initialize it.
1465 (fw_prop): If the variable has reached zero, stop propagating.
1466 (fwprop_addr): Ditto.
1468 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1471 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
1472 a FUNCTION_DECL, pass it as decl instead of origin to
1475 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
1477 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
1480 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
1483 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
1484 TARGET_VSX_SMALL_INTEGER.
1486 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1488 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1489 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
1491 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1493 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
1494 extraction from odd-numbered MSA register.
1496 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1499 * expmed.c (store_bit_field_1): Don't attempt to create
1500 a word subreg out of hard registers wider than word if they
1501 have HARD_REGNO_NREGS of 1 for their mode.
1504 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
1505 conversions to integer types wider than word and pointer.
1508 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
1509 (rtx_equal_for_cselib_p): Pass 0 to it.
1510 * cselib.c (cselib_hasher::equal): Likewise.
1511 (rtx_equal_for_cselib_1): Add depth argument. If depth
1512 is 128, don't look up VALUE locs and punt. Increment
1513 depth in recursive calls when walking VALUE locs.
1515 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
1517 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
1518 (make_gcov_file_name): Use the canonical path name for generating
1520 (read_line): Fix handling of files with ascii null bytes.
1522 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
1524 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
1525 to initialise a vector register instead
1526 of using a const_int.
1528 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1530 PR translation/80189
1531 * gimplify.c (omp_default_clause): Use %qs instead of %s in
1532 diagnostic messages.
1534 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
1537 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
1538 (dfp_diex_<mode>): Update mode of operand 1.
1539 * doc/extend.texi (dxex, dxexq): Document change to return type.
1540 (diex, diexq): Document change to argument type.
1542 2017-03-30 Martin Jambor <mjambor@suse.cz>
1545 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
1546 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
1547 it reflects the signature changes performed at the callee side.
1548 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
1549 to cgraph_build_function_type_skip_args.
1550 (build_function_decl_skip_args): Adjust call to the above function.
1552 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1555 * config/i386/sse.md
1556 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
1557 register as dest whenever it is a MEM not rtx_equal_p to the
1558 corresponding dup operand, and when forcing into reg move the
1559 reg into the memory afterwards.
1560 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
1561 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
1562 for the force_reg mode.
1563 (avx512vl_vextractf128<mode>): Use register as dest either
1564 always when a MEM, or when it is a MEM not rtx_equal_p to the
1565 corresponding dup operand, or even not when it is a CONST_VECTOR
1566 depending on the mode and lo vs. hi.
1567 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
1569 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
1570 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
1571 Likewise. Require that operands[2] is even.
1572 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
1573 Remove extraneous parens. Require that operands[2] is a multiple
1575 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
1576 operands[0] is a MEM if <mask_applied>, the predicates/constraints
1577 disallow memory then.
1579 2017-03-30 Richard Biener <rguenther@suse.de>
1581 PR tree-optimization/77498
1582 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
1583 to non-constants over backedges.
1585 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
1587 PR rtl-optimization/80233
1588 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
1589 as last_combined_insn. Do not test for BARRIER_P separately.
1591 2017-03-29 Andreas Schwab <schwab@suse.de>
1594 * calls.c (prepare_call_address): Convert funexp to Pmode before
1595 copying to temp reg.
1597 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1599 PR tree-optimization/80158
1600 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
1601 Handle possible future case of more than one alternate
1603 (replace_rhs_if_not_dup): Likewise.
1604 (replace_one_candidate): Likewise.
1606 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
1608 PR rtl-optimization/80193
1609 * ira.c (ira): Do not check allocation for LRA.
1611 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
1613 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
1614 (nvptx_output_simt_exit): Declare.
1615 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
1616 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
1617 (init_softstack_frame): Move initialization of crtl->is_leaf to...
1618 (nvptx_declare_function_name): ...here. Emit declaration of local
1619 memory space buffer for omp_simt_enter insn.
1620 (nvptx_output_unisimt_switch): New.
1621 (nvptx_output_softstack_switch): New.
1622 (nvptx_output_simt_enter): New.
1623 (nvptx_output_simt_exit): New.
1624 * config/nvptx/nvptx.h (struct machine_function): New fields
1625 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
1626 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
1627 (UNSPECV_SIMT_EXIT): Ditto.
1628 (omp_simt_enter_insn): New insn.
1629 (omp_simt_enter): New expansion.
1630 (omp_simt_exit): New insn.
1631 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
1633 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
1634 (expand_GOMP_SIMT_ENTER_ALLOC): New.
1635 (expand_GOMP_SIMT_EXIT): New.
1636 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
1637 (GOMP_SIMT_ENTER_ALLOC): Ditto.
1638 (GOMP_SIMT_EXIT): Ditto.
1639 * target-insns.def (omp_simt_enter): New insn.
1640 (omp_simt_exit): Ditto.
1641 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
1643 (lower_rec_simd_input_clauses): Implement SIMT privatization.
1644 (lower_rec_input_clauses): Likewise.
1645 (lower_lastprivate_clauses): Handle SIMT privatization.
1647 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
1648 (ompdevlow_adjust_simt_enter): New.
1649 (find_simtpriv_var_op): New.
1650 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
1651 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
1653 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
1654 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
1655 (copy_decl_for_dup_finish): Ditto.
1657 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
1659 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
1662 * config/i386/i386.c (ix86_option_override_internal): Always
1663 allow -mpreferred-stack-boundary=3 for 64-bit targets.
1665 2017-03-28 Bin Cheng <bin.cheng@arm.com>
1667 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
1669 2017-03-28 Bin Cheng <bin.cheng@arm.com>
1671 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
1672 mark new edge's irreducible flag accordign to it.
1673 (vect_do_peeling): Check loop preheader edge's irreducible flag
1674 and pass it to function slpeel_add_loop_guard.
1676 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
1678 PR tree-optimization/80218
1679 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
1680 Update block frequencies and counts.
1682 2017-03-28 Richard Biener <rguenther@suse.de>
1684 PR tree-optimization/78644
1685 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
1686 of a simplification result we may not use it at all.
1688 2017-03-28 Richard Biener <rguenther@suse.de>
1691 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
1692 without arguments, generate default definition of a SSA name.
1694 2017-03-28 Richard Biener <rguenther@suse.de>
1697 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
1698 TYPE_REF_CAN_ALIAS_ALL references.
1699 * fold-const.c (fold_indirect_ref_1): Likewise.
1701 2017-03-28 Martin Liska <mliska@suse.cz>
1704 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
1705 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
1707 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1708 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
1710 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
1711 (EXTRA_SPECS): Define.
1712 (SUBTARGET_EXTRA_SPECS): Likewise.
1713 (SUBTARGET_CPP_SPEC): Likewise.
1714 * config/arc/elf.h (EXTRA_SPECS): Renamed to
1715 SUBTARGET_EXTRA_SPECS.
1716 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
1718 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1720 * config/arc/simdext.md (vst64_insn): Update pattern.
1721 (vld32wh_insn): Likewise.
1722 (vld32wl_insn): Likewise.
1723 (vld64_insn): Likewise.
1724 (vld32_insn): Likewise.
1726 2017-03-28 Marek Polacek <polacek@redhat.com>
1729 * fold-const.c (fold_comparison): Use protected_set_expr_location
1730 instead of SET_EXPR_LOCATION.
1732 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
1734 * tree.c (add_expr): Avoid name lookup warning.
1736 2017-03-27 Jeff Law <law@redhat.com>
1738 PR tree-optimization/80216
1739 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
1740 function name. Limit recursion depth.
1741 (record_temporary_equivalences): Corresponding changes.
1743 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
1745 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
1748 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1751 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
1753 * cfgcleanup.c (reg_note_cfa_p): New array.
1754 (insns_have_identical_cfa_notes): New function.
1755 (old_insns_match_p): Don't cross-jump in between /f
1756 and non-/f instructions. If both i1 and i2 are frame related,
1757 verify all CFA notes, their order and content.
1759 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1762 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
1763 HImode and SImode with zero extend to DImode to one insn.
1764 (bswap<mode>2_extenddi): Likewise.
1765 (bswapsi2_extenddi): Likewise.
1766 (bswaphi2_extendsi): Likewise.
1767 (bswaphi2): Combine bswap HImode and SImode into one insn.
1768 Separate memory insns from swapping register.
1769 (bswapsi2): Likewise.
1770 (bswap<mode>2): Likewise.
1771 (bswaphi2_internal): Delete, no longer used.
1772 (bswapsi2_internal): Likewise.
1773 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
1774 store, and gpr<-gpr swap insns.
1775 (bswap<mode>2_store): Likewise.
1776 (bswaphi2_reg): Register only splitter, combine with the splitter.
1777 (bswaphi2 splitter): Likewise.
1778 (bswapsi2_reg): Likewise.
1779 (bswapsi2 splitter): Likewise.
1780 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
1781 the insns into load, store, and register/register insns.
1782 (bswapdi2_ldbrx): Likewise.
1783 (bswapdi2_load): Likewise.
1784 (bswapdi2_store): Likewise.
1785 (bswapdi2_reg): Likewise.
1787 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
1789 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
1790 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
1792 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1795 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
1797 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1798 special handling for target option conflicts between dform
1799 options (-mpower9-dform, -mpower9-dform-vector,
1800 -mpower9-dform-scalar) and -mno-direct-move.
1802 2017-03-27 Richard Biener <rguenther@suse.de>
1804 PR tree-optimization/80181
1805 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
1807 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1809 * config/arc/predicates.md (move_double_src_operand): Replace the
1810 call to move_double_src_operand with a call to address_operand.
1812 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1814 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
1815 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
1816 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
1818 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1820 * config/arc/predicates.md (long_immediate_loadstore_operand):
1821 Consider scaled addresses cases.
1823 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1825 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
1826 restored when in interrupt.
1827 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
1828 doesn't have delay slot.
1830 2017-03-27 Richard Biener <rguenther@suse.de>
1833 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
1834 inlined thunk clones.
1836 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1839 * asan.c (instrument_derefs): Copy over last operand from
1840 original COMPONENT_REF to the new COMPONENT_REF with
1841 DECL_BIT_FIELD_REPRESENTATIVE.
1842 * ubsan.c (instrument_object_size): Likewise.
1844 2017-03-27 Richard Biener <rguenther@suse.de>
1846 PR tree-optimization/80170
1847 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
1848 sure DR/SCEV didnt fold in constants we do not see when looking
1849 at the reference base alignment.
1851 2017-03-27 Richard Biener <rguenther@suse.de>
1854 * gimple-fold.c (fold_ctor_reference): Properly guard against
1855 NULL return value from canonicalize_constructor_val.
1857 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
1860 * config/i386/i386.c (ix86_expand_builtin)
1861 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
1862 flags reg setting and flags reg using instructions.
1863 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
1864 clobbering instructions to zero extend op2.
1866 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
1868 * doc/install.texi (Configuration) <--with-aix-soname>:
1869 Update link to AIX ld.
1871 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
1873 PR rtl-optimization/80160
1874 PR rtl-optimization/80159
1875 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
1876 reg_alternate_class into account.
1878 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
1881 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
1882 to consider in curr_insn_transform.
1884 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1886 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
1887 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
1888 and emit_mode_inner.
1890 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1892 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
1893 argument to the overloaded builtin variants. Use the new flag to
1894 deprecate certain builtin variants.
1895 * config/s390/s390-builtin-types.def: Add new builtin types.
1896 * config/s390/s390-builtins.h: Support new flags field for
1897 overloaded builtins.
1898 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
1899 (s390_macro_to_expand): Enable vector float data type.
1900 (s390_cpu_cpp_builtins_internal): Indicate support of the new
1901 builtins by incrementing the __VEC__ version number.
1902 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
1904 (s390_resolve_overloaded_builtin): Emit error messages depending
1905 on the builtin flags.
1906 * config/s390/s390.c (s390_expand_builtin): Support additional
1907 flags argument. Change error message to match the messages
1908 emitted in s390-c.c.
1909 * config/s390/s390.md: New UNSPEC_* constants.
1910 (op_type): Add new instruction types.
1911 * config/s390/vecintrin.h: Add new builtins and test data class
1913 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
1914 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
1915 (VEC_INEXACT, VEC_NOINEXACT): New constants.
1916 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
1917 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
1918 ("vec_mergel<mode>"): V_HW -> VEC_HW.
1920 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
1921 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
1922 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
1923 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
1925 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
1926 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
1927 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
1928 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
1930 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
1931 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
1932 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
1933 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
1934 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
1935 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
1936 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
1938 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
1939 ("vec_scatter_element<V_HW_4:mode>_DI")
1940 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
1941 ("vec_fpint<mode>", "vflls")
1942 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
1943 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
1944 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
1945 ("*vec_cmphe<mode>_cc"): ... these.
1947 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
1948 mode constant instead of magic value.
1950 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1952 * config/s390/s390.c (s390_expand_vec_compare): Support other
1953 vector floating point modes than just V2DF.
1954 (s390_expand_vcond): Likewise.
1955 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
1956 (s390_cannot_change_mode_class): Prevent mode changes between TF
1957 and V1TF in vector registers.
1958 * config/s390/s390.md (DF, SF): New mode attributes.
1959 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
1960 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
1961 SFmode support for VRs.
1962 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
1964 (VFT, VF_HW): New mode iterators.
1965 (vw, sdx): New mode attributes.
1966 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
1967 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
1968 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
1969 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
1970 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
1971 also the new vector floating point modes. Renaming to ...
1973 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
1974 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
1975 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
1976 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
1977 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
1978 ("vec_unordered<mode>"): ... these.
1980 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
1981 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
1982 ("*vec_extendv2df"): New insn definitions.
1984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1986 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
1987 ("mulditi3_2", "*muldi3_sign"): New patterns.
1988 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
1989 rename the pattern definition.
1991 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1993 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
1995 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
1997 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1999 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
2000 instruction if possible.
2001 * config/s390/vector.md (vec_halfnumelts): New mode
2003 ("*vec_vllezlf<mode>"): New pattern.
2005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2007 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
2008 ("popcountv4si2", "popcountv2di2"): Rename to ...
2009 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
2010 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
2012 ("popcount<mode>2_vxe"): New pattern.
2014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2016 * common/config/s390/s390-common.c (processor_flags_table): Add
2018 * config.gcc: Add arch12.
2019 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2020 Default to arch12 for unknown CPU model numbers.
2021 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
2022 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
2023 PROCESSOR_max sanity check.
2024 * config/s390/s390-opts.h (enum processor_type): Add
2026 * config/s390/s390.c (processor_table): Add arch12.
2027 (s390_expand_builtin): Add check for B_VXE flag.
2028 (s390_issue_rate): Add PROCESSOR_ARCH12.
2029 (s390_get_sched_attrmask): Likewise.
2030 (s390_get_unit_mask): Likewise.
2031 (s390_sched_score): Enable z13 scheduling for arch12.
2032 (s390_sched_reorder): Likewise.
2033 (s390_sched_variable_issue): Likewise.
2034 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
2036 (s390_tune_attr): Use z13 scheduling also for arch12.
2037 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
2038 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
2039 (TARGET_VXE_P): New macros.
2040 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
2041 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
2042 * config/s390/s390.opt: Add arch12 as processor_type.
2044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2046 * config/s390/s390.md
2047 ("fixuns_truncdddi2", "fixuns_trunctddi2")
2048 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
2049 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
2051 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
2052 Rename expanders to ...
2054 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
2055 ("fixuns_truncdddi2_emu"): ... these.
2057 ("fixuns_trunc<mode>si2_emu"): New expander.
2059 ("*fixuns_truncdfdi2_z13"): Rename to ...
2060 ("*fixuns_truncdfdi2_vx"): ... this.
2062 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2064 * config/s390/2964.md: Remove the single element vector compare
2065 instructions which are no longer used.
2066 * config/s390/s390.c (s390_select_ccmode): Remove handling of
2068 (s390_canonicalize_comparison): Remove handling of DFmode
2070 (s390_expand_vec_compare_scalar): Remove function.
2071 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
2072 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
2074 ("*cmp<mode>_ccs"): Add wfcdb instruction.
2076 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2078 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
2080 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
2081 will anyway by matched by mov<mode>_64dfp.
2083 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2085 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
2086 vlef/vstef. Add missing operand to vleif.
2088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2090 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
2091 pair for all vector types with 64 bit elements.
2092 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
2093 * config/s390/vector.md (V_HW_64): ... here.
2094 (V_128_NOSINGLE): New mode iterator.
2095 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
2096 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
2097 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
2098 ("*vec_load_pairv2di"): Change to ...
2099 ("*vec_load_pair<mode>"): ... this one.
2101 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2103 * config/s390/constraints.md: Add comments.
2104 (jKK): Reject element sizes > 8 bytes.
2105 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
2107 * config/s390/s390.md: Add the s_operand checks formerly in
2108 s390_split_ok_p to various splitters where they are still
2110 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
2111 for 128 bit vectors. Plus two splitters.
2113 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2115 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
2118 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2121 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
2122 error if the boundary argument is not constant.
2124 2017-03-24 Jakub Jelinek <jakub@redhat.com>
2126 PR rtl-optimization/80112
2127 * loop-doloop.c (doloop_condition_get): Don't check condition
2128 if cmp isn't SET with IF_THEN_ELSE src.
2130 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2132 PR tree-optimization/80158
2133 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
2134 replacing a candidate statement, also replace it for the
2135 candidate's alternate interpretation.
2136 (replace_rhs_if_not_dup): Likewise.
2137 (replace_one_candidate): Likewise.
2139 2017-03-24 Richard Biener <rguenther@suse.de>
2141 PR tree-optimization/80167
2142 * graphite-isl-ast-to-gimple.c
2143 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
2145 (translate_isl_ast_to_gimple::get_rename): Likewise.
2147 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
2149 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
2150 handling of certain combinations of target options, including the
2151 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
2152 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
2154 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2157 * config/arm/arm.md (*load_multiple): Add reload_completed to
2160 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2161 Richard Biener <rguenth@suse.de>
2163 PR tree-optimization/79908
2164 PR tree-optimization/80136
2165 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2166 been cast away, gimplify_and_add suffices.
2168 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
2170 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
2172 2017-03-23 Richard Biener <rguenther@suse.de>
2174 PR tree-optimization/80032
2175 * gimplify.c (gimple_push_cleanup): Forced unconditional
2176 cleanups still have to go to the conditional_cleanups
2179 2017-03-22 Jakub Jelinek <jakub@redhat.com>
2181 PR tree-optimization/80072
2182 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
2184 (next_operand_entry_id): Change type to unsigned int.
2185 (sort_by_operand_rank): Make sure to return the right return value
2186 even if unsigned fields are bigger than INT_MAX.
2187 (struct oecount): Change cnt and id type to unsigned int.
2188 (oecount_hasher::equal): Formatting fix.
2189 (oecount_cmp): Make sure to return the right return value
2190 even if unsigned fields are bigger than INT_MAX.
2191 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
2194 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
2195 TREE_READONLY on result if writing it more than once.
2198 * doc/invoke.texi (-fsanitize=thread): Document that with
2199 -fnon-call-exceptions atomics are not able to throw
2203 * tsan.c: Include tree-eh.h.
2204 (instrument_builtin_call): Call maybe_clean_eh_stmt or
2205 maybe_clean_or_replace_eh_stmt where needed.
2206 (instrument_memory_accesses): Add cfg_changed argument.
2207 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
2208 if it returned true.
2209 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
2211 PR rtl-optimization/63191
2212 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
2213 wrapper function, moved the whole old content into ...
2214 (ix86_delegitimize_address_1): ... this. New inline function.
2215 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
2216 true as last argument instead of ix86_delegitimize_address.
2218 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
2220 * config/aarch64/aarch64.c (generic_branch_cost): Copy
2221 cortexa57_branch_cost.
2223 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
2225 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
2227 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2230 * doc/md.texi (Constraints): Document wA constraint.
2231 * config/rs6000/constraints.md (wA): New.
2232 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
2233 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
2234 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
2235 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
2237 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
2240 * gimplify.c (is_oacc_declared): New function.
2241 (oacc_default_clause): Use it to set default flags for acc declared
2242 variables inside parallel regions.
2243 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
2245 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
2246 declare attribute to any decl as necessary.
2248 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
2251 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
2252 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
2253 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
2254 (arm_arch_lpae): This.
2255 * config/arm/arm.c (arm_arch7ve): Rename into ...
2256 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
2257 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
2260 2017-03-22 Martin Liska <mliska@suse.cz>
2263 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
2264 error message instead of an ICE.
2266 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2268 * doc/extend.texi (6.11 Additional Floating Types): Revise.
2270 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
2272 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2274 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2277 2017-03-21 Martin Sebor <msebor@redhat.com>
2279 * doc/extend.texi: Use "cannot" instead of "can't."
2280 * doc/hostconfig.texi: Same.
2281 * doc/install.texi: Same.
2282 * doc/invoke.texi: Same.
2283 * doc/loop.texi: Same.
2284 * doc/md.texi: Same.
2285 * doc/objc.texi: Same.
2286 * doc/rtl.texi: Same.
2287 * doc/tm.texi: Same.
2288 * doc/tm.texi.in: Same.
2289 * doc/trouble.texi: Same.
2291 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
2294 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
2295 (collect_checksum_attributes): Set it.
2296 (die_checksum_ordered): Use it.
2298 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2300 PR tree-optimization/79908
2301 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
2302 change: For a VA_ARG whose LHS has been cast away, use
2303 force_gimple_operand to construct the side effects.
2305 2017-03-21 David Malcolm <dmalcolm@redhat.com>
2307 PR translation/80001
2308 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
2309 more amenable to translation.
2310 (oacc_loop_auto_partitions): Likewise.
2312 2017-03-21 Marek Polacek <polacek@redhat.com>
2313 Martin Sebor <msebor@redhat.com>
2315 PR tree-optimization/80109
2316 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
2319 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2320 Segher Boessenkool <segher@kernel.crashing.org>
2323 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
2324 check reg_used_between_p between insn and one of succ or succ2
2325 depending on if succ is artificial insn not inserted into insn
2328 2017-03-21 Martin Liska <mliska@suse.cz>
2330 PR gcov-profile/80081
2331 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
2332 * doc/gcc.texi: Include gcov-dump stuff.
2333 * doc/gcov-dump.texi: New file.
2335 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
2337 PR rtl-optimization/79150
2338 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
2339 conditional jump, if the jump is the last insn of the loop.
2341 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2342 Richard Biener <rguenth@suse.de>
2344 PR tree-optimization/79908
2345 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2346 been cast away, use force_gimple_operand to construct the side
2349 2017-03-21 Martin Liska <mliska@suse.cz>
2352 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
2355 2017-03-21 Brad Spengler <spender@grsecurity.net>
2358 * plugin.c (htab_hash_plugin): New function.
2359 (add_new_plugin): Use it and adjust.
2360 (parse_plugin_arg_opt): Adjust.
2361 (init_one_plugin): Likewise.
2363 2017-03-21 Richard Biener <rguenther@suse.de>
2365 PR tree-optimization/80032
2366 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
2367 if set force the cleanup to happen unconditionally.
2368 (gimplify_target_expr): Push inserted clobbers with force_uncond
2369 to avoid them being removed by control-dependent DCE.
2371 2017-03-21 Richard Biener <rguenther@suse.de>
2373 PR tree-optimization/80122
2374 * tree-inline.c (copy_bb): Do not expans va-arg packs or
2375 va_arg_pack_len when the inlined call stmt requires pack
2377 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
2379 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2382 * tsan.c (instrument_builtin_call): If the memory model argument
2383 is not a constant, assume it is valid.
2386 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
2389 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2391 PR rtl-optimization/79910
2392 * combine.c (can_combine_p): Do not allow combining an I0 or I1
2393 if its dest is used by an insn before I2 (other than the combined
2394 insns themselves, which are properly handled already).
2396 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2399 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2401 * combine.c (record_used_regs): New static function.
2402 (try_combine): Handle situations where there is an additional
2403 instruction between I2 and I3 which needs to have a LOG_LINK
2407 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2409 * combine.c (try_combine): Delete redundant i1 test. Call
2410 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2412 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2415 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
2418 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2420 PR tree-optimization/80054
2421 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
2422 the optimization if a PHI or any of its arguments is not dominated
2423 by the candidate's basis. Use gphi* rather than gimple* as
2425 (replace_profitable_candidates): Clean up a gimple* variable that
2426 should be a gphi* variable.
2428 2017-03-20 Martin Sebor <msebor@redhat.com>
2431 * doc/extend.texi (attribute constructor): Document present limitation.
2433 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
2436 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
2437 __POWER9_VECTOR__ #ifdef control, change template definition to
2438 use Power9-specific built-in function.
2439 (vec_any_eq): Likewise.
2440 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
2441 to control outcomes from this test.
2442 (vector_ae_<mode>p): For VEC_F modes, likewise.
2444 2017-03-20 Ian Lance Taylor <iant@google.com>
2446 * config/i386/i386.c (ix86_function_regparm): Save an extra
2447 register for -fsplit-stack with DECL_STATIC_CHAIN.
2449 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2452 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
2453 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
2455 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2457 * config/riscv/riscv.c (riscv_print_operand): Use "fence
2459 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
2462 2017-03-20 Marek Polacek <polacek@redhat.com>
2465 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
2467 2017-03-20 Richard Biener <rguenther@suse.de>
2469 PR tree-optimization/80113
2470 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
2471 allocate extra SSA name for PHI def.
