1 2014-03-27 Marek Polacek <polacek@redhat.com>
4 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
7 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
9 * config/s390/s390.c (s390_can_use_return_insn): Check for
10 call-saved FPRs on 31 bit.
12 2014-03-27 Jakub Jelinek <jakub@redhat.com>
15 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
16 if they need regimplification, just drop them instead of
17 calling gimple_regimplify_operands on them.
19 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
22 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
23 (aarch64_frame_pointer_required): Adjust logic.
24 (aarch64_can_eliminate): Adjust logic.
25 (aarch64_override_options_after_change): Adjust logic.
27 2014-03-27 Dehao Chen <dehao@google.com>
29 * ipa-inline.c (early_inliner): Update node's inline info.
31 2014-03-26 Dehao Chen <dehao@google.com>
33 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
34 compiler inserted conditional jumps for NAN float check.
36 2014-03-26 Jakub Jelinek <jakub@redhat.com>
38 * ubsan.h (ubsan_create_data): Change second argument's type
39 to const location_t *.
40 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
42 (ubsan_create_data): Change second argument to const location_t *PLOC.
43 Create Loc field whenever PLOC is non-NULL.
44 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
45 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
49 * real.c (real_to_integer2): Change type of low to UHWI.
51 2014-03-26 Tobias Burnus <burnus@net-b.de>
53 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
54 (CILK_SELF_SPECS): New define.
55 (driver_self_specs): Use it.
57 2014-03-26 Richard Biener <rguenther@suse.de>
59 * tree-pretty-print.c (percent_K_format): Implement special
60 case for LTO and its stripped down BLOCK tree.
62 2014-03-26 Jakub Jelinek <jakub@redhat.com>
65 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
67 * tree-vrp.c (simplify_internal_call_using_ranges): If only
68 one range is range_int_cst_p, but not both, at least optimize
69 addition/subtraction of 0 and multiplication by 0 or 1.
70 * gimple-fold.c (gimple_fold_call): Fold
71 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
72 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
73 INTEGER_CSTs, try to fold at least x * 0 and y - y.
75 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
77 PR rtl-optimization/60452
78 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
79 <case REG>: Return 1 for invalid offsets from the frame pointer.
81 2014-03-26 Marek Polacek <polacek@redhat.com>
84 * doc/extend.texi (C Extensions): Mention variable-length arrays in
87 2014-03-26 Marek Polacek <polacek@redhat.com>
90 * doc/extend.texi (Designated Inits): Describe what happens to omitted
93 2014-03-26 Marek Polacek <polacek@redhat.com>
96 * ira-color.c (update_conflict_hard_regno_costs): Perform the
97 multiplication in unsigned type.
99 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
101 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
103 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
105 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
107 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
110 * cif-code.def (UNREACHABLE) New code.
111 * ipa-inline.c (inline_small_functions): Skip edges to __builtlin_unreachable.
112 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
113 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
114 predicate to __bulitin_unreachable.
115 (set_cond_stmt_execution_predicate): Fix issue when invert_tree_comparison
117 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
118 propagate to inline clones.
119 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
121 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
122 * cgraphclones.c (cgraph_clone_node): If call destination is already
123 ureachable, do not redirect it back.
124 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
127 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
129 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
130 Do not modify inline clones.
132 2014-03-25 Jakub Jelinek <jakub@redhat.com>
134 * config/i386/i386.md (general_sext_operand): New mode attr.
135 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
136 don't generate (sign_extend (const_int)).
137 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
138 operands[2]. Use We constraint instead of <i> and <general_sext_operand>
139 predicate instead of <general_operand>.
140 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
141 * config/i386/constraints.md (We): New constraint.
142 * config/i386/predicates.md (x86_64_sext_operand,
143 sext_operand): New predicates.
145 2014-03-25 Martin Jambor <mjambor@suse.cz>
148 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
149 inconsistent devirtualizations to __builtin_unreachable.
151 2014-03-25 Marek Polacek <polacek@redhat.com>
154 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
156 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
158 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
159 order of elements for big-endian.
161 2014-03-25 Richard Biener <rguenther@suse.de>
164 * gimplify-me.c (gimple_regimplify_operands): Update the
167 2014-03-25 Martin Jambor <mjambor@suse.cz>
170 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
171 (lto_output_varpool_node): Likewise.
172 (input_overwrite_node): Likewise.
173 (input_varpool_node): Likewise.
175 2014-03-25 Richard Biener <rguenther@suse.de>
177 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like
181 2014-03-25 Jakub Jelinek <jakub@redhat.com>
183 * combine.c (simplify_compare_const): Add MODE argument.
184 Handle mode_width 0 as very large mode_width.
185 (try_combine, simplify_comparison): Adjust callers.
187 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
188 type to avoid signed integer overflow.
189 * explow.c (plus_constant): Likewise.
191 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
193 * doc/generic.texi: Correct typos.
195 2014-03-24 Tobias Burnus <burnus@net-b.de>
197 * doc/invoke.texi (-flto): Expand section about
198 using static libraries with LTO.
200 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
202 PR rtl-optimization/60501
203 * optabs.def (addptr3_optab): New optab.
204 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
205 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
206 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
208 * lra.c (emit_add3_insn): Use the addptr pattern if available.
210 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
212 2014-03-24 Ulrich Drepper <drepper@gmail.com>
214 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
217 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
218 (_mm256_undefined_ps): Define.
219 (_mm256_undefined_pd): Define.
220 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
221 (_mm_undefined_pd): Define.
222 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
223 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
224 (_mm512_undefined_ps): Define.
225 (_mm512_undefined_pd): Define.
226 Use _mm*_undefined_*.
227 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
229 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
231 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
232 (lshr_simd): DI mode added.
233 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
234 (aarch64_ushr_simddi): Likewise.
235 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
236 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
237 (vshrd_n_u64): Likewise.
239 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
241 * Makefile.in (s-macro_list): Depend on cc1.
243 2014-03-23 Teresa Johnson <tejohnson@google.com>
245 * ipa-utils.c (ipa_print_order): Use specified dump file.
247 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
249 PR rtl-optimization/60601
250 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
252 * gcc.c (eval_spec_function): Initialize save_growing_value.
254 2014-03-22 Jakub Jelinek <jakub@redhat.com>
257 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
258 code == MINUS_EXPR, never swap op0 with op1.
260 * toplev.c (init_local_tick): Avoid signed integer multiplication
262 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
263 shift by first operand's bitsize.
265 2014-03-21 Jakub Jelinek <jakub@redhat.com>
268 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
270 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
271 TARGET_ISA_64BIT_P(x).
273 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
275 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
276 pattern for vector nor instead of subtract from splat(-1).
277 (altivec_expand_vec_perm_const_le): Likewise.
279 2014-03-21 Richard Henderson <rth@twiddle.net>
282 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
283 related insns after epilogue_completed.
285 2014-03-21 Martin Jambor <mjambor@suse.cz>
288 * cgraph.h (symtab_node): New flag body_removed.
289 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
290 when removing bodies.
291 * symtab.c (dump_symtab_base): Dump body_removed flag.
292 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
293 had their bodies removed.
295 2014-03-21 Martin Jambor <mjambor@suse.cz>
298 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
301 2014-03-21 Richard Biener <rguenther@suse.de>
303 PR tree-optimization/60577
304 * tree-core.h (struct tree_base): Document nothrow_flag use
305 in VAR_DECL_NONALIASED.
306 * tree.h (VAR_DECL_NONALIASED): New.
307 (may_be_aliased): Adjust.
308 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
310 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
312 * expr.c (expand_expr_real_1): Remove outdated comment.
314 2014-03-20 Jakub Jelinek <jakub@redhat.com>
317 * ira.c (adjust_cleared_regs): Call copy_rtx on
318 *reg_equiv[REGNO (loc)].src_p before passing it to
319 simplify_replace_fn_rtx.
322 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
323 into CONST, put pic register as first operand of PLUS. Use
324 gen_const_mem for both 32-bit and 64-bit PIC got loads.
326 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
328 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
330 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
332 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
333 around for store forwarding issue in the FPU on the UT699.
334 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
335 loads and operations if -mfix-ut699 is specified.
336 (divtf3_hq): Tweak attribute.
337 (sqrttf2_hq): Likewise.
339 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
341 * calls.c (store_one_arg): Remove incorrect const qualification on the
342 type of the temporary.
343 * cfgexpand.c (expand_return): Likewise.
344 * expr.c (expand_constructor): Likewise.
345 (expand_expr_real_1): Likewise.
347 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
349 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
352 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
355 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
357 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
359 * config/arm/aarch-common-protos.h
360 (alu_cost_table): Fix spelling of "extend".
361 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
363 2014-03-19 Richard Biener <rguenther@suse.de>
366 * tree-core.h (tree_type_common): Re-order pointer members
367 to reduce recursion depth during GC walks.
369 2014-03-19 Marek Polacek <polacek@redhat.com>
372 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
375 2014-03-19 Richard Biener <rguenther@suse.de>
378 * lto-streamer-in.c (input_function): In WPA stage do not drop
381 2014-03-19 Jakub Jelinek <jakub@redhat.com>
383 PR tree-optimization/60559
384 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
385 with build_zero_cst assignment.
387 2014-03-18 Kai Tietz <ktietz@redhat.com>
389 PR rtl-optimization/56356
390 * sdbout.c (sdbout_parms): Verify that parms'
391 incoming argument is valid.
392 (sdbout_reg_parms): Likewise.
394 2014-03-18 Richard Henderson <rth@redhat.com>
397 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
398 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
399 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
401 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
403 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
404 * doc/plugins.texi (Plugin callbacks): Mention
406 Italicize plugin event names in description. Explain that
407 PLUGIN_PRAGMAS has no sense for lto1. Explain
409 Remind that no GCC functions should be called after
411 Explain what pragmas with expansion are.
413 2014-03-18 Martin Liska <mliska@suse.cz>
415 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
416 gimple call statement is update.
417 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
418 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
420 2014-03-18 Jakub Jelinek <jakub@redhat.com>
423 * ubsan.c (ubsan_instrument_unreachable): Call
424 initialize_sanitizer_builtins.
425 (ubsan_pass): Likewise.
428 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
429 varpool_finalize_decl instead of rest_of_decl_compilation.
431 2014-03-18 Richard Biener <rguenther@suse.de>
433 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
434 by using bitmap_and_compl instead of bitmap_and_compl_into.
435 (df_rd_transfer_function): Likewise.
437 2014-03-18 Richard Biener <rguenther@suse.de>
439 * doc/lto.texi (fresolution): Fix typo.
441 2014-03-18 Richard Biener <rguenther@suse.de>
443 * doc/invoke.texi (flto): Update for changes in 4.9.
445 2014-03-18 Richard Biener <rguenther@suse.de>
447 * doc/loop.texi: Remove section on the removed lambda framework.
448 Update loop docs with recent changes in preserving loop structure.
450 2014-03-18 Richard Biener <rguenther@suse.de>
452 * doc/lto.texi (-fresolution): Document.
454 2014-03-18 Richard Biener <rguenther@suse.de>
456 * doc/contrib.texi: Adjust my name.
458 2014-03-18 Jakub Jelinek <jakub@redhat.com>
461 * internal-fn.c: Include diagnostic-core.h.
462 (expand_BUILTIN_EXPECT): New function.
463 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
464 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
465 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
466 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
468 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
469 Revert 3 argument __builtin_expect code.
470 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
471 * gimple-fold.c (gimple_fold_call): Likewise.
472 * tree.h (fold_builtin_expect): New prototype.
473 * builtins.c (build_builtin_expect_predicate): Add predictor
474 argument, if non-NULL, create 3 argument __builtin_expect.
475 (fold_builtin_expect): No longer static. Add ARG2 argument,
476 pass it through to build_builtin_expect_predicate.
477 (fold_builtin_2): Adjust caller.
478 (fold_builtin_3): Handle BUILT_IN_EXPECT.
479 * internal-fn.def (BUILTIN_EXPECT): New.
481 2014-03-18 Tobias Burnus <burnus@net-b.de>
484 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
485 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
486 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
488 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
491 * predict.c (combine_predictions_for_bb): Fix up formatting.
492 (expr_expected_value_1, expr_expected_value): Add predictor argument,
493 fill what it points to if non-NULL.
494 (tree_predict_by_opcode): Adjust caller, use the predictor.
495 * predict.def (PRED_COMPARE_AND_SWAP): Add.
497 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
499 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
500 proper constant for the store mode.
502 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
504 * symtab.c (change_decl_assembler_name): Fix transparent alias
507 2014-03-16 Renlin Li <Renlin.Li@arm.com>
509 * config/aarch64/aarch64.c: Correct the comments about the
510 aarch64 stack layout.
512 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
514 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
515 check for GF_OMP_FOR_KIND_FOR.
517 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
519 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
520 ymm and zmm register names.
522 2014-03-17 Jakub Jelinek <jakub@redhat.com>
525 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
526 note creation for the 2010-08-31 changes.
528 2014-03-17 Marek Polacek <polacek@redhat.com>
531 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
532 as -fno-tree-loop-vectorize.
533 (expand_omp_simd): Likewise.
535 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
537 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
538 (eligible_for_call_delay): New prototype.
539 * config/sparc/sparc.c (tls_call_delay): Rename into...
540 (eligible_for_call_delay): ...this. Return false if the instruction
541 cannot be put in the delay slot of a branch.
542 (eligible_for_restore_insn): Simplify.
543 (eligible_for_return_delay): Return false if the instruction cannot be
544 put in the delay slot of a branch and simplify.
545 (eligible_for_sibcall_delay): Return false if the instruction cannot be
546 put in the delay slot of a branch.
547 * config/sparc/sparc.md (fix_ut699): New attribute.
548 (tls_call_delay): Delete.
549 (in_call_delay): Reimplement.
550 (eligible_for_sibcall_delay): Rename into...
551 (in_sibcall_delay): ...this.
552 (eligible_for_return_delay): Rename into...
553 (in_return_delay): ...this.
554 (in_branch_delay): Reimplement.
555 (in_uncond_branch_delay): Delete.
556 (in_annul_branch_delay): Delete.
558 2014-03-14 Richard Henderson <rth@redhat.com>
561 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
562 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
563 (*floathi<X87MODEF>2_i387_with_temp): Remove.
564 (floathi splitters): Remove.
565 (float<SWI48x>xf2): New pattern.
566 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
567 code that tried to handle DImode for 32-bit, but which was excluded
568 by the pattern's condition. Drop allocation of stack temporary.
569 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
570 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
571 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
572 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
573 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
574 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
575 (*float<SWI48><MODEF>2_sse_interunit): Remove.
576 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
577 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
578 (*float<SWI48x><X87MODEF>2_i387): Remove.
579 (all float _with_temp splitters): Remove.
580 (*float<SWI48x><MODEF>2_i387): New pattern.
581 (*float<SWI48><MODEF>2_sse): New pattern.
582 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
583 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
585 2014-03-14 Jakub Jelinek <jakub@redhat.com>
586 Marek Polacek <polacek@redhat.com>
589 * common.opt (dump_base_name_prefixed): New Variable.
590 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
591 if x_dump_base_name_prefixed is already set, set it at the end.
593 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
595 PR rtl-optimization/60508
596 * lra-constraints.c (get_reload_reg): Add new parameter
598 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
599 Pass the new parameter values.
601 2014-03-14 Richard Biener <rguenther@suse.de>
603 * common.opt: Revert unintented changes from r205065.
606 2014-03-14 Richard Biener <rguenther@suse.de>
609 * cfghooks.c (split_block): Properly adjust all loops the
610 block was a latch of.
612 2014-03-14 Martin Jambor <mjambor@suse.cz>
615 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
618 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
621 * config/avr/avr.c (avr_set_current_function): Pass function name
622 through default_strip_name_encoding before sanity checking instead
623 of skipping the first char of the assembler name.
625 2014-03-13 Richard Henderson <rth@redhat.com>
628 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
629 (ix86_force_to_memory, ix86_free_from_memory): Remove.
630 * config/i386/i386-protos.h: Likewise.
631 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
632 in the expander instead of a splitter.
633 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
634 any possibility of requiring a memory.
635 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
636 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
637 (fp branch splitters): Update for ix86_split_fp_branch.
638 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
639 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
640 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
641 (*fop_<MODEF>_2_i387): Remove f/r alternative.
642 (*fop_<MODEF>_3_i387): Likewise.
643 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
644 (splitters for the fop_* register patterns): Remove.
645 (fscalexf4_i387): Rename from *fscalexf4_i387.
646 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
648 2014-03-13 Jakub Jelinek <jakub@redhat.com>
650 PR tree-optimization/59779
651 * tree-dfa.c (get_ref_base_and_extent): Use double_int
652 type for bitsize and maxsize instead of HOST_WIDE_INT.
654 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
656 PR rtl-optimization/57320
657 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
658 the CFG after thread_prologue_and_epilogue_insns.
660 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
662 PR rtl-optimization/57189
663 * lra-constraints.c (process_alt_operands): Disfavor spilling
666 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
668 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
671 2014-03-13 Jakub Jelinek <jakub@redhat.com>
673 PR tree-optimization/59025
675 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
676 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
678 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
681 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
682 calls of avr_out_plus_1.
684 2014-03-13 Bin Cheng <bin.cheng@arm.com>
686 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
687 BB's single pred and update the father loop's latch info later.
689 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
691 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
695 (VEC_base): Likewise.
696 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
697 registers, we need to swap double words in little endian mode.
699 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
700 to be a container mode for 128-bit integer operations added in ISA
701 2.07. Unlike TImode and PTImode, the preferred register set is
702 the Altivec/VMX registers for the 128-bit operations.
704 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
706 (rs6000_split_128bit_ok_p): Likewise.
