1 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3 * config/rs6000/constraints.md (wD constraint): New constraint to
4 match the constant integer to get the top DImode/DFmode out of a
5 vector in a VSX register.
7 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
8 match the constant integer to get the top DImode/DFmode out of a
9 vector in a VSX register.
11 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
14 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
18 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
20 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
21 Optimize vec_extract of 64-bit values, where the value being
22 extracted is in the top word, where we can use scalar
23 instructions. Add direct move and store support. Combine the big
24 endian/little endian vector select load support into a single
26 (vsx_extract_<mode>_internal1): Likewise.
27 (vsx_extract_<mode>_internal2): Likewise.
28 (vsx_extract_<mode>_load): Likewise.
29 (vsx_extract_<mode>_store): Likewise.
30 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
31 combined into vsx_extract_<mode>_load.
32 (vsx_extract_<mode>_one_le): Likewise.
34 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
35 define the top 64-bit vector element.
37 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
41 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
42 enable use of xxsldwi and xxpermdi builtin functions.
43 (vec_xxpermdi): Likewise.
45 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
46 Document use of vec_xxsldwi and vec_xxpermdi builtins.
48 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
50 PR rtl-optimization/60650
51 * lra-asign.c (find_hard_regno_for, spill_for): Add parameter
53 (find_spills_for): New.
54 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
55 Spill all pseudos on the second iteration.
57 2014-03-27 Marek Polacek <polacek@redhat.com>
60 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
63 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
65 * config/s390/s390.c (s390_can_use_return_insn): Check for
66 call-saved FPRs on 31 bit.
68 2014-03-27 Jakub Jelinek <jakub@redhat.com>
71 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
72 if they need regimplification, just drop them instead of
73 calling gimple_regimplify_operands on them.
75 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
78 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
79 (aarch64_frame_pointer_required): Adjust logic.
80 (aarch64_can_eliminate): Adjust logic.
81 (aarch64_override_options_after_change): Adjust logic.
83 2014-03-27 Dehao Chen <dehao@google.com>
85 * ipa-inline.c (early_inliner): Update node's inline info.
87 2014-03-26 Dehao Chen <dehao@google.com>
89 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
90 compiler inserted conditional jumps for NAN float check.
92 2014-03-26 Jakub Jelinek <jakub@redhat.com>
94 * ubsan.h (ubsan_create_data): Change second argument's type
95 to const location_t *.
96 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
98 (ubsan_create_data): Change second argument to const location_t *PLOC.
99 Create Loc field whenever PLOC is non-NULL.
100 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
101 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
105 * real.c (real_to_integer2): Change type of low to UHWI.
107 2014-03-26 Tobias Burnus <burnus@net-b.de>
109 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
110 (CILK_SELF_SPECS): New define.
111 (driver_self_specs): Use it.
113 2014-03-26 Richard Biener <rguenther@suse.de>
115 * tree-pretty-print.c (percent_K_format): Implement special
116 case for LTO and its stripped down BLOCK tree.
118 2014-03-26 Jakub Jelinek <jakub@redhat.com>
121 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
123 * tree-vrp.c (simplify_internal_call_using_ranges): If only
124 one range is range_int_cst_p, but not both, at least optimize
125 addition/subtraction of 0 and multiplication by 0 or 1.
126 * gimple-fold.c (gimple_fold_call): Fold
127 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
128 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
129 INTEGER_CSTs, try to fold at least x * 0 and y - y.
131 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
133 PR rtl-optimization/60452
134 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
135 <case REG>: Return 1 for invalid offsets from the frame pointer.
137 2014-03-26 Marek Polacek <polacek@redhat.com>
140 * doc/extend.texi (C Extensions): Mention variable-length arrays in
143 2014-03-26 Marek Polacek <polacek@redhat.com>
146 * doc/extend.texi (Designated Inits): Describe what happens to omitted
149 2014-03-26 Marek Polacek <polacek@redhat.com>
152 * ira-color.c (update_conflict_hard_regno_costs): Perform the
153 multiplication in unsigned type.
155 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
157 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
159 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
161 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
163 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
166 * cif-code.def (UNREACHABLE) New code.
167 * ipa-inline.c (inline_small_functions): Skip edges to __builtlin_unreachable.
168 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
169 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
170 predicate to __bulitin_unreachable.
171 (set_cond_stmt_execution_predicate): Fix issue when invert_tree_comparison
173 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
174 propagate to inline clones.
175 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
177 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
178 * cgraphclones.c (cgraph_clone_node): If call destination is already
179 ureachable, do not redirect it back.
180 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
183 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
185 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
186 Do not modify inline clones.
188 2014-03-25 Jakub Jelinek <jakub@redhat.com>
190 * config/i386/i386.md (general_sext_operand): New mode attr.
191 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
192 don't generate (sign_extend (const_int)).
193 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
194 operands[2]. Use We constraint instead of <i> and <general_sext_operand>
195 predicate instead of <general_operand>.
196 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
197 * config/i386/constraints.md (We): New constraint.
198 * config/i386/predicates.md (x86_64_sext_operand,
199 sext_operand): New predicates.
201 2014-03-25 Martin Jambor <mjambor@suse.cz>
204 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
205 inconsistent devirtualizations to __builtin_unreachable.
207 2014-03-25 Marek Polacek <polacek@redhat.com>
210 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
212 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
214 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
215 order of elements for big-endian.
217 2014-03-25 Richard Biener <rguenther@suse.de>
220 * gimplify-me.c (gimple_regimplify_operands): Update the
223 2014-03-25 Martin Jambor <mjambor@suse.cz>
226 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
227 (lto_output_varpool_node): Likewise.
228 (input_overwrite_node): Likewise.
229 (input_varpool_node): Likewise.
231 2014-03-25 Richard Biener <rguenther@suse.de>
233 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like
237 2014-03-25 Jakub Jelinek <jakub@redhat.com>
239 * combine.c (simplify_compare_const): Add MODE argument.
240 Handle mode_width 0 as very large mode_width.
241 (try_combine, simplify_comparison): Adjust callers.
243 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
244 type to avoid signed integer overflow.
245 * explow.c (plus_constant): Likewise.
247 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
249 * doc/generic.texi: Correct typos.
251 2014-03-24 Tobias Burnus <burnus@net-b.de>
253 * doc/invoke.texi (-flto): Expand section about
254 using static libraries with LTO.
256 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
258 PR rtl-optimization/60501
259 * optabs.def (addptr3_optab): New optab.
260 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
261 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
262 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
264 * lra.c (emit_add3_insn): Use the addptr pattern if available.
266 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
268 2014-03-24 Ulrich Drepper <drepper@gmail.com>
270 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
273 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
274 (_mm256_undefined_ps): Define.
275 (_mm256_undefined_pd): Define.
276 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
277 (_mm_undefined_pd): Define.
278 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
279 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
280 (_mm512_undefined_ps): Define.
281 (_mm512_undefined_pd): Define.
282 Use _mm*_undefined_*.
283 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
285 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
287 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
288 (lshr_simd): DI mode added.
289 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
290 (aarch64_ushr_simddi): Likewise.
291 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
292 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
293 (vshrd_n_u64): Likewise.
295 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
297 * Makefile.in (s-macro_list): Depend on cc1.
299 2014-03-23 Teresa Johnson <tejohnson@google.com>
301 * ipa-utils.c (ipa_print_order): Use specified dump file.
303 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
305 PR rtl-optimization/60601
306 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
308 * gcc.c (eval_spec_function): Initialize save_growing_value.
310 2014-03-22 Jakub Jelinek <jakub@redhat.com>
313 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
314 code == MINUS_EXPR, never swap op0 with op1.
316 * toplev.c (init_local_tick): Avoid signed integer multiplication
318 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
319 shift by first operand's bitsize.
321 2014-03-21 Jakub Jelinek <jakub@redhat.com>
324 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
326 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
327 TARGET_ISA_64BIT_P(x).
329 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
331 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
332 pattern for vector nor instead of subtract from splat(-1).
333 (altivec_expand_vec_perm_const_le): Likewise.
335 2014-03-21 Richard Henderson <rth@twiddle.net>
338 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
339 related insns after epilogue_completed.
341 2014-03-21 Martin Jambor <mjambor@suse.cz>
344 * cgraph.h (symtab_node): New flag body_removed.
345 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
346 when removing bodies.
347 * symtab.c (dump_symtab_base): Dump body_removed flag.
348 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
349 had their bodies removed.
351 2014-03-21 Martin Jambor <mjambor@suse.cz>
354 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
357 2014-03-21 Richard Biener <rguenther@suse.de>
359 PR tree-optimization/60577
360 * tree-core.h (struct tree_base): Document nothrow_flag use
361 in VAR_DECL_NONALIASED.
362 * tree.h (VAR_DECL_NONALIASED): New.
363 (may_be_aliased): Adjust.
364 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
366 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
368 * expr.c (expand_expr_real_1): Remove outdated comment.