2472 (add_close_phis_to_outer_loops): Likewise.
2473 (add_close_phis_to_merge_points): Likewise.
2474 (copy_loop_close_phi_args): Likewise.
2475 (copy_cond_phi_nodes): Likewise.
2477 2017-03-20 Martin Liska <mliska@suse.cz>
2480 * tree-chkp.c (chkp_build_returned_bound): Do not build
2481 returned bounds for a LHS that's not a BOUNDED_P type.
2483 2017-03-20 Martin Liska <mliska@suse.cz>
2487 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
2488 COMPLEX_CST and VECTOR_CST.
2490 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2493 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
2494 target operand. A new splitter adds the clobber statement in case
2495 the target operand is dead anyway.
2497 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
2499 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
2500 to age-old versions of binutils and glibc.
2502 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
2504 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
2506 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2508 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
2510 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2512 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
2513 requirement for binutils 2.13.
2515 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2517 * combine.c (try_combine): Delete redundant i1 test. Call
2518 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2520 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
2522 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
2523 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
2525 <riscv64-*-elf>: Re-arrange section
2526 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
2527 <riscv32-*-linux>: Likewise.
2528 <riscv64-*-elf>: Likewise
2529 <riscv64-*-linux>: Likewise.
2531 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
2534 * aarch64.opt(verbose-cost-dump): Fix typo.
2536 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
2539 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
2540 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
2542 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2544 * reload.c (find_reloads): When reloading a nonoffsettable address,
2545 use RELOAD_OTHER for it and its address reloads.
2547 PR rtl-optimization/79910
2548 * combine.c (record_used_regs): New static function.
2549 (try_combine): Handle situations where there is an additional
2550 instruction between I2 and I3 which needs to have a LOG_LINK
2553 2017-03-17 Jeff Law <law@redhat.com>
2555 PR tree-optimization/71437
2556 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
2557 conditional in the hash table first.
2558 (vrp_dom_walker::before_dom_children): Extract condition from
2559 ASSERT_EXPR. Record condition, its inverion and any implied
2562 2017-03-17 Marek Polacek <polacek@redhat.com>
2563 Markus Trippelsdorf <markus@trippelsdorf.de>
2565 PR tree-optimization/80079
2566 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
2569 2017-03-17 Richard Biener <rguenther@suse.de>
2572 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
2573 Properly verify the LHS before the RHS possibly claims to be
2575 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
2578 2017-03-17 Martin Jambor <mjambor@suse.cz>
2580 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
2581 (List of -O2 options): Likewise.
2582 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
2585 2017-03-17 Tom de Vries <tom@codesourcery.com>
2587 * gcov-dump.c (print_usage): Print bug_report_url.
2589 2017-03-17 Richard Biener <rguenther@suse.de>
2592 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
2593 (parser::peek): Likewise.
2595 2017-03-17 Richard Biener <rguenther@suse.de>
2597 PR tree-optimization/80048
2598 * sese.c (free_sese_info): Properly release rename_map and
2599 copied_bb_map elements.
2601 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
2603 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
2604 Add linked-list forward and backlinks. Insert on
2605 construction, remove on destruction.
2606 (class pass_store_merging): Add m_stores_head field.
2607 (pass_store_merging::terminate_and_process_all_chains):
2608 Iterate over m_stores_head list.
2609 (pass_store_merging::terminate_all_aliasing_chains):
2611 (pass_store_merging::execute): Check for debug stmts first.
2612 Push new chains onto the m_stores_head stack.
2614 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2617 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
2618 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
2619 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
2621 2017-03-16 Jeff Law <law@redhat.com>
2623 PR tree-optimization/71437
2624 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
2625 member function. Implementation moved into after_dom_children
2626 member function and into the threader's thread_outgoing_edges
2628 (dom_opt_dom_walker::after_dom_children): Simplify by moving
2629 some code into new thread_outgoing_edges.
2630 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
2631 definition. Simplify marker handling (do it here). Assume we always
2632 have the available expression and the const/copies tables.
2633 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
2635 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
2636 * tree-vrp.c (equiv_stack): No longer file scoped.
2637 (vrp_dom_walker): New class.
2638 (vrp_dom_walker::before_dom_children): New member function.
2639 (vrp_dom_walker::after_dom_children): Likewise.
2640 (identify_jump_threads): Setup domwalker. Use it rather than
2641 walking edges in a random order by hand. Simplify setup/finalization.
2642 (finalize_jump_threads): Remove.
2643 (vrp_finalize): Do not call identify_jump_threads here.
2644 (execute_vrp): Do it here instead and call thread_through_all_blocks
2647 PR tree-optimization/71437
2648 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
2650 (simplify_stmt_for_jump_threading): Add basic_block argument. All
2652 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
2653 (dom_opt_dom_walker::thread_across_edge): Remove
2654 handle_dominating_asserts argument. All callers changed.
2655 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
2656 changes. Remove calls to lhs_of_dominating_assert. Other
2657 uses of handle_dominating_asserts turn into unconditional code
2658 (simplify_control_stmt_condition_1): Likewise.
2659 (simplify_control_stmt_condition): Likewise.
2660 (thread_through_normal_block, thread_across_edge): Likewise.
2661 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
2662 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
2663 object if it is not an SSA_NAME.
2664 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
2665 before calling into the VRP specific simplifiers.
2666 (identify_jump_threads): Remove handle_dominating_asserts
2669 2017-03-16 Jakub Jelinek <jakub@redhat.com>
2672 * tree-diagnostic.c (default_tree_printer): No longer static.
2673 * tree-diagnostic.h (default_tree_printer): New prototype.
2675 2017-03-16 Tamar Christina <tamar.christina@arm.com>
2677 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
2678 Change ins into fmov.
2680 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2682 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
2683 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
2684 Use h_con constraint for operand 1.
2685 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
2686 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
2688 2017-03-15 Jeff Law <law@redhat.com>
2690 PR tree-optimization/71437
2691 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
2692 (record_temporary_equivalences): Use it.
2694 PR tree-optimization/71437
2695 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
2696 tree-ssa-scopedtables.
2697 (lookup_avail_expr, build_and_record_new_cond): Likewise.
2698 (record_conditions, record_cond, vuse_eq): Likewise.
2699 (record_edge_info): Adjust to API tweak of record_conditions.
2700 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
2701 (record_temporary_equivalences, optimize_stmt): Likewise.
2702 (eliminate_redundant_computations): Likewise.
2703 (record_equivalences_from_stmt): Likewise.
2704 * tree-ssa-scopedtables.c: Include options.h and params.h.
2705 (vuse_eq): New function, moved from tree-ssa-dom.c
2706 (build_and_record_new_cond): Likewise.
2707 (record_conditions): Likewise. Accept vector of conditions rather
2708 than edge_equivalence structure for first argument.
2709 for the first argument.
2710 (avail_exprs_stack::lookup_avail_expr): New member function, moved
2711 from tree-ssa-dom.c.
2712 (avail_exprs_stack::record_cond): Likewise.
2713 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
2714 from tree-ssa-dom.c.
2715 (avail_exprs_stack): Add new member functions lookup_avail_expr
2717 (record_conditions): Declare.
2719 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
2722 * lra-constraints.c (process_alt_operands): Increase reject for
2723 reloading an input/output operand.
2725 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2728 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
2729 insns to convert from signed/unsigned char/short to IEEE 128-bit
2731 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
2733 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
2736 * config/i386/i386.c (ix86_vector_duplicate_value): Create
2737 subreg of inner mode for values already in registers.
2739 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
2741 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
2742 iteration reg is used after the loop.
2744 2017-03-14 Martin Sebor <msebor@redhat.com>
2746 PR tree-optimization/79800
2747 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
2748 precision in negative-positive range.
2749 (format_floating): Call non-const overload with adjusted precision.
2751 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2754 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
2757 2017-03-14 Martin Sebor <msebor@redhat.com>
2760 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
2761 * builtins.def (aligned_alloc): Use it.
2764 * Makefile.in (GTFILES): Add calls.c.
2765 * calls.c: Include "gt-calls.h".
2767 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
2769 PR rtl-optimization/79728
2770 * regs.h (struct target_regs): New field
2771 x_contains_allocatable_regs_of_mode.
2772 (contains_allocatable_regs_of_mode): New macro.
2773 * reginfo.c (init_reg_sets_1): Initialize it, and change
2774 contains_reg_of_mode so it includes global regs as well.
2775 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
2776 rather than contains_regs_of_mode.
2778 2017-03-14 Martin Liska <mliska@suse.cz>
2780 * doc/invoke.texi: Document options that can't be combined with
2781 -fcheck-pointer-bounds.
2783 2017-03-14 Martin Liska <mliska@suse.cz>
2786 * doc/invoke.texi (-Wchkp): Document the option.
2788 2017-03-14 Martin Liska <mliska@suse.cz>
2790 * Makefile.in: Install gcov-dump.
2792 2017-03-14 Martin Liska <mliska@suse.cz>
2794 * multiple_target.c (expand_target_clones): Bail out for
2795 an invalid attribute.
2797 2017-03-14 Richard Biener <rguenther@suse.de>
2799 * alias.c (struct alias_set_entry): Pack properly.
2800 * cfgloop.h (struct loop): Likewise.
2801 * cse.c (struct set): Likewise.
2802 * ipa-utils.c (struct searchc_env): Likewise.
2803 * loop-invariant.c (struct invariant): Likewise.
2804 * lra-remat.c (struct cand): Likewise.
2805 * recog.c (struct change_t): Likewise.
2806 * rtl.h (struct address_info): Likewise.
2807 * symbol-summary.h (function_summary): Likewise.
2808 * tree-loop-distribution.c (struct partition): Likewise.
2809 * tree-object-size.c (struct object_size_info): Likewise.
2810 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
2811 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
2812 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
2813 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
2814 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
2815 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
2816 (struct _stmt_vec_info): Likewise.
2818 2017-03-14 Martin Liska <mliska@suse.cz>
2821 * multiple_target.c (create_dispatcher_calls): Check that
2822 a target can create a function dispatcher.
2824 2017-03-14 Martin Liska <mliska@suse.cz>
2827 * multiple_target.c (expand_target_clones): Drop local.local
2828 flag for default implementation.
2830 2017-03-14 Richard Biener <rguenther@suse.de>
2832 PR tree-optimization/80030
2833 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
2835 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
2837 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
2838 gcc_fallthrough() instead of __attribute__((fallthrough));
2840 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2842 * doc/gcc.texi: Remove "up" link to (DIR).
2843 * doc/gccint.texi: Ditto.
2845 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2847 * doc/install.texi (Specific) <avr>: Remove reference to
2850 2017-03-13 Jeff Law <law@redhat.com>
2852 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
2853 attribute rather than comments.
2855 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
2856 match_scratch operand is highest.
2858 2017-03-13 Martin Liska <mliska@suse.cz>
2861 * ipa-pure-const.c (warn_function_noreturn): If the declarations
2862 is a CHKP clone, use original declaration.
2864 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2866 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
2867 (arc_conditional_register_usage): Use a different allocation order
2868 when optimizing for size.
2869 * common/config/arc/arc-common.c (arc_option_optimization_table):
2870 Section anchors default on when optimizing for size.
2872 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2874 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
2876 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2878 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
2879 * config/arc/arc.md (cpu_facility): Add cd variant.
2880 (*movqi_insn): Add code density variant.
2881 (*movhi_insn): Likewise.
2882 (*movqi_insn): Likewise.
2883 (*addsi3_mixed): Likewise.
2884 (subsi3_insn): Likewise.
2886 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2888 * config/arc/arc.md (movsi_cond_exec): Update constraint.
2890 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2892 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
2893 expressions with MINUS and UNARY ops.
2895 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2898 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
2900 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
2901 between vec_select and vector argument.
2902 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
2903 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
2904 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
2905 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
2906 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
2907 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
2909 2017-03-13 Richard Biener <rguenther@suse.de>
2912 * params.def (vect-max-peeling-for-alignment): Fix typo.
2914 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2916 * doc/install.texi (Specific) <mips-*-*>: Remove description of
2917 issue that only occurred with binutils below 2.18.
2919 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2921 * doc/install.texi (Specific) <cris-axis-elf>: No longer
2922 refer to binutils 2.11/2.12 minimum.
2924 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2926 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
2927 ftp.kernel.org and simplify binutils requirement.
2929 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
2931 * doc/invoke.texi (Warning Options): Fix spelling of link-time
2933 (Optimize Options): Ditto. Also remove redundancy.
2935 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2937 PR translation/79848
2938 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
2940 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
2941 to G_ to avoid double translation.
2943 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2945 PR translation/79923
2946 * auto-profile.c (get_combined_location): Convert leading
2947 character of diagnostics to lower case and remove trailing period.
2948 (read_profile): Likewise for various diagnostics.
2949 * config/arm/arm.c (arm_option_override): Remove trailing period
2950 from various diagnostics.
2951 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
2952 (msp430_expand_delay_cycles): Likewise.
2954 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2957 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
2958 full command-line argument, rather than just "str".
2959 (aarch64_validate_march): Likewise.
2960 (aarch64_validate_mtune): Likewise.
2962 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
2964 PR rtl-optimization/78911
2965 * lra-assigns.c (must_not_spill_p): New function.
2966 (spill_for): Use it.
2968 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2970 PR tree-optimization/79981
2971 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
2972 ATOMIC_COMPARE_EXCHANGE ifn result.
2973 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
2974 IFN_ATOMIC_COMPARE_EXCHANGE.
2976 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2979 * opts.c (parse_sanitizer_options): Add missing question mark to
2980 "did you mean" message.
2982 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2984 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
2986 (VMULEUH_UNS): Likewise.
2987 (VMULOUB_UNS): Likewise.
2988 (VMULOUH_UNS): Likewise.
2989 * config/rs6000/rs6000.c (builtin_function_type): Remove
2990 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
2992 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2995 * read-rtl-function.c (function_reader::read_rtx_operand): Update
2996 x with result of extra_parsing_for_operand_code_0.
2997 (function_reader::extra_parsing_for_operand_code_0): Convert
2998 return type from void to rtx, returning x. When reading
2999 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
3000 larger size containing struct block_symbol.
3002 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
3004 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
3005 -mfloat128-hardware without -m64.
3007 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
3010 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
3011 entries to the case statement that marks unsigned arguments to
3012 overloaded functions.
3014 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
3016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
3017 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
3019 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
3022 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
3023 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
3025 2017-03-10 Martin Liska <mliska@suse.cz>
3029 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
3030 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
3033 2017-03-10 Olivier Hainque <hainque@adacore.com>
3035 * tree-switch-conversion (array_value_type): Start by resetting
3036 candidate type to it's main variant.
3038 2017-03-10 Jakub Jelinek <jakub@redhat.com>
3040 PR rtl-optimization/79909
3041 * combine.c (try_combine): Use simplify_replace_rtx on individual
3042 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
3043 of the whole CALL_INSN_FUNCTION_USAGE.
3045 PR tree-optimization/79972
3046 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
3047 get_range_info on SSA_NAMEs. Formatting fixes.
3049 2017-03-10 Richard Biener <rguenther@suse.de>
3050 Jakub Jelinek <jakub@redhat.com>
3052 PR tree-optimization/77975
3053 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
3054 edge to be constant.
3055 (get_val_for): For constant x return it. Formatting fix.
3056 (loop_niter_by_eval): Avoid pointless looping if the next iteration
3057 would use the same bases as the current one.
3059 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3061 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
3062 instead of vec_select for V1TImode.
3063 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
3065 (VSX_LE_128): Add V1TI to this mode iterator.
3066 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
3067 (*vsx_le_perm_store_<mode>): Likewise.
3068 (pre-reload splitter for VSX stores): Likewise.
3069 (post-reload splitter for VSX stores): Likewise.
3070 (*vsx_xxpermdi2_le_<mode>): Likewise.
3071 (*vsx_lxvd2x2_le_<mode>): Likewise.
3072 (*vsx_stxvd2x2_le_<mode>): Likewise.
3074 2017-03-09 Michael Eager <eager@eagercon.com>
3076 Correct failures with --enable-checking=yes,rtl.
3078 * config/microblaze/microblaze.c (microblaze_expand_shift):
3079 Replace GET_CODE test with CONST_INT_P and INTVAL test with
3080 test for const0_rtx.
3081 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
3082 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
3084 2017-03-09 Richard Biener <rguenther@suse.de>
3086 PR tree-optimization/79977
3087 * graphite-scop-detection.c (scop_detection::merge_sese):
3088 Handle the case of extra exits to blocks dominating the entry.
3090 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
3092 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3095 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
3097 PR rtl-optimization/79949
3098 * lra-constraints.c (process_alt_operands): Check memory when
3099 trying to predict a cycle. Print about the overall increase.
3101 2017-03-09 Richard Biener <rguenther@suse.de>
3104 * gimple-expr.c (useless_type_conversion_p): Preserve
3105 TYPE_SATURATING for fixed-point types.
3107 2017-03-09 Richard Biener <rguenther@suse.de>
3110 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
3111 alignment of BLKmode params.
3113 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3116 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
3117 (VALL_NO_V2Q): Likewise.
3119 * config/aarch64/aarch64-simd.md
3120 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
3122 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
3123 VALL_NO_V2Q mode iterator.
3124 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
3126 2017-03-09 Martin Liska <mliska@suse.cz>
3128 PR tree-optimization/79631
3129 * tree-chkp-opt.c (chkp_is_constant_addr): Call
3130 tree_int_cst_sign_bit just for INTEGER constants.
3132 2017-03-09 Martin Liska <mliska@suse.cz>
3136 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
3139 2017-03-09 Marek Polacek <polacek@redhat.com>
3142 * tree.c (inchash::add_expr): Handle TREE_VEC.
3144 2017-03-09 Martin Liska <mliska@suse.cz>
3147 (chkp_narrow_size_and_offset): New function.
3148 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
3149 (void chkp_parse_bit_field_ref): New function.
3150 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
3151 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
3153 2017-03-09 Martin Liska <mliska@suse.cz>
3156 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
3157 (chkp_find_bounds_1): Remove gcc_unreachable.
3159 2017-03-09 Jakub Jelinek <jakub@redhat.com>
3162 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
3163 BUILT_IN_SYNC*, determine the access type from the size suffix and
3164 always build a MEM_REF with that type. Handle forgotten
3165 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
3168 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
3169 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
3170 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
3171 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
3172 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
3173 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
3174 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
3175 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
3176 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
3177 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
3178 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
3179 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
3180 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
3181 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
3182 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
3183 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
3184 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
3185 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
3186 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
3187 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
3188 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
3189 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
3190 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
3191 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
3192 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
3193 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
3194 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
3195 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
3196 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
3197 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
3198 definitions outside of __OPTIMIZE__ guarded section.
3201 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
3202 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
3203 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
3204 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
3207 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3209 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
3210 ("vfenez<mode>"): Add missing constraints.
3212 2017-03-08 Martin Sebor <msebor@redhat.com>
3215 * config/nds32/nds32.c (nds32_option_override):
3216 Fix misspelled diagnostic.
3218 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3221 * gimplify.c (gimplify_omp_for): Replace index var in outer
3222 taskloop statement with an artificial variable and add
3223 OMP_CLAUSE_PRIVATE clause for it.
3225 2017-03-08 Richard Biener <rguenther@suse.de>
3227 PR tree-optimization/79955
3228 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
3229 for accesses that are completely outside of the variable.
3231 2017-03-08 Andrew Haley <aph@redhat.com>
3233 PR tree-optimization/79943
3234 * tree-ssa-loop-split.c (compute_new_first_bound): When
3235 calculating the new upper bound, (END-BEG) should be added, not
3238 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3240 * config/avr/avr.md (setmemhi): Make sure match_dup
3241 operand number comes before match_scratch.
3243 2017-03-08 Richard Biener <rguenther@suse.de>
3245 PR tree-optimization/79920
3246 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
3247 with ncopies == 1 to ...
3248 (vect_transform_slp_perm_load): ... here. Properly compute
3249 all element loads by iterating VF times over the group. Do
3250 not handle ncopies (computed in a broken way) in
3251 vect_create_mask_and_perm.
3253 2017-03-08 Jakub Jelinek <jakub@redhat.com>
3256 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
3257 is a uniform vector, use uniform_vector_p return value instead of
3258 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
3260 2017-03-07 Marek Polacek <polacek@redhat.com>
3263 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
3264 (alloca_call_type): Likewise.
3266 2017-03-07 Martin Liska <mliska@suse.cz>
3268 * gcov.c (process_args): Put comment to correct location.
3270 2017-03-07 Martin Liska <mliska@suse.cz>
3273 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
3274 Use array_at_struct_end_p instead of DECL_CHAIN (field).
3275 (chkp_narrow_bounds_for_field): Likewise.
3276 (chkp_parse_array_and_component_ref): Pass one more argument to
3279 2017-03-07 Richard Biener <rguenther@suse.de>
3281 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
3284 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
3286 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
3287 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
3289 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3292 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
3293 to end of description.
3294 (PARAM_MAX_STORES_TO_MERGE): Likewise.
3296 2017-03-07 Jakub Jelinek <jakub@redhat.com>
3298 PR rtl-optimization/79901
3299 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
3301 (*avx512f_<code><mode>3<mask_name>): ... this.
3302 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
3303 iterator instead of VI8_AVX2_AVX512BW.
3305 PR rtl-optimization/79901
3306 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
3307 min/max expander, expand it using expand_vec_cond_expr.
3310 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
3313 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3316 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
3317 to void * for PCH reasons.
3318 * dwarf2out.c (output_loc_operands, output_die): Cast
3319 v.val_vec.array to unsigned char *.
3321 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
3324 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
3327 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
3329 PR rtl-optimization/79571
3330 * lra-constraints.c (process_alt_operands): Calculate static
3331 reject and subtract it from overall when only addresses will be
3334 2017-03-06 Julia Koval <julia.koval@intel.com>
3337 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
3338 incoming stack boundary to 128 for 64-bit targets.
3340 2017-03-06 Richard Biener <rguenther@suse.de>
3342 PR tree-optimization/79894
3343 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
3344 to NULL after folding it.
3346 2017-03-06 Richard Biener <rguenther@suse.de>
3348 PR tree-optimization/79824
3349 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
3350 check disabling peeling for gaps.
3352 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
3354 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
3355 attributes): Document gettimeofday.
3357 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3359 * config/s390/s390.c (s390_option_override_internal): Set
3360 PARAM_MIN_VECT_LOOP_BOUND
3362 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3364 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
3365 * config/s390/s390.md: Likewise.
3367 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3370 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
3371 (<avx2_avx512>_perm<mode>): Rename to ...
3372 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
3374 (<avx512>_perm<mode>_mask): Rename to ...
3375 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
3377 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
3378 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
3379 instead of VI8F_256_512.
3380 (avx512f_perm<mode>): New define_expand.
3381 (avx512f_perm<mode>_mask): Likewise.
3382 (avx512f_perm<mode>_1<mask_name>): New define_insn.
3383 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
3385 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3387 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
3388 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
3390 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
3392 2017-03-06 Martin Liska <mliska@suse.cz>
3395 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
3396 when having a SSA NAME w/o VAR_DECL assigned to it.
3398 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3400 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
3401 msa_dpsub_<su>_d): Fix MODE for vec_select.
3403 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3405 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
3407 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
3409 2017-03-06 Richard Biener <rguenther@suse.de>
3411 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
3412 * plugin.c (register_plugin_info): Likewise.
3413 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
3415 2017-03-05 Jakub Jelinek <jakub@redhat.com>
3417 * config/i386/sse.md (sse_storehps, sse_storelps,
3418 avx_<castmode><avxsizesuffix>_<castmode>,
3419 avx512f_<castmode><avxsizesuffix>_<castmode>,
3420 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
3421 in condition that at least one operand is not a MEM.
3423 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3426 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
3427 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
3429 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
3430 gimple_call_nothrow_p flag based on whether original builtin can throw.
3431 If it can, emit following stmts on the fallthrough edge.
3432 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
3433 don't create new bb if inserting just debug stmts on the edge, try to
3434 insert them on the fallthru bb or just reset debug stmts.
3436 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
3439 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
3440 restore recog_data (including the operand rtxes inside it) around
3441 the call to get_insn_template.
3443 2017-03-03 Martin Sebor <msebor@redhat.com>
3445 PR tree-optimization/79699
3446 * context.c (context::~context): Free MPFR caches to avoid
3447 a memory leak on program exit.
3449 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3451 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3452 Use wide_int::ulow () instead of .elt (0).
3454 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3456 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
3457 (*pushxf): Limit oF constraint to 32bit targets and add oC
3458 constraint for 64bit targets.
3459 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
3460 (*pushdf): Change rmF constraint to rmC.
3462 2017-03-03 Martin Liska <mliska@suse.cz>
3464 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
3465 Remove unused variable.
3467 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3470 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
3471 is a memory operand, increase num_memory.
3472 (ix86_expand_args_builtin): Likewise.
3474 2017-03-03 Jan Hubicka <jh@suse.cz>
3477 * ipa-devirt.c (maybe_record_node): Properly handle
3478 __cxa_pure_virtual visibility.