708 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
709 macros for creating ISA 2.07 normal and overloaded builtin
710 functions with 3 arguments.
711 (BU_P8V_OVERLOAD_3): Likewise.
712 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
713 for use as overloaded functions.
714 (VPERM_1TI_UNS): Likewise.
715 (VSEL_1TI): Likewise.
716 (VSEL_1TI_UNS): Likewise.
717 (ST_INTERNAL_1ti): Likewise.
718 (LD_INTERNAL_1ti): Likewise.
719 (XXSEL_1TI): Likewise.
720 (XXSEL_1TI_UNS): Likewise.
721 (VPERM_1TI): Likewise.
722 (VPERM_1TI_UNS): Likewise.
723 (XXPERMDI_1TI): Likewise.
725 (LXVD2X_V1TI): Likewise.
726 (STXVD2X_V1TI): Likewise.
727 (VEC_INIT_V1TI): Likewise.
728 (VEC_SET_V1TI): Likewise.
729 (VEC_EXT_V1TI): Likewise.
730 (EQV_V1TI): Likewise.
731 (NAND_V1TI): Likewise.
732 (ORC_V1TI): Likewise.
733 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
734 added in ISA 2.07. Add both normal 'altivec' builtins, and the
738 (VADDEUQM): Likewise.
739 (VADDECUQ): Likewise.
740 (VSUBEUQM): Likewise.
741 (VSUBECUQ): Likewise.
743 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
744 __int128_t and __uint128_t types.
745 (__uint128_type): Likewise.
746 (altivec_categorize_keyword): Add support for vector __int128_t,
747 vector __uint128_t, vector __int128, and vector unsigned __int128
748 as a container type for TImode operations that need to be done in
749 VSX/Altivec registers.
750 (rs6000_macro_to_expand): Likewise.
751 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
752 to support 128-bit integer instructions vaddcuq, vadduqm,
753 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
754 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
756 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
757 for V1TImode, and set up preferences to use VSX/Altivec registers.
758 Setup VSX reload handlers.
759 (rs6000_debug_reg_global): Likewise.
760 (rs6000_init_hard_regno_mode_ok): Likewise.
761 (rs6000_preferred_simd_mode): Likewise.
762 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
763 (easy_altivec_constant): Likewise.
764 (output_vec_const_move): Likewise.
765 (rs6000_expand_vector_set): Convert V1TImode set and extract to
767 (rs6000_expand_vector_extract): Likewise.
768 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
770 (rs6000_const_vec): Add support for V1TImode.
771 (rs6000_emit_le_vsx_load): Swap double words when loading or
772 storing TImode/V1TImode.
773 (rs6000_emit_le_vsx_store): Likewise.
774 (rs6000_emit_le_vsx_move): Likewise.
775 (rs6000_emit_move): Add support for V1TImode.
776 (altivec_expand_ld_builtin): Likewise.
777 (altivec_expand_st_builtin): Likewise.
778 (altivec_expand_vec_init_builtin): Likewise.
779 (altivec_expand_builtin): Likewise.
780 (rs6000_init_builtins): Add support for V1TImode type. Add
781 support for ISA 2.07 128-bit integer builtins. Define type names
782 for the VSX/Altivec vector types.
783 (altivec_init_builtins): Add support for overloaded vector
784 functions with V1TImode type.
785 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
786 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
788 (rs6000_split_128bit_ok_p): Likewise.
789 (rs6000_handle_altivec_attribute): Create V1TImode from vector
790 __int128_t and vector __uint128_t.
792 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
800 (VS_scalar): Likewise.
801 (VS_double): Likewise.
802 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
804 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
805 we support the ISA 2.07 128-bit integer arithmetic instructions.
806 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
807 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
808 and TImode types for use with the builtin functions.
809 (V1TI_type_node): Likewise.
810 (unsigned_V1TI_type_node): Likewise.
811 (intTI_type_internal_node): Likewise.
812 (uintTI_type_internal_node): Likewise.
814 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
815 128-bit builtin functions.
816 (UNSPEC_VADDEUQM): Likewise.
817 (UNSPEC_VADDECUQ): Likewise.
818 (UNSPEC_VSUBCUQ): Likewise.
819 (UNSPEC_VSUBEUQM): Likewise.
820 (UNSPEC_VSUBECUQ): Likewise.
821 (VM): Add V1TImode to vector mode iterators.
824 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
825 (altivec_vaddcuq): Likewise.
826 (altivec_vsubuqm): Likewise.
827 (altivec_vsubcuq): Likewise.
828 (altivec_vaddeuqm): Likewise.
829 (altivec_vaddecuq): Likewise.
830 (altivec_vsubeuqm): Likewise.
831 (altivec_vsubecuq): Likewise.
833 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
835 (BOOL_128): Likewise.
836 (BOOL_REGS_OUTPUT): Likewise.
837 (BOOL_REGS_OP1): Likewise.
838 (BOOL_REGS_OP2): Likewise.
839 (BOOL_REGS_UNARY): Likewise.
840 (BOOL_REGS_AND_CR0): Likewise.
842 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
843 128-bit integer builtin support.
844 (vec_vadduqm): Likewise.
845 (vec_vaddecuq): Likewise.
846 (vec_vaddeuqm): Likewise.
847 (vec_vsubecuq): Likewise.
848 (vec_vsubeuqm): Likewise.
849 (vec_vsubcuq): Likewise.
850 (vec_vsubuqm): Likewise.
852 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
853 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
854 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
855 128-bit integer add/subtract to ISA 2.07.
857 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
859 * config/arc/arc.c (arc_predicate_delay_insns):
860 Fix third argument passed to conditionalize_nonjump.
862 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
864 * config/aarch64/aarch64-builtins.c
865 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
866 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
867 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
868 instead of __builtin_lfloor.
869 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
871 2014-03-12 Jakub Jelinek <jakub@redhat.com>
873 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
874 (tree_ssa_ifcombine_bb_1): New function.
875 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
876 is an empty forwarder block to then_bb or vice versa and then_bb
877 and else_bb are effectively swapped.
879 2014-03-12 Christian Bruel <christian.bruel@st.com>
882 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
883 REG_CFA_DEF_CFA note.
884 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
885 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
887 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
889 PR tree-optimization/60454
890 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
892 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
894 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
895 Do not define target_cpu_default2 to generic.
896 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
897 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
898 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
900 2014-03-12 Jakub Jelinek <jakub@redhat.com>
901 Marc Glisse <marc.glisse@inria.fr>
903 PR tree-optimization/60502
904 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
905 instead of build_low_bits_mask.
907 2014-03-12 Jakub Jelinek <jakub@redhat.com>
910 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
911 if there are multiple uses, but op doesn't live on E edge.
912 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
913 clobber stmts before __builtin_unreachable.
915 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
917 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
918 hard_frame_pointer_rtx.
919 * cse.c (cse_insn): Remove volatile check.
920 * cselib.c (cselib_process_insn): Likewise.
921 * dse.c (scan_insn): Likewise.
923 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
925 * config/arc/arc.c (conditionalize_nonjump): New function,
927 (arc_ifcvt): ... this.
928 (arc_predicate_delay_insns): Use it.
930 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
932 * config/arc/predicates.md (extend_operand): During/after reload,
933 allow const_int_operand.
934 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
935 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
936 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
938 (umulsi3_highpart_i): Likewise.
940 2014-03-11 Richard Biener <rguenther@suse.de>
942 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
943 Add asserts to guard possible wrong-code bugs.
945 2014-03-11 Richard Biener <rguenther@suse.de>
947 PR tree-optimization/60429
948 PR tree-optimization/60485
949 * tree-ssa-structalias.c (set_union_with_increment): Properly
950 take into account all fields that overlap the shifted vars.
951 (do_sd_constraint): Likewise.
952 (do_ds_constraint): Likewise.
953 (get_constraint_for_ptr_offset): Likewise.
955 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
957 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
958 (nios2_compute_frame_layout):
959 Add calculation of cfun->machine->fp_save_offset.
960 (nios2_expand_prologue): Correct setting of frame pointer register
962 (nios2_expand_epilogue): Update recovery of stack pointer from
963 frame pointer accordingly.
964 (nios2_initial_elimination_offset): Update calculation of offset
965 for eliminating to HARD_FRAME_POINTER_REGNUM.
967 2014-03-10 Jakub Jelinek <jakub@redhat.com>
970 * ipa.c (symtab_remove_unreachable_nodes): Don't call
971 cgraph_get_create_node on VAR_DECLs.
973 2014-03-10 Richard Biener <rguenther@suse.de>
976 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
978 2014-03-08 Douglas B Rupp <rupp@gnat.com>
980 * config/vms/vms.opt (vms_float_format): New variable.
982 2014-03-08 Tobias Burnus <burnus@net-b.de>
984 * doc/invoke.texi (-fcilkplus): Update implementation status.
986 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
987 Richard Biener <rguenther@suse.de>
989 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
990 consistently accross all TUs.
991 (run_gcc): Enable -fshort-double automatically at link at link-time
992 and disallow override.
994 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
997 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
998 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
999 if they can't be used.
1001 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1003 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1004 for Solaris 11/x86 ld.
1005 * configure: Regenerate.
1007 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1009 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1010 (LIB_TLS_SPEC): Save as ld_tls_libs.
1011 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1012 (HAVE_AS_IX86_TLSLDM): New test.
1013 * configure, config.in: Regenerate.
1014 * config/i386/i386.c (legitimize_tls_address): Fall back to
1015 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1016 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1017 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1018 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1020 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1022 * common.opt (fira-loop-pressure): Mark as optimization.
1024 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1026 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1027 an OpenMP mappable type.
1029 2014-03-06 Matthias Klose <doko@ubuntu.com>
1031 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1032 MULTILIB_OSDIRNAMES is not defined.
1034 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1035 Meador Inge <meadori@codesourcery.com>
1038 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1039 (arm_legitimize_address): Call legitimize_tls_address for any
1040 arm_tls_referenced_p expression, handle constant addend. Call it
1041 before testing for !TARGET_ARM.
1042 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1044 2014-03-06 Richard Biener <rguenther@suse.de>
1050 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1052 * tree-streamer.c (record_common_node): Assert we don't record
1053 nodes with type double.
1054 (preload_common_node): Skip type double, complex double and double
1055 pointer since it is now frontend dependent due to fshort-double option.
1057 2014-03-06 Richard Biener <rguenther@suse.de>
1059 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1060 or -fno-lto is specified and the linker has full plugin support.
1061 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1062 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1063 * lto-wrapper.c (merge_and_complain): Merge compile-time
1064 optimization levels.
1065 (run_gcc): And pass it through to the link options.
1067 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1071 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1073 * cselib.c (remove_useless_values): Skip to avoid quadratic
1074 behavior if the condition moved from...
1075 (cselib_process_insn): ... here holds.
1077 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1080 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1081 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1084 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1085 (TM_H): Add x86-tune.def.
1087 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1089 * config/aarch64/aarch64.c (generic_tunings):
1090 Use cortexa57_extra_costs.
1092 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1095 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1096 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1097 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1100 2014-03-04 Heiher <r@hev.cc>
1102 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1103 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1105 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1107 * config/i386/predicates.md (const2356_operand): Change to ...
1108 (const2367_operand): ... this.
1109 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1111 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1112 (*avx512pf_scatterpf<mode>sf): Ditto.
1113 (avx512pf_scatterpf<mode>df): Ditto.
1114 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1115 (*avx512pf_scatterpf<mode>df): Ditto.
1116 * config/i386/i386.c (ix86_expand_builtin): Update
1117 incorrect hint operand error message.
1119 2014-03-04 Richard Biener <rguenther@suse.de>
1121 * lto-section-in.c (lto_get_section_data): Fix const cast.
1123 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1125 * tree-streamer.c (record_common_node): Assert we don't record
1126 nodes with type double.
1127 (preload_common_node): Skip type double, complex double and double
1128 pointer since it is now frontend dependent due to fshort-double option.
1130 2014-03-04 Richard Biener <rguenther@suse.de>
1133 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1134 (lto_input_toplevel_asms): Likewise.
1135 * lto-section-in.c (lto_get_section_data): Instead do it here
1138 2014-03-04 Richard Biener <rguenther@suse.de>
1140 PR tree-optimization/60382
1141 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1142 dead PHIs a reduction.
1144 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1146 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1148 (_mm_prefetch): Move out of GCC target("sse") pragma.
1149 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1150 GCC target("prfchw") pragma.
1151 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1153 * config/i386/i386.c (ix86_option_override_internal): Enable
1154 -mprfchw with -mprefetchwt1.
1156 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1158 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1161 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1163 * opts.h (CL_PCH_IGNORE): Define.
1164 * targhooks.c (option_affects_pch_p):
1165 Return false for options that have CL_PCH_IGNORE set.
1166 * opt-functions.awk: Process PchIgnore.
1167 * doc/options.texi: Document PchIgnore.
1169 * config/arc/arc.opt (misize): Add PchIgnore property.
1171 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1173 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1174 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1175 constraint on constants to permit them being loaded into
1176 GENERAL_REGS or BASE_REGS.
1178 2014-03-03 Nick Clifton <nickc@redhat.com>
1180 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1181 anti-cacnonical alternatives.
1182 (negandhi3_real): New pattern.
1183 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1185 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1187 * config/avr/avr-mcus.def: Remove atxmega16x1.
1188 * config/avr/avr-tables.opt: Regenerate.
1189 * config/avr/t-multilib: Regenerate.
1190 * doc/avr-mmcu.texi: Regenerate.
1192 2014-03-03 Tobias Grosser <tobias@grosser.es>
1193 Mircea Namolaru <mircea.namolaru@inria.fr>
1195 PR tree-optimization/58028
1196 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1199 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1201 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1202 not handled by recognizers.
1204 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1207 * function.c (expand_function_end): Don't emit
1208 clobber_return_register sequence if clobber_after is a BARRIER.
1209 * cfgexpand.c (construct_exit_block): Append instructions before
1210 return_label to prev_bb.
1212 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1214 * config/rs6000/constraints.md: Document reserved use of "wc".
1216 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1219 * ipa.c (function_and_variable_visibility): When dissolving comdat
1220 group, also set all symbols to local.
1222 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1227 2013-12-14 Jan Hubicka <jh@suse.cz>
1229 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1231 2014-03-02 Jon Beniston <jon@beniston.com>
1237 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1238 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1239 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1240 (simple_return, *simple_return): New patterns
1241 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1242 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1244 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1246 * dwarf2out.c (gen_subprogram_die): Tidy.
1248 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1251 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1252 (*mov_t_msb_neg_negc): ... this new insn.
1254 2014-02-28 Jason Merrill <jason@redhat.com>
1257 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1260 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1263 * dwarf2out.c (decltype_auto_die): New static.
1264 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1265 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1266 (is_cxx_auto): Likewise.
1268 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1270 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1271 we are not using general regs only.
1273 2014-02-28 Richard Biener <rguenther@suse.de>
1276 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1277 previous fix and only allow to remove trivial pre-headers
1278 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1279 (remove_forwarder_block): Properly update the latch of a loop.
1281 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1284 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1285 (cselib_preserved_hash_table): New.
1286 (preserve_constants_and_equivs): Move preserved vals to it.
1287 (cselib_find_slot): Look it up first.
1288 (cselib_init): Initialize it.
1289 (cselib_finish): Release it.
1290 (dump_cselib_table): Dump it.
1292 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1295 * cselib.c (remove_useless_values): Skip to avoid quadratic
1296 behavior if the condition moved from...
1297 (cselib_process_insn): ... here holds.
1299 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1302 * var-tracking.c (vt_initialize): Apply the same condition to
1303 preserve the CFA base value.
1305 2014-02-28 Joey Ye <joey.ye@arm.com>
1308 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1309 if reload in progress or completed.
1311 2014-02-28 Tobias Burnus <burnus@net-b.de>
1314 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1317 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1319 * doc/tm.texi.in (Condition Code Status): Update documention for
1320 relative locations of cc0-setter and cc0-user.
1322 2014-02-27 Jeff Law <law@redhat.com>
1324 PR rtl-optimization/52714
1325 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1326 into two independent simple sets, if I3 is a jump, ensure the
1327 pattern we place into I3 is a (set (pc) ...).
1329 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1330 Jeff Law <law@redhat.com>
1332 PR rtl-optimization/49847
1333 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1334 are in different blocks.
1335 * doc/tm.texi (Condition Code Status): Update documention for
1336 relative locations of cc0-setter and cc0-user.
1338 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1341 * lra.c (lra_emit_add): Check SUBREG too.
1343 2014-02-27 Andreas Schwab <schwab@suse.de>
1345 * config/m68k/m68k.c (m68k_option_override): Disable
1346 -flive-range-shrinkage for classic m68k.
1347 (m68k_override_options_after_change): Likewise.
1349 2014-02-27 Marek Polacek <polacek@redhat.com>
1352 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1353 -Wmaybe-uninitialized.
1355 2014-02-27 Alan Modra <amodra@gmail.com>
1358 * reload1.c (emit_input_reload_insns): When reload_override_in,
1359 set old to rl->in_reg when rl->in_reg is a subreg.
1361 2014-02-26 Richard Biener <rguenther@suse.de>
1364 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1366 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1368 * common/config/i386/predicates.md (const1256_operand): Remove.
1369 (const2356_operand): New.
1370 (const_1_to_2_operand): Remove.
1371 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1372 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1373 (*avx512pf_gatherpf<mode>sf): Ditto.
1374 (avx512pf_gatherpf<mode>df): Ditto.
1375 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1376 (*avx512pf_gatherpf<mode>df): Ditto.
1377 (avx512pf_scatterpf<mode>sf): Ditto.