370 2014-03-20 Jakub Jelinek <jakub@redhat.com>
373 * ira.c (adjust_cleared_regs): Call copy_rtx on
374 *reg_equiv[REGNO (loc)].src_p before passing it to
375 simplify_replace_fn_rtx.
378 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
379 into CONST, put pic register as first operand of PLUS. Use
380 gen_const_mem for both 32-bit and 64-bit PIC got loads.
382 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
384 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
386 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
388 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
389 around for store forwarding issue in the FPU on the UT699.
390 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
391 loads and operations if -mfix-ut699 is specified.
392 (divtf3_hq): Tweak attribute.
393 (sqrttf2_hq): Likewise.
395 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
397 * calls.c (store_one_arg): Remove incorrect const qualification on the
398 type of the temporary.
399 * cfgexpand.c (expand_return): Likewise.
400 * expr.c (expand_constructor): Likewise.
401 (expand_expr_real_1): Likewise.
403 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
405 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
408 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
411 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
413 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
415 * config/arm/aarch-common-protos.h
416 (alu_cost_table): Fix spelling of "extend".
417 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
419 2014-03-19 Richard Biener <rguenther@suse.de>
422 * tree-core.h (tree_type_common): Re-order pointer members
423 to reduce recursion depth during GC walks.
425 2014-03-19 Marek Polacek <polacek@redhat.com>
428 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
431 2014-03-19 Richard Biener <rguenther@suse.de>
434 * lto-streamer-in.c (input_function): In WPA stage do not drop
437 2014-03-19 Jakub Jelinek <jakub@redhat.com>
439 PR tree-optimization/60559
440 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
441 with build_zero_cst assignment.
443 2014-03-18 Kai Tietz <ktietz@redhat.com>
445 PR rtl-optimization/56356
446 * sdbout.c (sdbout_parms): Verify that parms'
447 incoming argument is valid.
448 (sdbout_reg_parms): Likewise.
450 2014-03-18 Richard Henderson <rth@redhat.com>
453 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
454 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
455 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
457 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
459 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
460 * doc/plugins.texi (Plugin callbacks): Mention
462 Italicize plugin event names in description. Explain that
463 PLUGIN_PRAGMAS has no sense for lto1. Explain
465 Remind that no GCC functions should be called after
467 Explain what pragmas with expansion are.
469 2014-03-18 Martin Liska <mliska@suse.cz>
471 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
472 gimple call statement is update.
473 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
474 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
476 2014-03-18 Jakub Jelinek <jakub@redhat.com>
479 * ubsan.c (ubsan_instrument_unreachable): Call
480 initialize_sanitizer_builtins.
481 (ubsan_pass): Likewise.
484 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
485 varpool_finalize_decl instead of rest_of_decl_compilation.
487 2014-03-18 Richard Biener <rguenther@suse.de>
489 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
490 by using bitmap_and_compl instead of bitmap_and_compl_into.
491 (df_rd_transfer_function): Likewise.
493 2014-03-18 Richard Biener <rguenther@suse.de>
495 * doc/lto.texi (fresolution): Fix typo.
497 2014-03-18 Richard Biener <rguenther@suse.de>
499 * doc/invoke.texi (flto): Update for changes in 4.9.
501 2014-03-18 Richard Biener <rguenther@suse.de>
503 * doc/loop.texi: Remove section on the removed lambda framework.
504 Update loop docs with recent changes in preserving loop structure.
506 2014-03-18 Richard Biener <rguenther@suse.de>
508 * doc/lto.texi (-fresolution): Document.
510 2014-03-18 Richard Biener <rguenther@suse.de>
512 * doc/contrib.texi: Adjust my name.
514 2014-03-18 Jakub Jelinek <jakub@redhat.com>
517 * internal-fn.c: Include diagnostic-core.h.
518 (expand_BUILTIN_EXPECT): New function.
519 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
520 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
521 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
522 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
524 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
525 Revert 3 argument __builtin_expect code.
526 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
527 * gimple-fold.c (gimple_fold_call): Likewise.
528 * tree.h (fold_builtin_expect): New prototype.
529 * builtins.c (build_builtin_expect_predicate): Add predictor
530 argument, if non-NULL, create 3 argument __builtin_expect.
531 (fold_builtin_expect): No longer static. Add ARG2 argument,
532 pass it through to build_builtin_expect_predicate.
533 (fold_builtin_2): Adjust caller.
534 (fold_builtin_3): Handle BUILT_IN_EXPECT.
535 * internal-fn.def (BUILTIN_EXPECT): New.
537 2014-03-18 Tobias Burnus <burnus@net-b.de>
540 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
541 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
542 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
544 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
547 * predict.c (combine_predictions_for_bb): Fix up formatting.
548 (expr_expected_value_1, expr_expected_value): Add predictor argument,
549 fill what it points to if non-NULL.
550 (tree_predict_by_opcode): Adjust caller, use the predictor.
551 * predict.def (PRED_COMPARE_AND_SWAP): Add.
553 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
555 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
556 proper constant for the store mode.
558 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
560 * symtab.c (change_decl_assembler_name): Fix transparent alias
563 2014-03-16 Renlin Li <Renlin.Li@arm.com>
565 * config/aarch64/aarch64.c: Correct the comments about the
566 aarch64 stack layout.
568 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
570 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
571 check for GF_OMP_FOR_KIND_FOR.
573 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
575 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
576 ymm and zmm register names.
578 2014-03-17 Jakub Jelinek <jakub@redhat.com>
581 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
582 note creation for the 2010-08-31 changes.
584 2014-03-17 Marek Polacek <polacek@redhat.com>
587 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
588 as -fno-tree-loop-vectorize.
589 (expand_omp_simd): Likewise.
591 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
593 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
594 (eligible_for_call_delay): New prototype.
595 * config/sparc/sparc.c (tls_call_delay): Rename into...
596 (eligible_for_call_delay): ...this. Return false if the instruction
597 cannot be put in the delay slot of a branch.
598 (eligible_for_restore_insn): Simplify.
599 (eligible_for_return_delay): Return false if the instruction cannot be
600 put in the delay slot of a branch and simplify.
601 (eligible_for_sibcall_delay): Return false if the instruction cannot be
602 put in the delay slot of a branch.
603 * config/sparc/sparc.md (fix_ut699): New attribute.
604 (tls_call_delay): Delete.
605 (in_call_delay): Reimplement.
606 (eligible_for_sibcall_delay): Rename into...
607 (in_sibcall_delay): ...this.
608 (eligible_for_return_delay): Rename into...
609 (in_return_delay): ...this.
610 (in_branch_delay): Reimplement.
611 (in_uncond_branch_delay): Delete.
612 (in_annul_branch_delay): Delete.
614 2014-03-14 Richard Henderson <rth@redhat.com>
617 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
618 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
619 (*floathi<X87MODEF>2_i387_with_temp): Remove.
620 (floathi splitters): Remove.
621 (float<SWI48x>xf2): New pattern.
622 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
623 code that tried to handle DImode for 32-bit, but which was excluded
624 by the pattern's condition. Drop allocation of stack temporary.
625 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
626 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
627 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
628 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
629 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
630 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
631 (*float<SWI48><MODEF>2_sse_interunit): Remove.
632 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
633 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
634 (*float<SWI48x><X87MODEF>2_i387): Remove.
635 (all float _with_temp splitters): Remove.
636 (*float<SWI48x><MODEF>2_i387): New pattern.
637 (*float<SWI48><MODEF>2_sse): New pattern.
638 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
639 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
641 2014-03-14 Jakub Jelinek <jakub@redhat.com>
642 Marek Polacek <polacek@redhat.com>
645 * common.opt (dump_base_name_prefixed): New Variable.
646 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
647 if x_dump_base_name_prefixed is already set, set it at the end.
649 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
651 PR rtl-optimization/60508
652 * lra-constraints.c (get_reload_reg): Add new parameter
654 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
655 Pass the new parameter values.
657 2014-03-14 Richard Biener <rguenther@suse.de>
659 * common.opt: Revert unintented changes from r205065.
662 2014-03-14 Richard Biener <rguenther@suse.de>
665 * cfghooks.c (split_block): Properly adjust all loops the
666 block was a latch of.
668 2014-03-14 Martin Jambor <mjambor@suse.cz>
671 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
674 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
677 * config/avr/avr.c (avr_set_current_function): Pass function name
678 through default_strip_name_encoding before sanity checking instead
679 of skipping the first char of the assembler name.
681 2014-03-13 Richard Henderson <rth@redhat.com>
684 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
685 (ix86_force_to_memory, ix86_free_from_memory): Remove.
686 * config/i386/i386-protos.h: Likewise.
687 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
688 in the expander instead of a splitter.
689 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
690 any possibility of requiring a memory.
691 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
692 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
693 (fp branch splitters): Update for ix86_split_fp_branch.
694 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
695 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
696 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
697 (*fop_<MODEF>_2_i387): Remove f/r alternative.