3480 2017-03-03 Martin Liska <mliska@suse.cz>
3482 PR tree-optimization/79803
3483 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
3485 (pass_loop_prefetch::execute): Disabled optimization if an
3486 assumption about L1 cache size is not met.
3488 2017-03-03 Martin Liska <mliska@suse.cz>
3490 PR rtl-optimization/79574
3491 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
3492 (hash_scan_set): Likewise.
3493 (dump_hash_table): Likewise.
3494 (hoist_code): Likewise.
3496 2017-03-03 Richard Biener <rguenther@suse.de>
3498 * fixed-value.c (fixed_from_string): Restore use of elt (1)
3499 in place of uhigh ().
3500 (fixed_convert_from_real): Likewise.
3502 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3505 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
3507 2017-03-03 Richard Biener <rguenther@suse.de>
3510 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
3511 TYPE_OVERFLOW_UNDEFINED check.
3513 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3515 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
3517 (vector_ae_<mode>_p): Likewise.
3518 (vector_nez_<mode>_p): Likewise.
3519 (vector_ne_v2di_p): Likewise.
3520 (vector_ae_v2di_p): Likewise.
3521 (vector_ne_<mode>_p): Likewise.
3522 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
3524 (vsx_tsqrt<mode>2_fe): Likewise.
3526 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
3529 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
3531 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3533 PR rtl-optimization/79780
3534 * cprop.c (one_cprop_pass): When second and further conditional trap
3535 in a single basic block is turned into an unconditional trap, turn it
3536 into a deleted note to avoid RTL verification failures.
3538 2017-03-02 Richard Biener <rguenther@suse.de>
3540 * fold-const.c (const_binop): Use ulow () instead of elt (0).
3542 2017-03-02 Richard Biener <rguenther@suse.de>
3544 PR tree-optimization/79345
3546 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
3547 param and abort the walk, returning -1 if it is hit.
3548 (walk_aliased_vdefs): Take a limit param and pass it on.
3549 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
3550 defaulting to 0 and return a signed int.
3551 * tree-ssa-uninit.c (struct check_defs_data): New struct.
3552 (check_defs): New helper.
3553 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
3554 about uninitialized memory.
3555 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
3556 bogus uninitialized warning.
3557 (fixed_convert_from_real): Likewise.
3559 2017-03-02 Bin Cheng <bin.cheng@arm.com>
3561 PR tree-optimization/66768
3562 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
3563 iv_use if base object can't be determined.
3565 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3567 PR tree-optimization/79345
3568 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
3569 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
3570 (get_pattern_stats): Initialize it.
3571 * genemit.c (gen_expand): Verify match_scratch numbers come after
3572 match_operand/match_dup numbers.
3573 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
3574 match_scratch numbers.
3575 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
3577 * config/s390/s390.md (trunctdsd2): Likewise.
3579 2017-03-02 Richard Biener <rguenther@suse.de>
3581 * wide-int.h (wide_int_storage::operator=): Implement in terms
3584 2017-03-02 Richard Biener <rguenther@suse.de>
3586 PR tree-optimization/79777
3587 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
3588 the to insert expression to sth existing.
3590 2017-03-01 Martin Sebor <msebor@redhat.com>
3593 * gimple-ssa-sprintf.c
3594 (directive::known_width_and_precision): New function.
3595 (format_integer): Use it.
3596 (get_mpfr_format_length): Consider the full range of precision
3597 when computing %g output with the # flag. Set the likely byte
3598 count to 3 rather than 1 when precision is indeterminate.
3599 (format_floating): Correct the lower bound of precision.
3601 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3603 * doc/invoke.texi: Document default code model for 64-bit Linux.
3605 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3608 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
3609 udiv rather than div since input pattern is unsigned.
3611 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
3613 * config/i386/i386.c (print_reg): Warn for values of
3614 unsupported size in integer register.
3616 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3619 * config/rs6000/predicates.md (current_file_function_operand): Do
3620 not allow self calls to be local if the function is replaceable.
3622 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
3625 * config/rs6000/altivec.h (vec_ctz and others): Change the
3626 preprocessor macro that controls conditional compilation from
3627 _ARCH_PWR9 to __POWER9_VECTOR__.
3628 (vec_all_ne): Change parameterization of __altivec_scalar_pred
3629 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
3630 control (instead of _ARCH_PWR9 control) so that template
3631 definition uses power9-specific function.
3632 (vec_any_eq): Likewise.
3633 (vec_all_ne): Change macro definition to use a power9-specific
3634 expansion under #ifdef __POWER9_VECTOR__ control (instead of
3635 _ARCH_PWR9 control).
3636 (vec_any_eq) Likewise.
3637 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
3638 expansion for CMPNEF to remove support for xvcmpnesp instruction.
3639 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
3640 support for xvcmpnedp instruction.
3641 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
3642 macro expansion so that Power9 implementation of vec_all_ne does
3643 not use the AltiVec predicate framework.
3644 (VCMPNEH_P): Likewise.
3645 (VCMPNEW_P): Likewise.
3646 (VCMPNED_P): Likewise.
3647 (VCMPNEFP_P): Likewise.
3648 (VCMPNEDP_P): Likewise.
3649 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
3650 implementation of vec_any_eq to not use AltiVec predicate
3652 (VCMPAEH_P): Likewise.
3653 (VCMPAEW_P): Likewise.
3654 (VCMPAED_P): Likewise.
3655 (VCMPAEFP_P): Likewise.
3656 (VCMPAEDP_P): Likewise.
3657 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
3658 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
3659 not use the AltiVec predicate framework.
3660 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
3661 of vec_any_eq to not use AltiVec predicate framework.
3662 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
3663 support for predefined __POWER9_VECTOR__ macro to indicate that
3664 Power9 instruction selection is enabled.
3665 (altivec_overloaded_builtins): Remove extraneous
3666 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
3667 function argument types RS6000_BTI_bool_V16QI and
3668 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
3669 entry for overloaded function argument types RS6000_BTI_bool_V4SI
3670 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
3671 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
3672 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
3673 Power9 for implementations of vec_cmpne. Change the signature for
3674 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
3675 (representing vec_all_ne) to remove the previously described first
3676 argument of type RS6000_BTI_INTSI, as this was an artifact of
3677 reliance on the AltiVec predicate framework, which is no longer
3678 used in the implementation of these functions. Add
3679 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
3680 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
3681 since, unlike the AltiVec predicate framework implementation, we
3682 do not share function descriptors between vec_alle and vec_anyeq.
3683 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
3684 set of modes that receive special treatment even when
3685 TARGET_P9_VECTOR is true. The special treatment emits code that
3686 does not depend on Power9 instructions.
3687 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
3688 define_expand to not rely on AltiVec predicate framework.
3689 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3691 (vector_ne_v2di_p): Change this define_expand to not rely on
3692 AltiVec predicate framework.
3693 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
3695 (vector_ne_<mode>_p): Change this define_expand to not rely on
3696 AltiVec predicate framework.
3697 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3699 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
3700 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
3701 define_insn pattern.
3702 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
3703 define_insn pattern because the xvcmpne<VSs>. instruction is not
3705 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
3706 instruction is not supported.
3708 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3710 * config/nvptx/nvptx.c: Include intl.h.
3712 2017-03-01 Martin Jambor <mjambor@suse.cz>
3715 * ipa-prop.h (ipa_bits): Removed field known.
3716 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
3717 to pointers. Adjusted their comments to warn about their sharing.
3718 (ipcp_transformation_summary): Change bits to a vector of pointers.
3719 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
3720 (ipa_get_ipa_bits_for_value): Declare.
3721 * tree-vrp.h (value_range): Mark as GTY((for_user)).
3722 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
3723 (ipa_bits_hash_table): Likewise.
3724 (ipa_vr_ggc_hash_traits): Likewise.
3725 (ipa_vr_hash_table): Likewise.
3726 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
3727 being pointers and vr_known being removed.
3728 (ipa_set_jf_unknown): Likewise.
3729 (ipa_get_ipa_bits_for_value): New function.
3730 (ipa_set_jfunc_bits): Likewise.
3731 (ipa_get_value_range): New overloaded functions.
3732 (ipa_set_jfunc_vr): Likewise.
3733 (ipa_compute_jump_functions_for_edge): Use the above functions to
3734 construct bits and vr parts of jump functions.
3735 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
3736 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3738 (ipcp_grow_transformations_if_necessary): Also allocate
3739 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3741 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
3742 them. Fix too long lines.
3743 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
3744 vr_known being removed.
3745 (ipa_read_jump_function): Use new setter functions to construct bits
3746 and vr parts of jump functions or set them to NULL.
3747 (write_ipcp_transformation_info): Adjust for bits being pointers.
3748 (read_ipcp_transformation_info): Likewise.
3749 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
3751 Include gt-ipa-prop.h.
3752 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
3754 (ipcp_store_bits_results): Likewise.
3755 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
3756 Do not write to existing jump functions but use a temporary instead.
3758 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3761 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
3762 attempt to use its first operand as BIT_FIELD_REF base.
3764 2017-03-01 Richard Biener <rguenther@suse.de>
3767 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
3768 interpolating formula in wrapping arithmetic.
3769 (chrec_apply): Convert chrec_evaluate return value to wanted type.
3771 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3773 PR tree-optimization/79734
3774 * tree-vect-generic.c (expand_vector_condition): Optimize
3775 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
3776 Handle VEC_COND_EXPR where comparison has different inner width from
3779 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
3781 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
3782 markup, and similar issues. Remove @opindex entries for things
3783 that aren't options. Add missing -mmpy-option entries.
3785 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3787 PR tree-optimization/79737
3788 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
3789 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
3790 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
3791 instead of byte_size. Formatting fix.
3792 (shift_bytes_in_array_right): Formatting fix.
3794 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
3797 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
3798 condition on optimize for the leaf function test.
3800 2017-02-28 Martin Liska <mliska@suse.cz>
3803 * read-rtl-function.c (function_reader::handle_unknown_directive):
3804 Bail out when one uses -flto.
3806 2017-02-28 Martin Liska <mliska@suse.cz>
3808 * common.opt: Replace space with tabular for options of <number>
3810 * config/i386/i386.opt: Show <number> value for
3811 -mlarge-data-threshold.
3812 * opts.c (print_filtered_help): Do not display number in hexadecimal
3815 2017-02-28 Martin Liska <mliska@suse.cz>
3817 * common.opt: Fix --help=option -Q for options which are of
3820 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
3822 * config/i386/i386.c (print_reg): Error out for values
3823 of 8-bit size in invalid integer register.
3825 2017-02-28 Martin Sebor <msebor@redhat.com>
3827 PR tree-optimization/79691
3828 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
3830 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3833 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
3834 gcc_unreachable with output_operand_lossage.
3836 2017-02-28 Richard Biener <rguenther@suse.de>
3838 PR tree-optimization/79740
3839 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
3841 (visit_nary_op): Insert the nary into the hashtable if we
3842 pattern-matched sth.
3843 * tree-ssa-pre.c (eliminate_insert): Robustify.
3845 2017-02-28 Richard Biener <rguenther@suse.de>
3848 * fold-const.c (decode_field_reference): Reject out-of-bound
3851 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3853 * config/i386/i386.c: Include intl.h.
3854 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
3855 instead of just cond ? "..." : "...".
3856 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
3857 * coverage.c (read_counts_file): Likewise.
3858 * omp-offload.c: Include intl.h.
3859 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
3860 of just cond ? "..." : "...".
3861 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
3862 of just cond ? "..." : "...".
3864 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
3867 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
3869 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
3870 'tune for' CPU name.
3871 * config/arm/arm-cpu-data.h: Regenerated.
3873 2017-02-28 Richard Biener <rguenther@suse.de>
3875 PR tree-optimization/79732
3876 * tree-inline.c (expand_call_inline): Do not shadow var.
3878 2017-02-28 Richard Biener <rguenther@suse.de>
3880 PR tree-optimization/79723
3881 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
3882 address-space properly.
3884 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
3886 * doc/optinfo.texi (Optimization groups): Fix option used for
3888 * doc/invoke.texi (-fopt-info): Document "omp".
3889 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
3890 (OPTGROUP_ALL): Add OPTGROUP_OMP.
3891 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
3892 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
3893 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
3895 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
3897 * dumpfile.c (optgroup_options): Instead of "openmp", associate
3898 OPTGROUP_OMP with "omp".
3900 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
3903 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
3904 for arithmetic shift of unsigned V2DI.
3906 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
3908 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
3909 arc/linux.h headers.
3910 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
3911 (LINK_SPEC): Likewise.
3912 (ARC_TLS_EXTRA_START_SPEC): Likewise.
3913 (EXTRA_SPECS): Likewise.
3914 (STARTFILE_SPEC): Likewise.
3915 (ENDFILE_SPEC): Likewise.
3916 (LIB_SPEC): Likewise.
3917 (TARGET_SDATA_DEFAULT): Likewise.
3918 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
3919 (MULTILIB_DEFAULTS): Likewise.
3920 (DWARF2_UNWIND_INFO): Likewise.
3921 * config/arc/big.h: New file.
3922 * config/arc/elf.h: Likewise.
3923 * config/arc/linux.h: Likewise.
3924 * config/arc/t-uClibc: Remove.
3926 2017-02-27 Bin Cheng <bin.cheng@arm.com>
3928 PR tree-optimization/77536
3929 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
3930 (tree_transform_and_unroll_loop): Use above function to compute the
3931 estimated niter of unrolled loop and use it when scaling profile.
3932 Also use count info rather than frequency if it's non-zero.
3933 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
3934 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
3935 (vect_transform_loop): Call above function.
3937 2017-02-27 Richard Biener <rguenther@suse.de>
3939 PR tree-optimization/45397
3940 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
3941 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
3942 (visit_nary_op): Add pattern matching for CSEing sign-changed
3943 or truncated operations with wider ones.
3945 2017-02-27 Richard Biener <rguenther@suse.de>
3947 PR tree-optimization/79690
3948 * tree-vect-stmts.c (vectorizable_store): Use vector type
3949 built from the DR with address-space.
3951 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
3953 * doc/invoke.texi (Optimize Options): Refine the description
3954 of asan-use-after-return.
3956 2017-02-25 Alan Modra <amodra@gmail.com>
3958 PR rtl-optimization/79584
3959 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
3960 base, not ad->base_term, the reg within base. Remove assertion
3961 that ad->base == ad->base_term. Replace gen_int_mode using
3962 bogus mode with const0_rtx.
3964 2017-02-25 Jakub Jelinek <jakub@redhat.com>
3967 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
3968 FMA_EXPR like tcc_binary or tcc_unary.
3970 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
3973 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
3975 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
3976 (output_loc_operands): Handle DW_OP_call_ref and
3977 DW_OP_GNU_variable_value.
3978 (struct variable_value_struct): New type.
3979 (struct variable_value_hasher): Likewise.
3980 (variable_value_hash): New variable.
3981 (string_types): Remove.
3982 (copy_loc_descr): New function.
3983 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
3984 (prepend_loc_descr_to_each): New function.
3985 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
3986 instead of add_loc_descr_to_each if the first argument is single
3987 location list and the second has multiple.
3988 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
3989 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
3990 when looking for variable value which doesn't have other location info.
3991 (loc_list_from_tree): Formatting fix.
3992 (gen_array_type_die): Simplify DW_AT_string_length handling.
3993 (adjust_string_types): Remove.
3994 (gen_subprogram_die): Don't call adjust_string_types nor test/set
3995 string_types. Call resolve_variable_values.
3996 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
3997 (resolve_addr_in_expr): Likewise. Add A argument.
3998 (copy_deref_exprloc): Remove deref argument. Adjust for the
3999 original expression being DW_OP_GNU_variable_value with optionally
4000 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
4001 optionally after it.
4002 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
4003 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
4004 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
4005 (variable_value_hasher::hash, variable_value_hasher::equal): New
4007 (resolve_variable_value_in_expr, resolve_variable_value,
4008 resolve_variable_values, note_variable_value_in_expr,
4009 note_variable_value): New functions.
4010 (dwarf2out_early_finish): Call note_variable_value on all toplevel
4013 2017-02-24 Jakub Jelinek <jakub@redhat.com>
4016 * opts.h (handle_generated_option): Add GENERATED_P argument.
4017 * opts-common.c (handle_option): Adjust function comment.
4018 (handle_generated_option): Add GENERATED_P argument, pass it to
4020 (control_warning_option): Pass false to handle_generated_option
4022 * opts.c (maybe_default_option): Pass true to handle_generated_option
4024 * optc-gen.awk: Likewise.
4026 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4028 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
4029 a REG, look at the REG it is a SUBREG of.
4030 (splitter for cmpeqsi_t): Ditto.
4032 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4034 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
4035 the special USEs with the pattern of the insn, not the insn itself.
4037 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
4040 * doc/invoke.texi: Document -mload-store-pairs.
4042 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4043 Sandra Loosemore <sandra@codesourcery.com>
4045 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
4046 argument isn't a CONST_INT.
4047 (nios2_alternate_compare_const): Assert op is a CONST_INT.
4048 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
4049 (nios2_validate_compare): Bypass alternate compare logic if *op2
4051 (ldstwm_operation_p): Return false if first_base is not a REG or
4052 if first_offset is not a CONST_INT.
4054 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4056 * config/cris/cris.md: Use correct operand in a define_peephole2.
4058 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4060 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
4062 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
4064 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
4065 this_insn if it is an INSN or JUMP_INSN.
4066 (force_offsettable): Look at base, not at addr.
4067 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
4068 on things that aren't necessarily CONST_INTs.
4070 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
4072 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
4073 -mfpmath=sse is the default also for x86-32 targets with SSE2
4074 instruction set when @option{-ffast-math} is enabled
4076 2017-02-24 Jeff Law <law@redhat.com>
4078 PR rtl-optimizatoin/79286
4079 * ira.c (update_equiv_regs): Drop may_trap_p exception to
4082 2017-02-24 Richard Biener <rguenther@suse.de>
4084 PR tree-optimization/79389
4085 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
4088 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
4090 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
4091 function comment to reflect reality.
4092 (loop_exits_before_overflow): Fix typo in function description.
4094 2017-02-24 Richard Biener <rguenther@suse.de>
4096 PR tree-optimization/79389
4097 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
4098 properly that a threading opportunity exists. Detect conditional
4099 copy/constant propagation opportunities.
4101 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
4103 * config/visium/visium.md (type): Add trap.
4104 (b): New mode attribute.
4105 (*btst): Rename into...
4106 (*btst<mode>): ...this and adjust.
4107 (*cbranchsi4_btst_insn): Rename into...
4108 (*cbranch<mode>4_btst_insn): ...this and adjust.
4109 (trap): New define_insn.
4111 2017-02-23 Jakub Jelinek <jakub@redhat.com>
4113 PR tree-optimization/79389
4114 * ifcvt.c (struct noce_if_info): Add rev_cond field.
4115 (noce_reversed_cond_code): New function.
4116 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
4117 reversed_comparison_code. Formatting fix.
4118 (noce_try_store_flag): Test rev_cond != NULL in addition to
4119 reversed_comparison_code.
4120 (noce_try_store_flag_constants): Likewise.
4121 (noce_try_store_flag_mask): Likewise.
4122 (noce_try_addcc): Use rev_cond if non-NULL instead of
4123 reversed_comparison_code.
4124 (noce_try_cmove_arith): Likewise. Formatting fixes.
4125 (noce_try_minmax, noce_try_abs): Clear rev_cond.
4126 (noce_find_if_block): Initialize rev_cond.
4127 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
4128 instead of false as last argument never attempt to reverse it
4131 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4133 PR tree-optimization/79663
4134 * tree-predcom.c (combine_chains): Process refs in reverse order
4135 only for ZERO length chains, and add explaining comment.
4137 2017-02-23 Jeff Law <law@redhat.com>
4139 PR tree-optimization/79578
4140 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
4141 in call to operand_equal_p.
4143 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
4146 * config/i386/cpuid.h: Fix another undefined behavior.
4148 2017-02-23 Richard Biener <rguenther@suse.de>
4150 PR tree-optimization/79683
4151 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
4152 vector types for data-refs.
4154 2017-02-23 Martin Liska <mliska@suse.cz>
4156 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
4158 2017-02-23 Jakub Jelinek <jakub@redhat.com>
4161 * internal-fn.c (get_range_pos_neg): Moved to ...
4162 * tree.c (get_range_pos_neg): ... here. No longer static.
4163 * tree.h (get_range_pos_neg): New prototype.
4164 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
4165 are known to be in between 0 and signed maximum inclusive, try to
4166 expand both unsigned and signed divmod and use the cheaper one from
4169 2017-02-22 Jeff Law <law@redhat.com>
4171 PR tree-optimization/79578
4172 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
4173 to compare base operands.
4175 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
4178 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
4179 gpc_reg_operand instead of fpr_reg_operand.
4181 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
4183 * config/mips/mips.c (mips_return_in_memory): Force FP
4184 vector types to be returned in memory for o32 ABI.
4186 2017-02-22 Jakub Jelinek <jakub@redhat.com>
4188 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
4189 instead of DW_TAG_member for static data member declarations and don't
4190 set no_linkage_name for static inline data members.
4191 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
4194 2017-02-22 Martin Liska <mliska@suse.cz>
4196 * doc/invoke.texi: Replace inequality signs with square brackets
4199 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4201 PR tree-optimization/68644
4202 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
4204 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
4207 * lra-constraints.c (simplify_operand_subreg): Handle
4208 WORD_REGISTER_OPERATIONS targets.
4210 2017-02-22 Jakub Jelinek <jakub@redhat.com>
4213 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
4214 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
4215 elimination by swapping fld*.
4217 2017-02-22 Richard Biener <rguenther@suse.de>
4219 PR tree-optimization/79673
4220 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
4221 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
4222 irrelevant address-space qualifiers and avoiding a
4223 ADDR_SPACE_CONVERT_EXPR from fold_convert.
4225 2017-02-22 Richard Biener <rguenther@suse.de>
4227 PR tree-optimization/79666
4228 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
4229 to not symbolically negate if that may introduce undefined
4232 2017-02-22 Martin Liska <mliska@suse.cz>
4235 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
4236 * data-streamer-out.c (streamer_write_gcov_count_stream):
4238 * value-prof.c (stream_out_histogram_value): Make assert more
4239 precise based on type of counter.
4241 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
4244 * config/i386/i386.md (standard_x87sse_constant_load splitter):
4245 Use nonimmediate_operand instead of memory_operand for operand 1.
4246 (float-extend standard_x87sse_constant_load splitter): Ditto.
4248 2017-02-21 Jeff Law <law@redhat.com>
4250 PR tree-optimization/79621
4251 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
4252 blocks with edges to themselves.
4254 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4257 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
4258 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
4259 Use gimple_call_builtin_p.
4262 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
4263 on temporarily removed DEBUG_INSNs.
4265 PR tree-optimization/79649
4266 * tree-loop-distribution.c (classify_partition): Give up on
4267 non-generic address space loads/stores.
4269 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
4271 * doc/loop.texi (Loop manipulation): Remove nonexistent
4272 tree_ssa_loop_version from the documentation.
4273 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
4275 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4278 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
4279 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
4280 * config/rs6000/rs6000.c: Include except.h.
4281 (rs6000_expand_split_stack_prologue): Call
4282 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
4284 2017-02-21 Martin Jambor <mjambor@suse.cz>
4287 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
4290 2017-02-21 Martin Jambor <mjambor@suse.cz>
4292 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
4293 for backward compatibility only.
4294 * doc/invoke.texi (Option Summary): Remove all references to
4297 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
4301 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4303 * lra-constraints.c (curr_insn_transform): Handle
4304 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4306 2017-02-21 Martin Liska <mliska@suse.cz>
4308 * config/i386/i386.opt: Replace -masm-dialect with -masm.
4310 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
4312 PR translation/79638
4313 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
4315 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
4318 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
4319 (arm_function_ok_for_sibcall): Return false for an indirect call by
4320 descriptor if all the argument registers are used.
4321 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
4322 alignment of the function.
4324 2017-02-21 Jakub Jelinek <jakub@redhat.com>
4326 PR tree-optimization/61441
4327 * simplify-rtx.c (simplify_const_unary_operation): For
4328 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
4329 the sNaN unmodified.
4331 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4333 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
4334 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
4335 instead of SYSTEM_HEADER_DIR.
4337 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
4338 Martin Liška <mliska@suse.cz>
4340 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
4341 Fix typos and grammar, use active voice, and clarify.
4343 2017-02-20 Marek Polacek <polacek@redhat.com>
4346 * gimplify.c (gimplify_expr): Handle unused *&&L;.
4349 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
4351 2017-02-20 Jakub Jelinek <jakub@redhat.com>
4354 * config/i386/i386.c (ix86_expand_builtin): Handle
4355 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
4356 ix86_builtins_isa[fcode].isa as a requirement of those
4357 flags and any other flag in the bitmask.
4358 (ix86_init_mmx_sse_builtins): Use 0 instead of
4359 ~OPTION_MASK_ISA_64BIT as mask.
4360 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
4361 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
4362 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
4363 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
4365 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4368 * lra-constraints.c (split_reg): Check requested split mode
4369 is supported by the register.