1378 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1379 (*avx512pf_scatterpf<mode>sf): Ditto.
1380 (avx512pf_scatterpf<mode>df): Ditto.
1381 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1382 (*avx512pf_scatterpf<mode>df): Ditto.
1383 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1385 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1387 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1388 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1389 (_mm512_mask_testn_epi64_mask): Move to ...
1390 * config/i386/avx512cdintrin.h: Here.
1391 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1392 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1393 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1394 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1395 TARGET_AVX512F from TARGET_AVX512CD.
1397 2014-02-26 Richard Biener <rguenther@suse.de>
1400 * ipa.c (walk_polymorphic_call_targets): Properly guard
1401 call to inline_update_overall_summary.
1403 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1406 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1407 and latches only if requested. Fix latch if it is removed.
1408 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1409 LOOPS_HAVE_PREHEADERS.
1411 2014-02-25 Andrew Pinski <apinski@cavium.com>
1413 * builtins.c (expand_builtin_thread_pointer): Create a new target
1414 when the target is NULL.
1416 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1418 PR rtl-optimization/60317
1419 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1420 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1421 * lra-assigns.c: Include params.h.
1422 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1423 other reload pseudos considerations.
1425 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1427 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1428 to use canonical form for nor<mode>3.
1430 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1433 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1436 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1438 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1439 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1440 (ix86_handle_option): Handle OPT_mprefetchwt1.
1441 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1442 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1444 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1445 OPTION_MASK_ISA_PREFETCHWT1.
1446 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1447 (PTA_PREFETCHWT1): New.
1448 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1449 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1450 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1451 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1452 (*prefetch_avx512pf_<mode>_: Change into ...
1453 (*prefetch_prefetchwt1_<mode>: This.
1454 * config/i386/i386.opt (mprefetchwt1): New.
1455 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1456 (_mm_prefetch): Handle intent to write.
1457 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1459 2014-02-25 Richard Biener <rguenther@suse.de>
1462 * emit-rtl.c (mem_attrs_htab): Remove.
1463 (mem_attrs_htab_hash): Likewise.
1464 (mem_attrs_htab_eq): Likewise.
1465 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1466 (init_emit_once): Do not allocate mem_attrs_htab.
1468 2014-02-25 Richard Biener <rguenther@suse.de>
1471 * lto-opts.c (lto_write_options): Output non-explicit conservative
1472 -fwrapv, -fno-trapv and -fno-strict-overflow.
1473 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1474 (run_gcc): And pass them through.
1476 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1478 * sel-sched.c (calculate_new_fences): New parameter ptime.
1479 Calculate it as a maximum over all fence cycles.
1480 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1481 Print the final schedule timing when sched_verbose.
1483 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1485 PR rtl-optimization/60292
1486 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1487 bit fot the fence instruction.
1489 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1491 * calls.h: Fix typo in comment.
1493 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1495 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1496 adjusting offsetable addresses.
1498 2014-02-24 Guozhi Wei <carrot@google.com>
1500 * sparseset.h (sparseset_pop): Fix the wrong index.
1502 2014-02-24 Walter Lee <walt@tilera.com>
1504 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1505 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1507 * common/config/tilegx/tilegx-common.c
1508 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1509 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1511 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1512 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1513 (tilegx_gimplify_va_arg_expr): Handle big endian.
1514 (tilegx_expand_unaligned_load): Ditto.
1515 (tilegx_expand_unaligned_store): Ditto.
1516 (TARGET_RETURN_IN_MSB): New.
1517 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1518 (TARGET_ENDIAN_DEFAULT): New.
1519 (TARGET_BIG_ENDIAN): Handle big endian.
1520 (BYTES_BIG_ENDIAN): Ditto.
1521 (WORDS_BIG_ENDIAN): Ditto.
1522 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1525 * config/tilegx/tilegx.md (extv): Handle big endian.
1527 (insn_st<n>): Ditto.
1528 (insn_st<n>_add<bitsuffix>): Ditto.
1529 (insn_stnt<n>): Ditto.
1530 (insn_stnt<n>_add<bitsuffix>):Ditto.
1531 (vec_interleave_highv8qi): Handle big endian.
1532 (vec_interleave_highv8qi_be): New.
1533 (vec_interleave_highv8qi_le): New.
1534 (insn_v1int_h): Handle big endian.
1535 (vec_interleave_lowv8qi): Handle big endian.
1536 (vec_interleave_lowv8qi_be): New.
1537 (vec_interleave_lowv8qi_le): New.
1538 (insn_v1int_l): Handle big endian.
1539 (vec_interleave_highv4hi): Handle big endian.
1540 (vec_interleave_highv4hi_be): New.
1541 (vec_interleave_highv4hi_le): New.
1542 (insn_v2int_h): Handle big endian.
1543 (vec_interleave_lowv4hi): Handle big endian.
1544 (vec_interleave_lowv4hi_be): New.
1545 (vec_interleave_lowv4hi_le): New.
1546 (insn_v2int_l): Handle big endian.
1547 (vec_interleave_highv2si): Handle big endian.
1548 (vec_interleave_highv2si_be): New.
1549 (vec_interleave_highv2si_le): New.
1550 (insn_v4int_h): Handle big endian.
1551 (vec_interleave_lowv2si): Handle big endian.
1552 (vec_interleave_lowv2si_be): New.
1553 (vec_interleave_lowv2si_le): New.
1554 (insn_v4int_l): Handle big endian.
1555 * config/tilegx/tilegx.opt (mbig-endian): New option.
1556 (mlittle-endian): New option.
1557 * doc/install.texi: Document tilegxbe-linux.
1558 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1560 2014-02-24 Martin Jambor <mjambor@suse.cz>
1563 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1564 there are no parameter descriptors.
1566 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1568 PR rtl-optimization/60268
1569 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1570 initialization to ...
1571 (sched_rgn_init): ... here.
1572 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1574 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1576 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1579 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1581 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1584 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1586 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1587 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1589 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1591 * config/microblaze/predicates.md: Add cmp_op predicate.
1592 * config/microblaze/microblaze.md: Add branch_compare instruction
1593 which uses cmp_op predicate and emits cmp insn before branch.
1594 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1595 to microblaze_expand_conditional_branch and consolidate logic.
1596 (microblaze_expand_conditional_branch): emit branch_compare
1597 insn instead of handling cmp op separate from branch insn.
1599 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1601 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1604 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1606 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1607 define_insn with define_expand and new define_insn
1608 *altivec_lve<VI_char>x_internal.
1609 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1610 and new define_insn *altivec_stve<VI_char>x_internal.
1611 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1613 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1615 (altivec_expand_stvex_be): New function.
1617 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1619 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1620 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1621 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1622 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1624 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1627 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1628 instead of emit_move_insn.
1630 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1632 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1634 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1635 gen_altivec_vsumsws.
1637 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1639 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1640 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1641 (altivec_lvxl_<mode>): New define_expand incorporating
1642 -maltivec=be semantics where needed.
1643 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1644 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1645 semantics where needed.
1646 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1647 (altivec_stvx_<mode>): New define_expand incorporating
1648 -maltivec=be semantics where needed.
1649 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1650 VM2 iterator instead of V4SI.
1651 (altivec_stvxl_<mode>): New define_expand incorporating
1652 -maltivec=be semantics where needed.
1653 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1654 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1655 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1656 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1657 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1658 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1659 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1660 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1661 ALTIVEC_BUILTIN_STVXL.
1662 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1663 (altivec_expand_stvx_be): Likewise.
1664 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1665 (altivec_expand_lvx_be): Likewise.
1666 (altivec_expand_stvx_be): Likewise.
1667 (altivec_expand_builtin): Add cases for
1668 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1669 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1670 (altivec_init_builtins): Add definitions for
1671 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1672 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1674 2014-02-21 Catherine Moore <clm@codesourcery.com>
1676 * doc/invoke.texi (mvirt, mno-virt): Document.
1677 * config/mips/mips.opt (mvirt): New option.
1678 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1680 2014-02-21 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/60276
1683 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1684 (STMT_VINFO_MIN_NEG_DIST): New macro.
1685 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1686 STMT_VINFO_MIN_NEG_DIST.
1687 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1688 made for negative dependence distances still hold.
1690 2014-02-21 Richard Biener <rguenther@suse.de>
1693 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1694 DECL_INITIAL for globals not in the current function context.
1696 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1698 PR tree-optimization/56490
1699 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1700 * tree-ssa-uninit.c: Include params.h.
1701 (compute_control_dep_chain): Add num_calls argument, return false
1702 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1703 num_calls to recursive call.
1704 (find_predicates): Change dep_chain into normal array,
1705 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1706 variable and adjust compute_control_dep_chain caller.
1707 (find_def_preds): Likewise.
1709 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1711 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1712 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1714 2014-02-21 Nick Clifton <nickc@redhat.com>
1716 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1717 (pushhi1): Likewise.
1718 (popqi1): Add mode to pre_dec.
1721 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1723 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1724 mode for mask of V8SFmode permutation.
1726 2014-02-20 Richard Henderson <rth@redhat.com>
1729 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1730 a new pseudo for OLDVAL.
1732 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1735 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1736 gen_reg_rtx if d->testing_p.
1737 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1738 if d->testing_p and we will certainly return true.
1739 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1742 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1744 * emit-rtl.c (gen_reg_rtx): Assert that
1745 crtl->emit.regno_pointer_align_length is non-zero.
1747 2014-02-20 Richard Henderson <rth@redhat.com>
1750 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1751 on failure the store back into EXPECT.
1753 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1754 Sandra Loosemore <sandra@codesourcery.com>
1756 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1757 * config/nios2/nios2.c (nios2_function_profiler): Add
1758 -fPIC (flag_pic == 2) support.
1759 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1760 (nios2_large_offset_p): New function.
1761 (nios2_unspec_reloc_p): Move up position, update to use
1762 nios2_large_offset_p.
1763 (nios2_unspec_address): Remove function.
1764 (nios2_unspec_offset): New function.
1765 (nios2_large_got_address): New function.
1766 (nios2_got_address): Add large offset support.
1767 (nios2_legitimize_tls_address): Update usage of removed and new
1769 (nios2_symbol_binds_local_p): New function.
1770 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1771 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1772 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1773 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1774 processing for (const (unspec ...)).
1775 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1777 2014-02-20 Richard Biener <rguenther@suse.de>
1779 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1780 doing the substitution.
1781 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1783 2014-02-20 Martin Jambor <mjambor@suse.cz>
1786 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1787 info when checking whether lattices are bottom.
1789 2014-02-20 Richard Biener <rguenther@suse.de>
1792 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1795 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1798 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1799 parameter specifying the scaling.
1800 (inline_call): Update.
1801 (want_inline_recursively): Guard division by zero.
1802 (recursive_inlining): Update.
1803 * ipa-inline.h (clone_inlined_nodes): Update.
1805 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1808 * config/i386/i386.c (classify_argument): Pass structures of size
1809 64 bytes or less in register.
1811 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1812 Kirill Yukhin <kirill.yukhin@intel.com>
1814 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1815 (_mm_rcp28_round_ss): Ditto.
1816 (_mm_rsqrt28_round_sd): Ditto.
1817 (_mm_rsqrt28_round_ss): Ditto.
1818 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1819 (_mm_rcp14_round_ss): Ditto.
1820 (_mm_rsqrt14_round_sd): Ditto.
1821 (_mm_rsqrt14_round_ss): Ditto.
1822 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1823 the first input operand, get rid of match_dup.
1824 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1826 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1828 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1829 operand as the first input operand, set type attribute.
1830 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1832 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1833 operand as the first input operand, set type attribute.
1835 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1837 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1840 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1843 * config/i386/i386.c (construct_container): Remove TFmode check
1844 for X86_64_INTEGER_CLASS.
1846 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1849 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1850 only when -Wpsabi is enabled.
1852 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1855 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1856 passing arrays in registers are the same as for structs, so remove the
1857 special case for them.
1859 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1861 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1862 destination type, extract only the valid bits if the source type is not
1863 integral and has a different mode.
1865 2014-02-19 Richard Biener <rguenther@suse.de>
1868 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1871 2014-02-19 Richard Biener <rguenther@suse.de>
1874 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1875 (ipa_modify_call_arguments): Emit an argument load explicitely and
1876 preserve virtual SSA form there and for the replacement call.
1877 Do not update SSA form nor free dominance info.
1879 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1881 * ipa.c (function_and_variable_visibility): Also clear WEAK
1882 flag when disolving COMDAT_GROUP.
1884 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1886 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1887 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1889 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1890 do more sanity checks.
1891 (detect_type_change): Return true when giving up early.
1892 (compute_complex_assign_jump_func): Fix type parameter of
1893 ipa_set_ancestor_jf.
1894 (compute_complex_ancestor_jump_func): Likewise.
1895 (update_jump_functions_after_inlining): Fix updating of
1897 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1899 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1901 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1902 inline clones when edge disappears.
1904 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1907 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1908 Split 64-bit moves into 2 patterns. Do not allow the use of
1909 direct move for TDmode in little endian, since the decimal value
1910 has little endian bytes within a word, but the 64-bit pieces are
1911 ordered in a big endian fashion, and normal subreg's of TDmode are
1913 (mov<mode>_64bit_dm): Likewise.
1914 (movtd_64bit_nodm): Likewise.
1916 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1918 PR tree-optimization/60174
1919 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1920 statement of an SSA_NAME that occurs in an abnormal PHI node.
1922 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1925 * final.c (SEEN_BB): Remove.
1926 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1927 (final_scan_insn): Don't force_source_line on second
1928 NOTE_INSN_BASIC_BLOCK.
1930 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1933 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1934 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1935 (type_natural_mode): Warn ABI change when %zmm register is not
1936 available for AVX512F vector value passing.
1938 2014-02-18 Kai Tietz <ktietz@redhat.com>
1941 * config/i386/i386.c (ix86_expand_prologue): Use value in
1942 rax register as displacement when restoring %r10 or %rax.
1943 Fix wrong offset when restoring both registers.
1945 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1947 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1948 assertion with conditional return.
1950 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1951 Uros Bizjak <ubizjak@gmail.com>
1954 * config/i386/driver-i386.c (host_detect_local_cpu): If
1955 YMM state is not saved by the OS, also clear has_f16c. Move
1956 CPUID 0x80000001 handling before YMM state saving checking.
1958 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1960 PR rtl-optimization/58960
1961 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1962 factored out from ...
1963 (sched_init): ... here.
1964 (free_global_sched_pressure_data): New, factored out from ...
1965 (sched_finish): ... here.
1966 * sched-int.h (free_global_sched_pressure_data): Declare.
1967 * sched-rgn.c (nr_regions_initial): New static global.
1968 (haifa_find_rgns): Initialize it.
1969 (schedule_region): Disable sched-pressure for the newly
1972 2014-02-17 Richard Biener <rguenther@suse.de>
1974 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1975 release SSA defs of pattern stmts.
1977 2014-02-17 Richard Biener <rguenther@suse.de>
1979 * tree-inline.c (expand_call_inline): Release the virtual
1980 operand defined by the call we are about to inline.
1982 2014-02-17 Richard Biener <rguenther@suse.de>
1984 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1986 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1987 Ilya Tocar <ilya.tocar@intel.com>
1989 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1990 arguments order in builtin.
1991 (_mm512_permutexvar_epi64): Ditto.
1992 (_mm512_mask_permutexvar_epi64): Ditto
1993 (_mm512_maskz_permutexvar_epi32): Ditto
1994 (_mm512_permutexvar_epi32): Ditto
1995 (_mm512_mask_permutexvar_epi32): Ditto
1997 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1999 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2000 (p8_vmrgow): Likewise.
2002 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2004 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2007 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2010 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2011 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2012 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2013 using direct move instructions on ISA 2.07. Also adjust
2014 instruction length for 64-bit.
2015 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2016 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2018 2014-02-15 Alan Modra <amodra@gmail.com>
2022 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2023 find_replacement on parts of insn rtl that might be reloaded.
2025 2014-02-15 Richard Biener <rguenther@suse.de>
2027 PR tree-optimization/60183
2028 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2029 (tree_ssa_phiprop): Calculate and free post-dominators.
2031 2014-02-14 Jeff Law <law@redhat.com>
2033 PR rtl-optimization/60131
2034 * ree.c (get_extended_src_reg): New function.
2035 (combine_reaching_defs): Use it rather than assuming location of REG.
2036 (find_and_remove_re): Verify first operand of extension is
2037 a REG before adding the insns to the copy list.
2039 2014-02-14 Roland McGrath <mcgrathr@google.com>
2041 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2042 * configure: Regenerated.
2043 * config.in: Regenerated.
2044 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2045 instead of ASM_SHORT.
2047 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2048 Richard Earnshaw <rearnsha@arm.com>
2050 PR rtl-optimization/59535
2051 * lra-constraints.c (process_alt_operands): Encourage alternative
2052 when unassigned pseudo class is superset of the alternative class.
2053 (inherit_reload_reg): Don't inherit when optimizing for code size.
2054 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2055 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2056 modes not less than 4 for Thumb1.
2058 2014-02-14 Kyle McMartin <kyle@redhat.com>
2061 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2063 2014-02-14 Richard Biener <rguenther@suse.de>
2065 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2066 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2067 Do not strip INDIRECT_REFs.
2069 2014-02-14 Richard Biener <rguenther@suse.de>
2072 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2073 DECL_FUNCTION_SPECIFIC_TARGET.
2074 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2075 * tree-streamer-out.c (pack_ts_target_option): Remove.