698 (*fop_<MODEF>_3_i387): Likewise.
699 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
700 (splitters for the fop_* register patterns): Remove.
701 (fscalexf4_i387): Rename from *fscalexf4_i387.
702 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
704 2014-03-13 Jakub Jelinek <jakub@redhat.com>
706 PR tree-optimization/59779
707 * tree-dfa.c (get_ref_base_and_extent): Use double_int
708 type for bitsize and maxsize instead of HOST_WIDE_INT.
710 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
712 PR rtl-optimization/57320
713 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
714 the CFG after thread_prologue_and_epilogue_insns.
716 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
718 PR rtl-optimization/57189
719 * lra-constraints.c (process_alt_operands): Disfavor spilling
722 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
724 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
727 2014-03-13 Jakub Jelinek <jakub@redhat.com>
729 PR tree-optimization/59025
731 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
732 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
734 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
737 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
738 calls of avr_out_plus_1.
740 2014-03-13 Bin Cheng <bin.cheng@arm.com>
742 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
743 BB's single pred and update the father loop's latch info later.
745 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
747 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
751 (VEC_base): Likewise.
752 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
753 registers, we need to swap double words in little endian mode.
755 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
756 to be a container mode for 128-bit integer operations added in ISA
757 2.07. Unlike TImode and PTImode, the preferred register set is
758 the Altivec/VMX registers for the 128-bit operations.
760 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
762 (rs6000_split_128bit_ok_p): Likewise.
764 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
765 macros for creating ISA 2.07 normal and overloaded builtin
766 functions with 3 arguments.
767 (BU_P8V_OVERLOAD_3): Likewise.
768 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
769 for use as overloaded functions.
770 (VPERM_1TI_UNS): Likewise.
771 (VSEL_1TI): Likewise.
772 (VSEL_1TI_UNS): Likewise.
773 (ST_INTERNAL_1ti): Likewise.
774 (LD_INTERNAL_1ti): Likewise.
775 (XXSEL_1TI): Likewise.
776 (XXSEL_1TI_UNS): Likewise.
777 (VPERM_1TI): Likewise.
778 (VPERM_1TI_UNS): Likewise.
779 (XXPERMDI_1TI): Likewise.
781 (LXVD2X_V1TI): Likewise.
782 (STXVD2X_V1TI): Likewise.
783 (VEC_INIT_V1TI): Likewise.
784 (VEC_SET_V1TI): Likewise.
785 (VEC_EXT_V1TI): Likewise.
786 (EQV_V1TI): Likewise.
787 (NAND_V1TI): Likewise.
788 (ORC_V1TI): Likewise.
789 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
790 added in ISA 2.07. Add both normal 'altivec' builtins, and the
794 (VADDEUQM): Likewise.
795 (VADDECUQ): Likewise.
796 (VSUBEUQM): Likewise.
797 (VSUBECUQ): Likewise.
799 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
800 __int128_t and __uint128_t types.
801 (__uint128_type): Likewise.
802 (altivec_categorize_keyword): Add support for vector __int128_t,
803 vector __uint128_t, vector __int128, and vector unsigned __int128
804 as a container type for TImode operations that need to be done in
805 VSX/Altivec registers.
806 (rs6000_macro_to_expand): Likewise.
807 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
808 to support 128-bit integer instructions vaddcuq, vadduqm,
809 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
810 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
812 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
813 for V1TImode, and set up preferences to use VSX/Altivec registers.
814 Setup VSX reload handlers.
815 (rs6000_debug_reg_global): Likewise.
816 (rs6000_init_hard_regno_mode_ok): Likewise.
817 (rs6000_preferred_simd_mode): Likewise.
818 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
819 (easy_altivec_constant): Likewise.
820 (output_vec_const_move): Likewise.
821 (rs6000_expand_vector_set): Convert V1TImode set and extract to
823 (rs6000_expand_vector_extract): Likewise.
824 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
826 (rs6000_const_vec): Add support for V1TImode.
827 (rs6000_emit_le_vsx_load): Swap double words when loading or
828 storing TImode/V1TImode.
829 (rs6000_emit_le_vsx_store): Likewise.
830 (rs6000_emit_le_vsx_move): Likewise.
831 (rs6000_emit_move): Add support for V1TImode.
832 (altivec_expand_ld_builtin): Likewise.
833 (altivec_expand_st_builtin): Likewise.
834 (altivec_expand_vec_init_builtin): Likewise.
835 (altivec_expand_builtin): Likewise.
836 (rs6000_init_builtins): Add support for V1TImode type. Add
837 support for ISA 2.07 128-bit integer builtins. Define type names
838 for the VSX/Altivec vector types.
839 (altivec_init_builtins): Add support for overloaded vector
840 functions with V1TImode type.
841 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
842 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
844 (rs6000_split_128bit_ok_p): Likewise.
845 (rs6000_handle_altivec_attribute): Create V1TImode from vector
846 __int128_t and vector __uint128_t.
848 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
856 (VS_scalar): Likewise.
857 (VS_double): Likewise.
858 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
860 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
861 we support the ISA 2.07 128-bit integer arithmetic instructions.
862 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
863 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
864 and TImode types for use with the builtin functions.
865 (V1TI_type_node): Likewise.
866 (unsigned_V1TI_type_node): Likewise.
867 (intTI_type_internal_node): Likewise.
868 (uintTI_type_internal_node): Likewise.
870 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
871 128-bit builtin functions.
872 (UNSPEC_VADDEUQM): Likewise.
873 (UNSPEC_VADDECUQ): Likewise.
874 (UNSPEC_VSUBCUQ): Likewise.
875 (UNSPEC_VSUBEUQM): Likewise.
876 (UNSPEC_VSUBECUQ): Likewise.
877 (VM): Add V1TImode to vector mode iterators.
880 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
881 (altivec_vaddcuq): Likewise.
882 (altivec_vsubuqm): Likewise.
883 (altivec_vsubcuq): Likewise.
884 (altivec_vaddeuqm): Likewise.
885 (altivec_vaddecuq): Likewise.
886 (altivec_vsubeuqm): Likewise.
887 (altivec_vsubecuq): Likewise.
889 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
891 (BOOL_128): Likewise.
892 (BOOL_REGS_OUTPUT): Likewise.
893 (BOOL_REGS_OP1): Likewise.
894 (BOOL_REGS_OP2): Likewise.
895 (BOOL_REGS_UNARY): Likewise.
896 (BOOL_REGS_AND_CR0): Likewise.
898 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
899 128-bit integer builtin support.
900 (vec_vadduqm): Likewise.
901 (vec_vaddecuq): Likewise.
902 (vec_vaddeuqm): Likewise.
903 (vec_vsubecuq): Likewise.
904 (vec_vsubeuqm): Likewise.
905 (vec_vsubcuq): Likewise.
906 (vec_vsubuqm): Likewise.
908 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
909 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
910 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
911 128-bit integer add/subtract to ISA 2.07.
913 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
915 * config/arc/arc.c (arc_predicate_delay_insns):
916 Fix third argument passed to conditionalize_nonjump.
918 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
920 * config/aarch64/aarch64-builtins.c
921 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
922 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
923 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
924 instead of __builtin_lfloor.
925 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
927 2014-03-12 Jakub Jelinek <jakub@redhat.com>
929 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
930 (tree_ssa_ifcombine_bb_1): New function.
931 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
932 is an empty forwarder block to then_bb or vice versa and then_bb
933 and else_bb are effectively swapped.
935 2014-03-12 Christian Bruel <christian.bruel@st.com>
938 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
939 REG_CFA_DEF_CFA note.
940 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
941 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
943 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
945 PR tree-optimization/60454
946 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
948 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
950 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
951 Do not define target_cpu_default2 to generic.
952 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
953 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
954 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
956 2014-03-12 Jakub Jelinek <jakub@redhat.com>
957 Marc Glisse <marc.glisse@inria.fr>
959 PR tree-optimization/60502
960 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
961 instead of build_low_bits_mask.
963 2014-03-12 Jakub Jelinek <jakub@redhat.com>
966 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
967 if there are multiple uses, but op doesn't live on E edge.
968 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
969 clobber stmts before __builtin_unreachable.
971 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
973 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
974 hard_frame_pointer_rtx.
975 * cse.c (cse_insn): Remove volatile check.
976 * cselib.c (cselib_process_insn): Likewise.
977 * dse.c (scan_insn): Likewise.
979 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
981 * config/arc/arc.c (conditionalize_nonjump): New function,
983 (arc_ifcvt): ... this.
984 (arc_predicate_delay_insns): Use it.
986 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
988 * config/arc/predicates.md (extend_operand): During/after reload,
989 allow const_int_operand.
990 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
991 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
992 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
994 (umulsi3_highpart_i): Likewise.