4371 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4373 * lra-constraints.c (simplify_operand_subreg): Remove early
4376 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4379 * lra-constraints.c (curr_insn_transform): Tighten condition
4380 for converting SUBREG reloads from OP_OUT to OP_INOUT.
4382 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4385 * lra-constraints.c (curr_insn_transform): Handle
4386 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4388 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
4391 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
4393 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
4395 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
4398 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
4401 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4403 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
4406 2017-02-18 Jakub Jelinek <jakub@redhat.com>
4408 * final.c (last_columnnum, override_columnnum): New variables.
4409 (final_start_function): Set last_columnnum, pass it to begin_prologue
4410 hook and pass 0 to dwarf2out_begin_prologue.
4411 (final_scan_insn): Update override_columnnum. Pass last_columnnum
4412 to source_line debug hook.
4413 (notice_source_line): Compute last_columnnum and for debug_column_info
4414 return true on column changes.
4415 * debug.h (struct gcc_debug_hooks): Add column argument to
4416 source_line and begin_prologue hooks.
4417 (debug_nothing_int_charstar_int_bool): Remove prototype.
4418 (debug_nothing_int_int_charstar,
4419 debug_nothing_int_int_charstar_int_bool): New prototypes.
4420 (dwarf2out_begin_prologue): Add column argument.
4421 * debug.c (do_nothing_debug_hooks): Adjust source_line and
4422 begin_prologue hooks.
4423 (debug_nothing_int_charstar_int_bool): Remove.
4424 (debug_nothing_int_int_charstar,
4425 debug_nothing_int_int_charstar_int_bool): New functions.
4426 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
4427 through to dwarf2out_source_line.
4428 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
4429 (dwarf2out_source_line): Add column argument, emit it if requested.
4430 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
4432 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4433 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4434 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
4435 through to dwarf2out_begin_prologue.
4436 (vmsdbgout_source_line): Add column argument, pass it through to
4437 dwarf2out_source_line.
4438 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
4439 dbxout_source_line caller.
4440 (dbxout_source_line): Add column argument.
4442 * common.opt (gno-column-info, gcolumn-info): New options.
4443 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
4444 (check_die): Also test for multiple DW_AT_decl_column attributes.
4445 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
4446 DW_AT_decl_column if requested.
4447 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
4449 (gen_variable_die): Likewise.
4450 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
4451 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
4454 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
4455 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
4456 (ix86_handle_option): Handle OPT_m3dnowa.
4457 * doc/invoke.texi (-m3dnowa): Document.
4458 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
4459 -m3dnowa instead of -m3dnow -march=athlon.
4462 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
4463 instead of gcc_assert for K, r and R code checks. Formatting fixes.
4465 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4468 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
4469 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
4470 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
4471 generator for vsx_xxpermdi_<mode>_be.
4472 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
4473 force big-endian semantics.
4474 (vsx_xxpermdi_<mode>_be): New define_expand with same
4475 implementation as previous version of vsx_xxpermdi_<mode>.
4477 2017-02-17 Jakub Jelinek <jakub@redhat.com>
4479 PR tree-optimization/79327
4480 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
4481 variable, its initialization and use.
4483 2017-02-17 Julia Koval <julia.koval@intel.com>
4485 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
4486 (OPTION_MASK_ISA_PKU_UNSET): New.
4487 (ix86_handle_option): Handle -mrdpid.
4488 * config/i386/cpuid.h (bit_RDPID): New.
4489 * config/i386/driver-i386.c (host_detect_local_cpu):
4490 Detect RDPID feature.
4491 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
4492 * config/i386/i386-c.c (ix86_target_macros_internal):
4494 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
4495 (ix86_valid_target_attribute_inner_p): Add "rdpid".
4496 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
4497 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
4498 * config/i386/i386.md (define_insn "rdpid"): New.
4499 * config/i386/i386.opt Add -mrdpid.
4500 * config/i386/immintrin.h (_rdpid_u32): New.
4502 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
4504 PR rtl-optimization/79541
4505 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
4506 instead of transforming it into USE.
4508 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
4510 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
4511 If HONOR_SNANS (SFmode) force the input to a register.
4512 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
4513 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
4514 an frsp or similar insn.
4516 2017-02-17 Martin Liska <mliska@suse.cz>
4518 PR rtl-optimization/79577
4519 * params.def (selsched-max-sched-times): Increase minimum to 1.
4521 2017-02-17 Martin Liska <mliska@suse.cz>
4523 PR rtl-optimization/79574
4524 * gcse.c (want_to_gcse_p): Prevent integer overflow.
4526 2017-02-17 Martin Liska <mliska@suse.cz>
4528 PR tree-optimization/79529
4529 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
4530 ssa_defined_default_def_p to handle cases which are implicitly
4532 * tree-ssa.c (ssa_defined_default_def_p): New function.
4533 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
4534 which are implicitly defined.
4535 * tree-ssa.h (ssa_defined_default_def_p): Declare.
4537 2017-02-17 Richard Biener <rguenther@suse.de>
4540 * params.def (max-ssa-name-query-depth): Limit to 10.
4542 2017-02-17 Richard Biener <rguenther@suse.de>
4544 PR tree-optimization/79552
4545 * tree-ssa-structalias.c (visit_loadstore): Properly verify
4548 2017-02-17 Richard Biener <rguenther@suse.de>
4551 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
4553 2017-02-17 Marek Polacek <polacek@redhat.com>
4556 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
4557 (fold_negate_expr): New wrapper.
4559 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
4561 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
4562 Correct terminology and de-emphasize pre-standard behavior.
4564 2017-02-16 Alan Modra <amodra@gmail.com>
4566 PR rtl-optimization/79286
4567 * ira.c (def_dominates_uses): New function.
4568 (update_equiv_regs): Don't create an equivalence for insns that
4569 may trap where the register def does not dominate the use.
4571 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
4573 PR rtl-optimization/78127
4574 * lra.c (lra): Call lra_eliminate before finish the loop after
4577 2017-02-16 Richard Biener <rguenther@suse.de>
4579 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
4581 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
4582 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
4583 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
4584 (isl_val_int_from_wi): New function.
4585 (extract_affine_gmp): Rename to ...
4586 (extract_affine_wi): ... this, take a widest_int.
4587 (extract_affine_int): Just wrap extract_affine_wi.
4588 (add_param_constraints): Use isl_val_int_from_wi.
4589 (add_loop_constraints): Likewise, and extract_affine_wi.
4591 2017-02-15 Jeff Law <law@redhat.com>
4594 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
4595 ira_init_register_move_cost_if_necessary.
4597 2017-02-15 Martin Sebor <msebor@redhat.com>
4600 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
4601 removed in a prior commit.
4603 2017-02-15 Bin Cheng <bin.cheng@arm.com>
4605 PR tree-optimization/79347
4606 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
4607 counters during peeling.
4609 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
4611 * Makefile.in (site.exp): Remove "set ISLVER".
4613 2017-02-15 Jakub Jelinek <jakub@redhat.com>
4616 * real.c (real_from_integer): Call real_convert even for decimal.
4618 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
4621 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
4623 2017-02-14 Andrew Pinski <apinski@cavium.com>
4625 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
4626 cores and change the partno/implementer to be correct.
4627 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
4628 the 'B" as the implementer.
4629 * config/aarch64/aarch64-tune.md: Regenerate.
4631 2017-02-14 Carl Love <cel@us.ibm.com>
4633 * config/rs6000/rs6000.c: Add case statement entry to make the
4634 xvcvuxdsp built-in argument unsigned.
4635 * config/rs6000/vsx.md: Fix the source and return operand types so they
4636 match the instruction definitions from the ISA document. Fix typo
4637 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
4640 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
4643 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
4644 member early_clobber_alts.
4645 * lra-lives.c (reg_early_clobber_p): New.
4646 (process_bb_lives): Use it.
4647 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
4648 (debug_operand_data): Initialize early_clobber_alts.
4649 (setup_operand_alternative): Set up early_clobber_alts.
4650 (collect_non_operand_hard_regs): Ditto. Pass early clobber
4651 alternatives to new_insn_reg.
4652 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
4654 (lra_update_insn_regno_info): Pass the new arg.
4656 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4659 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
4660 (new_oacc_loop_raw): Don't clear already cleared fields.
4663 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
4664 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
4665 _mm512_prefetch_i64gather_ps): New inline functions and macros.
4667 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
4670 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
4672 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
4675 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
4676 the extra instruction to the right place to store 128-bit constant
4679 2017-02-14 Martin Sebor <msebor@redhat.com>
4682 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
4683 warning for strings of unknown length.
4685 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
4687 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
4689 2017-02-14 Jeff Law <law@redhat.com>
4692 * ira-costs.c (scan_one_insn): Initialize register move costs
4693 for pseudos seen in USE/CLOBBER insns.
4695 PR tree-optimization/79095
4696 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
4697 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
4698 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
4699 if the operands are known to be not equal, then the resulting range
4701 (intersect_ranges): If the new range is ~[0,0] and the old range is
4702 wide, then prefer ~[0,0].
4703 * tree-vrp.c (overflow_comparison_p_1): New function.
4704 (overflow_comparison_p): New function.
4705 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
4706 if NAME is used in an overflow test.
4707 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
4708 overflow check that can be expressed as an equality test, then adjust
4709 ops to be that equality test.
4711 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4713 * config/s390/s390-builtin-types.def: Remove flags argument.
4714 * config/s390/s390.c (s390_init_builtins): Likewise.
4716 2017-02-14 Martin Liska <mliska@suse.cz>
4718 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
4719 vector. Fix trailing white spaces.
4721 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
4723 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
4726 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4728 PR rtl-optimization/68664
4729 * config/arm/arm.c (arm_sched_can_speculate_insn):
4730 New function. Declare prototype.
4731 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4733 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4735 PR rtl-optimization/68664
4736 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
4738 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4740 2017-02-14 Amit Pawar <amit.pawar@amd.com>
4742 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
4743 max skip bytes for function, loop and jump.
4745 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4747 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
4748 ABS_EXPR for gimple dump.
4750 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4753 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
4755 PR tree-optimization/79408
4756 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
4757 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
4758 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
4759 also if rhs1 is INTEGER_CST.
4761 2017-02-14 Richard Biener <rguenther@suse.de>
4764 * tree-into-ssa.c (insert_phi_nodes): When the function can
4765 have abnormal edges rewrite SSA names with broken use-def
4766 dominance out of SSA and register them for PHI insertion.
4768 2017-02-13 Martin Sebor <msebor@redhat.com>
4771 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
4772 clearing info.nowrite flag when snprintf size argument is a range.
4774 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4776 * cprop.c (cprop_jump): Add missing space in string literal.
4777 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
4778 (get_constraint_for_component_ref): Likewise.
4779 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
4780 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
4781 * lra-constraints.c (process_alt_operands): Likewise.
4782 * ipa-inline.c (inline_small_functions): Likewise.
4783 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
4784 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4785 * trans-mem.c (diagnose_tm_1_op): Likewise.
4786 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
4787 (grid_parallel_clauses_gridifiable): Likewise.
4789 * config/nvptx/mkoffload.c (process): Add space in between
4792 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
4793 "MOD4_SSE_REGS" and "ALL_REGS".
4795 * spellcheck.c (test_data): Add , in between "foo" and "food".
4797 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4800 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
4801 boundary crossing check and subsequent code generation agree.
4803 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4805 * config/aarch64/aarch64.c (has_memory_op): Delete.
4806 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
4809 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4811 PR rtl-optimization/79388
4812 PR rtl-optimization/79450
4813 * combine.c (distribute_notes): When removing TEM_INSN for which
4814 corresponding dest has last value recorded, invalidate that last
4817 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4819 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
4820 of explicit '@'. Add missing assembly comment marker on branch costs
4823 2017-02-13 Nathan Sidwell <nathan@acm.org>
4825 * gengtype-lex.l (<in_struct>): Add '/'.
4827 2017-02-13 Martin Liska <mliska@suse.cz>
4830 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
4832 2017-02-13 Richard Biener <rguenther@suse.de>
4834 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
4836 * configure: Re-generate.
4837 * config.in: Likewise.
4838 * graphite-dependences.c: Simplify as if
4839 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
4840 * graphite-isl-ast-to-gimple.c: Likewise.
4841 * graphite-optimize-isl.c: Likewise.
4842 * graphite-poly.c: Likewise.
4843 * graphite-sese-to-poly.c: Likewise.
4844 * graphite.h: Likewise.
4845 * toplev.c: Include isl/version.h and use isl_version () for
4846 printing the ISL version.
4847 * doc/install.texi: Update ISL requirement.
4849 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4851 * doc/standards.texi (Standards): Update reference to
4854 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4856 * doc/extend.texi (Named Address Spaces): sourceware.org now
4858 * doc/install.texi (Binaries): Ditto.
4861 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4863 * doc/cpp.texi: Replace "stringify"/"stringification" with C
4864 standard terminology "stringize"/"stringizing" throughout.
4865 * doc/cppinternals.texi: Likewise.
4867 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4869 * doc/extend.texi: Fix some spelling mistakes and typos.
4870 * doc/invoke.texi: Likewise.
4872 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4875 * params.def (inline-min-speedup) Change from 10 to 8.
4877 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4879 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
4882 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4885 * ipa-inline-analysis.c (get_minimal_bb): New function.
4886 (record_modified): Use it.
4887 (remap_edge_change_prob): Handle also ancestor functions.
4889 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
4891 * doc/contrib.texi (Contributors): Remove broken link into
4892 the Mauve CVS repository.
4894 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4897 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
4898 result computation whenever lhs doesn't have vector mode, not
4899 just when it has BLKmode.
4901 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4903 * doc/makefile.texi (profiledbootstrap): Refer to the
4904 installation instructions only in textual form.
4906 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4909 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
4911 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4913 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
4914 (Specific): Update mingw-w64 reference.
4916 (Specific): Remove broken link to Renesas RX processor.
4918 2017-02-10 Richard Biener <rguenther@suse.de>
4920 * toplev.c (process_options): Do not mention obsolete graphite
4921 options when printing sorry message about missing graphite support.
4922 Mention -floop-nest-optimize.
4924 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
4926 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
4927 (vtst_p16): Likewise.
4928 (vtstq_p8): Likewise.
4929 (vtstq_p16): Likewise.
4931 (vtstq_p64): Likewise.
4932 * config/arm/arm_neon.h (vgetq_lane_p64): New.
4933 (vset_lane_p64): New.
4934 (vsetq_lane_p64): New.
4936 2017-02-10 Jakub Jelinek <jakub@redhat.com>
4938 PR tree-optimization/79411
4939 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
4940 stmt operands are SSA_NAMEs used in abnormal phis.
4941 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
4944 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4947 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
4950 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4952 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
4954 2017-02-09 Jakub Jelinek <jakub@redhat.com>
4956 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
4960 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
4961 not arbitrary TREE_CONSTANT.
4964 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
4965 "omp declare target link" attribute unless is_global_var.
4966 * omp-offload.c (find_link_var_op): Likewise.
4968 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
4969 Chung-Lin Tang <cltang@codesourcery.com>
4971 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
4973 (gimplify_adjust_omp_clauses): Don't delete TILE.
4974 (gimplify_omp_for): Deal with TILE.
4975 * internal-fn.c (expand_GOACC_TILE): New function.
4976 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
4978 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
4979 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
4981 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
4982 avoid DIV for outermost collapse var.
4983 (expand_oacc_for): Insert tile element loop as needed. Adjust.
4984 Remove out of date comments, fix whitespace.
4985 * omp-general.c (omp_extract_for_data): Deal with tiling.
4986 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
4987 adjust OLF_DIM_BASE value.
4988 (struct omp_for_data): Add tiling field.
4989 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
4990 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
4991 for auto loops. Remove default auto determining, moved to
4992 oacc_loop_fixed_partitions.
4993 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
4994 stmts, add e_mask field.
4995 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
4996 (oacc_thread_numbers): Use oacc_dim_call.
4997 (oacc_xform_tile): New.
4998 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
4999 (finish_oacc_loop): Adjust for ifns vector.
5000 (oacc_loop_discover_walk): Append loop abstraction sites to list,
5001 add case for GOACC_TILE fns.
5002 (oacc_loop_xform_loop): Delete.
5003 (oacc_loop_process): Iterate over call list directly, and add
5004 handling for GOACC_TILE fns.
5005 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
5007 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
5008 vector partitioning to outer loops. Assign 2 partitions to loops
5009 when available. Add TILE handling.
5010 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
5011 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
5012 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
5013 * tree.c (omp_clause_num_ops): Adjust TILE ops.
5014 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
5016 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
5018 * configure.ac (ACX_BUGURL): Update.
5019 * configure: Regenerate.
5021 2017-02-09 Richard Biener <rguenther@suse.de>
5023 PR tree-optimization/69823
5024 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
5025 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
5027 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
5029 * config/arc/arc-c.def: Add __NPS400__ definition.
5030 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
5031 (TARGET_NPS400): Define.
5033 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
5035 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
5037 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
5039 (arc_cpu_types): Fill the arch_info field with a pointer into the
5040 arc_arch_types table.
5041 (arc_selected_cpu): Declare.
5042 * config/arc/arc.c (arc_selected_cpu): Make global.
5043 (arc_selected_arch): Delete.
5044 (arc_base_cpu): Delete.
5045 (arc_override_options): Remove references to deleted variables,
5046 update access to arch information.
5047 (ARC_OPT): Update access to arch information.
5048 (ARC_OPTX): Likewise.
5049 * config/arc/arc.h (arc_base_cpu): Remove declaration.
5050 (TARGET_ARC600): Update access to arch information.
5051 (TARGET_ARC601): Likewise.
5052 (TARGET_ARC700): Likewise.
5053 (TARGET_EM): Likewise.
5054 (TARGET_HS): Likewise.
5055 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
5058 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
5061 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
5062 condition/operands for integer GE/LE/GEU/LEU operations.
5064 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
5066 PR translation/79397
5067 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
5070 2017-02-08 Martin Jambor <mjambor@suse.cz>
5073 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
5074 whether allocation happened.
5075 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
5076 nothing was allocated.
5078 2017-02-08 Jakub Jelinek <jakub@redhat.com>
5080 PR tree-optimization/79408
5081 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
5082 constant, but SSA_NAME with a known integer range, use the minimum
5083 of that range instead of op1 to determine if modulo can be replaced
5084 with its first operand.
5086 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5088 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
5090 2017-02-08 Richard Biener <rguenther@suse.de>
5092 PR tree-optimization/71824
5093 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
5094 Check all loops contained in the merged region.
5096 2017-02-07 Andrew Pinski <apinski@cavium.com>
5098 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
5100 2017-02-07 Andrew Pinski <apinski@cavium.com>
5102 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
5103 (thunderxt88): Likewise.
5104 (thunderxt81): Disable LSE and change v8.1 to v8.
5105 (thunderxt83): Likewise.
5107 2017-02-07 Jakub Jelinek <jakub@redhat.com>
5108 Richard Biener <rguenther@suse.de>
5111 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
5112 type from int to size_t.
5113 * ira-costs.c (struct_costs_size): Change type from int to size_t.
5115 2017-02-07 Jakub Jelinek <jakub@redhat.com>
5117 PR rtl-optimization/79386
5118 * cprop.c (bypass_conditional_jumps): Initialize
5119 bypass_last_basic_block already before splitting bbs after
5120 unconditional traps...
5121 (bypass_conditional_jumps): ... rather than here.
5124 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
5125 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
5126 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
5127 fix -masm=intel patterns.
5129 2017-02-07 Richard Biener <rguenther@suse.de>
5131 PR tree-optimization/79256
5133 * builtins.c (get_object_alignment_2): Use min_align_of_type
5134 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
5135 and ADJUST_FIELD_ALIGN.
5137 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
5139 * doc/tm.texi: Regenerate.
5140 * stor-layout.c (layout_decl): Adjust.
5141 (update_alignment_for_field): Likewise.
5142 (place_field): Likewise.
5143 (min_align_of_type): Likewise.
5144 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
5145 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
5146 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
5147 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
5148 * config/frv/frv.c (frv_adjust_field_align): Likewise.
5149 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
5150 * config/i386/i386.c (x86_field_alignment): Likewise.
5151 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
5152 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
5153 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
5154 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
5155 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
5156 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
5160 2017-01-30 Richard Biener <rguenther@suse.de>
5162 PR tree-optimization/79256
5163 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5164 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5167 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
5169 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
5170 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
5171 builtins to SImode and emit a zero-extend, if necessary.
5173 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5175 * docs/invoke.texi (RISC-V Options): Alphabetize.
5177 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5179 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
5182 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
5184 * config/riscv/riscv.c: New file.
5185 * gcc/common/config/riscv/riscv-common.c: Likewise.
5186 * config.gcc: Likewise.
5187 * config/riscv/constraints.md: Likewise.
5188 * config/riscv/elf.h: Likewise.
5189 * config/riscv/generic.md: Likewise.
5190 * config/riscv/linux.h: Likewise.
5191 * config/riscv/multilib-generator: Likewise.
5192 * config/riscv/peephole.md: Likewise.
5193 * config/riscv/pic.md: Likewise.
5194 * config/riscv/predicates.md: Likewise.
5195 * config/riscv/riscv-builtins.c: Likewise.
5196 * config/riscv/riscv-c.c: Likewise.
5197 * config/riscv/riscv-ftypes.def: Likewise.
5198 * config/riscv/riscv-modes.def: Likewise.
5199 * config/riscv/riscv-opts.h: Likewise.
5200 * config/riscv/riscv-protos.h: Likewise.
5201 * config/riscv/riscv.h: Likewise.
5202 * config/riscv/riscv.md: Likewise.
5203 * config/riscv/riscv.opt: Likewise.
5204 * config/riscv/sync.md: Likewise.
5205 * config/riscv/t-elf-multilib: Likewise.
5206 * config/riscv/t-linux: Likewise.
5207 * config/riscv/t-linux-multilib: Likewise.
5208 * config/riscv/t-riscv: Likewise.
5209 * configure.ac: Likewise.
5210 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
5211 Waterman as RISC-V maintainers.
5212 * doc/install.texi: Add RISC-V entries.
5213 * doc/invoke.texi: Add RISC-V options section.
5214 * doc/md.texi: Add RISC-V constraints section.
5215 * configure: Regenerated.
5217 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
5220 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
5221 false values to be constant vectors with all 0 or all 1 bits set.
5222 (vcondu<mode><mode>): Likewise.
5223 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
5225 (fpmask_comparison_operator): Update comment.
5226 (vecint_comparison_operator): New predicate.
5227 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
5228 vector conditionals when the true and false values are constant
5229 vectors with all 0 bits or all 1 bits set.
5231 2017-02-06 Martin Sebor <msebor@redhat.com>
5233 PR tree-optimization/79376
5234 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
5236 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
5238 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
5239 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
5240 to simplify split condition.
5242 2017-02-06 Jakub Jelinek <jakub@redhat.com>
5244 * omp-expand.c (oxpand_omp_atomic_fetch_op,
5245 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
5248 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
5250 PR rtl-optimization/68664
5251 * target.def (can_speculate_insn): New hook.
5252 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
5253 * doc/tm.texi: Regenerate.
5254 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
5255 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
5256 (rs6000_sched_can_speculate_insn): New function.
5258 2017-02-06 Jakub Jelinek <jakub@redhat.com>
5260 PR tree-optimization/79284
5261 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
5262 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
5263 vectorizable_mask_load_store, vectorizable_operation,
5264 vect_is_simple_cond, get_same_sized_vectype): Use it instead
5265 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
5266 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
5267 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
5268 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
5269 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
5270 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
5271 is_gimple_assign (stmt). Replace another such test with
5272 is_gimple_assign (stmt).
5274 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
5277 * config/avr/avr.c (rtl-iter.h): Include it.
5278 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
5279 (avr_legitimate_combined_insn): ...and implementation.
5281 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
5283 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
5284 * config/s390/s390.c (s390_const_operand_ok)
5285 (s390_canonicalize_comparison, s390_extract_part)
5286 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
5287 (s390_contiguous_bitmask_p, s390_rtx_costs)
5288 (legitimize_pic_address): Likewise.
5289 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
5290 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
5291 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
5292 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
5293 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
5295 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
5297 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
5298 REGNO($0) == REGNO($1).
5300 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5302 * config/s390/linux.h(SIZE_TYPE): Add comment.
5304 2017-02-06 Julian Brown <julian@codesourcery.com>
5305 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5306 Virendra Pathak <virendra.pathak@broadcom.com>
5308 * config/aarch64/aarch64-cores.def: Change the scheduler
5310 * config/aarch64/aarch64.md: Include thunderx2t99.md.
5311 * config/aarch64/thunderx2t99.md: New file.
5313 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5315 * doc/standards.texi (Go Language): Update link to language
5318 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
5320 * tree-eh.c (lower_resx): Sanitize profile.
5321 (cleanup_empty_eh_move_lp): Likewise.
5323 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
5326 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
5328 * cfgloopmanip.h (loop_version): Update prototype.
5329 * modulo-sched.c (sms_schedule): Update call of loop_version.
5330 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
5331 * tree-parloops.c (gen_parallel_loop): Likewise.
5332 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
5333 * tree-ssa-loop-split.c (split_loop): Likewise.