2076 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2077 (write_ts_function_decl_tree_pointers): Do not stream
2078 DECL_FUNCTION_SPECIFIC_TARGET.
2079 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2080 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2081 (lto_input_ts_function_decl_tree_pointers): Do not stream
2082 DECL_FUNCTION_SPECIFIC_TARGET.
2084 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2086 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2087 (get_initial_def_for_induction, vectorizable_induction): Ignore
2088 debug stmts when looking for exit_phi.
2089 (vectorizable_live_operation): Fix up condition.
2091 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2093 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2094 nreverse() because it changes the content of original tree list.
2096 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2098 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2099 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2101 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2103 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2104 GNU coding standards.
2106 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2109 * dwarf2out.c (gen_subprogram_die): Don't call
2110 add_calling_convention_attribute if subr_die is old_die.
2112 2014-02-13 Sharad Singhai <singhai@google.com>
2114 * doc/optinfo.texi: Fix order of nodes.
2116 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2118 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2119 operands[2], not operands[3].
2121 2014-02-13 Richard Biener <rguenther@suse.de>
2124 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2125 mention the possibility of an in-tree build.
2126 (CLooG): Update recommended version to 0.18.1, mention the
2127 possibility of an in-tree build and clarify that the ISL
2128 bundled with CLooG does not work.
2130 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2133 * expr.c (compress_float_constant): If x is a hard register,
2134 extend into a pseudo and then move to x.
2136 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2138 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2139 caused by bad second argument to warning_at() with -mhotpatch and
2140 nested functions (e.g. with gfortran).
2142 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2144 * opts.c (option_name): Remove "enabled by default" rider.
2146 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2148 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2150 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2151 Uros Bizjak <ubizjak@gmail.com>
2154 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2155 * configure: Regenerated.
2157 2014-02-12 Richard Biener <rguenther@suse.de>
2159 * vec.c (vec_prefix::calculate_allocation): Move as
2160 inline variant to vec.h.
2161 (vec_prefix::calculate_allocation_1): New out-of-line version.
2162 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2163 (vec_prefix::m_has_auto_buf): Rename to ...
2164 (vec_prefix::m_using_auto_storage): ... this.
2165 (vec_prefix::calculate_allocation): Inline the easy cases
2166 and dispatch to calculate_allocation_1 which doesn't need the
2168 (va_heap::reserve): Use gcc_checking_assert.
2169 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2170 m_using_auto_storage.
2171 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2173 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2174 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2175 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2177 2014-02-12 Richard Biener <rguenther@suse.de>
2179 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2180 when we found a dependence.
2182 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2184 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2186 (maybe_fold_stmt): ... into this new function.
2187 * omp-low.c (lower_omp): Update comment.
2189 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2192 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2195 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2197 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2198 identifiers in comments.
2199 (cortexa53_extra_costs): Likewise.
2200 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2201 (cortexa7_extra_costs): Likewise.
2202 (cortexa12_extra_costs): Likewise.
2203 (cortexa15_extra_costs): Likewise.
2204 (v7m_extra_costs): Likewise.
2206 2014-02-12 Richard Biener <rguenther@suse.de>
2209 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2210 of posix_memalign being successful.
2211 (lower_stmt): Restrict lowering of posix_memalign to when
2212 -ftree-bit-ccp is enabled.
2214 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2216 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2218 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2220 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2222 PR rtl-optimization/60116
2223 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2224 other_insn once the combination has been validated.
2226 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2229 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2231 * ipa-devirt.c: Include demangle.h
2232 (odr_violation_reported): New static variable.
2233 (add_type_duplicate): Update odr_violations.
2234 (maybe_record_node): Add completep parameter; update it.
2235 (record_target_from_binfo): Add COMPLETEP parameter;
2236 update it as needed.
2237 (possible_polymorphic_call_targets_1): Likewise.
2238 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2239 rename FINAL to COMPLETE.
2240 (record_targets_from_bases): Sanity check we found the binfo;
2241 fix COMPLETEP updating.
2242 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2243 parameter, fix computing of COMPLETEP.
2244 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2245 at LTO time do demangling.
2246 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2247 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2249 (gimple_get_virt_method_for_binfo): Likewise.
2250 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2251 gimple_get_virt_method_for_vtable): Update prototypes.
2253 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2256 * genautomata.c (add_presence_absence): Fix typo with
2257 {final_}presence_list.
2259 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2262 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2263 for VSX/Altivec vectors that land in GPR registers.
2265 2014-02-11 Richard Henderson <rth@redhat.com>
2266 Jakub Jelinek <jakub@redhat.com>
2269 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2270 around drhs if type conversion to lacc->type is not useless.
2272 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2274 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2276 (cortex-a57.cortex-a53): Likewise.
2277 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2279 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2281 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2284 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2286 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2287 add_options_for_arm_vfp3.
2289 2014-02-11 Jeff Law <law@redhat.com>
2292 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2293 object with an undesirable mode.
2295 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2298 * config/i386/sol2-9.h: New file.
2299 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2300 *-*-solaris2.9*): Use it.
2302 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2304 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2305 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2307 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2309 * config/microblaze/microblaze.c: Extend mcpu version format
2311 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2313 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2315 2014-02-10 Richard Henderson <rth@redhat.com>
2318 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2319 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2320 ms-abi vs -mno-accumulate-outgoing-args.
2321 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2322 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2325 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2328 * cfgexpand.c (expand_asm_operands): Attach source location to
2329 ASM_INPUT rtx objects.
2330 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2332 2014-02-10 Nick Clifton <nickc@redhat.com>
2334 * config/mn10300/mn10300.c (popcount): New function.
2335 (mn10300_expand_prologue): Include saved registers in stack usage
2338 2014-02-10 Jeff Law <law@redhat.com>
2341 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2342 when changing the SET_DEST of a prior insn to avoid an input reload.
2344 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2346 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2347 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2348 -mcall-openbsd, or -mcall-linux.
2349 (CC1_ENDIAN_BIG_SPEC): Remove.
2350 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2351 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2352 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2353 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2354 and %cc1_endian_default.
2355 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2357 2014-02-10 Richard Biener <rguenther@suse.de>
2359 PR tree-optimization/60115
2360 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2361 MEM_REF handling. Properly verify that the accesses are not
2362 out of the objects bound.
2364 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2366 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2369 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2371 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2372 proper constants and fix formatting.
2373 (possible_polymorphic_call_targets): Fix formatting.
2375 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2376 Ilya Tocar <ilya.tocar@intel.com>
2378 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2379 (_mm512_loadu_epi32): Renamed into...
2380 (_mm512_loadu_si512): This.
2381 (_mm512_storeu_epi32): Renamed into...
2382 (_mm512_storeu_si512): This.
2383 (_mm512_maskz_ceil_ps): Removed.
2384 (_mm512_maskz_ceil_pd): Ditto.
2385 (_mm512_maskz_floor_ps): Ditto.
2386 (_mm512_maskz_floor_pd): Ditto.
2387 (_mm512_floor_round_ps): Ditto.
2388 (_mm512_floor_round_pd): Ditto.
2389 (_mm512_ceil_round_ps): Ditto.
2390 (_mm512_ceil_round_pd): Ditto.
2391 (_mm512_mask_floor_round_ps): Ditto.
2392 (_mm512_mask_floor_round_pd): Ditto.
2393 (_mm512_mask_ceil_round_ps): Ditto.
2394 (_mm512_mask_ceil_round_pd): Ditto.
2395 (_mm512_maskz_floor_round_ps): Ditto.
2396 (_mm512_maskz_floor_round_pd): Ditto.
2397 (_mm512_maskz_ceil_round_ps): Ditto.
2398 (_mm512_maskz_ceil_round_pd): Ditto.
2399 (_mm512_expand_pd): Ditto.
2400 (_mm512_expand_ps): Ditto.
2401 * config/i386/i386.c (ix86_builtins): Remove
2402 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2403 (bdesc_args): Ditto.
2404 * config/i386/predicates.md (const1256_operand): New.
2405 (const_1_to_2_operand): Ditto.
2406 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2407 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2408 (*avx512pf_gatherpf<mode>sf): Ditto.
2409 (avx512pf_gatherpf<mode>df): Ditto.
2410 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2411 (*avx512pf_gatherpf<mode>df): Ditto.
2412 (avx512pf_scatterpf<mode>sf): Ditto.
2413 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2414 (*avx512pf_scatterpf<mode>sf): Ditto.
2415 (avx512pf_scatterpf<mode>df): Ditto.
2416 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2417 (*avx512pf_scatterpf<mode>df): Ditto.
2418 (avx512f_expand<mode>): Removed.
2419 (<shift_insn><mode>3<mask_name>): Change predicate type.
2421 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2423 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2424 not at the end of datarefs vector use ordered_remove to avoid
2425 reordering datarefs vector.
2428 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2429 mark local addressable non-static vars as GOVD_PRIVATE
2430 instead of GOVD_LOCAL.
2431 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2432 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2436 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2437 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2438 assume_aligned or alloc_align attributes.
2439 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2440 arguments. Handle also assume_aligned and alloc_align attributes.
2441 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2442 calls to functions with assume_aligned or alloc_align attributes.
2443 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2445 2014-02-08 Terry Guo <terry.guo@arm.com>
2447 * doc/invoke.texi: Document ARM -march=armv7e-m.
2449 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2451 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2452 flag on __cilkrts_rethrow builtin.
2455 * ipa-cp.c (determine_versionability): Fail at -O0
2456 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2457 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2460 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2463 * tree-inline.c (copy_forbidden): Fail for
2464 __attribute__((optimize (0))) functions.
2466 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2468 * varpool.c: Include pointer-set.h.
2469 (varpool_remove_unreferenced_decls): Variables in other partitions
2470 will not be output; be however careful to not lose information
2473 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2475 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2476 lookup in the vtable constructor.
2478 2014-02-07 Jeff Law <law@redhat.com>
2481 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2482 define_insn_and_split.
2484 * ipa-inline.c (inline_small_functions): Fix typos.
2486 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2488 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2489 (s390_can_use_return_insn): Declare.
2490 * config/s390/s390.h (EPILOGUE_USES): Define.
2491 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2493 (s390_chunkify_start): Handle return JUMP_LABELs.
2494 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2495 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2496 (s390_can_use_return_insn): New functions.
2497 (s390_fix_long_loop_prediction): Handle conditional returns.
2498 (TARGET_SET_UP_BY_PROLOGUE): Define.
2499 * config/s390/s390.md (ANY_RETURN): New code iterator.
2500 (*creturn, *csimple_return, return, simple_return): New patterns.
2502 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2504 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2505 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2506 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2507 REG_CFA_RESTORE list when deciding not to restore a register.
2509 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2511 * config/s390/s390.c: Include tree-pass.h and context.h.
2512 (s390_early_mach): New function, split out from...
2513 (s390_emit_prologue): ...here.
2514 (pass_data_s390_early_mach): New pass structure.
2515 (pass_s390_early_mach): New class.
2516 (s390_option_override): Create and register early_mach pass.
2517 Move to end of file.
2519 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2521 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2522 to match for the exit block.
2524 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2526 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2527 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2528 Reject misaligned operands.
2530 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2532 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2534 2014-02-07 Richard Biener <rguenther@suse.de>
2537 * gimple-low.c (lower_builtin_posix_memalign): New function.
2538 (lower_stmt): Call it to lower posix_memalign in a way
2539 to make alignment info accessible.
2541 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2544 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2545 __builtin_setjmp_receiver.
2547 2014-02-07 Richard Biener <rguenther@suse.de>
2550 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2551 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2552 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2553 Handle BUILT_IN_POSIX_MEMALIGN.
2554 (find_func_clobbers): Likewise.
2555 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2556 (call_may_clobber_ref_p_1): Likewise.
2558 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2561 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2564 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2567 * lto-cgraph.c (lto_output_node): Use
2568 symtab_get_symbol_partitioning_class.
2569 (lto_output_varpool_node): likewise.
2570 (symtab_get_symbol_partitioning_class): Move here from
2572 * cgraph.h (symbol_partitioning_class): Likewise.
2573 (symtab_get_symbol_partitioning_class): Declare.
2575 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2577 * ggc.h (ggc_internal_cleared_alloc): New macro.
2578 * vec.h (vec_safe_copy): Handle memory stats.
2579 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2580 * target-globals.c (save_target_globals): Likewise.
2582 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2585 * expr.c (emit_move_resolve_push): Export; be bit more selective
2586 on when to clear alias set.
2587 * expr.h (emit_move_resolve_push): Declare.
2588 * function.h (struct function): Add tail_call_marked.
2589 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2590 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2591 * config/i386/i386.md (TImode move expander): De not call
2593 (FP push expanders): Preserve memory attributes.
2594 * config/i386/sse.md (push<mode>1): Remove.
2595 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2596 (ix86_expand_push): Remove.
2597 * config/i386/mmx.md (push<mode>1): Remove.
2599 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2601 PR rtl-optimization/60030
2602 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2603 lopart with paradoxical subreg before shifting it up by hprec.
2605 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2607 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2608 Remove extra newline at end of file.
2609 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2610 (arm_issue_rate): Handle cortexa57.
2611 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2612 (cortex-a57.cortex-a53): Likewise.
2614 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2617 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2618 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2620 (arm_expand_prologue): Adjust all callers.
2621 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2622 info, registers also at the lowest numbered registers side. Use
2623 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2627 * var-tracking.c (adjust_mems): Before adding a SET to
2628 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2630 2014-02-06 Alan Modra <amodra@gmail.com>
2633 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2634 change SDmode to DDmode when lra_in_progress.
2636 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2639 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2640 free_data_ref on the dr first, and before goto again also set dr
2641 to the next dr. For simd_lane_access, free old datarefs[i] before
2642 overwriting it. For get_vectype_for_scalar_type failure, don't
2643 free_data_ref if simd_lane_access.
2645 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2648 * tree.h (opts_for_fn): New inline function.
2649 (opt_for_fn): Define.
2650 * config/i386/i386.c (ix86_function_regparm): Use
2651 opt_for_fn (decl, optimize) instead of optimize.
2653 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2655 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2656 for SYMBOL_REF in large memory model.
2658 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2660 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2662 (cortex-a57): Likewise.
2663 (cortex-a57.cortex-a53): Likewise.
2665 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2666 Kugan Vivekanandarajah <kuganv@linaro.org>
2668 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2670 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2672 2014-02-06 Richard Biener <rguenther@suse.de>
2674 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2675 set_loop_copy and initialize_original_copy_tables.
2677 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2679 * config/aarch64/aarch64-simd.md
2680 (aarch64_ashr_simddi): Change QI to SI.
2682 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2683 Jakub Jelinek <jakub@redhat.com>
2686 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2689 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2691 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2692 CODE_FOR_altivec_vpku[hw]um to
2693 CODE_FOR_altivec_vpku[hw]um_direct.
2694 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2695 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2696 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2697 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2699 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2701 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2702 generation for -maltivec=be.
2703 (altivec_vsumsws): Simplify redundant test.
2705 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2707 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2708 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2709 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2710 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2711 gen_altivec_vpkuwum.
2712 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2714 (altivec_vpks<VI_char>ss): Likewise.
2715 (altivec_vpks<VI_char>us): Likewise.
2716 (altivec_vpku<VI_char>us): Likewise.
2717 (altivec_vpku<VI_char>um): Likewise.
2718 (altivec_vpku<VI_char>um_direct): New (copy of
2719 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2721 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2722 target is little endian and -maltivec=be is not specified.
2723 (*altivec_vupkhs<VU_char>_direct): New (copy of
2724 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2725 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2726 target is little endian and -maltivec=be is not specified.
2727 (*altivec_vupkls<VU_char>_direct): New (copy of
2728 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2729 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2730 little endian and -maltivec=be is not specified.
2731 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2732 little endian and -maltivec=be is not specified.
2734 2014-02-05 Richard Henderson <rth@redhat.com>
2737 * combine-stack-adj.c: Revert r206943.
2738 * sched-int.h (struct deps_desc): Add last_args_size.
2739 * sched-deps.c (init_deps): Initialize it.
2740 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2741 contain REG_ARGS_SIZE notes.
2743 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2745 * lto-cgraph.c (asm_nodes_output): Make global.
2746 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2747 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2748 (driver_handle_option): Handle OPT_fwpa.
2750 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2753 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2754 a comment typo and formatting issue. If odr_hash hasn't been
2755 created, return vNULL and set *completep to false.
2758 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2759 bb with no successors.
2761 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2764 * doc/invoke.texi (-march): Clarify documentation for ARM.
2768 2014-02-05 Richard Biener <rguenther@suse.de>
2770 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2771 when not vectorizing because of too many alias checks.
2772 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2773 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2775 2014-02-05 Nick Clifton <nickc@redhat.com>
2777 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2778 accept extended registers in any mode when compiling for the MN10300.
2780 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2782 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2783 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2784 sanitization attributes.
2785 (can_inline_edge_p): Likewise.
2786 (sanitize_attrs_match_for_inline_p): New function.
2788 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2790 * ipa-prop.c (detect_type_change): Shor circuit testing of
2791 type changes on THIS pointer.
2793 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2796 * config/pa/pa.c (legitimize_tls_address): Return original address
2797 if not passed a SYMBOL_REF rtx.
2798 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2800 (pa_emit_move_sequence): Simplify TLS source operands.
2801 (pa_legitimate_constant_p): Reject all TLS constants.
2802 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2803 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2805 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2807 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2808 groups when we know they are controlled by LTO.
2809 * varasm.c (default_binds_local_p_1): If object is in other partition,
2810 it will be resolved locally.
2812 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2814 * config/host-linux.c (linux_gt_pch_use_address): Don't
2815 use SSIZE_MAX because it is not always defined.
2817 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2820 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2821 threshold for pseudo splitting.
2822 (update_ebb_live_info): Process call argument hard registers and
2823 hard registers from insn definition too.
2824 (max_small_class_regs_num): New constant.