996 2014-03-11 Richard Biener <rguenther@suse.de>
998 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
999 Add asserts to guard possible wrong-code bugs.
1001 2014-03-11 Richard Biener <rguenther@suse.de>
1003 PR tree-optimization/60429
1004 PR tree-optimization/60485
1005 * tree-ssa-structalias.c (set_union_with_increment): Properly
1006 take into account all fields that overlap the shifted vars.
1007 (do_sd_constraint): Likewise.
1008 (do_ds_constraint): Likewise.
1009 (get_constraint_for_ptr_offset): Likewise.
1011 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1013 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1014 (nios2_compute_frame_layout):
1015 Add calculation of cfun->machine->fp_save_offset.
1016 (nios2_expand_prologue): Correct setting of frame pointer register
1018 (nios2_expand_epilogue): Update recovery of stack pointer from
1019 frame pointer accordingly.
1020 (nios2_initial_elimination_offset): Update calculation of offset
1021 for eliminating to HARD_FRAME_POINTER_REGNUM.
1023 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1026 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1027 cgraph_get_create_node on VAR_DECLs.
1029 2014-03-10 Richard Biener <rguenther@suse.de>
1032 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1034 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1036 * config/vms/vms.opt (vms_float_format): New variable.
1038 2014-03-08 Tobias Burnus <burnus@net-b.de>
1040 * doc/invoke.texi (-fcilkplus): Update implementation status.
1042 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1043 Richard Biener <rguenther@suse.de>
1045 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1046 consistently accross all TUs.
1047 (run_gcc): Enable -fshort-double automatically at link at link-time
1048 and disallow override.
1050 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1053 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1054 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1055 if they can't be used.
1057 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1059 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1060 for Solaris 11/x86 ld.
1061 * configure: Regenerate.
1063 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1065 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1066 (LIB_TLS_SPEC): Save as ld_tls_libs.
1067 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1068 (HAVE_AS_IX86_TLSLDM): New test.
1069 * configure, config.in: Regenerate.
1070 * config/i386/i386.c (legitimize_tls_address): Fall back to
1071 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1072 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1073 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1074 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1076 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1078 * common.opt (fira-loop-pressure): Mark as optimization.
1080 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1082 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1083 an OpenMP mappable type.
1085 2014-03-06 Matthias Klose <doko@ubuntu.com>
1087 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1088 MULTILIB_OSDIRNAMES is not defined.
1090 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1091 Meador Inge <meadori@codesourcery.com>
1094 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1095 (arm_legitimize_address): Call legitimize_tls_address for any
1096 arm_tls_referenced_p expression, handle constant addend. Call it
1097 before testing for !TARGET_ARM.
1098 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1100 2014-03-06 Richard Biener <rguenther@suse.de>
1106 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1108 * tree-streamer.c (record_common_node): Assert we don't record
1109 nodes with type double.
1110 (preload_common_node): Skip type double, complex double and double
1111 pointer since it is now frontend dependent due to fshort-double option.
1113 2014-03-06 Richard Biener <rguenther@suse.de>
1115 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1116 or -fno-lto is specified and the linker has full plugin support.
1117 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1118 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1119 * lto-wrapper.c (merge_and_complain): Merge compile-time
1120 optimization levels.
1121 (run_gcc): And pass it through to the link options.
1123 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1127 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1129 * cselib.c (remove_useless_values): Skip to avoid quadratic
1130 behavior if the condition moved from...
1131 (cselib_process_insn): ... here holds.
1133 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1136 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1137 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1140 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1141 (TM_H): Add x86-tune.def.
1143 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1145 * config/aarch64/aarch64.c (generic_tunings):
1146 Use cortexa57_extra_costs.
1148 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1151 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1152 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1153 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1156 2014-03-04 Heiher <r@hev.cc>
1158 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1159 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1161 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1163 * config/i386/predicates.md (const2356_operand): Change to ...
1164 (const2367_operand): ... this.
1165 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1167 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1168 (*avx512pf_scatterpf<mode>sf): Ditto.
1169 (avx512pf_scatterpf<mode>df): Ditto.
1170 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1171 (*avx512pf_scatterpf<mode>df): Ditto.
1172 * config/i386/i386.c (ix86_expand_builtin): Update
1173 incorrect hint operand error message.
1175 2014-03-04 Richard Biener <rguenther@suse.de>
1177 * lto-section-in.c (lto_get_section_data): Fix const cast.
1179 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1181 * tree-streamer.c (record_common_node): Assert we don't record
1182 nodes with type double.
1183 (preload_common_node): Skip type double, complex double and double
1184 pointer since it is now frontend dependent due to fshort-double option.
1186 2014-03-04 Richard Biener <rguenther@suse.de>
1189 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1190 (lto_input_toplevel_asms): Likewise.
1191 * lto-section-in.c (lto_get_section_data): Instead do it here
1194 2014-03-04 Richard Biener <rguenther@suse.de>
1196 PR tree-optimization/60382
1197 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1198 dead PHIs a reduction.
1200 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1202 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1204 (_mm_prefetch): Move out of GCC target("sse") pragma.
1205 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1206 GCC target("prfchw") pragma.
1207 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1209 * config/i386/i386.c (ix86_option_override_internal): Enable
1210 -mprfchw with -mprefetchwt1.
1212 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1214 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1217 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1219 * opts.h (CL_PCH_IGNORE): Define.
1220 * targhooks.c (option_affects_pch_p):
1221 Return false for options that have CL_PCH_IGNORE set.
1222 * opt-functions.awk: Process PchIgnore.
1223 * doc/options.texi: Document PchIgnore.
1225 * config/arc/arc.opt (misize): Add PchIgnore property.
1227 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1229 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1230 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1231 constraint on constants to permit them being loaded into
1232 GENERAL_REGS or BASE_REGS.
1234 2014-03-03 Nick Clifton <nickc@redhat.com>
1236 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1237 anti-cacnonical alternatives.
1238 (negandhi3_real): New pattern.
1239 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1241 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1243 * config/avr/avr-mcus.def: Remove atxmega16x1.
1244 * config/avr/avr-tables.opt: Regenerate.
1245 * config/avr/t-multilib: Regenerate.
1246 * doc/avr-mmcu.texi: Regenerate.
1248 2014-03-03 Tobias Grosser <tobias@grosser.es>
1249 Mircea Namolaru <mircea.namolaru@inria.fr>
1251 PR tree-optimization/58028
1252 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1255 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1257 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1258 not handled by recognizers.
1260 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1263 * function.c (expand_function_end): Don't emit
1264 clobber_return_register sequence if clobber_after is a BARRIER.
1265 * cfgexpand.c (construct_exit_block): Append instructions before
1266 return_label to prev_bb.
1268 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1270 * config/rs6000/constraints.md: Document reserved use of "wc".
1272 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1275 * ipa.c (function_and_variable_visibility): When dissolving comdat
1276 group, also set all symbols to local.
1278 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1283 2013-12-14 Jan Hubicka <jh@suse.cz>
1285 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1287 2014-03-02 Jon Beniston <jon@beniston.com>
1293 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1294 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1295 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1296 (simple_return, *simple_return): New patterns
1297 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1298 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1300 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1302 * dwarf2out.c (gen_subprogram_die): Tidy.
1304 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1307 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1308 (*mov_t_msb_neg_negc): ... this new insn.
1310 2014-02-28 Jason Merrill <jason@redhat.com>
1313 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1316 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1319 * dwarf2out.c (decltype_auto_die): New static.
1320 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1321 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1322 (is_cxx_auto): Likewise.
1324 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1326 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1327 we are not using general regs only.
1329 2014-02-28 Richard Biener <rguenther@suse.de>
1332 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1333 previous fix and only allow to remove trivial pre-headers
1334 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1335 (remove_forwarder_block): Properly update the latch of a loop.
1337 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1340 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1341 (cselib_preserved_hash_table): New.
1342 (preserve_constants_and_equivs): Move preserved vals to it.
1343 (cselib_find_slot): Look it up first.
1344 (cselib_init): Initialize it.
1345 (cselib_finish): Release it.
1346 (dump_cselib_table): Dump it.
1348 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1351 * cselib.c (remove_useless_values): Skip to avoid quadratic
1352 behavior if the condition moved from...
1353 (cselib_process_insn): ... here holds.
1355 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1358 * var-tracking.c (vt_initialize): Apply the same condition to
1359 preserve the CFA base value.
1361 2014-02-28 Joey Ye <joey.ye@arm.com>
1364 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1365 if reload in progress or completed.
1367 2014-02-28 Tobias Burnus <burnus@net-b.de>
1370 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1373 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1375 * doc/tm.texi.in (Condition Code Status): Update documention for
1376 relative locations of cc0-setter and cc0-user.
1378 2014-02-27 Jeff Law <law@redhat.com>
1380 PR rtl-optimization/52714
1381 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1382 into two independent simple sets, if I3 is a jump, ensure the
1383 pattern we place into I3 is a (set (pc) ...).
1385 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1386 Jeff Law <law@redhat.com>
1388 PR rtl-optimization/49847
1389 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1390 are in different blocks.