5334 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
5335 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
5337 2017-02-05 Martin Liska <mliska@suse.cz>
5340 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
5342 (print_operand_address): Initialize a struct to zero.
5344 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5346 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
5347 garbage collector only in textual form.
5349 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5351 * doc/extend.texi (x86 specific memory model extensions for
5352 transactional memory): Simplify a phrase.
5354 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
5357 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
5358 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
5359 (atomic_storedi_1): Likewise.
5361 2017-02-04 Jakub Jelinek <jakub@redhat.com>
5363 PR tree-optimization/79338
5364 * tree-parloops.c (gather_scalar_reductions): Don't call
5365 vect_analyze_loop_form for loop->inner before destroying loop's
5368 2017-02-03 Martin Sebor <msebor@redhat.com>
5370 PR tree-optimization/79327
5371 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
5372 when precision has resulted in leading zeros.
5373 (format_integer): Adjust the likely counter to assume an unknown
5374 argument that may be zero is non-zero.
5376 2017-02-03 Jason Merrill <jason@redhat.com>
5379 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
5380 avoid copying non-taken branch.
5382 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5384 PR tree-optimization/79340
5385 * tree-vect-loop.c (vectorizable_reduction): Release
5386 vec_defs elements after safe_splicing them into other vectors.
5389 PR tree-optimization/79327
5390 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
5391 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
5393 (format_integer): Use wide_int_to_tree instead of build_int_cst
5394 + to_?hwi. If argmin is NULL, just set argmin and argmax to
5395 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
5396 of shortest and longest sequence.
5398 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
5400 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
5401 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
5403 2017-02-03 Walter Lee <walt@tilera.com>
5406 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
5407 after initial stackframe link reg save.
5408 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
5410 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5413 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
5414 wu for stxssp alternative.
5416 2017-02-03 Martin Sebor <msebor@redhat.com>
5418 PR tree-optimization/79352
5419 * gimple-fold.c (get_range_strlen): Add argument.
5420 (get_range_strlen): Change return type to bool.
5421 (get_maxval_strlen): Pass in a dummy argument.
5422 * gimple-fold.h (get_range_strlen): Change return type to bool.
5423 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
5424 * tree.h (array_at_struct_end_p): Add argument.
5425 * tree.c (array_at_struct_end_p): Handle it.
5427 2017-02-03 Martin Liska <mliska@suse.cz>
5430 * multiple_target.c (create_dispatcher_calls): Redirect edge
5431 from a caller of a dispatcher.
5432 (expand_target_clones): Make the clones local.
5433 (ipa_target_clone): Do both target clones and resolvers.
5434 (ipa_dispatcher_calls): Remove the pass.
5435 (pass_dispatcher_calls::gate): Likewise.
5436 (make_pass_dispatcher_calls): Likewise.
5437 * passes.def (pass_target_clone): Put as very first IPA early
5440 2017-02-03 Martin Liska <mliska@suse.cz>
5442 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
5443 in case of a function with ifunc attribute.
5445 2017-02-03 Martin Liska <mliska@suse.cz>
5447 * cgraph.c (cgraph_node::dump): Dump function version info.
5448 * symtab.c (symtab_node::dump_base): Add missing new line.
5450 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5452 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
5453 (ifcombine_ifandif): Use it.
5455 2017-02-03 Martin Liska <mliska@suse.cz>
5457 * doc/invoke.texi: Document default value for
5458 use-after-scope-direct-emission-threshold.
5460 2017-02-03 Martin Liska <mliska@suse.cz>
5462 PR tree-optimization/79339
5463 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
5464 (format_floating): Likewise.
5466 2017-02-03 Martin Liska <mliska@suse.cz>
5469 * ipa-prop.c (ipa_node_params_t::insert): Remove current
5471 (ipa_node_params_t::remove): Likewise.
5472 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
5473 initialization from removed ipa_node_params_t::insert.
5474 (ipa_node_params::~ipa_node_params): Move from removed
5475 ipa_node_params_t::release.
5476 * symbol-summary.h (symbol_summary::m_released): New member.
5477 Do not release a summary twice. Do not allow to call finalizer
5478 for types of a summary that live in GGC memory.
5480 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5482 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
5485 2017-02-02 Martin Sebor <msebor@redhat.com>
5488 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
5489 (format_string): Tighten up the range of output for non-constant
5490 strings and correct the expected range for wide non-constant strings.
5492 2017-02-02 Martin Sebor <msebor@redhat.com>
5494 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
5497 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
5499 (-fdump-tree-@var): Add to index and document how to come up
5500 with pass-specific option and dump file names.
5501 (-fdump-passes): Clarify where to look for output.
5503 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5506 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
5507 statistics of the analyzed path; allow threading for speed when
5508 any of BBs along the path are optimized for speed.
5510 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
5513 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
5514 settings of the virtual registers.
5517 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
5519 * explow.c (get_dynamic_stack_size): Take known alignment of stack
5520 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
5523 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5525 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
5526 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
5528 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5530 * config/s390/s390.md: Add missing comments with the expanded
5532 * config/s390/vector.md: Likewise.
5533 * config/s390/vx-builtins.md: Likewise.
5535 2017-02-02 Jakub Jelinek <jakub@redhat.com>
5538 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
5539 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
5540 conditions on a single line.
5542 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5544 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5545 __S390_VX__ to __VX__.
5547 2017-02-01 Andrew Pinski <apinski@cavium.com>
5549 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
5550 stmt_info to record_stmt_cost.
5551 (vect_get_known_peeling_cost): Pass stmt_info if known to
5553 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
5554 cpu_vector_cost field into
5555 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
5556 field into vec_int_stmt_cost and vec_fp_stmt_cost.
5557 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
5558 splitting of scalar_stmt_cost and vec_stmt_cost.
5559 (thunderx_vector_cost): Likewise.
5560 (cortexa57_vector_cost): LIkewise.
5561 (exynosm1_vector_cost): Likewise.
5562 (xgene1_vector_cost): Likewise.
5563 (thunderx2t99_vector_cost): Improve after the splitting of the two
5565 (aarch64_builtin_vectorization_cost): Update for the splitting of
5566 scalar_stmt_cost and vec_stmt_cost.
5568 2017-02-01 Torvald Riegel <triegel@redhat.com>
5569 Richard Henderson <rth@redhat.com>
5571 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
5572 conditional on existance of a fast atomic load.
5573 * optabs-query.c (can_atomic_load_p): New function.
5574 * optabs-query.h (can_atomic_load_p): Declare it.
5575 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
5576 no fast atomic load is available for the particular size of access.
5577 (expand_atomic_compare_and_swap): Likewise.
5578 (expand_atomic_load): Likewise.
5579 (expand_atomic_store): Likewise.
5580 (expand_atomic_fetch_op): Likewise.
5581 * testsuite/lib/target-supports.exp
5582 (check_effective_target_sync_int_128): Remove x86 because it provides
5583 no fast atomic load.
5584 (check_effective_target_sync_int_128_runtime): Likewise.
5586 2017-02-01 Richard Biener <rguenther@suse.de>
5588 * graphite.c: Include tree-vectorizer.h for find_loop_location.
5589 (graphite_transform_loops): Provide opt-info for optimized nests.
5590 * tree-parloop.c (parallelize_loops): Provide opt-info for
5593 2017-02-01 Richard Biener <rguenther@suse.de>
5596 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
5599 2017-02-01 Richard Biener <rguenther@suse.de>
5601 PR tree-optimization/71824
5602 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
5603 Verify the loops are valid in the merged SESE region.
5604 (scop_detection::can_represent_loop_1): Check analyzing the
5605 evolution of the number of iterations in the region succeeds.
5607 2017-01-31 Ian Lance Taylor <iant@golang.org>
5609 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
5610 REG_ARGS_SIZE note to 32-bit push insns and call insn.
5612 2017-01-31 David Malcolm <dmalcolm@redhat.com>
5614 PR preprocessor/79210
5615 * input.c (get_substring_ranges_for_loc): Replace line_width
5616 assertion with error-handling.
5618 2017-01-31 Richard Biener <rguenther@suse.de>
5620 PR tree-optimization/77318
5621 * graphite-sese-to-poly.c (extract_affine): Fix assert.
5622 (create_pw_aff_from_tree): Take loop parameter.
5623 (add_condition_to_pbb): Pass loop of the condition to
5624 create_pw_aff_from_tree.
5626 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5628 * config/s390/s390.c (s390_asan_shadow_offset): New function.
5629 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
5631 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5635 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
5637 (convert_int_to_float128): Likewise.
5638 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
5639 (convert_int_to_float128): Likewise.
5640 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
5641 (UNSPEC_IEEE128_CONVERT): Likewise.
5642 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
5643 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
5644 Use local variables for IBM extended format.
5645 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
5646 (fix_trunc<mode>si2_fprs): Likewise.
5647 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
5648 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
5649 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
5650 to know that we can now have integers of all sizes in vector
5652 (fix<uns>_<mode>di2_hw): Likewise.
5653 (float<uns>_<mode>si2_hw): Likewise.
5654 (fix_<mode>si2_hw): Likewise.
5655 (fixuns_<mode>si2_hw): Likewise.
5656 (float<uns>_<mode>di2_hw): Likewise.
5657 (float_<mode>di2_hw): Likewise.
5658 (float_<mode>si2_hw): Likewise.
5659 (floatuns_<mode>di2_hw): Likewise.
5660 (floatuns_<mode>si2_hw): Likewise.
5661 (xscvqp<su>wz_<mode>): Delete, no longer used.
5662 (xscvqp<su>dz_<mode>): Likewise.
5663 (xscv<su>dqp_<mode>): Likewise.
5664 (ieee128_mfvsrd_64bit): Likewise.
5665 (ieee128_mfvsrd_32bit): Likewise.
5666 (ieee128_mfvsrwz): Likewise.
5667 (ieee128_mtvsrw): Likewise.
5668 (ieee128_mtvsrd_64bit): Likewise.
5669 (ieee128_mtvsrd_32bit): Likewise.
5671 2017-01-31 Martin Liska <mliska@suse.cz>
5674 * ipa-prop.c (ipa_free_all_node_params): Call release method
5675 instead of ~sumbol_summary to not to trigger double times
5678 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
5680 PR tree-optimization/71691
5681 * bitmap.h (class auto_bitmap): New.
5682 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
5683 is_maybe_undefined instead of ssa_undefined_value_p.
5685 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5687 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5688 __S390_ARCH_LEVEL__ to __ARCH__.
5690 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5692 PR tree-optimization/79267
5693 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
5694 if should_remove_lhs_p is true.
5696 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
5699 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
5700 (add_alignment_attribute): New.
5701 (base_type_die): Add alignment attribute.
5702 (subrange_type_die): Likewise.
5703 (modified_type_die): Likewise.
5704 (gen_array_type_die): Likewise.
5705 (gen_descr_array_type_die: Likewise.
5706 (gen_enumeration_type_die): Likewise.
5707 (gen_subprogram_die): Likewise.
5708 (gen_variable_die): Likewise.
5709 (gen_field_die): Likewise.
5710 (gen_ptr_to_mbr_type_die): Likewise.
5711 (gen_struct_or_union_type_die): Likewise.
5712 (gen_subroutine_type_die): Likewise.
5713 (gen_typedef_die): Likewise.
5714 (base_type_cmp): Compare alignment attribute.
5716 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5719 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
5720 (setb_unsigned) New pattern for setb with CCUNS.
5721 * config/rs6000/rs6000.c (expand_block_compare): Use a different
5722 subfc./subfe sequence to avoid overflow problems. Generate a
5723 shorter sequence with cmpld/setb for power9.
5724 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
5725 for generating subfc. instruction.
5726 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
5727 now uses this instruction.
5729 2017-01-30 Ian Lance Taylor <iant@google.com>
5732 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
5733 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
5735 2017-01-30 Martin Sebor <msebor@redhat.com>
5737 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
5738 Move constant to the right of a relational operator.
5739 (get_mpfr_format_length, format_character, format_string): Ditto.
5740 (should_warn_p, maybe_warn): Same.
5742 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
5744 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5747 * asan.c (get_translation_unit_decl): Remove function.
5748 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
5750 2017-01-30 Martin Liska <mliska@suse.cz>
5752 PR gcov-profile/79259
5753 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
5756 2017-01-30 Martin Liska <mliska@suse.cz>
5759 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
5760 Initialize variables with NULL value.
5762 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
5765 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
5767 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
5769 2017-01-30 Richard Biener <rguenther@suse.de>
5771 PR tree-optimization/79276
5772 * tree-vrp.c (process_assert_insertions): Properly adjust common
5773 when removing a duplicate.
5775 * gcc.dg/torture/pr79276.c: New testcase.
5777 2017-01-30 Richard Biener <rguenther@suse.de>
5779 PR tree-optimization/79256
5780 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5781 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5783 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
5785 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
5788 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
5789 ("*r<noxa>sbg_<mode>_sll_bitmask")
5790 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
5791 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
5792 Use contiguous_bitmask_nowrap_operand.
5794 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5797 * config/rs6000/altivec.h (vec_xl): Revise #define.
5798 (vec_xst): Likewise.
5800 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
5802 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
5804 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
5806 PR rtl-optimization/79194
5807 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
5808 traps before call to bypass_conditional_jumps.
5810 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5812 PR tree-optimization/71374
5813 * lra-constraints.c (check_conflict_input_operands): New.
5814 (match_reload): Use it.
5816 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5819 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
5820 account to calculate conflict_set.
5822 2017-01-27 Bin Cheng <bin.cheng@arm.com>
5824 PR rtl-optimization/78559
5825 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
5826 other_insn in combine.
5828 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
5830 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
5831 uint16_type_node for BT_UINT16.
5833 2017-01-27 David Malcolm <dmalcolm@redhat.com>
5835 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
5836 "RTL Tests" to menu.
5837 (GIMPLE Tests): New node.
5838 (RTL Tests): New node.
5840 2017-01-27 Richard Biener <rguenther@suse.de>
5842 PR tree-optimization/79245
5843 * tree-loop-distribution.c (distribute_loop): Apply cost
5844 modeling also to detected patterns.
5846 2017-01-27 Richard Biener <rguenther@suse.de>
5848 PR tree-optimization/71433
5849 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
5850 (compare_assert_loc): New function.
5851 (process_assert_insertions): Sort and optimize assert locations
5852 to remove duplicates and push down identical assertions on
5853 edges to their destination block.
5855 2017-01-27 Richard Biener <rguenther@suse.de>
5857 PR tree-optimization/79244
5858 * tree-vrp.c (remove_range_assertions): Forcefully propagate
5859 out SSA names even if abnormal.
5861 2017-01-27 Jakub Jelinek <jakub@redhat.com>
5863 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
5864 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
5865 instead of MPFR_RNDN.
5867 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
5870 * arm.c (arm_option_override): Don't call build_target_option_node
5871 until after doing all option overrides.
5872 (arm_valid_target_attribute_tree): Likewise.
5874 2017-01-27 Martin Liska <mliska@suse.cz>
5876 * doc/invoke.texi (-fprofile-arcs): Document profiling support
5877 for {cd}tors and C++ {cd}tors.
5879 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
5881 * config/s390/s390.md ("*setmem_long_and")
5882 ("*setmem_long_and_31z"): Use zero_extend instead of and.
5884 2017-01-26 Martin Sebor <msebor@redhat.com>
5886 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
5889 2017-01-26 Martin Sebor <msebor@redhat.com>
5891 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
5892 HAVE_DFmode before using XFmode or DFmode.
5893 (parse_directive): Avoid using the z length modifier to avoid
5894 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
5897 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
5898 to accept adjustment as an array.
5899 (get_int_range): New function.
5900 (struct directive): Make width and prec arrays.
5901 (directive::set_width, directive::set_precision): Call get_int_range.
5902 (format_integer, format_floating): Handle width and precision ranges.
5903 (format_string, parse_directive): Same.
5905 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5908 * dwarf2out.c (generate_skeleton_bottom_up): For children with
5909 comdat_type_p set, just clone them, but keep the children in the
5913 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
5914 which have direct callers with -fvar-tracking-assignments enabled
5916 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
5917 inside of type units.
5919 2017-01-26 Martin Sebor <msebor@redhat.com>
5922 * gimple-ssa-sprintf.c (struct result_range): Add likely and
5924 (struct format_result): Replace number_chars, number_chars_min,
5925 and number_chars_max with a single member of struct result_range.
5927 (format_result::operator+=): Adjust.
5928 (struct fmtresult): Remove bounded. Handle likely and unlikely
5930 (fmtresult::adjust_for_width_or_precision): New function.
5931 (fmtresult:type_max_digits): New function.
5932 (bytes_remaining): Handle likely and unlikely counters.
5933 (min_bytes_remaining): Remove.
5934 (format_percent): Simplify.
5935 (format_integer, format_floating): Set likely and unlikely counters.
5936 (get_string_length, format_character, format_string): Same.
5937 (format_plain, should_warn_p): New function.
5938 (maybe_warn): Call should_warn_p. Update diagnostic messages
5939 and handle those for all directives, including plain strings.
5940 (format_directive): Handle likely and unlikely counters.
5941 Remove unnecessary quoting from diagnostics. Add an informational
5943 (add_bytes): Remove.
5944 (pass_sprintf_length::compute_format_length): Simplify.
5945 (try_substitute_return_value): Handle likely and unlikely counters.
5947 2017-01-26 Carl Love <cel@us.ibm.com>
5949 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
5950 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
5952 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
5955 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
5956 endianess for subregs into account.
5957 * lra-constraints.c (lra_constraints): Do risky transformations
5958 always on the first iteration.
5959 * lra-lives.c (check_pseudos_live_through_calls): Add arg
5960 last_call_used_reg_set.
5961 (process_bb_lives): Define and use last_call_used_reg_set.
5962 * lra.c (lra): Always continue after lra_constraints on the first
5965 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
5967 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
5969 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
5971 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5973 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
5974 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
5975 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
5976 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
5977 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
5978 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
5979 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
5980 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
5981 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
5983 2017-01-26 Marek Polacek <polacek@redhat.com>
5986 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
5987 for the third operand.
5989 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5992 * omp-low.c (struct omp_context): Add simt_stmt field.
5993 (scan_omp_for): Return omp_context *.
5994 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
5995 context to the _simt_ SIMD stmt.
5996 (lower_omp_for): For combined SIMD with sibling _simt_
5997 SIMD, make sure to use the same decls in _looptemp_
5998 clauses as in the sibling.
6000 2017-01-26 David Sherwood <david.sherwood@arm.com>
6003 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
6006 2017-01-26 Jakub Jelinek <jakub@redhat.com>
6009 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
6010 emit fld b; fld a; if possible.
6012 * brig-builtins.def: Update copyright years.
6013 * config/arm/arm_acle_builtins.def: Update copyright years.
6015 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
6018 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
6019 constraint instead of o for the stxsd instruction.
6021 2017-01-25 Carl Love <cel@us.ibm.com>
6023 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
6024 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
6026 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
6028 * doc/invoke.texi (C++ Dialect Options): Fix typo.
6030 2017-01-25 Richard Biener <rguenther@suse.de>
6032 PR tree-optimization/69264
6033 * target.def (vector_alignment_reachable): Improve documentation.
6034 * doc/tm.texi: Regenerate.
6035 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
6037 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
6038 earlier changes with respect to TYPE_USER_ALIGN.
6039 (vector_alignment_reachable_p): Likewise. Improve dumping.
6041 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6044 * config/arm/arm.md (xordi3): Force constant operand into a register
6047 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6049 * doc/invoke.texi (-fstore-merging): Correct default optimization
6050 levels at which it is enabled.
6051 (-O): Move -fstore-merging from list to...
6054 2017-01-25 Richard Biener <rguenther@suse.de>
6057 * omp-expand.c: Include debug.h.
6058 (expand_omp_taskreg): Make sure to generate early debug before
6059 outlining anything from a function.
6060 (expand_omp_target): Likewise.
6061 (grid_expand_target_grid_body): Likewise.
6063 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
6066 * asan.c (get_translation_unit_decl): New function.
6067 (asan_add_global): Extract modules file name from globals
6068 TRANSLATION_UNIT_DECL name.
6070 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
6073 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
6074 for long calls with APCS frame and VFP.
6076 2017-01-24 David Malcolm <dmalcolm@redhat.com>
6078 * cfg.c (original_copy_tables_initialized_p): New function.
6079 * cfg.h (original_copy_tables_initialized_p): New decl.
6080 * cfgrtl.c (relink_block_chain): Guard the call to
6081 free_original_copy_tables with a call to
6082 original_copy_tables_initialized_p.
6083 * cgraph.h (symtab_node::native_rtl_p): New decl.
6084 * cgraphunit.c (symtab_node::native_rtl_p): New function.
6085 (symtab_node::needed_p): Don't assert for early assembly output
6086 for __RTL functions.
6087 (cgraph_node::finalize_function): Set "force_output" for __RTL
6089 (cgraph_node::analyze): Bail out early for __RTL functions.
6090 (analyze_functions): Update assertion to support __RTL functions.
6091 (cgraph_node::expand): Bail out early for __RTL functions.
6092 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
6094 * function.h (struct function): Update comment for field
6096 * gimple-expr.c: Include "tree-pass.h".
6097 (gimple_has_body_p): Return false for __RTL functions.
6098 * Makefile.in (OBJS): Add run-rtl-passes.o.
6099 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
6101 (gcc::pass_manager::get_clean_slate): New accessor.
6102 * passes.c: Include "insn-addr.h".
6103 (should_skip_pass_p): Add logging. Update logic for running
6104 "expand" to be compatible with both __GIMPLE and __RTL. Guard
6105 property-provider override so it is only done for gimple passes.
6107 (skip_pass): New function.
6108 (execute_one_pass): Call skip_pass when skipping passes.
6109 * read-md.c (md_reader::read_char): Support filtering
6110 the input to a subset of line numbers.
6111 (md_reader::md_reader): Initialize fields
6112 m_first_line and m_last_line.
6113 (md_reader::read_file_fragment): New function.
6114 * read-md.h (md_reader::read_file_fragment): New decl.
6115 (md_reader::m_first_line): New field.
6116 (md_reader::m_last_line): New field.
6117 * read-rtl-function.c (function_reader::create_function): Only
6118 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
6119 curr_properties. Set DECL_INITIAL to a dummy block.
6120 (read_rtl_function_body_from_file_range): New function.
6121 * read-rtl-function.h (read_rtl_function_body_from_file_range):
6123 * run-rtl-passes.c: New file.
6124 * run-rtl-passes.h: New file.
6126 2017-01-24 Jeff Law <law@redhat.com>
6128 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
6131 2017-01-24 Bin Cheng <bin.cheng@arm.com>
6133 PR tree-optimization/79159
6134 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
6135 (record_nonwrapping_iv): Improve boundary using above function if no
6136 value range information.
6138 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
6139 Martin Jambor <mjambor@suse.cz>
6141 * brig-builtins.def: New file.
6142 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
6143 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
6144 (DEF_HSAIL_SAT_BUILTIN): Likewise.
6145 (DEF_HSAIL_INTR_BUILTIN): Likewise.
6146 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
6147 * builtin-types.def (BT_INT8): New.
6148 (BT_INT16): Likewise.
6149 (BT_UINT8): Likewise.
6150 (BT_UINT16): Likewise.
6151 (BT_FN_ULONG): Likewise.
6152 (BT_FN_UINT_INT): Likewise.
6153 (BT_FN_UINT_ULONG): Likewise.
6154 (BT_FN_UINT_LONG): Likewise.
6155 (BT_FN_UINT_PTR): Likewise.
6156 (BT_FN_ULONG_PTR): Likewise.
6157 (BT_FN_INT8_FLOAT): Likewise.
6158 (BT_FN_INT16_FLOAT): Likewise.
6159 (BT_FN_UINT32_FLOAT): Likewise.
6160 (BT_FN_UINT16_FLOAT): Likewise.
6161 (BT_FN_UINT8_FLOAT): Likewise.
6162 (BT_FN_UINT64_FLOAT): Likewise.
6163 (BT_FN_UINT16_UINT32): Likewise.
6164 (BT_FN_UINT32_UINT16): Likewise.
6165 (BT_FN_UINT16_UINT16_UINT16): Likewise.
6166 (BT_FN_INT_PTR_INT): Likewise.
6167 (BT_FN_UINT_PTR_UINT): Likewise.
6168 (BT_FN_LONG_PTR_LONG): Likewise.
6169 (BT_FN_ULONG_PTR_ULONG): Likewise.
6170 (BT_FN_VOID_UINT64_UINT64): Likewise.
6171 (BT_FN_UINT8_UINT8_UINT8): Likewise.
6172 (BT_FN_INT8_INT8_INT8): Likewise.
6173 (BT_FN_INT16_INT16_INT16): Likewise.
6174 (BT_FN_INT_INT_INT): Likewise.
6175 (BT_FN_UINT_FLOAT_UINT): Likewise.
6176 (BT_FN_FLOAT_UINT_UINT): Likewise.
6177 (BT_FN_ULONG_UINT_UINT): Likewise.
6178 (BT_FN_ULONG_UINT_PTR): Likewise.
6179 (BT_FN_ULONG_ULONG_ULONG): Likewise.
6180 (BT_FN_UINT_UINT_UINT): Likewise.