2825 (inherit_in_ebb): Update live hard regs through EBBs. Update
2826 reloads_num only for small register classes. Don't split for
2829 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2832 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2835 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2837 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2840 2014-02-04 Marek Polacek <polacek@redhat.com>
2842 * gdbinit.in (pel): Define.
2844 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2846 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2849 2014-02-04 Richard Biener <rguenther@suse.de>
2852 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2853 in function context local.
2854 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2855 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2856 similar to LTO_imported_decl_ref.
2858 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2860 PR tree-optimization/60002
2861 * cgraphclones.c (build_function_decl_skip_args): Clear
2864 PR tree-optimization/60023
2865 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2866 false to gsi_replace.
2867 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2868 has been in some EH region and vec_stmt could throw, add
2869 vec_stmt into the same EH region.
2870 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2871 has no lhs, ignore it.
2872 * internal-fn.c (expand_MASK_LOAD): Likewise.
2875 * tree-inline.c (copy_forbidden): Fail for
2876 __attribute__((optimize (0))) functions.
2879 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2880 is set, copy one less argument.
2881 (expand_simd_clones): Don't subtract clone_info->inbranch
2882 from simd_clone_struct_alloc argument.
2884 PR rtl-optimization/57915
2885 * recog.c (simplify_while_replacing): If all unary/binary/relational
2886 operation arguments are constant, attempt to simplify those.
2889 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2890 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2892 2014-02-04 Richard Biener <rguenther@suse.de>
2894 PR tree-optimization/60012
2895 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2896 TBAA disambiguation to all DDRs.
2898 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2901 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2902 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2904 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2907 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2909 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2911 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2912 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2914 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2917 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2918 to figure out targets of polymorphic calls with known decl.
2919 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2920 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2921 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2922 (get_polymorphic_call_info): ... here.
2923 (get_polymorphic_call_info_from_invariant): New function.
2925 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2927 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2928 lookup via vtable pointer; check for type consistency
2929 and turn inconsitent facts into UNREACHABLE.
2930 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2931 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2932 type inconsistent querries; return UNREACHABLE instead.
2934 2014-02-03 Richard Henderson <rth@twiddle.net>
2937 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2938 already processed this node.
2939 (normalize_one_pred_1): Pass along mark_set.
2940 (normalize_one_pred): Create and destroy a pointer_set_t.
2941 (normalize_one_pred_chain): Likewise.
2943 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2945 PR gcov-profile/58602
2946 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2948 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2951 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2952 -fno-devirtualize; try to devirtualize by the knowledge of
2953 virtual table pointer given by aggregate propagation.
2954 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2955 (ipa_print_node_jump_functions): Dump also offset that
2956 is relevant for polymorphic calls.
2957 (determine_known_aggregate_parts): Add arg_type parameter; use it
2958 instead of determining the type from pointer type.
2959 (ipa_compute_jump_functions_for_edge): Update call of
2960 determine_known_aggregate_parts.
2961 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2962 (gimple_get_virt_method_for_binfo): ... here; simplify using
2963 vtable_pointer_value_to_vtable.
2964 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2965 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2966 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2967 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2969 (vtable_pointer_value_to_binfo): ... here.
2970 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2972 2014-02-03 Teresa Johnson <tejohnson@google.com>
2974 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2975 redef of outer loop index variable.
2977 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2981 * doc/extend.texi (Function Attributes): Typo.
2983 2014-02-03 Cong Hou <congh@google.com>
2985 PR tree-optimization/60000
2986 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2987 if the vectorized statement is a store. A store statement can only
2988 appear at the end of pattern statements.
2990 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2992 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2993 (ix86_option_override_internal): Default long double to 64-bit for
2994 32-bit Bionic and to 128-bit for 64-bit Bionic.
2996 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2997 TARGET_LONG_DOUBLE_128 is true.
2998 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3000 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3001 (mlong-double-64): Negate -mlong-double-128.
3002 (mlong-double-128): New option.
3004 * config/i386/i386-c.c (ix86_target_macros): Define
3005 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3007 * doc/invoke.texi: Document -mlong-double-128.
3009 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3011 PR rtl-optimization/60024
3012 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3014 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3016 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3018 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3020 PR rtl-optimization/57662
3021 * sel-sched.c (code_motion_path_driver): Do not mark already not
3022 existing blocks in the visiting bitmap.
3024 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3026 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3027 on the insn being emitted.
3029 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3030 Will Deacon <will.deacon@arm.com>
3032 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3034 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3036 * config/arm/arm-tables.opt: Regenerate.
3038 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3040 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3041 for vector types other than V16QImode.
3042 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3043 define_expand, and call altivec_expand_vec_perm_le when producing
3044 code with little endian element order.
3045 (*altivec_vperm_<mode>_internal): New insn having previous
3046 behavior of altivec_vperm_<mode>.
3047 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3048 altivec_expand_vec_perm_le when producing code with little endian
3050 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3051 behavior of altivec_vperm_<mode>_uns.
3053 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3055 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3056 (altivec_vsumsws): Add handling for -maltivec=be with a little
3058 (altivec_vsumsws_direct): New.
3059 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3060 gen_altivec_vsumsws.
3062 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3064 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3065 vtable_pointer_value_to_binfo): New functions.
3066 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3067 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3069 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3071 * config/nios2/nios2.md (load_got_register): Initialize GOT
3072 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3073 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3075 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3077 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3078 preserverd by passthrough, do not propagate the type.
3080 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3082 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3083 (mips_atomic_assign_expand_fenv): New function.
3084 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3086 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3088 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3089 (__builtin_mips_set_fcsr): Likewise.
3090 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3091 MIPS_USI_FTYPE_VOID.
3092 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3093 (mips16_expand_set_fcsr): Likewise.
3094 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3095 (mips16_set_fcsr_stub): Likewise.
3096 (mips16_get_fcsr_one_only_stub): New class.
3097 (mips16_set_fcsr_one_only_stub): Likewise.
3098 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3099 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3100 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3101 (hard_float): New availability predicate.
3102 (mips_builtins): Add get_fcsr and set_fcsr.
3103 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3104 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3105 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3106 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3107 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3110 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3112 * config/mips/mips.c (mips_one_only_stub): New class.
3113 (mips_need_mips16_rdhwr_p): Replace with...
3114 (mips16_rdhwr_stub): ...this new variable.
3115 (mips16_stub_call_address): New function.
3116 (mips16_rdhwr_one_only_stub): New class.
3117 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3118 (mips_output_mips16_rdhwr): Delete.
3119 (mips_finish_stub): New function.
3120 (mips_code_end): Use it to handle rdhwr stubs.
3122 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3125 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3126 when calculating size of integer atomic types.
3128 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3130 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3132 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3134 PR tree-optimization/60003
3135 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3136 * profile.c (branch_prob): Use gimple_call_builtin_p
3137 to check for BUILT_IN_SETJMP_RECEIVER.
3138 * tree-inline.c (copy_bb): Call notice_special_calls.
3140 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3143 * lra-constraints.c (process_alt_operands): Update reload_sum only
3146 2014-01-31 Richard Henderson <rth@redhat.com>
3149 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3150 until after else_eh is processed.
3152 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3154 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3155 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3156 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3157 in smmintrin.h, remove them.
3158 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3159 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3160 * config/i386/i386.md (ROUND_SAE): Fix value.
3161 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3162 (const48_operand): New.
3163 * config/i386/subst.md (round), (round_expand): Use
3164 const_4_or_8_to_11_operand.
3165 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3167 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3169 * config/i386/constraints.md (Yk): Swap meaning with k.
3170 * config/i386/i386.md (movhi_internal): Change Yk to k.
3171 (movqi_internal): Ditto.
3172 (*k<logic><mode>): Ditto.
3175 (kandn<mode>): Ditto.
3176 (*<code>hi_1): Ditto.
3177 (*<code>qi_1): Ditto.
3178 (kxnor<mode>): Ditto.
3179 (kortestzhi): Ditto.
3180 (kortestchi): Ditto.
3182 (*one_cmplhi2_1): Ditto.
3183 (*one_cmplqi2_1): Ditto.
3184 * config/i386/sse.md (): Change k to Yk.
3185 (avx512f_load<mode>_mask): Ditto.
3186 (avx512f_blendm<mode>): Ditto.
3187 (avx512f_store<mode>_mask): Ditto.
3188 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3189 (avx512f_storedqu<mode>_mask): Ditto.
3190 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3192 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3193 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3194 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3195 (avx512f_maskcmp<mode>3): Ditto.
3196 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3197 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3198 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3199 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3200 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3201 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3202 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3203 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3204 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3205 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3206 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3207 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3208 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3209 (vec_extract_lo_<mode>_maskm): Ditto.
3210 (vec_extract_hi_<mode>_maskm): Ditto.
3211 (avx512f_vternlog<mode>_mask): Ditto.
3212 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3213 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3214 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3215 (avx512f_<code>v8div16qi2_mask): Ditto.
3216 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3217 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3218 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3219 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3220 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3221 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3222 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3223 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3224 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3225 (avx512cd_maskb_vec_dupv8di): Ditto.
3226 (avx512cd_maskw_vec_dupv16si): Ditto.
3227 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3228 (avx512f_vpermi2var<mode>3_mask): Ditto.
3229 (avx512f_vpermi2var<mode>3_mask): Ditto.
3230 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3231 (*avx512f_gathersi<mode>): Ditto.
3232 (*avx512f_gathersi<mode>_2): Ditto.
3233 (*avx512f_gatherdi<mode>): Ditto.
3234 (*avx512f_gatherdi<mode>_2): Ditto.
3235 (*avx512f_scattersi<mode>): Ditto.
3236 (*avx512f_scatterdi<mode>): Ditto.
3237 (avx512f_compress<mode>_mask): Ditto.
3238 (avx512f_compressstore<mode>_mask): Ditto.
3239 (avx512f_expand<mode>_mask): Ditto.
3240 * config/i386/subst.md (mask): Change k to Yk.
3241 (mask_scalar_merge): Ditto.
3244 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3246 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3248 2014-01-31 Richard Biener <rguenther@suse.de>
3251 * builtins.c (fold_builtin_memory_op): Make sure to not
3252 use a floating-point mode or a boolean or enumeral type for
3255 2014-01-30 DJ Delorie <dj@redhat.com>
3257 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3258 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3259 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3260 whenever main() has an epilogue.
3262 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3264 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3265 unused variable "field".
3266 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3267 (vsx_mergeh_<mode>): Likewise.
3268 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3269 (altivec_vmrghh): Likewise.
3270 (altivec_vmrghw): Likewise.
3271 (altivec_vmrglb): Likewise.
3272 (altivec_vmrglh): Likewise.
3273 (altivec_vmrglw): Likewise.
3274 (altivec_vspltb): Add missing uses.
3275 (altivec_vsplth): Likewise.
3276 (altivec_vspltw): Likewise.
3277 (altivec_vspltsf): Likewise.
3279 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3282 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3283 frame related instructions.
3285 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3287 PR rtl-optimization/59959
3288 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3289 any reload of register whose subreg is invalid.
3291 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3293 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3294 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3295 Add missing return type - void.
3297 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3299 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3300 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3301 remove element index adjustment for endian (now handled in vsx.md
3303 (altivec_expand_vec_perm_const): Use
3304 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3305 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3306 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3307 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3308 define_expand and a new define_insn *altivec_vspltb_internal;
3309 adjust for -maltivec=be on a little endian target.
3310 (altivec_vspltb_direct): New.
3311 (altivec_vsplth): Divide into a define_expand and a new
3312 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3313 little endian target.
3314 (altivec_vsplth_direct): New.
3315 (altivec_vspltw): Divide into a define_expand and a new
3316 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3317 little endian target.
3318 (altivec_vspltw_direct): New.
3319 (altivec_vspltsf): Divide into a define_expand and a new
3320 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3321 a little endian target.
3323 2014-01-30 Richard Biener <rguenther@suse.de>
3325 PR tree-optimization/59993
3326 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3327 can propagate form the earlier stmt and avoid the transform
3328 when the intermediate result is needed.
3330 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3332 * README.Portability: Fix typo.
3334 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3336 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3337 comparison_operator with ordered_comparison_operator.
3339 2014-01-30 Nick Clifton <nickc@redhat.com>
3341 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3342 Rename to mn10300_store_multiple_regs.
3343 * config/mn10300/mn10300.c: Likewise.
3344 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3345 store_multiple_regs.
3346 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3347 Call mn10300_store_multiple_regs.
3349 2014-01-30 Nick Clifton <nickc@redhat.com>
3350 DJ Delorie <dj@redhat.com>
3352 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3353 %fp 2 to keep registers after it properly word-aligned.
3354 (rl78_alloc_physical_registers_umul): Handle the case where both
3355 input operands are the same.
3357 2014-01-30 Richard Biener <rguenther@suse.de>
3359 PR tree-optimization/59903
3360 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3363 2014-01-30 Jason Merrill <jason@redhat.com>
3366 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3369 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3371 2014-01-30 Richard Biener <rguenther@suse.de>
3373 PR tree-optimization/59951
3374 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3376 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3379 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3380 SFmode to DFmode case.
3382 2014-01-29 DJ Delorie <dj@redhat.com>
3384 * config/msp430/msp430.opt (-minrt): New.
3385 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3387 (ENDFILE_SPEC): Likewise.
3389 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3391 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3392 (estimate_function_body_sizes): Use it.
3394 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3397 * dwarf2out.c (is_cxx_auto): New.
3398 (is_base_type): Use it.
3399 (gen_type_die_with_usage): Likewise.
3401 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3403 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3404 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3405 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3406 -maltivec=be with LE targets.
3407 (vsx_mergeh_<mode>): Likewise.
3408 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3409 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3410 (altivec_vmrghb): Replace with define_expand and new
3411 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3412 (altivec_vmrghb_direct): New define_insn.
3413 (altivec_vmrghh): Replace with define_expand and new
3414 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3415 (altivec_vmrghh_direct): New define_insn.
3416 (altivec_vmrghw): Replace with define_expand and new
3417 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3418 (altivec_vmrghw_direct): New define_insn.
3419 (*altivec_vmrghsf): Adjust for endianness.
3420 (altivec_vmrglb): Replace with define_expand and new
3421 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3422 (altivec_vmrglb_direct): New define_insn.
3423 (altivec_vmrglh): Replace with define_expand and new
3424 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3425 (altivec_vmrglh_direct): New define_insn.
3426 (altivec_vmrglw): Replace with define_expand and new
3427 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3428 (altivec_vmrglw_direct): New define_insn.
3429 (*altivec_vmrglsf): Adjust for endianness.
3430 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3431 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3432 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3433 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3434 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3435 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3436 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3437 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3439 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3441 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3442 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3445 2014-01-29 Richard Biener <rguenther@suse.de>
3447 PR tree-optimization/58742
3448 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3449 associate_pointerplus_align.
3450 (associate_pointerplus_diff): New function.
3451 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3452 and associate_pointerplus_diff.
3454 2014-01-29 Richard Biener <rguenther@suse.de>
3456 * lto-streamer.h (LTO_major_version): Bump to 3.
3457 (LTO_minor_version): Reset to 0.
3459 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3461 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3462 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3463 (arm_file_start): Generate correct asm header for armv7ve.
3464 * config/arm/bpabi.h: Add multilib support for armv7ve.
3465 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3466 and cortex-a15 to armv7ve.
3467 * config/arm/t-aprofile: Add multilib support for armv7ve.
3468 * doc/invoke.texi: Document -march=armv7ve.
3470 2014-01-29 Richard Biener <rguenther@suse.de>
3472 PR tree-optimization/58742
3473 * tree-ssa-forwprop.c (associate_plusminus): Return true
3474 if we changed sth, defer EH cleanup to ...
3475 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3476 (simplify_mult): New function.
3478 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3481 PR tree-optimization/59920
3482 * tree.c (build_common_builtin_nodes): Remove
3483 __builtin_setjmp_dispatcher initialization.
3484 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3485 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3486 instead of gsi_after_labels + manually skipping debug stmts.
3487 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3488 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3489 * tree-inline.c (copy_edges_for_bb): Remove
3490 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3491 argument. Ignore computed_goto_p stmts. Don't call
3492 make_abnormal_goto_edges. If a call might need abnormal edges
3493 for non-local gotos, see if it already has an edge to
3494 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3495 with true argument, don't do anything then, otherwise add
3496 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3497 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3499 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3500 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3501 (lower_stmt): Don't set data->calls_builtin_setjmp.
3502 (lower_builtin_setjmp): Adjust comment.
3503 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3504 * tree-cfg.c (found_computed_goto): Remove.
3505 (factor_computed_gotos): Remove.
3506 (make_goto_expr_edges): Return bool, true for computed gotos.
3507 Don't call make_abnormal_goto_edges.
3508 (build_gimple_cfg): Don't set found_computed_goto, don't call
3509 factor_computed_gotos.
3510 (computed_goto_p): No longer static.
3511 (make_blocks): Don't set found_computed_goto.
3512 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3513 (make_edges): If make_goto_expr_edges returns true, push bb
3514 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3515 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3516 vector. Record mapping between bbs and OpenMP regions if there
3517 are any, adjust make_gimple_omp_edges caller. Call
3518 handle_abnormal_edges.
3519 (make_abnormal_goto_edges): Remove.
3520 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3521 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3522 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3523 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3524 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3525 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3526 filling *region also set *region_idx to (*region)->entry->index.
3529 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3530 For REGs set ORIGINAL_REGNO.
3532 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3534 * doc/md.texi: Mention that a target shouldn't implement
3535 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3538 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3540 PR tree-optimization/59594
3541 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3542 a copy of the datarefs vector rather than the vector itself.