1391 * doc/tm.texi (Condition Code Status): Update documention for
1392 relative locations of cc0-setter and cc0-user.
1394 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1397 * lra.c (lra_emit_add): Check SUBREG too.
1399 2014-02-27 Andreas Schwab <schwab@suse.de>
1401 * config/m68k/m68k.c (m68k_option_override): Disable
1402 -flive-range-shrinkage for classic m68k.
1403 (m68k_override_options_after_change): Likewise.
1405 2014-02-27 Marek Polacek <polacek@redhat.com>
1408 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1409 -Wmaybe-uninitialized.
1411 2014-02-27 Alan Modra <amodra@gmail.com>
1414 * reload1.c (emit_input_reload_insns): When reload_override_in,
1415 set old to rl->in_reg when rl->in_reg is a subreg.
1417 2014-02-26 Richard Biener <rguenther@suse.de>
1420 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1422 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1424 * common/config/i386/predicates.md (const1256_operand): Remove.
1425 (const2356_operand): New.
1426 (const_1_to_2_operand): Remove.
1427 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1428 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1429 (*avx512pf_gatherpf<mode>sf): Ditto.
1430 (avx512pf_gatherpf<mode>df): Ditto.
1431 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1432 (*avx512pf_gatherpf<mode>df): Ditto.
1433 (avx512pf_scatterpf<mode>sf): Ditto.
1434 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1435 (*avx512pf_scatterpf<mode>sf): Ditto.
1436 (avx512pf_scatterpf<mode>df): Ditto.
1437 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1438 (*avx512pf_scatterpf<mode>df): Ditto.
1439 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1441 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1443 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1444 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1445 (_mm512_mask_testn_epi64_mask): Move to ...
1446 * config/i386/avx512cdintrin.h: Here.
1447 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1448 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1449 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1450 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1451 TARGET_AVX512F from TARGET_AVX512CD.
1453 2014-02-26 Richard Biener <rguenther@suse.de>
1456 * ipa.c (walk_polymorphic_call_targets): Properly guard
1457 call to inline_update_overall_summary.
1459 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1462 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1463 and latches only if requested. Fix latch if it is removed.
1464 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1465 LOOPS_HAVE_PREHEADERS.
1467 2014-02-25 Andrew Pinski <apinski@cavium.com>
1469 * builtins.c (expand_builtin_thread_pointer): Create a new target
1470 when the target is NULL.
1472 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1474 PR rtl-optimization/60317
1475 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1476 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1477 * lra-assigns.c: Include params.h.
1478 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1479 other reload pseudos considerations.
1481 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1483 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1484 to use canonical form for nor<mode>3.
1486 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1489 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1492 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1494 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1495 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1496 (ix86_handle_option): Handle OPT_mprefetchwt1.
1497 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1498 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1500 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1501 OPTION_MASK_ISA_PREFETCHWT1.
1502 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1503 (PTA_PREFETCHWT1): New.
1504 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1505 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1506 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1507 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1508 (*prefetch_avx512pf_<mode>_: Change into ...
1509 (*prefetch_prefetchwt1_<mode>: This.
1510 * config/i386/i386.opt (mprefetchwt1): New.
1511 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1512 (_mm_prefetch): Handle intent to write.
1513 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1515 2014-02-25 Richard Biener <rguenther@suse.de>
1518 * emit-rtl.c (mem_attrs_htab): Remove.
1519 (mem_attrs_htab_hash): Likewise.
1520 (mem_attrs_htab_eq): Likewise.
1521 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1522 (init_emit_once): Do not allocate mem_attrs_htab.
1524 2014-02-25 Richard Biener <rguenther@suse.de>
1527 * lto-opts.c (lto_write_options): Output non-explicit conservative
1528 -fwrapv, -fno-trapv and -fno-strict-overflow.
1529 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1530 (run_gcc): And pass them through.
1532 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1534 * sel-sched.c (calculate_new_fences): New parameter ptime.
1535 Calculate it as a maximum over all fence cycles.
1536 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1537 Print the final schedule timing when sched_verbose.
1539 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1541 PR rtl-optimization/60292
1542 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1543 bit fot the fence instruction.
1545 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1547 * calls.h: Fix typo in comment.
1549 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1551 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1552 adjusting offsetable addresses.
1554 2014-02-24 Guozhi Wei <carrot@google.com>
1556 * sparseset.h (sparseset_pop): Fix the wrong index.
1558 2014-02-24 Walter Lee <walt@tilera.com>
1560 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1561 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1563 * common/config/tilegx/tilegx-common.c
1564 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1565 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1567 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1568 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1569 (tilegx_gimplify_va_arg_expr): Handle big endian.
1570 (tilegx_expand_unaligned_load): Ditto.
1571 (tilegx_expand_unaligned_store): Ditto.
1572 (TARGET_RETURN_IN_MSB): New.
1573 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1574 (TARGET_ENDIAN_DEFAULT): New.
1575 (TARGET_BIG_ENDIAN): Handle big endian.
1576 (BYTES_BIG_ENDIAN): Ditto.
1577 (WORDS_BIG_ENDIAN): Ditto.
1578 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1581 * config/tilegx/tilegx.md (extv): Handle big endian.
1583 (insn_st<n>): Ditto.
1584 (insn_st<n>_add<bitsuffix>): Ditto.
1585 (insn_stnt<n>): Ditto.
1586 (insn_stnt<n>_add<bitsuffix>):Ditto.
1587 (vec_interleave_highv8qi): Handle big endian.
1588 (vec_interleave_highv8qi_be): New.
1589 (vec_interleave_highv8qi_le): New.
1590 (insn_v1int_h): Handle big endian.
1591 (vec_interleave_lowv8qi): Handle big endian.
1592 (vec_interleave_lowv8qi_be): New.
1593 (vec_interleave_lowv8qi_le): New.
1594 (insn_v1int_l): Handle big endian.
1595 (vec_interleave_highv4hi): Handle big endian.
1596 (vec_interleave_highv4hi_be): New.
1597 (vec_interleave_highv4hi_le): New.
1598 (insn_v2int_h): Handle big endian.
1599 (vec_interleave_lowv4hi): Handle big endian.
1600 (vec_interleave_lowv4hi_be): New.
1601 (vec_interleave_lowv4hi_le): New.
1602 (insn_v2int_l): Handle big endian.
1603 (vec_interleave_highv2si): Handle big endian.
1604 (vec_interleave_highv2si_be): New.
1605 (vec_interleave_highv2si_le): New.
1606 (insn_v4int_h): Handle big endian.
1607 (vec_interleave_lowv2si): Handle big endian.
1608 (vec_interleave_lowv2si_be): New.
1609 (vec_interleave_lowv2si_le): New.
1610 (insn_v4int_l): Handle big endian.
1611 * config/tilegx/tilegx.opt (mbig-endian): New option.
1612 (mlittle-endian): New option.
1613 * doc/install.texi: Document tilegxbe-linux.
1614 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1616 2014-02-24 Martin Jambor <mjambor@suse.cz>
1619 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1620 there are no parameter descriptors.
1622 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1624 PR rtl-optimization/60268
1625 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1626 initialization to ...
1627 (sched_rgn_init): ... here.
1628 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1630 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1632 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1635 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1637 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1640 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1642 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1643 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1645 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1647 * config/microblaze/predicates.md: Add cmp_op predicate.
1648 * config/microblaze/microblaze.md: Add branch_compare instruction
1649 which uses cmp_op predicate and emits cmp insn before branch.
1650 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1651 to microblaze_expand_conditional_branch and consolidate logic.
1652 (microblaze_expand_conditional_branch): emit branch_compare
1653 insn instead of handling cmp op separate from branch insn.
1655 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1657 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1660 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1662 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1663 define_insn with define_expand and new define_insn
1664 *altivec_lve<VI_char>x_internal.
1665 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1666 and new define_insn *altivec_stve<VI_char>x_internal.
1667 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1669 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1671 (altivec_expand_stvex_be): New function.
1673 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1675 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1676 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1677 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1678 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1680 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1683 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1684 instead of emit_move_insn.
1686 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1688 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1690 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1691 gen_altivec_vsumsws.
1693 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1695 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1696 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1697 (altivec_lvxl_<mode>): New define_expand incorporating
1698 -maltivec=be semantics where needed.
1699 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1700 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1701 semantics where needed.
1702 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1703 (altivec_stvx_<mode>): New define_expand incorporating
1704 -maltivec=be semantics where needed.
1705 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1706 VM2 iterator instead of V4SI.
1707 (altivec_stvxl_<mode>): New define_expand incorporating
1708 -maltivec=be semantics where needed.
1709 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1710 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1711 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1712 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1713 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1714 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1715 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1716 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1717 ALTIVEC_BUILTIN_STVXL.
1718 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1719 (altivec_expand_stvx_be): Likewise.
1720 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1721 (altivec_expand_lvx_be): Likewise.
1722 (altivec_expand_stvx_be): Likewise.