6181 (BT_FN_VOID_UINT_PTR): Likewise.
6182 (BT_FN_UINT_UINT_PTR: Likewise.
6183 (BT_FN_UINT32_UINT64_PTR): Likewise.
6184 (BT_FN_INT_INT_UINT_UINT): Likewise.
6185 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
6186 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
6187 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
6188 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
6189 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
6190 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
6191 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
6192 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
6193 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
6194 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
6195 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
6196 * doc/frontends.texi: List BRIG FE.
6197 * doc/install.texi (Testing): Add BRIG tesring requirements.
6198 * doc/invoke.texi (Overall Options): Mention BRIG.
6199 * doc/standards.texi (Standards): Doucment BRIG HSA version.
6201 2017-01-24 Richard Biener <rguenther@suse.de>
6203 PR translation/79208
6204 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
6206 2017-01-24 Martin Jambor <mjambor@suse.cz>
6209 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
6210 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
6213 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
6216 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
6217 casts from signed to unsigned really don't have a range.
6219 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
6221 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
6222 GMP_RNDx for compatiblity.
6224 2017-01-24 Martin Liska <mliska@suse.cz>
6227 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
6228 that would prevent us to call alloca with -1 as argument.
6230 2017-01-24 Jakub Jelinek <jakub@redhat.com>
6232 * dwarf2out.c (output_compilation_unit_header, output_file_names):
6233 Avoid -Wformat-security warning.
6235 2017-01-23 Andrew Pinski <apinski@cavium.com>
6237 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
6240 2017-01-23 Martin Sebor <msebor@redhat.com>
6243 * gimple-ssa-sprintf.c (warn_level): New global.
6244 (format_integer): Use it here and throughout the rest of the file.
6245 Use the same switch to compute sign as base.
6246 (maybe_warn): New function.
6247 (format_directive): Factor out warnings into maybe_warn.
6248 Add debugging output. Use warn_level.
6249 (add_bytes): Use warn_level.
6250 (pass_sprintf_length::compute_format_length): Add debugging output.
6251 (try_substitute_return_value): Same.
6252 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
6255 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
6256 (struct fmtresult, format_integer, format_floating): Adjust.
6257 (fmtresult::fmtresult): Set max correctly in two argument ctor.
6258 (get_string_length, format_string,format_directive): Same.
6259 (pass_sprintf_length::compute_format_length): Same.
6260 (try_substitute_return_value): Simplify slightly.
6263 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
6264 (fmtresult::operator+=): Outlined.
6265 (struct fmtresult): Add ctors.
6266 (struct conversion_spec): Rename...
6267 (struct directive): ...to this. Add and remove data members.
6268 (directive::set_width, directive::set_precision): New functions.
6269 (format_percent): Use fmtresult ctor.
6270 (get_width_and_precision): Remove.
6271 (format_integer): Make naming changes. Avoid computing width and
6273 (format_floating): Same. Adjust indentation.
6274 (format_character, format_none): New functions.
6275 (format_string): Moved character handling to format_character.
6276 (format_directive): Remove arguments, change return type.
6277 (parse_directive): New function.
6278 (pass_sprintf_length::compute_format_length): Move directive
6279 parsing to parse_directive.
6281 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6283 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
6284 (assign_assembler_name_if_needed): ... this.
6285 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
6286 (assign_assembler_name_if_needed): ... this.
6287 (free_lang_data_in_cgraph): Adjust callers.
6288 * cgraphunit.c (cgraph_node::analyze): Likewise.
6289 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
6292 2017-01-23 Richard Biener <rguenther@suse.de>
6294 PR tree-optimization/79088
6295 PR tree-optimization/79188
6296 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
6297 resetting loop bounds after last path deletion. Reset loop
6298 bounds of the target loop, make code match the comments.
6299 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
6300 Make sure loops need no fixups.
6302 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
6304 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
6305 exponent support with double type for first argument.
6306 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
6307 type returned by __builtin_vec_extract_sig,
6308 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
6309 functions from "vector int" to "vector unsigned int" or from
6310 "vector long long int" to "vector unsigned long long int".
6311 Changed type returned by __builtin_vec_extract_exp,
6312 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
6313 functions from "vector int" to "vector unsigned int" or from
6314 "vector long long int" to "vector unsigned long long int".
6315 Changed return type of __builtin_vec_test_data_class,
6316 __builtin_vec_test_data_class_sp, and
6317 __builtin_vec_test_data_class_dp from "vector int" to
6318 "vector bool int" or from "vector long long int" to "vector bool
6319 long long int" and changed second argument type from "unsigned
6320 int" to "int". Added new overloaded function forms "vector float
6321 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
6322 "vector float __builtin_vec_insert_exp_sp (vector float, vector
6323 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
6324 double, vector unsigned long long int)" and "vector double
6325 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
6326 long int)". Changed return type of
6327 __builtin_scalar_test_data_class and
6328 __builtin_scalar_test_data_class_sp and
6329 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
6330 int" and changed second argument from "unsigned int" to "int".
6331 Changed type returned by __builtin_scalar_test_neg,
6332 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
6333 from "int" to "bool int". Added new overloaded function form
6334 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
6335 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
6336 exponent double-precision with floating point first argument.
6337 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
6338 documentation of scalar_test_data_class, scalar_test_neg,
6339 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
6340 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
6341 vec_test_data_class built-in functions to reflect refinements in
6342 their type signatures.
6344 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
6346 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
6348 (aarch64_elf_asm_destructor): Likewise.
6350 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
6352 PR rtl-optimization/78634
6353 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
6354 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
6355 * ifcvt.c (noce_try_cmove): Add missing cost check.
6357 PR rtl-optimization/71724
6358 * combine.c (if_then_else_cond): Look for situations where it is
6359 beneficial to undo the work of one of the recursive calls.
6361 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6363 PR tree-optimization/70754
6364 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6365 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6366 combined stmt before it if not NULL.
6367 (combine_chains): Process refs reversely and compute dominance point
6370 2017-01-23 Martin Liska <mliska@suse.cz>
6372 PR tree-optimization/79196
6373 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
6374 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
6376 (strlen_optimize_stmt): Call the renamed function.
6378 2017-01-23 Michael Matz <matz@suse.de>
6380 PR tree-optimization/78384
6381 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
6383 2017-01-23 Richard Biener <rguenther@suse.de>
6385 PR tree-optimization/79186
6386 * tree-vrp.c (register_new_assert_for): Make sure we've seen
6387 both incoming edges before moving an assert.
6389 2017-01-23 Martin Jambor <mjambor@suse.cz>
6391 * ipa-prop.c (load_from_param_1): Removed.
6392 (load_from_unmodified_param): Bits from load_from_param_1 put back
6394 (load_from_param): Removed.
6395 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
6396 with stmt. Reverted back to use of load_from_unmodified_param.
6398 2017-01-23 Martin Jambor <mjambor@suse.cz>
6401 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
6402 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
6403 field a pointer to garbage collected vector, mark lattices and
6404 ipcp_orig_node with GTY((skip)).
6405 (ipa_get_param_count): Adjust to descriptors being a pointer.
6406 (ipa_get_param): Likewise.
6407 (ipa_get_type): Likewise.
6408 (ipa_get_param_move_cost): Likewise.
6409 (ipa_set_param_used): Likewise.
6410 (ipa_get_controlled_uses): Likewise.
6411 (ipa_set_controlled_uses): Likewise.
6412 (ipa_is_param_used): Likewise.
6413 (ipa_node_params_t): Move into garbage collector. New methods insert
6415 (ipa_node_params_sum): Annotate wth GTY(()).
6416 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
6418 (ipa_load_from_parm_agg): Adjust declaration.
6419 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
6420 * ipa-profile.c (ipa_profile): Likewise.
6421 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
6422 (ipa_populate_param_decls): Make descriptors parameter garbage
6424 (ipa_dump_param): Adjust to descriptors being a pointer.
6425 (ipa_alloc_node_params): Likewise.
6426 (ipa_initialize_node_params): Likewise.
6427 (load_from_param_1): Make descriptors parameter garbage collected.
6428 (load_from_unmodified_param): Likewise.
6429 (load_from_param): Likewise.
6430 (ipa_load_from_parm_agg): Likewise.
6431 (ipa_node_params::~ipa_node_params): Removed.
6432 (ipa_free_all_node_params): Remove call to delete operator.
6433 (ipa_node_params_t::insert): New.
6434 (ipa_node_params_t::remove): Likewise.
6435 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
6436 copy known_csts and known_contexts vectors.
6437 (ipa_read_node_info): Adjust to descriptors being a pointer.
6438 (ipcp_modif_dom_walker): Make m_descriptors field garbage
6440 (ipcp_transform_function): Make descriptors variable garbage
6443 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
6445 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
6446 * config/i386/avx512dqintrin.h: Ditto.
6447 * config/i386/avx512fintrin.h: Ditto.
6448 * gcc/config/i386/i386.c: Handle new builtins.
6449 * config/i386/i386-builtin.def: Add new builtins.
6450 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
6451 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
6453 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6454 Martin Liska <mliska@suse.cz>
6456 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
6457 * asan.c (asan_expand_poison_ifn): Support stores and use
6458 appropriate ASAN report function.
6459 * internal-fn.c (expand_ASAN_POISON_USE): New function.
6460 * internal-fn.def (ASAN_POISON_USE): Declare.
6461 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
6462 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
6463 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
6464 ASAN_POISON calls w/o LHS.
6465 * tree-ssa.c (execute_update_addresses_taken): Create clobber
6466 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
6467 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
6468 * gimplify.c (asan_poison_variables): Add attribute
6469 use_after_scope_memory to variables that really needs to live
6471 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
6472 having the attribute.
6474 2017-01-23 Martin Liska <mliska@suse.cz>
6476 * asan.c (create_asan_shadow_var): New function.
6477 (asan_expand_poison_ifn): Likewise.
6478 * asan.h (asan_expand_poison_ifn): New declaration.
6479 * internal-fn.c (expand_ASAN_POISON): Likewise.
6480 * internal-fn.def (ASAN_POISON): New builtin.
6481 * sanopt.c (pass_sanopt::execute): Expand
6482 asan_expand_poison_ifn.
6483 * tree-inline.c (copy_decl_for_dup_finish): Make function
6485 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
6486 * tree-ssa.c (is_asan_mark_p): New function.
6487 (execute_update_addresses_taken): Rewrite local variables
6488 (identified just by use-after-scope as addressable) into SSA.
6490 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6492 * doc/install.texi (Specific): opensource.apple.com uses https
6493 now. Remove trailing slash.
6495 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6497 * README.Portability: Remove note on an Irix compatibility issue.
6499 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
6501 * gcov.c (INCLUDE_ALGORITHM): Define.
6502 (INCLUDE_VECTOR): Define.
6503 No longer include <vector> and <algorithm> directly.
6505 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
6507 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
6509 * doc/invoke.texi (Code Gen Options): Ditto.
6511 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
6514 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
6516 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
6518 rtl-optimization/79125
6519 * cprop.c (local_cprop_pass): Handle cases where we make an
6522 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
6526 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
6527 read from, for big endian.
6529 2017-01-20 Jiong Wang <jiong.wang@arm.com>
6531 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
6532 register pauth builtins for LP64 only.
6534 2017-01-20 Marek Polacek <polacek@redhat.com>
6537 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
6540 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6542 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
6544 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
6545 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
6546 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
6548 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6551 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
6552 in big-endian mode when they are not a single duplicated value.
6554 2017-01-20 Richard Biener <rguenther@suse.de>
6556 * BASE-VER: Bump to 7.0.1.
6558 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6560 * omp-low.c (omplow_simd_context): New struct. Use it...
6561 (lower_rec_simd_input_clauses): ...here and...
6562 (lower_rec_input_clauses): ...here to hold common data. Adjust all
6563 references to idx, lane, max_vf, is_simt.
6565 2017-01-20 Graham Markall <graham.markall@embecosm.com>
6567 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
6570 2017-01-20 Martin Jambor <mjambor@suse.cz>
6572 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
6573 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
6575 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
6576 (GTFILES): Rename hsa.c to hsa-common.c.
6577 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
6578 * hsa-dump.c: Likewise.
6579 * hsa-gen.c: Likewise.
6580 * hsa-regalloc.c: Likewise.
6581 * ipa-hsa.c: Likewise.
6582 * omp-expand.c: Likewise.
6583 * omp-low.c: Likewise.
6584 * toplev.c: Likewise.
6586 2017-01-20 Marek Polacek <polacek@redhat.com>
6589 * doc/invoke.texi: Document -Wduplicated-branches.
6590 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
6591 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
6592 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
6593 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
6594 return 0 only when not OEP_LEXICOGRAPHIC.
6595 (fold_build_cleanup_point_expr): Use the expression
6596 location when building CLEANUP_POINT_EXPR.
6597 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
6598 * tree.c (add_expr): Handle error_mark_node.
6600 2017-01-20 Martin Liska <mliska@suse.cz>
6603 * tree-profile.c (init_ic_make_global_vars): Do not call
6605 (gimple_init_gcov_profiler): Likewise.
6607 2017-01-20 Martin Liska <mliska@suse.cz>
6610 * cgraph.h (maybe_create_reference): Remove argument and
6612 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
6614 * ipa-cp.c (create_specialized_node): Likewise.
6615 * symtab.c (symtab_node::maybe_create_reference): Handle
6616 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
6618 2017-01-20 Martin Liska <mliska@suse.cz>
6620 * read-rtl-function.c (function_reader::create_function): Use
6621 build_decl instread of build_decl_stat.
6623 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
6625 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
6626 * config/i386/avx512dqintrin.h: Ditto.
6627 * config/i386/avx512fintrin.h: Ditto.
6628 * config/i386/i386-builtin-types.def: Add new types.
6629 * gcc/config/i386/i386.c: Handle new types.
6630 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
6631 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
6632 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
6633 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
6634 (__builtin_ia32_kshiftridi): New.
6635 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
6637 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
6641 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
6642 define to rs6000_init_stack_protect_guard.
6643 (rs6000_init_stack_protect_guard): New function.
6645 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6646 Yunqiang Su <yunqiang.su@imgtec.com>
6648 * config.gcc (supported_defaults): Add madd4.
6649 (with_madd4): Add validation.
6650 (all_defaults): Add madd4.
6651 * config/mips/mips.opt (mmadd4): New option.
6652 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6654 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6656 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
6657 (ISA_HAS_FUSED_MADD4): Likewise.
6658 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
6659 * gcc/doc/install.texi (--with-madd4): Document the new option.
6661 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6663 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
6664 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
6665 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
6666 (aarch64_init_pauth_hint_builtins): New.
6667 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
6668 (aarch64_expand_builtin): Expand new builtins.
6670 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6672 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
6673 * combine-stack-adj.c (no_unhandled_cfa): Handle
6674 REG_CFA_TOGGLE_RA_MANGLE.
6675 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
6676 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
6677 info for return address signing.
6678 (aarch64_expand_epilogue): Likewise.
6680 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6682 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
6683 * config/aarch64/aarch64-protos.h
6684 (aarch64_return_address_signing_enabled): New declaration.
6685 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
6687 (aarch64_expand_prologue): Sign return address before it's pushed onto
6689 (aarch64_expand_epilogue): Authenticate return address fetched from
6691 (aarch64_override_options): Sanity check for ILP32 and ISA level.
6692 (aarch64_attributes): New function attributes for "sign-return-address".
6693 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
6694 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
6695 ("*do_return"): Generate combined instructions according to key index.
6696 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
6697 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
6699 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
6700 * config/aarch64/aarch64.opt (msign-return-address=): New.
6701 * doc/extend.texi (AArch64 Function Attributes): Documents
6702 "sign-return-address=".
6703 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
6705 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6707 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
6708 overall option summary.
6710 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6712 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
6713 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
6714 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
6715 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
6717 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
6719 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
6720 -mpower9-minmax by default for -mcpu=power9.
6721 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
6722 128-bit floating point.
6724 2017-01-20 Alan Modra <amodra@gmail.com>
6726 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
6727 optimizing for size.
6729 2017-01-20 Alan Modra <amodra@gmail.com>
6732 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
6733 for strcmp and strncmp from corresponding builtin decl.
6735 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6737 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
6738 instead of i386/rtems-64.h.
6739 * config/i386/rtems-64.h: Remove.
6741 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6745 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
6747 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
6749 2017-01-19 Tamar Christina <tamar.christina@arm.com>
6751 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6752 Change int to HOST_WIDE_INT.
6753 * config/aarch64/aarch64-protos.h
6754 (aarch64_simd_gen_const_vector_dup): Likewise.
6755 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
6757 2017-01-19 David Malcolm <dmalcolm@redhat.com>
6759 * langhooks-def.h (lhd_type_for_size): New decl.
6760 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
6761 * langhooks.c (lhd_type_for_size): New function, taken from
6764 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
6766 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
6767 define_bypass for CR latency.
6768 (power9-cracked-alu): Update bypass latency and remove power9-branch.
6769 (power9-alu2): Add define_bypass for CR latency.
6771 (power9-mul): Update insn latency.
6772 (power9-mul-compare): Update insn latency, bypass latency and remove
6775 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6777 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
6779 * config/aarch64/aarch64.md
6780 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
6781 aarch64_nopcrelative_literal_loads.
6782 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
6784 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
6786 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
6788 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
6790 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
6793 * config.gcc (supported_defaults): Add lxc1-sxc1.
6794 (with_lxc1_sxc1): Add validation.
6795 (all_defaults): Add lxc1-sxc1.
6796 * config/mips/mips.opt (mlxc1-sxc1): New option.
6797 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6799 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6800 __mips_no_lxc1_sxc1.
6801 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
6802 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
6803 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
6805 2017-01-19 Richard Biener <rguenther@suse.de>
6807 PR tree-optimization/72488
6808 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
6809 sure to restore SSA info.
6810 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
6812 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
6814 PR rtl-optimization/79121
6815 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
6816 of the inner type when shifting an extended value.
6818 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6821 * symtab.c (symtab_node::equal_address_to): Fix comparing of
6822 interposable aliases.
6824 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
6827 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
6828 Use the evmergelohi instruction.
6829 (mov_si<mode>_e500_subreg4_2_le): Likewise.
6830 (mov_sitf_e500_subreg8_2_be): Likewise.
6831 (mov_sitf_e500_subreg12_2_le): Likewise.
6832 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
6833 (mov_si<mode>_e500_subreg4_2_be): Likewise.
6834 (mov_sitf_e500_subreg8_2_le): Likewise.
6835 (mov_sitf_e500_subreg12_2_be): Likewise.
6837 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6839 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
6840 attribute from vecsimple to vecperm.
6841 (altivec_vbpermq2): Likewise.
6843 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6846 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
6848 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6849 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
6850 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
6851 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
6852 case where N arg is SIZE_MAX.
6853 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
6854 (cmpstrsi): Add pattern.
6856 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6858 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6859 __builtin_vec_revb builtins.
6860 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
6861 built-in functions to support generation of the ISA 3.0 XXBR<x>
6862 vector byte reverse instructions.
6863 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
6864 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
6865 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
6866 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
6867 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
6868 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
6869 (P9V_BUILTIN_VEC_REVB): Likewise.
6870 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
6871 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
6872 (p9_xxbrq_v16qi): Likewise.
6873 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
6874 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
6875 (p9_xxbrh_v8hi): Likewise.
6876 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
6877 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
6878 vec_revb built-in functions.
6880 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
6882 PR rtl-optimization/78952
6883 * config/i386/i386.md (any_extract): New code iterator.
6884 (*insvqi_2): Use any_extract for source operand.
6885 (*insvqi_3): Use any_shiftrt for source operand.
6887 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
6889 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
6891 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
6893 2017-01-18 Matthias Klose <doko@ubuntu.com>
6895 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
6897 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6899 * config/rs6000/altivec.h (vec_bperm): Change #define.
6900 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
6901 (altivec_vbpermq2): New define_insn.
6902 (altivec_vbpermd): Likewise.
6903 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
6905 (VBPERMD): Likewise.
6906 (VBPERM): New polymorphic function interface.
6907 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
6908 Add entries for P9V_BUILTIN_VEC_VBPERM.
6909 * doc/extend.texi: Add interfaces for vec_bperm.
6911 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6913 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
6914 first letter of error messages.
6915 (s390_resolve_overloaded_builtin): Likewise.
6916 * config/s390/s390.c (s390_expand_builtin): Likewise.
6917 (s390_invalid_arg_for_unprototyped_fn): Likewise.
6918 (s390_valid_target_attribute_inner_p): Likewise.
6919 * config/s390/s390.md ("tabort"): Likewise.
6921 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
6923 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
6924 (ISA_AVOID_DIV_HILO): New macro.
6925 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
6926 (ISA_HAS_DDIV): Likewise.
6928 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6930 * doc/invoke.texi (fabi-version): Correct number of occurrences.
6932 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6934 * doc/invoke.texi (fabi-version): Spelling fix.
6936 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6939 * doc/invoke.texi (fabi-version): Mention mangling fix for
6942 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6945 * doc/invoke.texi (fabi-version): Document discriminator mangling.
6947 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
6950 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
6951 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
6953 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
6955 (stack_protect_test): Ditto.
6956 * config/rs6000/rs6000.opt (mstack-protector-guard=,
6957 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
6959 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
6960 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
6961 -mstack-protector-guard-offset=.
6962 (RS/6000 and PowerPC Options): Ditto.
6964 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6966 * config/i386/i386.h (MASK_CLASS_P): New define.
6967 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
6968 there are no registers from different register sets also when
6969 mask registers are used. Update function comment.
6970 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
6971 to (*k/*r) and (*k/*km) alternatives.
6973 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
6975 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
6976 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
6977 (EH_RETURN_HANDLER_RTX): New define.
6978 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
6979 Force frame pointer in EH return functions.
6980 (aarch64_expand_epilogue): Add barrier for eh_return.
6981 (aarch64_final_eh_return_addr): Remove.
6982 (aarch64_eh_return_handler_rtx): New function.
6983 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
6985 (aarch64_eh_return_handler_rtx): New prototype.
6987 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6989 * config/rs6000/altivec.h (vec_rlmi): New #define.
6990 (vec_vrlnm): Likewise.
6991 (vec_rlnm): Likewise.
6992 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
6993 (UNSPEC_VRLNM): Likewise.
6994 (VIlong): New mode iterator.
6995 (altivec_vrl<VI_char>mi): New define_insn.
6996 (altivec_vrl<VI_char>nm): Likewise.
6997 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
7000 (RLNM): New polymorphic function entry.
7001 (VRLWMI): New monomorphic function entry.
7003 (RLMI): New polymorphic function entry.
7004 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
7005 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
7006 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
7009 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7012 * dwarf2out.c (field_byte_offset): Restore the
7013 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
7014 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
7015 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
7018 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
7021 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
7023 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7026 * dwarf2out.c (add_data_member_location_attribute): For constant
7027 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
7028 instead of DW_AT_data_member_location, DW_AT_bit_offset and
7029 DW_AT_byte_size attributes.
7031 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
7033 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
7034 after forcing to constant memory when the code model is medium.
7036 2017-01-17 Julia Koval <julia.koval@intel.com>
7039 * config/i386/avx512fintrin.h
7040 (_mm512_i32gather_ps): Change __addr type to void const*.
7041 (_mm512_mask_i32gather_ps): Ditto.
7042 (_mm512_i32gather_pd): Ditto.
7043 (_mm512_mask_i32gather_pd): Ditto.
7044 (_mm512_i64gather_ps): Ditto.
7045 (_mm512_mask_i64gather_ps): Ditto.
7046 (_mm512_i64gather_pd): Ditto.
7047 (_mm512_mask_i64gather_pd): Ditto.
7048 (_mm512_i32gather_epi32): Ditto.
7049 (_mm512_mask_i32gather_epi32): Ditto.
7050 (_mm512_i32gather_epi64): Ditto.
7051 (_mm512_mask_i32gather_epi64): Ditto.
7052 (_mm512_i64gather_epi32): Ditto.
7053 (_mm512_mask_i64gather_epi32): Ditto.
7054 (_mm512_i64gather_epi64): Ditto.
7055 (_mm512_mask_i64gather_epi64): Ditto.
7056 (_mm512_i32scatter_ps): Change __addr type to void*.
7057 (_mm512_mask_i32scatter_ps): Ditto.
7058 (_mm512_i32scatter_pd): Ditto.
7059 (_mm512_mask_i32scatter_pd): Ditto.
7060 (_mm512_i64scatter_ps): Ditto.
7061 (_mm512_mask_i64scatter_ps): Ditto.
7062 (_mm512_i64scatter_pd): Ditto.
7063 (_mm512_mask_i64scatter_pd): Ditto.
7064 (_mm512_i32scatter_epi32): Ditto.
7065 (_mm512_mask_i32scatter_epi32): Ditto.
7066 (_mm512_i32scatter_epi64): Ditto.
7067 (_mm512_mask_i32scatter_epi64): Ditto.
7068 (_mm512_i64scatter_epi32): Ditto.
7069 (_mm512_mask_i64scatter_epi32): Ditto.
7070 (_mm512_i64scatter_epi64): Ditto.
7071 (_mm512_mask_i64scatter_epi64): Ditto.
7072 * config/i386/avx512pfintrin.h
7073 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
7074 (_mm512_mask_prefetch_i32gather_ps): Ditto.