3544 2014-01-28 Jason Merrill <jason@redhat.com>
3547 * dwarf2out.c (auto_die): New static.
3548 (gen_type_die_with_usage): Handle C++1y 'auto'.
3549 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3552 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3555 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3556 (SPEC_X32): Likewise.
3557 (SPEC_64): Likewise.
3558 * config/i386/i386.c (ix86_option_override_internal): Turn off
3559 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3561 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3562 * config/i386/i386.h (TARGET_16BIT): New macro.
3563 (TARGET_16BIT_P): Likewise.
3564 * config/i386/i386.opt: Add m16.
3565 * doc/invoke.texi: Document -m16.
3567 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3569 PR preprocessor/59935
3570 * input.c (location_get_source_line): Bail out on when line number
3571 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3573 2014-01-28 Richard Biener <rguenther@suse.de>
3575 PR tree-optimization/58742
3576 * tree-ssa-forwprop.c (associate_plusminus): Handle
3577 pointer subtraction of the form (T)(P + A) - (T)P.
3579 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3581 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3584 2014-01-28 Richard Biener <rguenther@suse.de>
3587 2014-01-28 Richard Biener <rguenther@suse.de>
3589 PR rtl-optimization/45364
3590 PR rtl-optimization/59890
3591 * var-tracking.c (local_get_addr_clear_given_value): Handle
3592 already cleared slot.
3593 (val_reset): Handle not allocated local_get_addr_cache.
3594 (vt_find_locations): Use post-order on the inverted CFG.
3596 2014-01-28 Richard Biener <rguenther@suse.de>
3598 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3600 2014-01-28 Richard Biener <rguenther@suse.de>
3602 PR rtl-optimization/45364
3603 PR rtl-optimization/59890
3604 * var-tracking.c (local_get_addr_clear_given_value): Handle
3605 already cleared slot.
3606 (val_reset): Handle not allocated local_get_addr_cache.
3607 (vt_find_locations): Use post-order on the inverted CFG.
3609 2014-01-28 Alan Modra <amodra@gmail.com>
3611 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3612 * configure.ac <recursive call for build != host>: Define
3613 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3614 and LD_FOR_BUILD too.
3615 * configure: Regenerate.
3617 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3619 * config/i386/i386.c (get_builtin_code_for_version): Separate
3620 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3621 Broadwell from Haswell.
3623 2014-01-27 Steve Ellcey <sellcey@mips.com>
3625 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3626 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3627 * config/mips/mips.c (mips_option_override): Change setting
3629 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3630 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3631 Change from Mask to Var.
3633 2014-01-27 Jeff Law <law@redhat.com>
3635 * ipa-inline.c (inline_small_functions): Fix typo.
3637 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3639 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3640 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3641 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3642 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3643 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3644 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3645 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3646 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3647 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3648 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3649 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3650 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3651 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3652 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3653 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3654 (_mm512_storeu_epi64): Ditto.
3655 (_mm512_cmpge_epi32_mask): Ditto.
3656 (_mm512_cmpge_epu32_mask): Ditto.
3657 (_mm512_cmpge_epi64_mask): Ditto.
3658 (_mm512_cmpge_epu64_mask): Ditto.
3659 (_mm512_cmple_epi32_mask): Ditto.
3660 (_mm512_cmple_epu32_mask): Ditto.
3661 (_mm512_cmple_epi64_mask): Ditto.
3662 (_mm512_cmple_epu64_mask): Ditto.
3663 (_mm512_cmplt_epi32_mask): Ditto.
3664 (_mm512_cmplt_epu32_mask): Ditto.
3665 (_mm512_cmplt_epi64_mask): Ditto.
3666 (_mm512_cmplt_epu64_mask): Ditto.
3667 (_mm512_cmpneq_epi32_mask): Ditto.
3668 (_mm512_cmpneq_epu32_mask): Ditto.
3669 (_mm512_cmpneq_epi64_mask): Ditto.
3670 (_mm512_cmpneq_epu64_mask): Ditto.
3671 (_mm512_expand_pd): Ditto.
3672 (_mm512_expand_ps): Ditto.
3673 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3674 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3675 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3676 * config/i386/i386.c (ix86_builtins): Add
3677 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3678 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3679 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3680 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3681 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3682 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3683 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3684 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3685 IX86_BUILTIN_PMOVUSQW512_MEM.
3686 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3687 __builtin_ia32_pmovsqd512mem_mask,
3688 __builtin_ia32_pmovqd512mem_mask,
3689 __builtin_ia32_pmovusqw512mem_mask,
3690 __builtin_ia32_pmovsqw512mem_mask,
3691 __builtin_ia32_pmovqw512mem_mask,
3692 __builtin_ia32_pmovusdw512mem_mask,
3693 __builtin_ia32_pmovsdw512mem_mask,
3694 __builtin_ia32_pmovdw512mem_mask,
3695 __builtin_ia32_pmovqb512mem_mask,
3696 __builtin_ia32_pmovusqb512mem_mask,
3697 __builtin_ia32_pmovsqb512mem_mask,
3698 __builtin_ia32_pmovusdb512mem_mask,
3699 __builtin_ia32_pmovsdb512mem_mask,
3700 __builtin_ia32_pmovdb512mem_mask.
3701 (bdesc_args): Add __builtin_ia32_expanddf512,
3702 __builtin_ia32_expandsf512.
3703 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3704 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3705 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3706 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3707 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3708 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3709 (avx512f_<code>v8div16qi2_mask_store): This.
3710 (avx512f_expand<mode>): New.
3712 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3714 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3716 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3717 (_mm512_prefetch_i32scatter_pd): Ditto.
3718 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3719 (_mm512_prefetch_i64scatter_pd): Ditto.
3720 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3721 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3722 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3723 (_mm512_prefetch_i32scatter_ps): Ditto.
3724 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3725 (_mm512_prefetch_i64scatter_ps): Ditto.
3726 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3727 * config/i386/i386-builtin-types.def: Define
3728 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3729 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3730 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3731 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3732 IX86_BUILTIN_SCATTERPFQPD.
3733 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3734 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3735 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3736 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3737 __builtin_ia32_scatterpfqps.
3738 (ix86_expand_builtin): Expand new built-ins.
3739 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3740 fix memory access data type.
3741 (*avx512pf_gatherpf<mode>_mask): Ditto.
3742 (*avx512pf_gatherpf<mode>): Ditto.
3743 (avx512pf_scatterpf<mode>): Ditto.
3744 (*avx512pf_scatterpf<mode>_mask): Ditto.
3745 (*avx512pf_scatterpf<mode>): Ditto.
3746 (GATHER_SCATTER_SF_MEM_MODE): New.
3747 (avx512pf_gatherpf<mode>df): Ditto.
3748 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3749 (*avx512pf_scatterpf<mode>df): Ditto.
3751 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3754 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3755 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3758 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3760 * common/config/arm/arm-common.c
3761 (arm_rewrite_mcpu): Handle multiple names.
3763 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3765 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3767 * gimple-builder.h (create_gimple_tmp): Delete.
3769 2014-01-27 Christian Bruel <christian.bruel@st.com>
3771 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3774 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3776 * config/pa/pa.md (call): Generate indirect long calls to non-local
3777 functions when outputing 32-bit code.
3778 (call_value): Likewise except for special call to buggy powf function.
3780 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3781 portable runtime and PIC indirect calls.
3782 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3783 and PIC call sequences. Use ldo instead of blr to set return register
3784 in PIC call sequence.
3786 2014-01-25 Walter Lee <walt@tilera.com>
3788 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3789 avoid clobbering a live register.
3791 2014-01-25 Walter Lee <walt@tilera.com>
3793 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3794 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3795 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3796 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3798 2014-01-25 Walter Lee <walt@tilera.com>
3800 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3801 arguments on even registers.
3802 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3804 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3805 (BIGGEST_ALIGNMENT): Ditto.
3806 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3808 2014-01-25 Walter Lee <walt@tilera.com>
3810 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3811 insns before bundling.
3812 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3814 2014-01-25 Walter Lee <walt@tilera.com>
3816 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3817 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3818 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3820 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3822 * config/mips/constraints.md (kl): Delete.
3823 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3824 define expands, using...
3825 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3826 instructions for MIPS16.
3827 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3828 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3830 2014-01-25 Walter Lee <walt@tilera.com>
3832 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3836 2014-01-25 Walter Lee <walt@tilera.com>
3838 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3839 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3841 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3843 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3846 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3848 * print-rtl.c (in_call_function_usage): New var.
3849 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3850 EXPR_LIST mode as mode and not as reg note name.
3853 * cfgloopmanip.c (copy_loop_info): If
3854 loop->warned_aggressive_loop_optimizations, make sure
3855 the flag is set in target loop too.
3857 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3859 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3861 * builtins.def: Likewise.
3862 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3863 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3864 * ira.c (ira_setup_eliminable_regset): Likewise.
3865 * omp-low.c (gate_expand_omp): Likewise.
3866 (execute_lower_omp): Likewise.
3867 (diagnose_sb_0): Likewise.
3868 (gate_diagnose_omp_blocks): Likewise.
3869 (simd_clone_clauses_extract): Likewise.
3872 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3874 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3875 correction for little endian...
3876 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3879 2014-01-24 Jeff Law <law@redhat.com>
3881 PR tree-optimization/59919
3882 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3883 for non-returning calls.
3885 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3887 * common/config/aarch64/aarch64-common.c
3888 (aarch64_rewrite_mcpu): Handle multiple names.
3889 * config/aarch64/aarch64.h
3890 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3892 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3894 * input.c (add_file_to_cache_tab): Handle the case where fopen
3897 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3900 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3901 from push operand if code of push isn't PRE_DEC.
3903 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3906 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3907 -mquad-memory-atomic. Update -mquad-memory documentation to say
3908 it is only used for non-atomic loads/stores.
3910 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3911 -mquad-memory or -mquad-memory-atomic switches.
3913 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3914 -mquad-memory-atomic to ISA 2.07 support.
3916 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3917 to separate support of normal quad word memory operations (ldq, stq)
3918 from the atomic quad word memory operations.
3920 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3921 support to separate non-atomic quad word operations from atomic
3922 quad word operations. Disable non-atomic quad word operations in
3923 little endian mode so that we don't have to swap words after the
3924 load and before the store.
3925 (quad_load_store_p): Add comment about atomic quad word support.
3926 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3927 options printed with -mdebug=reg.
3929 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3930 -mquad-memory-atomic as the test for whether we have quad word
3931 atomic instructions.
3932 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3933 or -mp8-vector are used, allow byte/half-word atomic operations.
3935 * config/rs6000/sync.md (load_lockedti): Insure that the address
3936 is a proper indexed or indirect address for the lqarx instruction.
3937 On little endian systems, swap the hi/lo registers after the lqarx
3939 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3940 insure the address is valid for the lqarx instruction.
3941 (store_conditionalti): Insure that the address is a proper indexed
3942 or indirect address for the stqcrx. instruction. On little endian
3943 systems, swap the hi/lo registers before doing the stqcrx.
3945 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3946 insure the address is valid for the stqcrx. instruction.
3948 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3949 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3950 type of quad memory support is available.
3952 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3955 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3956 there is a danger of looping.
3958 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3960 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3961 force flag_ira_loop_pressure if set via command line.
3963 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3965 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3966 (ashr_simd): New builtin handling DI mode.
3967 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3968 (aarch64_sshr_simddi): New match pattern.
3969 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3970 (vshrd_n_s64): Likewise.
3971 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3973 2014-01-23 Nick Clifton <nickc@redhat.com>
3975 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3976 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3977 favour of mcu specific scripts.
3978 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3981 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3982 Alex Velenko <Alex.Velenko@arm.com>
3984 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3985 (vaddv_s16): Likewise.
3986 (vaddv_s32): Likewise.
3987 (vaddv_u8): Likewise.
3988 (vaddv_u16): Likewise.
3989 (vaddv_u32): Likewise.
3990 (vaddvq_s8): Likewise.
3991 (vaddvq_s16): Likewise.
3992 (vaddvq_s32): Likewise.
3993 (vaddvq_s64): Likewise.
3994 (vaddvq_u8): Likewise.
3995 (vaddvq_u16): Likewise.
3996 (vaddvq_u32): Likewise.
3997 (vaddvq_u64): Likewise.
3998 (vaddv_f32): Likewise.
3999 (vaddvq_f32): Likewise.
4000 (vaddvq_f64): Likewise.
4001 (vmaxv_f32): Likewise.
4002 (vmaxv_s8): Likewise.
4003 (vmaxv_s16): Likewise.
4004 (vmaxv_s32): Likewise.
4005 (vmaxv_u8): Likewise.
4006 (vmaxv_u16): Likewise.
4007 (vmaxv_u32): Likewise.
4008 (vmaxvq_f32): Likewise.
4009 (vmaxvq_f64): Likewise.
4010 (vmaxvq_s8): Likewise.
4011 (vmaxvq_s16): Likewise.
4012 (vmaxvq_s32): Likewise.
4013 (vmaxvq_u8): Likewise.
4014 (vmaxvq_u16): Likewise.
4015 (vmaxvq_u32): Likewise.
4016 (vmaxnmv_f32): Likewise.
4017 (vmaxnmvq_f32): Likewise.
4018 (vmaxnmvq_f64): Likewise.
4019 (vminv_f32): Likewise.
4020 (vminv_s8): Likewise.
4021 (vminv_s16): Likewise.
4022 (vminv_s32): Likewise.
4023 (vminv_u8): Likewise.
4024 (vminv_u16): Likewise.
4025 (vminv_u32): Likewise.
4026 (vminvq_f32): Likewise.
4027 (vminvq_f64): Likewise.
4028 (vminvq_s8): Likewise.
4029 (vminvq_s16): Likewise.
4030 (vminvq_s32): Likewise.
4031 (vminvq_u8): Likewise.
4032 (vminvq_u16): Likewise.
4033 (vminvq_u32): Likewise.
4034 (vminnmv_f32): Likewise.
4035 (vminnmvq_f32): Likewise.
4036 (vminnmvq_f64): Likewise.
4038 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4040 * config/aarch64/aarch64-simd.md
4041 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4042 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4043 (*aarch64_mul3_elt<mode>): Likewise.
4044 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4045 (*aarch64_mul3_elt_to_64v2df): Likewise.
4046 (*aarch64_mla_elt<mode>): Likewise.
4047 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4048 (*aarch64_mls_elt<mode>): Likewise.
4049 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4050 (*aarch64_fma4_elt<mode>): Likewise.
4051 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4052 (*aarch64_fma4_elt_to_64v2df): Likewise.
4053 (*aarch64_fnma4_elt<mode>): Likewise.
4054 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4055 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4056 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4057 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4058 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4059 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4060 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4061 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4062 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4064 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4066 * config/aarch64/aarch64-simd.md
4067 (aarch64_be_checked_get_lane<mode>): New define_expand.
4068 * config/aarch64/aarch64-simd-builtins.def
4069 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4070 New builtin definition.
4071 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4072 Use new safe be builtin.
4074 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4076 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4078 (aarch64_be_st1<mode>): Likewise.
4079 (aarch_ld1<VALL:mode>): Define_expand modified.
4080 (aarch_st1<VALL:mode>): Likewise.
4081 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4082 (UNSPEC_ST1): Likewise.
4084 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4086 * config/microblaze/microblaze.md: Add trap insn and attribute
4088 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4090 PR preprocessor/58580
4091 * input.h (location_get_source_line): Take an additional line_size
4093 (void diagnostics_file_cache_fini): Declare new function.
4094 * input.c (struct fcache): New type.
4095 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4096 New static constants.
4097 (diagnostic_file_cache_init, total_lines_num)
4098 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4099 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4100 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4101 (get_next_line, read_next_line, goto_next_line, read_line_num):
4102 New static function definitions.
4103 (diagnostic_file_cache_fini): New function.
4104 (location_get_source_line): Take an additional output line_len
4105 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4107 * diagnostic.c (diagnostic_finish): Call
4108 diagnostic_file_cache_fini.
4109 (adjust_line): Take an additional input parameter for the length
4110 of the line, rather than calculating it with strlen.
4111 (diagnostic_show_locus): Adjust the use of
4112 location_get_source_line and adjust_line with respect to their new
4113 signature. While displaying a line now, do not stop at the first
4114 null byte. Rather, display the zero byte as a space and keep
4115 going until we reach the size of the line.
4116 * Makefile.in: Add vec.o to OBJS-libcommon
4118 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4119 Ilya Tocar <ilya.tocar@intel.com>
4121 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4122 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4123 (__builtin_ia32_kmov16): Ditto.
4124 * config/i386/i386.md (UNSPEC_KMOV): New.
4127 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4129 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4130 (_mm512_storeu_si512): Ditto.
4132 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4135 * rtl.h (get_referenced_operands): Declare.
4136 * recog.c (get_referenced_operands): New function.
4137 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4138 operands have been referenced when recording LO_SUM references.
4140 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4142 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4144 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4146 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4147 Enable for generic and recent AMD targets.
4149 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4151 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4152 ARG_SIZE note when adjustment was eliminated.
4154 2014-01-22 Jeff Law <law@redhat.com>
4156 PR tree-optimization/59597
4157 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4158 in file. Accept new argument REGISTERING and use it to modify
4159 dump output appropriately.
4160 (register_jump_thread): Corresponding changes.
4161 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4162 thread paths involving joiner blocks. Add code to dump cancelled
4163 jump threading paths.
4165 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4167 PR rtl-optimization/59477
4168 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4169 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4171 2014-01-22 Tom Tromey <tromey@redhat.com>
4173 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4175 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4177 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4179 PR rtl-optimization/59896
4180 * lra-constraints.c (process_alt_operands): Check unused note for
4181 matched operands of insn with no output reloads.