1723 (altivec_expand_builtin): Add cases for
1724 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1725 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1726 (altivec_init_builtins): Add definitions for
1727 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1728 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1730 2014-02-21 Catherine Moore <clm@codesourcery.com>
1732 * doc/invoke.texi (mvirt, mno-virt): Document.
1733 * config/mips/mips.opt (mvirt): New option.
1734 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1736 2014-02-21 Richard Biener <rguenther@suse.de>
1738 PR tree-optimization/60276
1739 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1740 (STMT_VINFO_MIN_NEG_DIST): New macro.
1741 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1742 STMT_VINFO_MIN_NEG_DIST.
1743 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1744 made for negative dependence distances still hold.
1746 2014-02-21 Richard Biener <rguenther@suse.de>
1749 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1750 DECL_INITIAL for globals not in the current function context.
1752 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1754 PR tree-optimization/56490
1755 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1756 * tree-ssa-uninit.c: Include params.h.
1757 (compute_control_dep_chain): Add num_calls argument, return false
1758 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1759 num_calls to recursive call.
1760 (find_predicates): Change dep_chain into normal array,
1761 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1762 variable and adjust compute_control_dep_chain caller.
1763 (find_def_preds): Likewise.
1765 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1767 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1768 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1770 2014-02-21 Nick Clifton <nickc@redhat.com>
1772 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1773 (pushhi1): Likewise.
1774 (popqi1): Add mode to pre_dec.
1777 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1779 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1780 mode for mask of V8SFmode permutation.
1782 2014-02-20 Richard Henderson <rth@redhat.com>
1785 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1786 a new pseudo for OLDVAL.
1788 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1791 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1792 gen_reg_rtx if d->testing_p.
1793 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1794 if d->testing_p and we will certainly return true.
1795 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1798 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1800 * emit-rtl.c (gen_reg_rtx): Assert that
1801 crtl->emit.regno_pointer_align_length is non-zero.
1803 2014-02-20 Richard Henderson <rth@redhat.com>
1806 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1807 on failure the store back into EXPECT.
1809 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1810 Sandra Loosemore <sandra@codesourcery.com>
1812 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1813 * config/nios2/nios2.c (nios2_function_profiler): Add
1814 -fPIC (flag_pic == 2) support.
1815 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1816 (nios2_large_offset_p): New function.
1817 (nios2_unspec_reloc_p): Move up position, update to use
1818 nios2_large_offset_p.
1819 (nios2_unspec_address): Remove function.
1820 (nios2_unspec_offset): New function.
1821 (nios2_large_got_address): New function.
1822 (nios2_got_address): Add large offset support.
1823 (nios2_legitimize_tls_address): Update usage of removed and new
1825 (nios2_symbol_binds_local_p): New function.
1826 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1827 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1828 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1829 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1830 processing for (const (unspec ...)).
1831 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1833 2014-02-20 Richard Biener <rguenther@suse.de>
1835 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1836 doing the substitution.
1837 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1839 2014-02-20 Martin Jambor <mjambor@suse.cz>
1842 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1843 info when checking whether lattices are bottom.
1845 2014-02-20 Richard Biener <rguenther@suse.de>
1848 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1851 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1854 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1855 parameter specifying the scaling.
1856 (inline_call): Update.
1857 (want_inline_recursively): Guard division by zero.
1858 (recursive_inlining): Update.
1859 * ipa-inline.h (clone_inlined_nodes): Update.
1861 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1864 * config/i386/i386.c (classify_argument): Pass structures of size
1865 64 bytes or less in register.
1867 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1868 Kirill Yukhin <kirill.yukhin@intel.com>
1870 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1871 (_mm_rcp28_round_ss): Ditto.
1872 (_mm_rsqrt28_round_sd): Ditto.
1873 (_mm_rsqrt28_round_ss): Ditto.
1874 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1875 (_mm_rcp14_round_ss): Ditto.
1876 (_mm_rsqrt14_round_sd): Ditto.
1877 (_mm_rsqrt14_round_ss): Ditto.
1878 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1879 the first input operand, get rid of match_dup.
1880 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1882 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1884 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1885 operand as the first input operand, set type attribute.
1886 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1888 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1889 operand as the first input operand, set type attribute.
1891 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1893 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1896 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1899 * config/i386/i386.c (construct_container): Remove TFmode check
1900 for X86_64_INTEGER_CLASS.
1902 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1905 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1906 only when -Wpsabi is enabled.
1908 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1911 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1912 passing arrays in registers are the same as for structs, so remove the
1913 special case for them.
1915 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1917 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1918 destination type, extract only the valid bits if the source type is not
1919 integral and has a different mode.
1921 2014-02-19 Richard Biener <rguenther@suse.de>
1924 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1927 2014-02-19 Richard Biener <rguenther@suse.de>
1930 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1931 (ipa_modify_call_arguments): Emit an argument load explicitely and
1932 preserve virtual SSA form there and for the replacement call.
1933 Do not update SSA form nor free dominance info.
1935 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1937 * ipa.c (function_and_variable_visibility): Also clear WEAK
1938 flag when disolving COMDAT_GROUP.
1940 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1942 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1943 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1945 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1946 do more sanity checks.
1947 (detect_type_change): Return true when giving up early.
1948 (compute_complex_assign_jump_func): Fix type parameter of
1949 ipa_set_ancestor_jf.
1950 (compute_complex_ancestor_jump_func): Likewise.
1951 (update_jump_functions_after_inlining): Fix updating of
1953 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1955 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1957 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1958 inline clones when edge disappears.
1960 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1963 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1964 Split 64-bit moves into 2 patterns. Do not allow the use of
1965 direct move for TDmode in little endian, since the decimal value
1966 has little endian bytes within a word, but the 64-bit pieces are
1967 ordered in a big endian fashion, and normal subreg's of TDmode are
1969 (mov<mode>_64bit_dm): Likewise.
1970 (movtd_64bit_nodm): Likewise.
1972 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1974 PR tree-optimization/60174
1975 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1976 statement of an SSA_NAME that occurs in an abnormal PHI node.
1978 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1981 * final.c (SEEN_BB): Remove.
1982 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1983 (final_scan_insn): Don't force_source_line on second
1984 NOTE_INSN_BASIC_BLOCK.
1986 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1989 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1990 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1991 (type_natural_mode): Warn ABI change when %zmm register is not
1992 available for AVX512F vector value passing.
1994 2014-02-18 Kai Tietz <ktietz@redhat.com>
1997 * config/i386/i386.c (ix86_expand_prologue): Use value in
1998 rax register as displacement when restoring %r10 or %rax.
1999 Fix wrong offset when restoring both registers.
2001 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2003 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2004 assertion with conditional return.
2006 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2007 Uros Bizjak <ubizjak@gmail.com>
2010 * config/i386/driver-i386.c (host_detect_local_cpu): If
2011 YMM state is not saved by the OS, also clear has_f16c. Move
2012 CPUID 0x80000001 handling before YMM state saving checking.
2014 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2016 PR rtl-optimization/58960
2017 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2018 factored out from ...
2019 (sched_init): ... here.
2020 (free_global_sched_pressure_data): New, factored out from ...
2021 (sched_finish): ... here.
2022 * sched-int.h (free_global_sched_pressure_data): Declare.
2023 * sched-rgn.c (nr_regions_initial): New static global.
2024 (haifa_find_rgns): Initialize it.
2025 (schedule_region): Disable sched-pressure for the newly
2028 2014-02-17 Richard Biener <rguenther@suse.de>
2030 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2031 release SSA defs of pattern stmts.
2033 2014-02-17 Richard Biener <rguenther@suse.de>
2035 * tree-inline.c (expand_call_inline): Release the virtual
2036 operand defined by the call we are about to inline.
2038 2014-02-17 Richard Biener <rguenther@suse.de>
2040 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2042 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2043 Ilya Tocar <ilya.tocar@intel.com>
2045 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2046 arguments order in builtin.
2047 (_mm512_permutexvar_epi64): Ditto.
2048 (_mm512_mask_permutexvar_epi64): Ditto
2049 (_mm512_maskz_permutexvar_epi32): Ditto
2050 (_mm512_permutexvar_epi32): Ditto
2051 (_mm512_mask_permutexvar_epi32): Ditto
2053 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2055 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2056 (p8_vmrgow): Likewise.
2058 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2060 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2063 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2066 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2067 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2068 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2069 using direct move instructions on ISA 2.07. Also adjust
2070 instruction length for 64-bit.
2071 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2072 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2074 2014-02-15 Alan Modra <amodra@gmail.com>
2078 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2079 find_replacement on parts of insn rtl that might be reloaded.
2081 2014-02-15 Richard Biener <rguenther@suse.de>
2083 PR tree-optimization/60183
2084 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2085 (tree_ssa_phiprop): Calculate and free post-dominators.
2087 2014-02-14 Jeff Law <law@redhat.com>
2089 PR rtl-optimization/60131
2090 * ree.c (get_extended_src_reg): New function.