7075 (_mm512_mask_prefetch_i64gather_pd): Ditto.
7076 (_mm512_mask_prefetch_i64gather_ps): Ditto.
7077 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
7078 (_mm512_prefetch_i32scatter_ps): Ditto.
7079 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
7080 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
7081 (_mm512_prefetch_i64scatter_pd): Ditto.
7082 (_mm512_prefetch_i64scatter_ps): Ditto.
7083 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
7084 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
7085 * config/i386/avx512vlintrin.h
7086 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
7087 (_mm_mmask_i32gather_ps): Ditto.
7088 (_mm256_mmask_i32gather_pd): Ditto.
7089 (_mm_mmask_i32gather_pd): Ditto.
7090 (_mm256_mmask_i64gather_ps): Ditto.
7091 (_mm_mmask_i64gather_ps): Ditto.
7092 (_mm256_mmask_i64gather_pd): Ditto.
7093 (_mm_mmask_i64gather_pd): Ditto.
7094 (_mm256_mmask_i32gather_epi32): Ditto.
7095 (_mm_mmask_i32gather_epi32): Ditto.
7096 (_mm256_mmask_i32gather_epi64): Ditto.
7097 (_mm_mmask_i32gather_epi64): Ditto.
7098 (_mm256_mmask_i64gather_epi32): Ditto.
7099 (_mm_mmask_i64gather_epi32): Ditto.
7100 (_mm256_mmask_i64gather_epi64): Ditto.
7101 (_mm_mmask_i64gather_epi64): Ditto.
7102 (_mm256_i32scatter_ps): Change __addr type to void*.
7103 (_mm256_mask_i32scatter_ps): Ditto.
7104 (_mm_i32scatter_ps): Ditto.
7105 (_mm_mask_i32scatter_ps): Ditto.
7106 (_mm256_i32scatter_pd): Ditto.
7107 (_mm256_mask_i32scatter_pd): Ditto.
7108 (_mm_i32scatter_pd): Ditto.
7109 (_mm_mask_i32scatter_pd): Ditto.
7110 (_mm256_i64scatter_ps): Ditto.
7111 (_mm256_mask_i64scatter_ps): Ditto.
7112 (_mm_i64scatter_ps): Ditto.
7113 (_mm_mask_i64scatter_ps): Ditto.
7114 (_mm256_i64scatter_pd): Ditto.
7115 (_mm256_mask_i64scatter_pd): Ditto.
7116 (_mm_i64scatter_pd): Ditto.
7117 (_mm_mask_i64scatter_pd): Ditto.
7118 (_mm256_i32scatter_epi32): Ditto.
7119 (_mm256_mask_i32scatter_epi32): Ditto.
7120 (_mm_i32scatter_epi32): Ditto.
7121 (_mm_mask_i32scatter_epi32): Ditto.
7122 (_mm256_i32scatter_epi64): Ditto.
7123 (_mm256_mask_i32scatter_epi64): Ditto.
7124 (_mm_i32scatter_epi64): Ditto.
7125 (_mm_mask_i32scatter_epi64): Ditto.
7126 (_mm256_i64scatter_epi32): Ditto.
7127 (_mm256_mask_i64scatter_epi32): Ditto.
7128 (_mm_i64scatter_epi32): Ditto.
7129 (_mm_mask_i64scatter_epi32): Ditto.
7130 (_mm256_i64scatter_epi64): Ditto.
7131 (_mm256_mask_i64scatter_epi64): Ditto.
7132 (_mm_i64scatter_epi64): Ditto.
7133 (_mm_mask_i64scatter_epi64): Ditto.
7134 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
7135 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
7136 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
7137 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
7138 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
7139 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
7140 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
7141 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
7142 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
7143 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
7144 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
7145 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
7146 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
7147 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
7148 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
7149 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
7150 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
7151 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
7152 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
7153 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
7154 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
7155 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
7156 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
7157 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
7158 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
7159 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
7160 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
7161 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
7162 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
7163 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
7164 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
7165 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
7166 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
7167 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
7168 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
7169 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
7170 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
7171 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
7172 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
7173 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
7174 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
7175 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
7176 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
7177 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
7178 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
7179 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
7180 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
7181 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
7182 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
7183 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
7184 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
7185 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
7186 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
7187 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
7188 definitions accordingly.
7190 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
7191 Kuan-Lin Chen <kuanlinchentw@gmail.com>
7194 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
7197 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
7200 * ira-conflicts.c (ira_build_conflicts): Update total conflict
7201 hard regs for inner regno.
7203 2017-01-17 Martin Liska <mliska@suse.cz>
7206 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
7207 assumption and add comment.
7209 2017-01-17 Nathan Sidwell <nathan@acm.org>
7211 * ipa-visibility.c (localize_node): New function, broken out of ...
7212 (function_and_variable_visibility): ... here. Call it.
7214 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
7217 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
7218 correctly set frequency of oudgoing edge.
7219 (duplicate_thread_path): Fix profile updating.
7221 2017-01-17 Jakub Jelinek <jakub@redhat.com>
7224 * configure.ac: Add GCC_BASE_VER.
7225 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
7226 version from BASE-VER file.
7227 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
7228 (gcc.o): Depend on $(BASEVER).
7229 * common.opt (dumpfullversion): New option.
7230 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
7231 * doc/invoke.texi: Document -dumpfullversion.
7232 * doc/install.texi: Document --with-gcc-major-version-only.
7233 * configure: Regenerated.
7235 2017-01-17 Richard Biener <rguenther@suse.de>
7237 PR tree-optimization/71433
7238 * tree-vrp.c (register_new_assert_for): Merge same asserts
7239 on all incoming edges.
7240 (process_assert_insertions_for): Handle insertions at the
7243 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
7245 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
7246 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
7248 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
7251 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
7254 2017-01-17 Alan Modra <amodra@gmail.com>
7257 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
7258 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
7259 symbolic stack limit when pic.
7261 2017-01-16 Martin Sebor <msebor@redhat.com>
7263 PR tree-optimization/78608
7264 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
7266 2017-01-16 Jeff Law <law@redhat.com>
7269 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
7270 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
7271 for several include directories that may be relative to sysroot.
7272 * config/i386/x-mingw32 (gplus_includedir): Define.
7273 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
7274 (native_system_includedir): Likewise.
7275 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
7276 override if TARGET_SYSTEM_ROOT is defined.
7277 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
7279 PR tree-optimization/79090
7280 PR tree-optimization/33562
7281 PR tree-optimization/61912
7282 PR tree-optimization/77485
7283 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
7284 and computed trims into the dump file.
7286 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
7288 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
7290 2017-01-16 Jakub Jelinek <jakub@redhat.com>
7293 * gimplify.c (gimplify_init_constructor): If want_value and
7294 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
7298 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
7299 sequence. Formatting fixes.
7300 (doloop_optimize): Formatting fixes.
7303 * gcc.c (debug_level_greater_than_spec_func): New function.
7304 (static_spec_functions): Add debug-level-gt spec function.
7305 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
7307 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
7308 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
7309 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
7310 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
7311 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
7312 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
7314 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
7316 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
7317 QImode fixups to general and mask registers only.
7319 2017-01-16 Carl Love <cel@us.ibm.com>
7321 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
7322 for built-in functions
7323 vector signed char vec_nabs (vector signed char)
7324 vector signed short vec_nabs (vector signed short)
7325 vector signed int vec_nabs (vector signed int)
7326 vector signed long long vec_nabs (vector signed long long)
7327 vector float vec_nabs (vector float)
7328 vector double vec_nabs (vector double)
7329 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
7331 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
7332 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
7333 * doc/extend.texi: Update the documentation file for the new built-in
7336 2017-01-16 Martin Sebor <msebor@redhat.com>
7338 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
7341 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7343 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
7344 UNSPEC_VSX__XXSPLTD to require special splat handling.
7346 2017-01-16 David Malcolm <dmalcolm@redhat.com>
7349 * system.h: Poison strndup.
7351 2017-01-16 Alan Modra <amodra@gmail.com>
7354 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
7357 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
7359 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
7361 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
7363 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
7364 call recog here. Assert that INSN_CODE (insn) is non-negative.
7366 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7369 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
7371 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
7372 in the currently scheduled RTL fragment.
7374 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7376 PR rtl-optimization/78751
7377 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
7380 2017-01-14 Jeff Law <law@redhat.com>
7382 PR tree-optimization/79090
7383 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
7384 variable length stores.
7385 (compute_trims): Delete dead assignment to *trim_tail.
7386 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
7389 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
7391 PR rtl-optimization/78626
7392 PR rtl-optimization/78727
7393 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
7394 of a block, and split such blocks after everything else is finished.
7396 2017-01-14 Alan Modra <amodra@gmail.com>
7399 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
7400 target legitimate_combined_insn.
7401 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
7402 (rs6000_legitimate_combined_insn): New function.
7403 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
7405 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
7406 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
7407 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
7409 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
7411 * doc/frontends.texi (G++ and GCC): Remove references to Java.
7413 2017-01-13 Jeff Law <law@redhat.com>
7415 PR tree-optimization/33562
7416 PR tree-optimization/61912
7417 PR tree-optimization/77485
7418 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
7420 (delete_dead_assignment): Likewise.
7421 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
7422 statement to delete_dead_call and delete_dead_assignment.
7424 2017-01-13 David Malcolm <dmalcolm@redhat.com>
7427 * substring-locations.c (format_warning_va): Strengthen case 1 so
7428 that both endpoints of the substring must be within the format
7429 range for just the substring to be printed.
7431 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
7433 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
7434 * config/i386/i386.c (ix86_target_string): Add missing options
7435 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
7436 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
7437 flags_other and ix86_target_other to flags2_other. Display unknown
7439 (ix86_valid_target_attribute_inner_p): Add missing options and
7440 reorder options by implied ISAs, as in ix86_target_string.
7442 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7444 * hash-table.h (hash_table::too_empty_p): New function.
7445 (hash_table::expand): Use it.
7446 (hash_table::traverse): Likewise.
7447 (hash_table::empty_slot): Use sizeof (value_type) instead of
7448 sizeof (PTR) to convert bytes to elements. Shrink the table
7449 if the current size is excessive for the current number of
7452 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7454 * ira-costs.c (record_reg_classes): Break from the inner loop
7455 early once alt_fail is known to be true. Update outer loop
7456 handling accordingly.
7458 2017-01-13 Jeff Law <law@redhat.com>
7460 * tree-ssa-dse.c (decrement_count): New function.
7461 (increment_start_addr, maybe_trim_memstar_call): Likewise.
7462 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
7463 when we know the partially dead statement is a mem* function.
7465 PR tree-optimization/61912
7466 PR tree-optimization/77485
7467 * tree-ssa-dse.c: Include expr.h.
7468 (maybe_trim_constructor_store): New function.
7469 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
7471 PR tree-optimization/33562
7472 PR tree-optimization/61912
7473 PR tree-optimization/77485
7474 * doc/invoke.texi: Document new dse-max-object-size param.
7475 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
7476 * tree-ssa-dse.c: Include params.h.
7477 (dse_store_status): New enum.
7478 (initialize_ao_ref_for_dse): New, partially extracted from
7480 (valid_ao_ref_for_dse, normalize_ref): New.
7481 (setup_live_bytes_from_ref, compute_trims): Likewise.
7482 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
7483 (maybe_trim_partially_dead_store): Likewise.
7484 (maybe_trim_complex_store): Likewise.
7485 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
7486 Track what bytes live from the original store. Return tri-state
7487 for dead, partially dead or live.
7488 (dse_dom_walker): Add constructor, destructor and new private members.
7489 (delete_dead_call, delete_dead_assignment): New extracted from
7491 (dse_optimize_stmt): Make a member of dse_dom_walker.
7492 Use initialize_ao_ref_for_dse.
7494 PR tree-optimization/33562
7495 PR tree-optimization/61912
7496 PR tree-optimization/77485
7497 * sbitmap.h (bitmap_count_bits): Prototype.
7498 (bitmap_clear_range, bitmap_set_range): Likewise.
7499 * sbitmap.c (bitmap_clear_range): New function.
7500 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
7502 2017-01-13 Martin Liska <mliska@suse.cz>
7505 * function.c (set_cfun): Add new argument force.
7506 * function.h (set_cfun): Likewise.
7507 * ipa-inline-transform.c (inline_call): Use the function when
7508 strict alising from is dropped for function we inline to.
7510 2017-01-13 Richard Biener <rguenther@suse.de>
7512 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
7513 for dumping GIMPLE INTEGER_CSTs.
7515 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7517 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
7518 to 201112L since C++17.
7520 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7523 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
7524 if -fsanitize=kernel-address is present.
7526 2017-01-13 Richard Biener <rguenther@suse.de>
7528 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
7529 as _Literal ( type ) number in case usual suffixes do not
7530 preserve all information.
7532 2017-01-13 Richard Biener <rguenther@suse.de>
7534 PR tree-optimization/77283
7535 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
7536 and ssa-iterators.h.
7537 (is_feasible_trace): Implement a cost model based on joiner
7540 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7543 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
7544 char or short to __float128/_Float128 directly.
7546 2017-01-12 Martin Sebor <msebor@redhat.com>
7548 to -Wformat-overflow.
7549 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
7550 (min_bytes_remaining): Same.
7551 (get_string_length): Same.
7552 (format_string): Same.
7553 (format_directive): Same.
7555 (pass_sprintf_length::handle_gimple_call): Same.
7557 2017-01-12 Jakub Jelinek <jakub@redhat.com>
7559 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
7560 info.nowrite calls with no lhs that can't throw. Return bool
7561 whether gsi_remove has been called or not.
7562 (pass_sprintf_length::handle_gimple_call): Return bool whether
7563 try_substitute_return_value called gsi_remove. Formatting fix.
7564 (pass_sprintf_length::execute): Don't use gsi_remove if
7565 handle_gimple_call returned true.
7568 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
7569 be removed due to side-effects, don't remove following barrier nor
7570 turn the successor edge into fallthru edge.
7572 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7575 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
7576 element-reversing loads and stores as not swappable.
7578 2017-01-12 Nathan Sidwell <nathan@acm.org>
7579 Nicolai Stange <nicstange@gmail.com>
7581 * combine.c (try_combine): Don't ignore result of overlap checking
7582 loop. Combine overlap & asm check into single loop.
7584 2017-01-12 Richard Biener <rguenther@suse.de>
7586 * tree-pretty-print.c (dump_generic_node): Provide -gimple
7587 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
7589 2017-01-12 Richard Biener <rguenther@suse.de>
7591 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
7592 and TS_TARGET_OPTION directly derive from TS_BASE.
7593 * tree-core.h (tree_optimization_option): Derive from tree_base.
7594 (tree_target_option): Likewise.
7596 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
7598 * config/i386/i386.c (memory_address_length): Increase len
7599 only when rip_relative_addr_p returns false.
7601 2017-01-11 Julia Koval <julia.koval@intel.com>
7603 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
7604 (OPTION_MASK_ISA_SGX_SET): New.
7605 (ix86_handle_option): Handle OPT_msgx.
7606 * config.gcc: Added sgxintrin.h.
7607 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
7608 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
7609 * config/i386/i386.c (ix86_target_string): Add -msgx.
7611 (ix86_option_override_internal): Handle new options.
7612 (ix86_valid_target_attribute_inner_p): Add sgx.
7613 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
7614 * config/i386/i386.opt: Add msgx.
7615 * config/i386/sgxintrin.h: New file.
7616 * config/i386/x86intrin.h: Add sgxintrin.h.
7618 2017-01-11 Jakub Jelinek <jakub@redhat.com>
7621 * fold-const.c (maybe_nonzero_address): Return 1 for function
7623 (tree_single_nonzero_warnv_p): Don't handle function local objects
7627 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
7630 2017-01-11 David Malcolm <dmalcolm@redhat.com>
7633 * opts.c: Include "spellcheck.h"
7634 (struct string_fragment): New struct.
7635 (struct edit_distance_traits<const string_fragment &>): New
7637 (get_closest_sanitizer_option): New function.
7638 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
7640 2017-01-11 Jakub Jelinek <jakub@redhat.com>
7642 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
7644 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
7645 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
7646 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
7647 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
7648 for initial die_offset if dwarf_split_debug_info.
7649 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
7650 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
7652 (output_skeleton_debug_sections): Formatting fix. Use
7653 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
7654 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
7656 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
7658 * config/arm/cortex-a53.md: Add bypasses for
7660 (cortex_a53_r2f): Only use for transfers.
7661 (cortex_a53_f2r): Likewise.
7662 (cortex_a53_r2f_cvt): Add reservation for conversions.
7663 (cortex_a53_f2r_cvt): Likewise.
7665 2017-01-11 Tamar Christina <tamar.christina@arm.com>
7667 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
7668 to all inlined functions, change static to extern.
7670 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
7673 * config/arm/arm.c (legitimize_pic_address): Handle reference to
7675 (arm_assemble_integer): Likewise.
7677 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
7679 * config.gcc: Use new awk script to check CPU, FPU and architecture
7680 parameters for --with-... options.
7681 * config/arm/parsecpu.awk: New file
7682 * config/arm/arm-cpus.in: New file.
7683 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
7685 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
7687 * config/arm/t-arm: Update dependency rules.
7688 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
7689 of processing .def files.
7690 * config/arm/genopt.sh: Deleted.
7691 * config/arm/gentune.sh: Deleted.
7692 * config/arm/arm-cores.def: Deleted.
7693 * config/arm/arm-arches.def: Deleted.
7694 * config/arm/arm-fpus.def: Deleted.
7695 * config/arm/arm-tune.md: Regenerated.
7696 * config/arm/arm-tables.opt: Regenerated.
7697 * config/arm/arm-cpu.h: New generated file.
7698 * config/arm/arm-cpu-data.h: New generated file.
7699 * config/arm/arm-cpu-cdata.h: New generated file.
7701 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
7704 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
7706 (input_varpool_node): Unpack dynamically_initialized bit.
7708 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
7710 PR rtl-optimization/79032
7711 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
7712 the alignment of the adjusted memory reference against that of MODE,
7713 instead of the alignment of the original memory reference.
7715 2017-01-11 Martin Jambor <mjambor@suse.cz>
7717 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
7719 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
7720 decorated functions.
7722 2017-01-11 Richard Biener <rguenther@suse.de>
7724 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
7725 set range/nonnull info for PHI results. Do not set it on
7726 stmts marked for removal.
7728 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
7730 * expr.c (store_field): In the bitfield case, fetch the return value
7731 from the registers before applying a single big-endian adjustment.
7732 Always do a final load for a BLKmode value not larger than a word.
7734 2017-01-10 David Malcolm <dmalcolm@redhat.com>
7737 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7738 that we correctly handle column numbers greater than
7739 LINE_MAP_MAX_COLUMN_NUMBER.
7741 2017-01-10 Martin Sebor <msebor@redhat.com>
7744 * gimple-ssa-sprintf.c (get_destination_size): Call
7745 {init,fini}object_sizes.
7746 * tree-object-size.c (addr_object_size): Adjust.
7747 (pass_through_call): Adjust.
7748 (pass_object_sizes::execute): Adjust.
7749 * tree-object-size.h (fini_object_sizes): Declare.
7751 2017-01-10 Martin Sebor <msebor@redhat.com>
7753 PR tree-optimization/78775
7754 * builtins.c (get_size_range): Move...
7755 * calls.c: ...to here.
7756 (alloc_max_size): Accept zero argument.
7757 (operand_signed_p): Remove.
7758 (maybe_warn_alloc_args_overflow): Call get_size_range.
7759 * calls.h (get_size_range): Declare.
7761 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
7763 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
7764 from TI's devices.csv file as of September 2016.
7765 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
7767 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
7769 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
7770 * doc/invoke.texi: Likewise.
7771 * doc/md.texi: Likewise.
7772 * doc/objc.texi: Likewise.
7774 2017-01-10 Joshua Conner <joshconner@google.com>
7776 * config/arm/fuchsia-elf.h: New file.
7777 * config/fuchsia.h: New file.
7778 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
7779 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
7781 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
7783 2016-01-10 Richard Biener <rguenther@suse.de>
7785 PR tree-optimization/79034
7786 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
7787 Propagate out degenerate PHIs in the joiner.
7789 2017-01-10 Martin Liska <mliska@suse.cz>
7791 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
7792 (sort_congruence_classes_by_decl_uid): Likewise.
7793 (sort_congruence_class_groups_by_decl_uid): Likewise.
7794 (sem_item_optimizer::merge_classes): Sort class, groups in these
7795 classes and members in the groups by DECL_UID of declarations.
7796 This would make merge operations stable.
7798 2017-01-10 Martin Liska <mliska@suse.cz>
7800 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
7801 usage of m_classes_vec.
7802 (sem_item_optimizer::~sem_item_optimizer): Likewise.
7803 (sem_item_optimizer::get_group_by_hash): Likewise.
7804 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
7805 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
7806 (sem_item_optimizer::verify_classes): Likewise.
7807 (sem_item_optimizer::process_cong_reduction): Likewise.
7808 (sem_item_optimizer::dump_cong_classes): Likewise.
7809 (sem_item_optimizer::merge_classes): Likewise.
7810 * ipa-icf.h (congruence_class_hash): Rename from
7811 congruence_class_group_hash. Remove declaration of m_classes_vec.
7813 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
7815 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
7816 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
7817 * config.gcc: Add avx512vpopcntdqintrin.h.
7818 * config/i386/avx512vpopcntdqintrin.h: New.
7819 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
7820 * config/i386/i386-builtin-types.def: Add new types.
7821 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
7822 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
7823 __builtin_ia32_vpopcountq_v8di_mask): New.
7824 * config/i386/i386-c.c (ix86_target_macros_internal): Define
7825 __AVX512VPOPCNTDQ__.
7826 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
7827 (PTA_AVX512VPOPCNTDQ): Define.
7828 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
7829 TARGET_AVX512VPOPCNTDQ_P): Define.
7830 * config/i386/i386.opt: Add mavx512vpopcntdq.
7831 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
7832 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
7834 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7837 * predict.def (PRED_CALL): Set to 67.
7839 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
7841 * expr.c (store_field): In the bitfield case, if the value comes from
7842 a function call and is of an aggregate type returned in registers, do
7843 not modify the field mode; extract the value in all cases if the mode
7844 is BLKmode and the size is not larger than a word.
7846 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
7849 * config/i386/cpuid.h: Fix undefined behavior.
7851 2017-01-04 Jeff Law <law@redhat.com>
7853 PR tree-optimization/79007
7854 PR tree-optimization/67955
7855 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
7856 conservative for pt.null when flag_non_call_exceptions is on.
7858 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7860 PR translation/79019
7861 PR translation/79020
7862 * params.def (PARAM_INLINE_MIN_SPEEDUP,
7863 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
7864 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
7866 * config/avr/avr.opt (maccumulate-args): Likewise.
7867 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
7868 * common.opt (freport-bug): Likewise.
7869 * cif-code.def (CIF_FINAL_ERROR): Likewise.
7870 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
7871 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
7872 translatable string.
7873 * config/i386/i386.c (function_value_32): Likewise.
7874 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
7875 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
7877 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
7878 * common/config/msp430/msp430-common.c (msp430_handle_option):
7880 * symtab.c (symtab_node::verify_base): Likewise.
7881 * opts.c (set_debug_level): Likewise.
7882 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
7883 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
7884 missing whitespace to translatable strings.
7885 * config/avr/avr.md (bswapsi2): Fix typo in comment.
7886 * config/sh/superh.h: Likewise.
7887 * config/i386/xopintrin.h: Likewise.
7888 * config/i386/znver1.md: Likewise.
7889 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
7890 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
7891 * double-int.h (struct double_int): Likewise.
7892 * double-int.c (div_and_round_double): Likewise.
7893 * wide-int.cc: Likewise.
7894 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
7895 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
7896 * cfgcleanup.c (crossjumps_occured): Renamed to ...
7897 (crossjumps_occurred): ... this.
7898 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
7901 PR tree-optimization/78899
7902 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
7903 returning bool return struct loop *, NULL for failure and the new
7905 (versionable_outer_loop_p): Don't version outer loop if it has
7906 dont_vectorized bit set.
7907 (tree_if_conversion): When versioning outer loop, ensure
7908 tree_if_conversion is performed also on the inner loop of the
7909 non-vectorizable outer loop copy.
7910 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
7911 LOOP_VECTORIZED in inner loop of the scalar outer loop and
7912 prevent vectorization of it.
7913 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
7914 the outer loop vectorization of the non-scalar version is attempted
7915 before vectorization of the inner loop in scalar version. If
7916 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
7917 vectorization of its inner loop.
7918 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
7919 has 2 inner loops, rename also on edges from bb whose single pred
7920 is outer_loop->header. Fix typo in function comment.
7922 2017-01-09 Martin Sebor <msebor@redhat.com>
7925 * asan.c (asan_emit_stack_protection): Increase local buffer size
7926 to avoid snprintf truncation warning.
7928 2017-01-09 Andrew Pinski <apinski@cavium.com>
7930 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
7931 to reference thunderx2t99 for the tuning structure
7932 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
7934 (thunderx2t99_extra_costs): This.
7935 * config/aarch64/aarch64-tune.md: Regenerate.