4183 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4185 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4186 (mips_move_from_gpr_cost): Likewise.
4188 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4190 PR rtl-optimization/59858
4191 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4192 ira_class_hard_regs_num.
4193 (process_alt_operands): Increase reject for dying matched operand.
4195 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4198 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4199 smaller than size, perform several stores or loads and stores
4200 at dst + count - size to store or copy all of size bytes, rather
4201 than just last modesize bytes.
4203 2014-01-20 DJ Delorie <dj@redhat.com>
4205 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4206 that CLOBBERs are REGs before propogating their values.
4208 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4211 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4212 (cgraph_inline_failed_type): New function.
4213 * cgraph.h (DEFCIFCODE): Add type.
4214 (cgraph_inline_failed_type_t): New enum.
4215 (cgraph_inline_failed_type): New prototype.
4216 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4217 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4218 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4219 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4220 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4221 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4222 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4223 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4224 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4225 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4226 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4227 OPTIMIZATION_MISMATCH.
4228 * tree-inline.c (expand_call_inline): Emit errors during
4229 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4231 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4234 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4235 mode attribute in insn output.
4237 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4239 * output.h (output_constant): Delete.
4240 * varasm.c (output_constant): Make private.
4242 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4244 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4246 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4249 * tree.h (fold_builtin_strcat): New prototype.
4250 * builtins.c (fold_builtin_strcat): No longer static. Add len
4251 argument, if non-NULL, don't call c_strlen. Optimize
4252 directly into __builtin_memcpy instead of __builtin_strcpy.
4253 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4254 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4256 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4258 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4259 for SImode_address_operand operands, having only a REG argument.
4261 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4263 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4264 loader name using mbig-endian.
4265 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4267 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4269 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4273 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4275 * config/aarch64/aarch64-protos.h
4276 (aarch64_cannot_change_mode_class_ptr): Declare.
4277 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4278 aarch64_cannot_change_mode_class_ptr): New.
4279 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4280 backend hook aarch64_cannot_change_mode_class.
4282 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4284 * common/config/aarch64/aarch64-common.c
4285 (aarch64_handle_option): Don't handle any option order logic here.
4286 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4287 selected_cpu, warn on architecture version mismatch.
4288 (aarch64_override_options): Fix parsing order for option strings.
4290 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4291 Iain Sandoe <iain@codesourcery.com>
4294 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4295 warning. Amend comment to reflect current functionality.
4297 2014-01-20 Richard Biener <rguenther@suse.de>
4300 * builtins.c (fold_builtin_strcat): Remove case better handled
4301 by tree-ssa-strlen.c.
4303 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4305 * config/aarch64/aarch64.opt
4306 (mcpu, march, mtune): Make case-insensitive.
4308 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4311 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4312 if operands[1] is a REG or ZERO_EXTEND of a REG.
4314 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4316 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4318 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4320 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4321 long non-pic millicode calls.
4323 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4325 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4327 2014-01-19 Kito Cheng <kito@0xlab.org>
4329 * builtins.c (expand_movstr): Check movstr expand done or fail.
4331 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4332 H.J. Lu <hongjiu.lu@intel.com>
4335 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4336 to DImode for zero-extended addresses.
4338 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4340 PR rtl-optimization/57763
4341 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4342 on the new indirect jump_insn and increment LABEL_NUSES (label).
4344 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4348 * config.gcc (x86_archs): New variable.
4349 (x86_64_archs): Likewise.
4350 (x86_cpus): Likewise.
4351 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4352 --with-arch/--with-cpu= options.
4353 Support --with-arch=/--with-cpu={nehalem,westmere,
4354 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4356 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4358 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4359 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4361 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4363 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4365 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4368 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4369 clear cpp_get_options (parse_in)->warn_unused_macros for
4370 ix86_target_macros_internal with cpp_define.
4372 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4374 * jump.c (delete_related_insns): Keep (use (insn))s.
4375 * reorg.c (redundant_insn): Check for barriers too.
4377 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4379 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4381 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4383 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4384 call to $$dyncall when TARGET_LONG_CALLS is true.
4386 2014-01-17 Jeff Law <law@redhat.com>
4388 * ree.c (combine_set_extension): Temporarily disable test for
4389 changing number of hard registers.
4391 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4394 * ipa-inline-analysis.c (inline_free_summary):
4395 Do not free summary of aliases.
4397 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4400 * gimplify.c (gimplify_expr): Use create_tmp_var
4401 instead of create_tmp_var_raw. If cond doesn't have
4402 integral type, don't add the IFN_ANNOTATE builtin at all.
4404 2014-01-17 Martin Jambor <mjambor@suse.cz>
4407 * ipa-cp.c (prev_edge_clone): New variable.
4408 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4409 Also resize prev_edge_clone vector.
4410 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4411 (ipcp_edge_removal_hook): New function.
4412 (ipcp_driver): Register ipcp_edge_removal_hook.
4414 2014-01-17 Andrew Pinski <apinski@cavium.com>
4415 Steve Ellcey <sellcey@mips.com>
4418 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4421 2014-01-17 Jeff Law <law@redhat.com>
4424 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4425 so that pass_ccp runs first.
4427 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4429 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4430 (ix86_adjust_cost): Use !TARGET_XXX.
4431 (do_reorder_for_imul): Likewise.
4432 (swap_top_of_ready_list): Likewise.
4433 (ix86_sched_reorder): Likewise.
4435 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4437 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4438 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4439 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4440 (intel_memset): New. Duplicate slm_memset.
4441 (intel_cost): New. Duplicate slm_cost.
4442 (m_INTEL): New macro.
4443 (processor_target_table): Add "intel".
4444 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4445 with PROCESSOR_INTEL for "intel".
4446 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4447 PROCESSOR_SILVERMONT.
4448 (ix86_issue_rate): Likewise.
4449 (ix86_adjust_cost): Likewise.
4450 (ia32_multipass_dfa_lookahead): Likewise.
4451 (swap_top_of_ready_list): Likewise.
4452 (ix86_sched_reorder): Likewise.
4453 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4454 instead of TARGET_OPT_AGU.
4455 * config/i386/i386.h (TARGET_INTEL): New.
4456 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4457 (processor_type): Add PROCESSOR_INTEL.
4458 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4459 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4461 2014-01-17 Marek Polacek <polacek@redhat.com>
4464 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4467 2014-01-17 Richard Biener <rguenther@suse.de>
4469 PR tree-optimization/46590
4470 * opts.c (default_options_table): Add entries for
4471 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4472 all enabled at -O1 but not for -Og.
4473 * common.opt (fbranch-count-reg): Remove Init(1).
4474 (fmove-loop-invariants): Likewise.
4475 (ftree-pta): Likewise.
4477 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4479 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4480 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4481 decls to at least the GCC 4.8 used alignments.
4484 * tree-nested.c (convert_nonlocal_reference_stmt,
4485 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4486 of GIMPLE_BIND stmts, adjust associated decls.
4488 2014-01-17 Richard Biener <rguenther@suse.de>
4490 PR tree-optimization/46590
4491 * vec.h (vec<>::bseach): New member function implementing
4492 binary search according to C89 bsearch.
4493 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4494 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4495 bitmap pointer again. Make accesses_in_loop a flat array.
4496 (mem_ref_obstack): New global.
4497 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4498 (mark_ref_stored): Likewise.
4499 (ref_indep_loop_p_2): Likewise.
4500 (set_ref_stored_in_loop): New helper function.
4501 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4502 (memref_free): Adjust.
4503 (record_mem_ref_loc): Simplify.
4504 (gather_mem_refs_stmt): Adjust.
4505 (sort_locs_in_loop_postorder_cmp): New function.
4506 (analyze_memory_references): Sort accesses_in_loop after
4507 loop postorder number.
4508 (find_ref_loc_in_loop_cmp): New function.
4509 (for_all_locs_in_loop): Find relevant cluster of locs in
4510 accesses_in_loop and iterate without recursion.
4511 (execute_sm): Avoid uninit warning.
4512 (struct ref_always_accessed): Simplify.
4513 (ref_always_accessed::operator ()): Likewise.
4514 (ref_always_accessed_p): Likewise.
4515 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4516 loop postorder numbers here.
4517 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4520 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4523 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4524 on decls for which assemble_alias has been called.
4526 2014-01-17 Nick Clifton <nickc@redhat.com>
4528 * config/msp430/msp430.opt: (mcpu): New option.
4529 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4530 (msp430_option_override): Parse target_cpu. If the MCU name
4531 matches a generic string, clear target_mcu.
4532 (msp430_attr): Allow numeric interrupt values up to 63.
4533 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4534 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4536 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4538 * config/msp430/msp430.md (popm): Use %J rather than %I.
4539 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4540 (addhi_cy_i): Use immediate_operand for operand 2.
4541 * doc/invoke.texi: Document -mcpu option.
4543 2014-01-17 Richard Biener <rguenther@suse.de>
4545 PR rtl-optimization/38518
4546 * df.h (df_analyze_loop): Declare.
4547 * df-core.c: Include cfgloop.h.
4548 (df_analyze_1): Split out main part of df_analyze.
4549 (df_analyze): Adjust.
4550 (loop_inverted_post_order_compute): New function.
4551 (loop_post_order_compute): Likewise.
4552 (df_analyze_loop): New function avoiding whole-function
4554 * loop-invariant.c (find_defs): Use df_analyze_loop.
4555 (find_invariants): Adjust.
4556 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4558 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4560 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4561 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4563 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4565 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4566 traversal when removing references.
4568 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4571 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4573 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4576 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4577 pushing a reload for an autoinc when we had previously reloaded an
4578 inner part of the address.
4580 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4582 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4584 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4585 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4586 when not giving up or versioning for alias only because of
4588 (vect_analyze_data_ref_dependences): Set to true.
4589 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4591 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4592 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4596 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4599 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4600 operand 0 predicate for gathers, use a new pseudo as subtarget.
4602 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4605 * lra-constraints.c (process_alt_operands): Add printing debug info.
4606 Check absence of input/output reloads for matched operands too.
4608 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4610 PR rtl-optimization/59835
4611 * ira.c (ira_init_register_move_cost): Increase cost for
4614 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4616 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4618 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4621 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4622 non-register objects. Use gen_(high/low)part more consistently.
4625 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4628 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4629 endian support, remove tests for WORDS_BIG_ENDIAN.
4630 (p8_mfvsrd_3_<mode>): Likewise.
4631 (reload_gpr_from_vsx<mode>): Likewise.
4632 (reload_gpr_from_vsxsf): Likewise.
4633 (p8_mfvsrd_4_disf): Likewise.
4635 2014-01-16 Richard Biener <rguenther@suse.de>
4637 PR rtl-optimization/46590
4638 * lcm.c (compute_antinout_edge): Use postorder iteration.
4639 (compute_laterin): Use inverted postorder iteration.
4641 2014-01-16 Nick Clifton <nickc@redhat.com>
4644 * varasm.c (output_constant): Return the number of bytes actually
4646 (output_constructor_array_range): Update the field size with the
4647 number of bytes emitted by output_constant.
4648 (output_constructor_regular_field): Likewise. Also do not
4649 complain if the total number of bytes emitted is now greater
4650 than the expected fieldpos.
4651 * output.h (output_constant): Update prototype and descriptive comment.
4653 2014-01-16 Marek Polacek <polacek@redhat.com>
4656 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4657 it is error_mark_node.
4659 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4661 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4662 VALID_AVX256_REG_OR_OI_MODE.
4664 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4666 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4667 current procedure should be profiled.
4669 2014-01-15 Andrew Pinski <apinski@cavium.com>
4671 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4672 of moving from/to the STACK_REG register class.
4674 2014-01-15 Richard Henderson <rth@redhat.com>
4677 * reginfo.c (global_regs_decl): Globalize.
4678 * rtl.h (global_regs_decl): Declare.
4679 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4680 reserved via global_regs.
4682 2014-01-15 Teresa Johnson <tejohnson@google.com>
4684 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4686 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4688 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4689 and vmulosh rather than call gen_vec_widen_smult_*.
4690 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4691 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4692 (vec_widen_smult_even_v16qi): Likewise.
4693 (vec_widen_umult_even_v8hi): Likewise.
4694 (vec_widen_smult_even_v8hi): Likewise.
4695 (vec_widen_umult_odd_v16qi): Likewise.
4696 (vec_widen_smult_odd_v16qi): Likewise.
4697 (vec_widen_umult_odd_v8hi): Likewise.
4698 (vec_widen_smult_odd_v8hi): Likewise.
4699 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4700 vmuloub rather than call gen_vec_widen_umult_*.
4701 (vec_widen_umult_lo_v16qi): Likewise.
4702 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4703 vmulosb rather than call gen_vec_widen_smult_*.
4704 (vec_widen_smult_lo_v16qi): Likewise.
4705 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4706 rather than call gen_vec_widen_umult_*.
4707 (vec_widen_umult_lo_v8hi): Likewise.
4708 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4709 rather than call gen_vec_widen_smult_*.
4710 (vec_widen_smult_lo_v8hi): Likewise.
4712 2014-01-15 Jeff Law <law@redhat.com>
4714 PR tree-optimization/59747
4715 * ree.c (find_and_remove_re): Properly handle case where a second
4716 eliminated extension requires widening a copy created for elimination
4717 of a prior extension.
4718 (combine_set_extension): Ensure that the number of hard regs needed
4719 for a destination register does not change when we widen it.
4721 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4723 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4724 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4725 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4726 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4727 (avr-*-rtems*): Likewise.
4728 (bfin*-rtems*): Likewise.
4729 (moxie-*-rtems*): Likewise.
4730 (h8300-*-rtems*): Likewise.
4731 (i[34567]86-*-rtems*): Likewise.
4732 (lm32-*-rtems*): Likewise.
4733 (m32r-*-rtems*): Likewise.
4734 (m68k-*-rtems*): Likewise.
4735 (microblaze*-*-rtems*): Likewise.
4736 (mips*-*-rtems*): Likewise.
4737 (powerpc-*-rtems*): Likewise.
4738 (sh-*-rtems*): Likewise.
4739 (sparc-*-rtems*): Likewise.
4740 (sparc64-*-rtems*): Likewise.
4741 (v850-*-rtems*): Likewise.
4742 (m32c-*-rtems*): Likewise.
4744 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4746 PR rtl-optimization/59511
4747 * ira.c (ira_init_register_move_cost): Use memory costs for some
4748 cases of register move cost calculations.
4749 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4750 instead of BB frequency.
4751 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4752 * lra-assigns.c (find_hard_regno_for): Ditto.
4754 2014-01-15 Richard Biener <rguenther@suse.de>
4756 PR tree-optimization/59822
4757 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4758 (vectorizable_load): Use it to hoist defs of uses of invariant
4759 loads out of the loop.
4761 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4762 Kugan Vivekanandarajah <kuganv@linaro.org>
4765 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4768 2014-01-15 Richard Biener <rguenther@suse.de>
4770 PR rtl-optimization/59802
4771 * lcm.c (compute_available): Use inverted postorder to seed
4772 the initial worklist.
4774 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4777 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4778 ADDR_REGS for invalid symrefs in non-PIC code.
4780 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4783 * builtins.c (determine_block_size): Initialize *probable_max_size
4784 even if len_rtx is CONST_INT.
4786 2014-01-14 Andrew Pinski <apinski@cavium.com>
4788 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4789 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4790 (cortexa53_tunings): Likewise.
4791 (aarch64_sched_issue_rate): New function.
4792 (TARGET_SCHED_ISSUE_RATE): Define.
4794 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4796 * ira-costs.c (find_costs_and_classes): Add missed
4797 ira_init_register_move_cost_if_necessary.
4799 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4802 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4804 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4807 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4808 to indicate if type is used for function return value. Warn ABI
4809 change if the vector mode isn't available for function return value.
4810 (ix86_function_arg_advance): Pass false to type_natural_mode.
4811 (ix86_function_arg): Likewise.
4812 (ix86_gimplify_va_arg): Likewise.
4813 (function_arg_32): Don't warn ABI change.
4814 (ix86_function_value): Pass true to type_natural_mode.
4815 (ix86_return_in_memory): Likewise.
4816 (ix86_struct_value_rtx): Removed.
4817 (TARGET_STRUCT_VALUE_RTX): Likewise.
4819 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4821 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4822 converting a conditional jump into a conditional return.
4824 2014-01-14 Richard Biener <rguenther@suse.de>
4826 PR tree-optimization/58921
4827 PR tree-optimization/59006
4828 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4829 hoisting invariant stmts.
4830 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4831 invariant loads on the preheader edge if possible.
4833 2014-01-14 Joey Ye <joey.ye@arm.com>
4835 * doc/plugin.texi (Building GCC plugins): Update to C++.
4837 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4839 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4840 (_mm_rcp28_round_ss): Ditto.
4841 (_mm_rsqrt28_round_sd): Ditto.
4842 (_mm_rsqrt28_round_ss): Ditto.
4843 (_mm_rcp28_sd): Ditto.
4844 (_mm_rcp28_ss): Ditto.
4845 (_mm_rsqrt28_sd): Ditto.
4846 (_mm_rsqrt28_ss): Ditto.
4847 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4848 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4849 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4850 (IX86_BUILTIN_RCP28SD): Ditto.
4851 (IX86_BUILTIN_RCP28SS): Ditto.
4852 (IX86_BUILTIN_RSQRT28SD): Ditto.
4853 (IX86_BUILTIN_RSQRT28SS): Ditto.
4854 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4855 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4856 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4857 (ix86_expand_special_args_builtin): Expand new FTYPE.
4858 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4859 (srcp14<mode>): Make insn unary.
4860 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4861 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4862 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4863 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4864 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4865 Fix rounding: make it SAE only.