2091 (combine_reaching_defs): Use it rather than assuming location of REG.
2092 (find_and_remove_re): Verify first operand of extension is
2093 a REG before adding the insns to the copy list.
2095 2014-02-14 Roland McGrath <mcgrathr@google.com>
2097 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2098 * configure: Regenerated.
2099 * config.in: Regenerated.
2100 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2101 instead of ASM_SHORT.
2103 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2104 Richard Earnshaw <rearnsha@arm.com>
2106 PR rtl-optimization/59535
2107 * lra-constraints.c (process_alt_operands): Encourage alternative
2108 when unassigned pseudo class is superset of the alternative class.
2109 (inherit_reload_reg): Don't inherit when optimizing for code size.
2110 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2111 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2112 modes not less than 4 for Thumb1.
2114 2014-02-14 Kyle McMartin <kyle@redhat.com>
2117 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2119 2014-02-14 Richard Biener <rguenther@suse.de>
2121 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2122 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2123 Do not strip INDIRECT_REFs.
2125 2014-02-14 Richard Biener <rguenther@suse.de>
2128 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2129 DECL_FUNCTION_SPECIFIC_TARGET.
2130 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2131 * tree-streamer-out.c (pack_ts_target_option): Remove.
2132 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2133 (write_ts_function_decl_tree_pointers): Do not stream
2134 DECL_FUNCTION_SPECIFIC_TARGET.
2135 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2136 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2137 (lto_input_ts_function_decl_tree_pointers): Do not stream
2138 DECL_FUNCTION_SPECIFIC_TARGET.
2140 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2142 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2143 (get_initial_def_for_induction, vectorizable_induction): Ignore
2144 debug stmts when looking for exit_phi.
2145 (vectorizable_live_operation): Fix up condition.
2147 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2149 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2150 nreverse() because it changes the content of original tree list.
2152 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2154 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2155 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2157 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2159 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2160 GNU coding standards.
2162 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2165 * dwarf2out.c (gen_subprogram_die): Don't call
2166 add_calling_convention_attribute if subr_die is old_die.
2168 2014-02-13 Sharad Singhai <singhai@google.com>
2170 * doc/optinfo.texi: Fix order of nodes.
2172 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2174 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2175 operands[2], not operands[3].
2177 2014-02-13 Richard Biener <rguenther@suse.de>
2180 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2181 mention the possibility of an in-tree build.
2182 (CLooG): Update recommended version to 0.18.1, mention the
2183 possibility of an in-tree build and clarify that the ISL
2184 bundled with CLooG does not work.
2186 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2189 * expr.c (compress_float_constant): If x is a hard register,
2190 extend into a pseudo and then move to x.
2192 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2194 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2195 caused by bad second argument to warning_at() with -mhotpatch and
2196 nested functions (e.g. with gfortran).
2198 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2200 * opts.c (option_name): Remove "enabled by default" rider.
2202 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2204 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2206 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2207 Uros Bizjak <ubizjak@gmail.com>
2210 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2211 * configure: Regenerated.
2213 2014-02-12 Richard Biener <rguenther@suse.de>
2215 * vec.c (vec_prefix::calculate_allocation): Move as
2216 inline variant to vec.h.
2217 (vec_prefix::calculate_allocation_1): New out-of-line version.
2218 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2219 (vec_prefix::m_has_auto_buf): Rename to ...
2220 (vec_prefix::m_using_auto_storage): ... this.
2221 (vec_prefix::calculate_allocation): Inline the easy cases
2222 and dispatch to calculate_allocation_1 which doesn't need the
2224 (va_heap::reserve): Use gcc_checking_assert.
2225 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2226 m_using_auto_storage.
2227 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2229 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2230 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2231 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2233 2014-02-12 Richard Biener <rguenther@suse.de>
2235 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2236 when we found a dependence.
2238 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2240 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2242 (maybe_fold_stmt): ... into this new function.
2243 * omp-low.c (lower_omp): Update comment.
2245 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2248 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2251 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2253 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2254 identifiers in comments.
2255 (cortexa53_extra_costs): Likewise.
2256 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2257 (cortexa7_extra_costs): Likewise.
2258 (cortexa12_extra_costs): Likewise.
2259 (cortexa15_extra_costs): Likewise.
2260 (v7m_extra_costs): Likewise.
2262 2014-02-12 Richard Biener <rguenther@suse.de>
2265 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2266 of posix_memalign being successful.
2267 (lower_stmt): Restrict lowering of posix_memalign to when
2268 -ftree-bit-ccp is enabled.
2270 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2272 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2274 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2276 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2278 PR rtl-optimization/60116
2279 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2280 other_insn once the combination has been validated.
2282 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2285 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2287 * ipa-devirt.c: Include demangle.h
2288 (odr_violation_reported): New static variable.
2289 (add_type_duplicate): Update odr_violations.
2290 (maybe_record_node): Add completep parameter; update it.
2291 (record_target_from_binfo): Add COMPLETEP parameter;
2292 update it as needed.
2293 (possible_polymorphic_call_targets_1): Likewise.
2294 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2295 rename FINAL to COMPLETE.
2296 (record_targets_from_bases): Sanity check we found the binfo;
2297 fix COMPLETEP updating.
2298 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2299 parameter, fix computing of COMPLETEP.
2300 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2301 at LTO time do demangling.
2302 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2303 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2305 (gimple_get_virt_method_for_binfo): Likewise.
2306 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2307 gimple_get_virt_method_for_vtable): Update prototypes.
2309 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2312 * genautomata.c (add_presence_absence): Fix typo with
2313 {final_}presence_list.
2315 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2318 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2319 for VSX/Altivec vectors that land in GPR registers.
2321 2014-02-11 Richard Henderson <rth@redhat.com>
2322 Jakub Jelinek <jakub@redhat.com>
2325 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2326 around drhs if type conversion to lacc->type is not useless.
2328 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2330 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2332 (cortex-a57.cortex-a53): Likewise.
2333 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2335 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2337 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2340 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2342 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2343 add_options_for_arm_vfp3.
2345 2014-02-11 Jeff Law <law@redhat.com>
2348 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2349 object with an undesirable mode.
2351 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2354 * config/i386/sol2-9.h: New file.
2355 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2356 *-*-solaris2.9*): Use it.
2358 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2360 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2361 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2363 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2365 * config/microblaze/microblaze.c: Extend mcpu version format
2367 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2369 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2371 2014-02-10 Richard Henderson <rth@redhat.com>
2374 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2375 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2376 ms-abi vs -mno-accumulate-outgoing-args.
2377 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2378 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2381 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2384 * cfgexpand.c (expand_asm_operands): Attach source location to
2385 ASM_INPUT rtx objects.
2386 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2388 2014-02-10 Nick Clifton <nickc@redhat.com>
2390 * config/mn10300/mn10300.c (popcount): New function.
2391 (mn10300_expand_prologue): Include saved registers in stack usage
2394 2014-02-10 Jeff Law <law@redhat.com>
2397 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2398 when changing the SET_DEST of a prior insn to avoid an input reload.
2400 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2402 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2403 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2404 -mcall-openbsd, or -mcall-linux.
2405 (CC1_ENDIAN_BIG_SPEC): Remove.
2406 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2407 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2408 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2409 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2410 and %cc1_endian_default.
2411 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2413 2014-02-10 Richard Biener <rguenther@suse.de>
2415 PR tree-optimization/60115
2416 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2417 MEM_REF handling. Properly verify that the accesses are not
2418 out of the objects bound.
2420 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2422 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2425 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2427 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2428 proper constants and fix formatting.
2429 (possible_polymorphic_call_targets): Fix formatting.
2431 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2432 Ilya Tocar <ilya.tocar@intel.com>
2434 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2435 (_mm512_loadu_epi32): Renamed into...
2436 (_mm512_loadu_si512): This.
2437 (_mm512_storeu_epi32): Renamed into...
2438 (_mm512_storeu_si512): This.
2439 (_mm512_maskz_ceil_ps): Removed.
2440 (_mm512_maskz_ceil_pd): Ditto.
2441 (_mm512_maskz_floor_ps): Ditto.
2442 (_mm512_maskz_floor_pd): Ditto.
2443 (_mm512_floor_round_ps): Ditto.
2444 (_mm512_floor_round_pd): Ditto.
2445 (_mm512_ceil_round_ps): Ditto.
2446 (_mm512_ceil_round_pd): Ditto.
2447 (_mm512_mask_floor_round_ps): Ditto.
2448 (_mm512_mask_floor_round_pd): Ditto.
2449 (_mm512_mask_ceil_round_ps): Ditto.
2450 (_mm512_mask_ceil_round_pd): Ditto.
2451 (_mm512_maskz_floor_round_ps): Ditto.
2452 (_mm512_maskz_floor_round_pd): Ditto.
2453 (_mm512_maskz_ceil_round_ps): Ditto.
2454 (_mm512_maskz_ceil_round_pd): Ditto.