7936 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
7937 (vulcan_addrcost_table): This.
7938 (vulcan_regmove_cost): Rename to ...
7939 (thunderx2t99_regmove_cost): This.
7940 (vulcan_vector_cost): Rename to ...
7941 (thunderx2t99_vector_cost): this.
7942 (vulcan_branch_cost): Rename to ...
7943 (thunderx2t99_branch_cost): This.
7944 (vulcan_tunings): Rename to ...
7945 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
7946 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
7948 2017-01-09 Martin Jambor <mjambor@suse.cz>
7952 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
7953 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
7954 (propagate_vr_accross_jump_function): Use the above function for all
7955 value range computations for pass-through jump functions and type
7956 converasion from explicit value range values.
7957 (ipcp_propagate_stage): Do not attempt to deduce types of formal
7958 parameters from TYPE_ARG_TYPES.
7959 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
7960 (ipa_write_node_info): Stream type of the actual argument.
7961 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
7963 2017-01-09 Martin Liska <mliska@suse.cz>
7966 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
7967 (lookup_compiler): Do not show error message with have_E.
7969 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7971 PR tree-optimization/78938
7972 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
7973 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
7974 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
7975 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
7978 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7980 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
7983 2017-01-09 Richard Biener <rguenther@suse.de>
7985 PR tree-optimization/78997
7986 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
7987 name condition properly.
7989 2017-01-09 Richard Biener <rguenther@suse.de>
7992 * dwarf2out.c (is_cxx): New overload with context.
7993 (is_naming_typedef_decl): Use it.
7995 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7997 * invoke.texi (Option Summary): Correct spacing in option lists
7998 and add line breaks to fix over-long lines.
8000 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8004 * extend.texi (Common Variable Attributes): Add xref to GCC
8005 Internals manual to explain mode attribute keywords.
8007 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
8010 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
8011 and Preprocessor Options.
8012 (Options for Linking): Document -pthread here....
8013 (RS/6000 and PowerPC Options): ...not here.
8014 (Solaris 2 Options): ...or here.
8015 * doc/cppopts.texi: Document -pthread.
8017 2017-01-08 Martin Sebor <msebor@redhat.com>
8020 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
8021 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
8022 New member functions.
8023 (format_directive): Used them.
8025 (pass_sprintf_length::handle_gimple_call): Same.
8026 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
8027 to avoid truncation for any argument.
8028 (extract_affine_mul): Same.
8029 * tree.c (get_file_function_name): Same.
8031 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8034 * predict.def (PRED_INDIR_CALL): Set to 86.
8036 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8038 PR preprocessor/54124
8039 * doc/cppopts.texi: Reformat -d subtable to list the full name
8040 of the options. Add cross-reference to the docs for the general
8041 compiler -d options.
8042 * doc/invoke.texi (Developer Options): Add cross-reference to the
8043 preprocessor-specific -d option documentation.
8045 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8047 PR preprocessor/13498
8048 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
8049 redudant material, and reflect new command-line options.
8050 (System Headers): Likewise.
8052 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
8054 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
8055 -isystem, and -idirafter. Copy-edit.
8056 * doc/cppopts.texi: Copy-edit. Remove contradiction about
8057 default for -ftrack-macro-expansion. Delete obsolete and
8058 badly-formatted implementation details about -fdebug-cpp output.
8059 * doc/cppwarnopts.texi: Copy-edit.
8061 2017-01-07 David Malcolm <dmalcolm@redhat.com>
8064 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
8065 that the transition from a max line width >= 1<<10 to narrower
8066 lines works correctly.
8068 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
8070 * doc/options.texi (PerFunction): New.
8071 * opt-functions.awk (switch_flags): Map both Optimization and
8072 PerFunction to CL_OPTIMIZATION.
8073 * opth-gen.awk: Test for PerFunction flag along with
8075 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
8076 it only when the latter is present. Skip those that don't in
8077 the hash function generator.
8078 * common.opt (fvar-tracking): Mark as PerFunction instead of
8080 (fvar-tracking-assignments): Likewise.
8081 (fvar-tracking-assignments-toggle): Likewise.
8082 (fvar-tracking-uninit): Likewise.
8084 2017-01-07 Jakub Jelinek <jakub@redhat.com>
8086 PR translation/79018
8087 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
8090 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
8093 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
8094 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
8095 TARGET_LONG_JUMP_TABLE_OFFSETS.
8096 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
8097 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
8098 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
8099 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
8100 * config/m68k/m68k.md (tablejump expander): Likewise.
8101 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
8102 TARGET_LONG_JUMP_TABLE_OFFSETS.
8103 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
8104 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
8106 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
8107 David Holsgrove <david.holsgrove@xilinx.com>
8109 * common/config/microblaze/microblaze-common.c
8110 (TARGET_EXCEPT_UNWIND_INFO): Remove.
8111 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
8113 * config/microblaze/microblaze.c (microblaze_must_save_register)
8114 (microblaze_expand_epilogue, microblaze_return_addr): Handle
8116 (microblaze_eh_return): New function.
8117 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
8118 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
8119 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
8120 * config/microblaze/microblaze.md (eh_return): New pattern.
8122 2017-01-06 Jakub Jelinek <jakub@redhat.com>
8124 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
8125 GCC_DIAGNOSTIC_STRINGIFY): Define.
8127 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
8129 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8131 * config/arm/arm.md (<mcrr>): New.
8133 * config/arm/arm.c (arm_arch5te): New.
8134 (arm_option_override): Set arm_arch5te.
8135 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
8137 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
8138 (arm_mcrr_qualifiers): ... this. New.
8139 (MRRC_QUALIFIERS): Define to...
8140 (arm_mrrc_qualifiers): ... this. New.
8141 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
8143 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
8144 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
8145 (MRRCI, mrrc, MRRC): New.
8146 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
8147 VUNSPEC_MRRC2): New.
8149 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8151 * config/arm/arm.md (<mcr>): New.
8153 * config/arm/arm.c (arm_coproc_builtin_available): Add
8154 support for mcr, mrc, mcr2 and mrc2.
8155 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
8156 (arm_mcr_qualifiers): ... this. New.
8157 (MRC_QUALIFIERS): Define to ...
8158 (arm_mrc_qualifiers): ... this. New.
8159 (MCR_QUALIFIERS): Define to ...
8160 (arm_mcr_qualifiers): ... this. New.
8161 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
8163 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
8164 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
8165 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
8168 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8170 * config/arm/arm.md (*ldc): New.
8174 * config/arm/arm.c (arm_coproc_builtin_available): Add
8175 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
8176 (arm_coproc_ldc_stc_legitimate_address): New.
8177 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
8178 'qualifier_const_pointer'.
8179 (LDC_QUALIFIERS): Define to...
8180 (arm_ldc_qualifiers): ... this. New.
8181 (STC_QUALIFIERS): Define to...
8182 (arm_stc_qualifiers): ... this. New.
8183 * config/arm/arm-protos.h
8184 (arm_coproc_ldc_stc_legitimate_address): New.
8185 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
8186 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
8187 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
8188 stc2, stcl, stc2l): New.
8189 * config/arm/constraints.md (Uz): New.
8190 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
8191 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
8192 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
8193 VUNSPEC_STC2L): New.
8195 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8197 * config/arm/arm.md (<cdp>): New.
8198 * config/arm/arm.c (neon_const_bounds): Rename this ...
8199 (arm_const_bounds): ... this.
8200 (arm_coproc_builtin_available): New.
8201 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
8202 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
8203 (CDP_QUALIFIERS): Define to...
8204 (arm_cdp_qualifiers): ... this. New.
8206 (arm_expand_builtin_args): Add case for 6 arguments.
8207 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
8208 (arm_const_bounds): ... this.
8209 (arm_coproc_builtin_available): New.
8210 * config/arm/arm_acle.h (__arm_cdp): New.
8212 * config/arm/arm_acle_builtins.def (cdp): New.
8214 * config/arm/iterators.md (CDPI,CDP,cdp): New.
8215 * config/arm/neon.md: Rename all 'neon_const_bounds' to
8217 * config/arm/types.md (coproc): New.
8218 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
8219 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
8220 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
8221 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
8223 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8225 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
8226 (UBINOP_QUALIFIERS): New.
8228 (acle_builtin_data): New. Change comment.
8229 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
8230 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
8231 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
8232 arm_acle_builtins.def.
8233 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
8234 (arm_init_acle_builtins): New.
8235 (CRC32_BUILTIN): Remove.
8236 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
8237 crc32cb, crc32ch and crc32cw.
8238 (arm_init_crc32_builtins): Remove.
8239 (arm_init_builtins): Use arm_init_acle_builtins rather
8240 than arm_init_crc32_builtins.
8241 (arm_expand_acle_builtin): New.
8242 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
8243 * config/arm/arm_acle_builtins.def: New.
8245 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
8247 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
8248 (arm_builtin_datum): ... this.
8249 (arm_init_neon_builtin): Rename to ...
8250 (arm_init_builtin): ... this. Add a new parameters PREFIX
8251 and USE_SIG_IN_NAME.
8252 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
8253 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
8254 'arm_builtin_datum'.
8255 (arm_init_vfp_builtins): Likewise.
8256 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
8257 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
8258 (arm_expand_neon_args): Rename to ...
8259 (arm_expand_builtin_args): ... this. Rename builtin_arg
8260 enum values and differentiate between ARG_BUILTIN_MEMORY
8261 and ARG_BUILTIN_NEON_MEMORY.
8262 (arm_expand_neon_builtin_1): Rename to ...
8263 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
8264 values, arm_expand_builtin_args and add bool parameter NEON.
8265 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
8266 (arm_expand_vfp_builtin): Likewise.
8267 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
8269 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8272 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
8273 * predict.c (tree_estimate_probability_bb): Reverse direction of
8274 polymorphic call predictor.
8276 2017-01-06 David Malcolm <dmalcolm@redhat.com>
8278 * passes.c (execute_one_pass): Split out pass-skipping logic into...
8279 (determine_pass_name_match): ...this new function and...
8280 (should_skip_pass_p): ...this new function.
8282 2017-01-06 Nathan Sidwell <nathan@acm.org>
8284 * ipa-visibility.c (function_and_variable_visibility): Reformat
8285 comments and long lines. Remove extrneous if.
8286 * symtab.c (symtab_node::make_decl_local): Fix code format.
8287 (symtab_node::set_section_for_node): Fix comment typo.
8289 2017-01-06 Martin Liska <mliska@suse.cz>
8292 * lra-constraints.c: Rename invariant to lra_invariant.
8293 * predict.c (set_even_probabilities): Initialize e to NULL.
8295 2017-01-05 Martin Sebor <msebor@redhat.com>
8297 PR tree-optimization/78910
8298 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
8299 (format_integer): Correct off-by-one error in the handling
8300 of precision with negative numbers in signed conversions..
8302 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
8304 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
8306 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8308 PR tree-optimization/71016
8309 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
8310 factor_out_conditional_conversion. Formatting fix.
8311 (factor_out_conditional_conversion): Add cond_stmt argument.
8312 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
8313 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
8316 2017-01-05 David Malcolm <dmalcolm@redhat.com>
8318 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
8319 read-rtl-function.o, and selftest-rtl.o.
8320 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
8321 (selftest::aarch64_test_loading_full_dump): New function.
8322 (selftest::aarch64_run_selftests): New function.
8323 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
8324 selftest::aarch64_run_selftests.
8325 * config/i386/i386.c
8326 (selftest::ix86_test_loading_dump_fragment_1): New function.
8327 (selftest::ix86_test_loading_call_insn): New function.
8328 (selftest::ix86_test_loading_full_dump): New function.
8329 (selftest::ix86_test_loading_unspec): New function.
8330 (selftest::ix86_run_selftests): Call the new functions.
8331 * emit-rtl.c (maybe_set_max_label_num): New function.
8332 * emit-rtl.h (maybe_set_max_label_num): New decl.
8333 * function.c (instantiate_decls): Guard call to
8334 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
8335 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
8337 * gensupport.c (gen_reader::gen_reader): Pass "false"
8338 for new "compact" param of rtx_reader.
8339 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
8340 rather than an empty string for NULL strings.
8341 * read-md.c: Potentially include config.h rather than bconfig.h.
8342 Wrap include of errors.h with #ifdef GENERATOR_FILE.
8343 (have_error): New global, copied from errors.c.
8344 (md_reader::read_name): Rename to...
8345 (md_reader::read_name_1): ...this, adding "out_loc" param,
8346 and converting "missing name or number" to returning false, rather
8348 (md_reader::read_name): Reimplement in terms of read_name_1.
8349 (md_reader::read_name_or_nil): New function.
8350 (md_reader::read_string): Handle "(nil)" by returning NULL.
8351 (md_reader::md_reader): Add new param "compact".
8352 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
8353 (md_reader::read_file): New method.
8354 * read-md.h (md_reader::md_reader): Add new param "compact".
8355 (md_reader::read_file): New method.
8356 (md_reader::is_compact): New accessor.
8357 (md_reader::read_name): Convert return type from void to file_location.
8358 (md_reader::read_name_or_nil): New decl.
8359 (md_reader::read_name_1): New decl.
8360 (md_reader::m_compact): New field.
8361 (noop_reader::noop_reader): Pass "false" for new "compact" param
8363 (rtx_reader::rtx_reader): Add new "compact" param.
8364 (rtx_reader::read_rtx_operand): Make virtual and convert return
8365 type from void to rtx.
8366 (rtx_reader::read_until): New decl.
8367 (rtx_reader::handle_any_trailing_information): New virtual function.
8368 (rtx_reader::postprocess): New virtual function.
8369 (rtx_reader::finalize_string): New virtual function.
8370 (rtx_reader::m_in_call_function_usage): New field.
8371 (rtx_reader::m_reuse_rtx_by_id): New field.
8372 * read-rtl-function.c: New file.
8373 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
8374 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
8375 (selftest::verify_three_block_rtl_cfg): New decl.
8376 * read-rtl-function.h: New file.
8377 * read-rtl.c: Potentially include config.h rather than bconfig.h.
8378 For host, include function.h, memmodel.h, and emit-rtl.h.
8379 (one_time_initialization): New function.
8380 (struct compact_insn_name): New struct.
8381 (compact_insn_names): New array.
8382 (find_code): Handle insn codes in compact dumps.
8383 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
8384 (bind_subst_iter_and_attr): Likewise.
8385 (add_condition_to_string): Likewise.
8386 (add_condition_to_rtx): Likewise.
8387 (apply_attribute_uses): Likewise.
8388 (add_current_iterators): Likewise.
8389 (apply_iterators): Likewise.
8390 (initialize_iterators): Guard usage of apply_subst_iterator with
8391 #ifdef GENERATOR_FILE.
8392 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
8393 (md_reader::read_mapping): Likewise.
8394 (add_define_attr_for_define_subst): Likewise.
8395 (add_define_subst_attr): Likewise.
8396 (read_subst_mapping): Likewise.
8397 (check_code_iterator): Likewise.
8398 (rtx_reader::read_rtx): Likewise. Move one-time initialization
8400 (one_time_initialization): New function.
8401 (rtx_reader::read_until): New method.
8402 (read_flags): New function.
8403 (parse_reg_note_name): New function.
8404 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
8405 Handle reuse_rtx ids.
8406 Wrap iterator lookup within #ifdef GENERATOR_FILE.
8407 Add parsing support for RTL dumps, mirroring the special-cases in
8408 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
8409 values, and calling handle_any_trailing_information.
8410 (rtx_reader::read_rtx_operand): Convert return type from void
8411 to rtx, returning return_rtx. Handle case 'e'. Call
8412 finalize_string on XSTR and XTMPL fields.
8413 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
8414 "(nil)" values were omitted. Call the postprocess vfunc on the
8416 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
8417 class ctor. Initialize m_in_call_function_usage. Call
8418 one_time_initialization.
8419 * rtl-tests.c (selftest::test_uncond_jump): Call
8420 set_new_first_and_last_insn.
8421 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
8422 * selftest-rtl.c: New file.
8423 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
8424 (selftest::get_insn_by_uid): New decl.
8425 * selftest-run-tests.c (selftest::run_tests): Call
8426 read_rtl_function_c_tests.
8427 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
8428 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
8431 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
8433 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
8434 operands in a special way. Assert that pos+len <= mode precision.
8436 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8438 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
8439 3 argument Alias with unlimited for the negative form.
8440 (fno-vect-cost-model): Removed.
8442 2017-01-05 Martin Liska <mliska@suse.cz>
8444 * hsa-gen.c (gen_hsa_divmod): New function.
8445 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
8447 2017-01-05 Martin Liska <mliska@suse.cz>
8450 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
8453 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8455 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
8456 small constant length operands.
8458 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8460 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
8461 between loop iterations.
8463 2017-01-05 Martin Liska <mliska@suse.cz>
8466 * gimplify.c (gimplify_decl_expr): Compare to
8467 asan_poisoned_variables instread of checking flags.
8468 (gimplify_target_expr): Likewise.
8469 (gimplify_expr): Likewise.
8470 (gimplify_function_tree): Conditionally initialize
8471 asan_poisoned_variables.
8473 2017-01-04 Jeff Law <law@redhat.com>
8475 PR tree-optimizatin/78812
8476 * rtl.h (contains_mem_rtx_p): Prototype.
8477 * ifcvt.c (containts_mem_rtx_p): Move from here to...
8478 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
8479 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
8480 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
8481 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
8483 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8485 * input.c (assert_char_at_range): Default-initialize actual_range.
8487 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8489 * df-scan.c (df_ref_create_structure): Make regno unsigned,
8490 to match the caller.
8492 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8494 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
8495 insns after final jump in test to emit dummy move.
8497 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8499 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
8500 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
8502 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8504 * multiple_target.c (create_dispatcher_calls): Init e_next.
8505 * tree-ssa-loop-split.c (split_loop): Init border.
8506 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
8509 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
8514 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
8515 (altivec_register_operand): Do not return true if the operand
8516 contains a SUBREG mixing SImode and SFmode.
8517 (vsx_register_operand): Likewise.
8518 (vsx_reg_sfsubreg_ok): New predicate.
8519 (vfloat_operand): Do not return true if the operand contains a
8520 SUBREG mixing SImode and SFmode.
8521 (vint_operand): Likewise.
8522 (vlogical_operand): Likewise.
8523 (gpc_reg_operand): Likewise.
8524 (int_reg_operand): Likewise.
8525 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
8526 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
8527 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
8529 (rs6000_emit_move_si_sf_subreg): New helper function.
8530 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
8531 fixup SUBREGs involving SImode and SFmode.
8532 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
8533 numbers for the new peephole2 optimization.
8534 (peephole2 for SFmode unions): New peephole2 to optimize cases in
8535 the GLIBC math library that do AND/IOR/XOR operations on single
8536 precision floating point.
8537 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
8538 target macros to say whether we need to avoid SUBREGs mixing
8540 (TARGET_ALLOW_SF_SUBREG): Likewise.
8541 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
8542 (UNSPEC_SI_FROM_SF): Likewise.
8543 (iorxor): Change spacing.
8544 (and_ior_xor): New iterator for AND, IOR, and XOR.
8545 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
8546 (movdi_from_sf_zero_ext): Likewise.
8547 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
8548 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
8549 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
8550 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
8551 (fms<mode>4): Likewise.
8552 (fnma<mode>4): Likewise.
8553 (fnms<mode>4): Likewise.
8554 (nfma<mode>4): Likewise.
8555 (nfms<mode>4): Likewise.
8557 2017-01-04 Marek Polacek <polacek@redhat.com>
8560 * doc/invoke.texi: Document -Wpointer-compare.
8562 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8564 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
8567 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
8568 descriptions for -gdwarf-5 and emit them as uleb128 instead of
8571 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
8574 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
8575 documentation of the powerpc_popcntb_ok attribute.
8576 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
8577 code to issue warning messages if a requested CPU configuration is
8578 not supported by the binary (assembler and loader) toolchain.
8579 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
8580 made to define a built-in function that has been disabled.
8581 (paired_init_builtins): Add assertion to prevent ICE if attempt is
8582 made to define a built-in function that has been disabled.
8583 (altivec_init_builtins): Add comment explaining why definition
8584 of the DST built-in functions is not preceded by an assertion
8585 check. Add assertions to prevent ICE if attempts are made to
8586 define an altivec predicate or an abs* built-in function that has
8588 (htm_init_builtins): Add comment explaining why definition of the
8589 htm built-in functions is not preceded by an assertion check.
8591 2017-01-04 Jeff Law <law@redhat.com>
8593 PR tree-optimizatin/67955
8594 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
8595 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
8596 the points-to solution does not include pt_null. Use DECL_PT_UID
8599 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
8601 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
8602 Use gen_int_mode instead of gen_lopwart for const_int operands.
8604 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8606 PR tree-optimization/71563
8607 * match.pd: Simplify X << Y into X if Y is known to be 0 or
8608 out of range value - has low bits known to be zero.
8610 2017-01-04 Alan Modra <amodra@gmail.com>
8612 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
8613 * configure: Regenerate.
8614 * config.in: Regenerate.
8616 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8619 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
8620 a substring of the message, but strcmp with the whole message. Ifdef
8621 ENABLE_NLS, translate the message first using dgettext.
8623 2017-01-03 Jeff Law <law@redhat.com>
8625 PR tree-optimizatin/78856
8626 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
8627 (mark_threaded_blocks): Remove code to truncate thread paths that
8628 cross multiple loop headers. Instead invalidate the cached loop
8629 iteration information and handle case of a thread path walking
8630 into an irreducible region.
8632 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8635 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
8636 assertions. Add support for doing the signbit if the IEEE 128-bit
8637 floating point value is in a GPR.
8638 * config/rs6000/rs6000.md (Fsignbit): Delete.
8639 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
8640 Update the length attribute if the value is in a GPR.
8641 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
8642 the sign or zero extension instruction, since the value is always 0/1.
8643 (signbit<mode>2_dm2): Delete using <Fsignbit>.
8646 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
8647 extracting SImode to a GPR register so that we can generate a
8648 store, limit the vector to be in a traditional Altivec register
8649 for the vextuwrx instruction.
8651 2017-01-03 Ian Lance Taylor <iant@google.com>
8653 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
8655 2017-01-03 Martin Sebor <msebor@redhat.com>
8657 PR tree-optimization/78696
8658 * gimple-ssa-sprintf.c (format_floating): Correct handling of
8659 precision. Use MPFR for %f for greater fidelity. Correct handling
8661 (pass_sprintf_length::compute_format_length): Set width and precision
8662 specified by asrerisk to void_node for vararg functions.
8663 (try_substitute_return_value): Adjust dump output.
8665 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
8667 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
8669 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
8671 * doc/invoke.texi (SPARC options): Document -mlra as the default.
8672 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
8673 -mlra/-mno-lra was passed to the compiler.
8675 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
8677 PR rtl-optimization/65618
8678 * emit-rtl.c (try_split): Move initialization of "before" and
8679 "after" to just before the call to emit_insn_after_setloc.
8681 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
8683 * doc/md.texi (Standard Names): Remove reference to Java frontend.
8685 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
8687 * dwarf2out.c (gen_enumeration_type_die): When
8688 -gno-strict-dwarf, add a DW_AT_encoding attribute.
8690 2017-01-03 Jakub Jelinek <jakub@redhat.com>
8692 PR tree-optimization/78965
8693 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
8694 Change first argument from const call_info & to call_info &. For %n
8695 set info.nowrite to false.
8698 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
8699 possibly throwing calls.
8701 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
8702 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
8703 and fns handling, rather than in a separate case SSA_NAME.
8705 2017-01-02 Jeff Law <law@redhat.com>
8707 * config/darwin-driver.c (darwin_driver_init): Const-correctness
8708 fixes for first_period and second_period variables.
8710 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
8713 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
8714 (*insvqi_1): New insn pattern.
8715 (*insvqi_1_mem_rex64): Ditto.
8717 (*insvqi_3): Rename from *insvqi.
8719 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
8721 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
8723 * doc/cfg.texi (Edges): Remove reference to Java.
8724 (Maintaining the CFG): Ditto.
8726 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8729 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
8730 transparent aliases.
8732 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8735 * predict.def (PRED_CALL): Update hitrate.
8736 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
8737 * predict.c (tree_estimate_probability_bb): Split CALL predictor
8738 into direct/indirect/polymorphic variants.
8740 2017-01-01 Jakub Jelinek <jakub@redhat.com>
8742 Update copyright years.
8744 * gcc.c (process_command): Update copyright notice dates.
8745 * gcov-dump.c (print_version): Ditto.
8746 * gcov.c (print_version): Ditto.
8747 * gcov-tool.c (print_version): Ditto.
8748 * gengtype.c (create_file): Ditto.
8749 * doc/cpp.texi: Bump @copying's copyright year.
8750 * doc/cppinternals.texi: Ditto.
8751 * doc/gcc.texi: Ditto.
8752 * doc/gccint.texi: Ditto.
8753 * doc/gcov.texi: Ditto.
8754 * doc/install.texi: Ditto.
8755 * doc/invoke.texi: Ditto.
8757 Copyright (C) 2017 Free Software Foundation, Inc.
8759 Copying and distribution of this file, with or without modification,
8760 are permitted in any medium without royalty provided the copyright
8761 notice and this notice are preserved.