4866 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4868 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4870 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4871 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4872 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4873 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4874 (round_saeonly_mask_scalar_operand4): Ditto.
4875 (round_saeonly_mask_scalar_op3): Ditto.
4876 (round_saeonly_mask_scalar_op4): Ditto.
4878 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4880 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4881 Implement -maltivec=be for vec_insert and vec_extract.
4883 2014-01-10 DJ Delorie <dj@redhat.com>
4885 * config/msp430/msp430.md (call_internal): Don't allow memory
4886 references with SP as the base register.
4887 (call_value_internal): Likewise.
4888 * config/msp430/constraints.md (Yc): New. For memory references
4889 that don't use SP as a base register.
4891 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4892 "an integer without a # prefix"
4893 * config/msp430/msp430.md (epilogue_helper): Use it.
4895 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4898 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4899 AVX512F gather builtins.
4900 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4901 on gather decls with INTEGER_TYPE masktype.
4902 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4903 directly into the builtin rather than hoisting it before loop.
4905 PR tree-optimization/59387
4906 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4907 (scev_const_prop): If folded_casts and type has undefined overflow,
4908 use force_gimple_operand instead of force_gimple_operand_gsi and
4909 for each added stmt if it is assign with
4910 arith_code_with_undefined_signed_overflow, call
4911 rewrite_to_defined_overflow.
4912 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4913 gimple-fold.h instead.
4914 (arith_code_with_undefined_signed_overflow,
4915 rewrite_to_defined_overflow): Moved to ...
4916 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4917 rewrite_to_defined_overflow): ... here. No longer static.
4918 Include gimplify-me.h.
4919 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4920 rewrite_to_defined_overflow): New prototypes.
4922 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4924 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4926 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4928 * builtins.c (get_object_alignment_2): Minor tweak.
4929 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4931 2014-01-13 Christian Bruel <christian.bruel@st.com>
4933 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4934 optimized non constant lengths.
4936 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4939 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4940 load as __atomic_load_N if possible.
4942 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4944 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4946 (rs6000_expand_builtin): Adjust call.
4948 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4951 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4952 * config/rs6000/rs6000.c: Include target-globals.h.
4953 (rs6000_set_current_function): Instead of doing target_reinit
4954 unconditionally, use save_target_globals_default_opts and
4955 restore_target_globals.
4957 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4959 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4960 (rs6000_expand_builtin): Handle mffs and mtfsf.
4961 (rs6000_init_builtins): Define mffs and mtfsf.
4962 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4963 (rs6000_mffs): New pattern.
4964 (rs6000_mtfsf): New pattern.
4966 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4968 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4969 Start narrowing with START. Apply candidate-use pair
4970 and check overall cost in narrowing.
4971 (iv_ca_prune): Pass new argument.
4973 2014-01-10 Jeff Law <law@redhat.com>
4976 * ree.c (combine_reaching_defs): Ensure the defining statement
4977 occurs before the extension when optimizing extensions with
4978 different source and destination hard registers.
4980 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4983 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4984 vtables into the type inheritance graph.
4986 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4988 PR rtl-optimization/59754
4989 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4990 modes in the REGNO != REGNO case.
4992 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4994 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4996 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4998 PR tree-optimization/59745
4999 * tree-predcom.c (tree_predictive_commoning_loop): Call
5000 free_affine_expand_cache if giving up because components is NULL.
5002 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5003 GC in payload of target_globals struct instead of allocating them on
5004 the heap and the larger structs separately using GC.
5005 * target-globals.h (struct target_globals): Make regs, hard_regs,
5006 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5007 of GTY((skip)) and change type to void *.
5008 (reset_target_globals): Cast loads from those fields to corresponding
5011 2014-01-10 Steve Ellcey <sellcey@mips.com>
5014 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5015 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5016 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5018 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5021 * aarch64-modes.def (CC_Zmode): New flags mode.
5022 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5023 represents an equality.
5024 (aarch64_get_condition_code): Handle CC_Zmode.
5025 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5027 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5029 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5030 extraction in good case.
5032 2014-01-10 Richard Biener <rguenther@suse.de>
5034 PR tree-optimization/59374
5035 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5036 checking after SLP discovery. Mark stmts not participating
5037 in any SLP instance properly.
5039 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5041 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5042 when handling a SET rtx.
5044 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5046 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5047 (cortex-a57): Likewise.
5048 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5050 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5052 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5053 non-iwmmxt builtins.
5055 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5059 * ipa-devirt.c record_target_from_binfo): Take as argument
5060 stack of binfos and lookup matching one for virtual inheritance.
5061 (possible_polymorphic_call_targets_1): Update.
5063 2014-01-10 Huacai Chen <chenhc@lemote.com>
5065 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5066 kernel strings for Loongson-2E/2F/3A.
5068 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5071 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5072 is_gimple_call before calling gimple_call_internal_p.
5074 2014-01-09 Steve Ellcey <sellcey@mips.com>
5076 * Makefile.in (TREE_FLOW_H): Remove.
5077 (TREE_SSA_H): Add file names from tree-flow.h.
5078 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5079 * tree.h: Remove tree-flow.h reference.
5080 * hash-table.h: Remove tree-flow.h reference.
5081 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5082 reference with tree-ssa-loop.h.
5084 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5086 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5087 default element-order behavior for -maltivec.
5088 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5089 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5090 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5091 when targeting big endian, at least for now.
5092 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5094 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5097 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5098 var satisfies use_register_for_decl, just take into account type
5099 alignment, rather than decl alignment.
5101 PR tree-optimization/59622
5102 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5103 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5104 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5105 Don't devirtualize for inplace at all. For targets.length () == 1,
5106 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5108 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5110 * config/i386/i386.md (cpu): Remove the unused btver1.
5112 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5114 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5116 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5119 * tree-core.h (struct target_globals): New forward declaration.
5120 (struct tree_target_option): Add globals field.
5121 * tree.h (TREE_TARGET_GLOBALS): Define.
5122 (prepare_target_option_nodes_for_pch): New prototype.
5123 * target-globals.h (struct target_globals): Define even if
5125 * tree.c (prepare_target_option_node_for_pch,
5126 prepare_target_option_nodes_for_pch): New functions.
5127 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5128 * config/i386/i386.c: Include target-globals.h.
5129 (ix86_set_current_function): Instead of doing target_reinit
5130 unconditionally, use save_target_globals_default_opts and
5131 restore_target_globals.
5133 2014-01-09 Richard Biener <rguenther@suse.de>
5135 PR tree-optimization/59715
5136 * tree-cfg.h (split_critical_edges): Declare.
5137 * tree-cfg.c (split_critical_edges): Export.
5138 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5140 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5142 * cfgexpand.c (expand_stack_vars): Optionally disable
5143 asan stack protection.
5144 (expand_used_vars): Likewise.
5145 (partition_stack_vars): Likewise.
5146 * asan.c (asan_emit_stack_protection): Optionally disable
5147 after return stack usage.
5148 (instrument_derefs): Optionally disable memory access instrumentation.
5149 (instrument_builtin_call): Likewise.
5150 (instrument_strlen_call): Likewise.
5151 (asan_protect_global): Optionally disable global variables protection.
5152 * doc/invoke.texi: Added doc for new options.
5153 * params.def: Added new options.
5154 * params.h: Likewise.
5156 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5158 PR rtl-optimization/59724
5159 * ifcvt.c (cond_exec_process_if_block): Don't call
5160 flow_find_head_matching_sequence with 0 longest_match.
5161 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5162 non-active insns if !stop_after.
5163 (try_head_merge_bb): Revert 2014-01-07 changes.
5165 2014-01-08 Jeff Law <law@redhat.com>
5167 * ree.c (get_sub_rtx): New function, extracted from...
5168 (merge_def_and_ext): Here.
5169 (combine_reaching_defs): Use get_sub_rtx.
5171 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5173 * cgraph.h (varpool_variable_node): Do not choke on null node.
5175 2014-01-08 Catherine Moore <clm@codesourcery.com>
5177 * config/mips/mips.md (simple_return): Attempt to use JRC
5179 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5181 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5183 PR rtl-optimization/59137
5184 * reorg.c (steal_delay_list_from_target): Call update_block for
5186 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5188 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5191 two duplicate entries.
5193 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5196 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5198 * config/mips/mips.c (mips_truncated_op_cost): New function.
5199 (mips_rtx_costs): Adjust test for BADDU.
5200 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5202 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5204 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5205 (*baddu_si): ...this new pattern.
5207 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5210 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5212 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5215 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5217 (expand_expr, expand_normal): Adjust.
5218 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5219 inner_reference_p. Use inner_reference_p to expand inner references.
5220 (store_expr): Adjust.
5221 * cfgexpand.c (expand_call_stmt): Adjust.
5223 2014-01-08 Rong Xu <xur@google.com>
5225 * gcov-io.c (gcov_var): Move from gcov-io.h.
5226 (gcov_position): Ditto.
5227 (gcov_is_error): Ditto.
5228 (gcov_rewrite): Ditto.
5229 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5230 only part to libgcc/libgcov.h.
5232 2014-01-08 Marek Polacek <polacek@redhat.com>
5235 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5237 2014-01-08 Marek Polacek <polacek@redhat.com>
5240 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5242 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5244 PR rtl-optimization/59649
5245 * stor-layout.c (get_mode_bounds): For BImode return
5246 0 and STORE_FLAG_VALUE.
5248 2014-01-08 Richard Biener <rguenther@suse.de>
5251 * gimple.h (is_gimple_builtin_call): Remove.
5252 (gimple_builtin_call_types_compatible_p): New.
5253 (gimple_call_builtin_p): New overload.
5254 * gimple.c (is_gimple_builtin_call): Remove.
5255 (validate_call): Rename to ...
5256 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5258 (validate_type): New static function.
5259 (gimple_call_builtin_p): New overload and adjust.
5260 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5261 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5262 (gimple_fold_stmt_to_constant_1): Likewise.
5263 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5265 2014-01-08 Richard Biener <rguenther@suse.de>
5268 * gimplify.c (gimplify_expr): Gimplify register-register type
5269 VIEW_CONVERT_EXPRs to separate stmts.
5271 2014-01-07 Jeff Law <law@redhat.com>
5274 * ree.c (combine_set_extension): Handle case where source
5275 and destination registers in an extension insn are different.
5276 (combine_reaching_defs): Allow source and destination registers
5277 in extension to be different under limited circumstances.
5278 (add_removable_extension): Remove restriction that the
5279 source and destination registers in the extension are the same.
5280 (find_and_remove_re): Emit a copy from the extension's
5281 destination to its source after the defining insn if
5282 the source and destination registers are different.
5285 * ifcvt.c (merge_if_block): If we are merging a block with more than
5286 one successor with a block with no successors, remove any BARRIER
5287 after the second block.
5289 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5291 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5293 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5296 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5297 for 14-bit register offsets when INT14_OK_STRICT is false.
5299 2014-01-07 Roland Stigge <stigge@antcom.de>
5300 Michael Meissner <meissner@linux.vnet.ibm.com>
5303 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5304 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5306 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5308 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5311 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5313 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5314 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5315 rtx is const0_rtx or not.
5317 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5320 * target-globals.c (save_target_globals): Remove this_fn_optab
5322 * toplev.c: Include optabs.h.
5323 (target_reinit): Temporarily restore the global options if another
5324 set of options are in force.
5326 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5328 PR rtl-optimization/58668
5329 * cfgcleanup.c (flow_find_cross_jump): Don't count
5330 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5331 to determine what is counted.
5332 (flow_find_head_matching_sequence): Use active_insn_p to determine
5334 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5336 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5337 determine what is counted.
5339 PR tree-optimization/59643
5340 * tree-predcom.c (split_data_refs_to_components): If one dr is
5341 read and one write, determine_offset fails and the write isn't
5342 in the bad component, just put the read into the bad component.
5344 2014-01-07 Mike Stump <mikestump@comcast.net>
5345 Jakub Jelinek <jakub@redhat.com>
5348 * tree-core.h (struct tree_optimization_option): Change optabs
5349 type from unsigned char * to void *.
5350 * optabs.c (init_tree_optimization_optabs): Adjust
5351 TREE_OPTIMIZATION_OPTABS initialization.
5353 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5356 * config/i386/i386.h (struct machine_function): Add
5357 no_drap_save_restore field.
5358 * config/i386/i386.c (ix86_save_reg): Use
5359 !cfun->machine->no_drap_save_restore instead of
5360 crtl->stack_realign_needed.
5361 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5362 this function clears frame_pointer_needed. Set
5363 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5364 and DRAP reg is needed.
5366 2014-01-06 Marek Polacek <polacek@redhat.com>
5369 * doc/implement-c.texi: Mention that other integer types are
5370 permitted as bit-field types in strictly conforming mode.
5372 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5374 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5377 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5379 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5381 2014-01-06 Martin Jambor <mjambor@suse.cz>
5384 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5386 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5388 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5392 * var-tracking.c (add_stores): Preserve the value of the source even if
5393 we don't record the store.
5395 2014-01-06 Terry Guo <terry.guo@arm.com>
5397 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5399 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5402 * config/darwin.c (darwin_function_section): Adjust return values to
5403 correspond to optimisation changes made in r206070.
5405 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5407 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5408 from prefetch_block tune setting.
5409 (nocona_cost): Correct size of prefetch block to 64.
5411 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5413 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5414 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5415 used to save the static chain register in the computation of the offset
5416 from which the FP registers need to be restored.
5418 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5420 PR tree-optimization/59519
5421 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5422 ICE if get_current_def (current_new_name) is already non-NULL, as long
5423 as it is a phi result of some other phi in *new_exit_bb that has
5426 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5427 or vmovdqu* for misaligned_operand.
5428 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5429 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5430 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5431 aligned_mem for AVX512F masked aligned load and store builtins and for
5434 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5436 PR tree-optimization/59651
5437 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5438 Address range for negative step should be added by TYPE_SIZE_UNIT.
5440 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5442 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5443 overlapping registers also for registers other than the stack pointer.
5445 2014-01-03 Marek Polacek <polacek@redhat.com>
5448 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5451 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5454 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5457 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5458 (push splitter): Use <P:MODE_SIZE> instead of
5459 GET_MODE_SIZE (<P:MODE>mode).
5460 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5461 (mov -1, reg peephole2): Likewise.
5462 * config/i386/sse.md (*mov<mode>_internal,
5463 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5464 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5465 *<code><mode>3, *andnot<mode>3<mask_name>,
5466 <mask_codefor><code><mode>3<mask_name>): Likewise.
5467 * config/i386/subst.md (mask_mode512bit_condition,
5468 sd_mask_mode512bit_condition): Likewise.
5470 2014-01-02 Xinliang David Li <davidxl@google.com>
5472 PR tree-optimization/59303
5473 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5474 (dump_predicates): Better output format.
5475 (pred_equal_p): New function.
5476 (is_neq_relop_p): Ditto.
5477 (is_neq_zero_form_p): Ditto.
5478 (pred_expr_equal_p): Ditto.
5479 (pred_neg_p): Ditto.
5480 (simplify_pred): Ditto.
5481 (simplify_preds_2): Ditto.
5482 (simplify_preds_3): Ditto.
5483 (simplify_preds_4): Ditto.
5484 (simplify_preds): Ditto.
5486 (push_to_worklist): Ditto.
5487 (get_pred_info_from_cmp): Ditto.
5488 (is_degenerated_phi): Ditto.
5489 (normalize_one_pred_1): Ditto.
5490 (normalize_one_pred): Ditto.
5491 (normalize_one_pred_chain): Ditto.
5492 (normalize_preds): Ditto.
5493 (normalize_cond_1): Remove function.
5494 (normalize_cond): Ditto.
5495 (is_gcond_subset_of): Ditto.
5496 (is_subset_of_any): Ditto.
5497 (is_or_set_subset_of): Ditto.
5498 (is_and_set_subset_of): Ditto.
5499 (is_norm_cond_subset_of): Ditto.
5500 (pred_chain_length_cmp): Ditto.
5501 (convert_control_dep_chain_into_preds): Type change.
5502 (find_predicates): Ditto.
5503 (find_def_preds): Ditto.
5504 (destroy_predicates_vecs): Ditto.
5505 (find_matching_predicates_in_rest_chains): Ditto.
5506 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5507 (is_pred_expr_subset): Ditto.
5508 (is_pred_chain_subset_of): Ditto.
5509 (is_included_in): Ditto.
5510 (is_superset_of): Ditto.
5512 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5514 Update copyright years.
5516 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5518 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5519 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5520 config/arc/arc.md, config/arc/arc.opt,
5521 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5522 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5523 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5524 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5525 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5526 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5527 vtable-verify.c, vtable-verify.h: Use the standard form for the
5530 2014-01-02 Tobias Burnus <burnus@net-b.de>
5532 * gcc.c (process_command): Update copyright notice dates.
5533 * gcov-dump.c: Ditto.
5535 * doc/cpp.texi: Bump @copying's copyright year.
5536 * doc/cppinternals.texi: Ditto.
5537 * doc/gcc.texi: Ditto.
5538 * doc/gccint.texi: Ditto.
5539 * doc/gcov.texi: Ditto.
5540 * doc/install.texi: Ditto.
5541 * doc/invoke.texi: Ditto.
5543 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5545 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5547 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5549 * config/i386/sse.md (*mov<mode>_internal): Guard
5550 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5552 PR rtl-optimization/59647
5553 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5554 new_rtx into UNSIGNED_FLOAT rtxes.
5556 Copyright (C) 2014 Free Software Foundation, Inc.
5558 Copying and distribution of this file, with or without modification,
5559 are permitted in any medium without royalty provided the copyright
5560 notice and this notice are preserved.