2455 (_mm512_expand_pd): Ditto.
2456 (_mm512_expand_ps): Ditto.
2457 * config/i386/i386.c (ix86_builtins): Remove
2458 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2459 (bdesc_args): Ditto.
2460 * config/i386/predicates.md (const1256_operand): New.
2461 (const_1_to_2_operand): Ditto.
2462 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2463 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2464 (*avx512pf_gatherpf<mode>sf): Ditto.
2465 (avx512pf_gatherpf<mode>df): Ditto.
2466 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2467 (*avx512pf_gatherpf<mode>df): Ditto.
2468 (avx512pf_scatterpf<mode>sf): Ditto.
2469 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2470 (*avx512pf_scatterpf<mode>sf): Ditto.
2471 (avx512pf_scatterpf<mode>df): Ditto.
2472 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2473 (*avx512pf_scatterpf<mode>df): Ditto.
2474 (avx512f_expand<mode>): Removed.
2475 (<shift_insn><mode>3<mask_name>): Change predicate type.
2477 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2479 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2480 not at the end of datarefs vector use ordered_remove to avoid
2481 reordering datarefs vector.
2484 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2485 mark local addressable non-static vars as GOVD_PRIVATE
2486 instead of GOVD_LOCAL.
2487 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2488 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2492 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2493 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2494 assume_aligned or alloc_align attributes.
2495 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2496 arguments. Handle also assume_aligned and alloc_align attributes.
2497 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2498 calls to functions with assume_aligned or alloc_align attributes.
2499 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2501 2014-02-08 Terry Guo <terry.guo@arm.com>
2503 * doc/invoke.texi: Document ARM -march=armv7e-m.
2505 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2507 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2508 flag on __cilkrts_rethrow builtin.
2511 * ipa-cp.c (determine_versionability): Fail at -O0
2512 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2513 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2516 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2519 * tree-inline.c (copy_forbidden): Fail for
2520 __attribute__((optimize (0))) functions.
2522 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2524 * varpool.c: Include pointer-set.h.
2525 (varpool_remove_unreferenced_decls): Variables in other partitions
2526 will not be output; be however careful to not lose information
2529 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2531 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2532 lookup in the vtable constructor.
2534 2014-02-07 Jeff Law <law@redhat.com>
2537 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2538 define_insn_and_split.
2540 * ipa-inline.c (inline_small_functions): Fix typos.
2542 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2544 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2545 (s390_can_use_return_insn): Declare.
2546 * config/s390/s390.h (EPILOGUE_USES): Define.
2547 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2549 (s390_chunkify_start): Handle return JUMP_LABELs.
2550 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2551 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2552 (s390_can_use_return_insn): New functions.
2553 (s390_fix_long_loop_prediction): Handle conditional returns.
2554 (TARGET_SET_UP_BY_PROLOGUE): Define.
2555 * config/s390/s390.md (ANY_RETURN): New code iterator.
2556 (*creturn, *csimple_return, return, simple_return): New patterns.
2558 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2560 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2561 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2562 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2563 REG_CFA_RESTORE list when deciding not to restore a register.
2565 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2567 * config/s390/s390.c: Include tree-pass.h and context.h.
2568 (s390_early_mach): New function, split out from...
2569 (s390_emit_prologue): ...here.
2570 (pass_data_s390_early_mach): New pass structure.
2571 (pass_s390_early_mach): New class.
2572 (s390_option_override): Create and register early_mach pass.
2573 Move to end of file.
2575 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2577 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2578 to match for the exit block.
2580 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2582 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2583 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2584 Reject misaligned operands.
2586 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2588 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2590 2014-02-07 Richard Biener <rguenther@suse.de>
2593 * gimple-low.c (lower_builtin_posix_memalign): New function.
2594 (lower_stmt): Call it to lower posix_memalign in a way
2595 to make alignment info accessible.
2597 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2600 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2601 __builtin_setjmp_receiver.
2603 2014-02-07 Richard Biener <rguenther@suse.de>
2606 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2607 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2608 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2609 Handle BUILT_IN_POSIX_MEMALIGN.
2610 (find_func_clobbers): Likewise.
2611 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2612 (call_may_clobber_ref_p_1): Likewise.
2614 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2617 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2620 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2623 * lto-cgraph.c (lto_output_node): Use
2624 symtab_get_symbol_partitioning_class.
2625 (lto_output_varpool_node): likewise.
2626 (symtab_get_symbol_partitioning_class): Move here from
2628 * cgraph.h (symbol_partitioning_class): Likewise.
2629 (symtab_get_symbol_partitioning_class): Declare.
2631 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2633 * ggc.h (ggc_internal_cleared_alloc): New macro.
2634 * vec.h (vec_safe_copy): Handle memory stats.
2635 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2636 * target-globals.c (save_target_globals): Likewise.
2638 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2641 * expr.c (emit_move_resolve_push): Export; be bit more selective
2642 on when to clear alias set.
2643 * expr.h (emit_move_resolve_push): Declare.
2644 * function.h (struct function): Add tail_call_marked.
2645 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2646 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2647 * config/i386/i386.md (TImode move expander): De not call
2649 (FP push expanders): Preserve memory attributes.
2650 * config/i386/sse.md (push<mode>1): Remove.
2651 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2652 (ix86_expand_push): Remove.
2653 * config/i386/mmx.md (push<mode>1): Remove.
2655 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2657 PR rtl-optimization/60030
2658 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2659 lopart with paradoxical subreg before shifting it up by hprec.
2661 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2663 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2664 Remove extra newline at end of file.
2665 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2666 (arm_issue_rate): Handle cortexa57.
2667 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2668 (cortex-a57.cortex-a53): Likewise.
2670 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2673 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2674 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2676 (arm_expand_prologue): Adjust all callers.
2677 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2678 info, registers also at the lowest numbered registers side. Use
2679 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2683 * var-tracking.c (adjust_mems): Before adding a SET to
2684 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2686 2014-02-06 Alan Modra <amodra@gmail.com>
2689 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2690 change SDmode to DDmode when lra_in_progress.
2692 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2695 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2696 free_data_ref on the dr first, and before goto again also set dr
2697 to the next dr. For simd_lane_access, free old datarefs[i] before
2698 overwriting it. For get_vectype_for_scalar_type failure, don't
2699 free_data_ref if simd_lane_access.
2701 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2704 * tree.h (opts_for_fn): New inline function.
2705 (opt_for_fn): Define.
2706 * config/i386/i386.c (ix86_function_regparm): Use
2707 opt_for_fn (decl, optimize) instead of optimize.
2709 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2711 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2712 for SYMBOL_REF in large memory model.
2714 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2716 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2718 (cortex-a57): Likewise.
2719 (cortex-a57.cortex-a53): Likewise.
2721 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2722 Kugan Vivekanandarajah <kuganv@linaro.org>
2724 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2726 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2728 2014-02-06 Richard Biener <rguenther@suse.de>
2730 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2731 set_loop_copy and initialize_original_copy_tables.
2733 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2735 * config/aarch64/aarch64-simd.md
2736 (aarch64_ashr_simddi): Change QI to SI.
2738 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2739 Jakub Jelinek <jakub@redhat.com>
2742 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2745 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2747 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2748 CODE_FOR_altivec_vpku[hw]um to
2749 CODE_FOR_altivec_vpku[hw]um_direct.
2750 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2751 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2752 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2753 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2755 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2757 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2758 generation for -maltivec=be.
2759 (altivec_vsumsws): Simplify redundant test.
2761 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2763 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2764 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2765 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2766 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2767 gen_altivec_vpkuwum.
2768 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2770 (altivec_vpks<VI_char>ss): Likewise.
2771 (altivec_vpks<VI_char>us): Likewise.
2772 (altivec_vpku<VI_char>us): Likewise.
2773 (altivec_vpku<VI_char>um): Likewise.
2774 (altivec_vpku<VI_char>um_direct): New (copy of
2775 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2777 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2778 target is little endian and -maltivec=be is not specified.
2779 (*altivec_vupkhs<VU_char>_direct): New (copy of
2780 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2781 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2782 target is little endian and -maltivec=be is not specified.
2783 (*altivec_vupkls<VU_char>_direct): New (copy of
2784 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2785 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2786 little endian and -maltivec=be is not specified.
2787 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2788 little endian and -maltivec=be is not specified.
2790 2014-02-05 Richard Henderson <rth@redhat.com>
2793 * combine-stack-adj.c: Revert r206943.
2794 * sched-int.h (struct deps_desc): Add last_args_size.
2795 * sched-deps.c (init_deps): Initialize it.
2796 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2797 contain REG_ARGS_SIZE notes.
2799 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2801 * lto-cgraph.c (asm_nodes_output): Make global.
2802 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2803 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2804 (driver_handle_option): Handle OPT_fwpa.
2806 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2809 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2810 a comment typo and formatting issue. If odr_hash hasn't been
2811 created, return vNULL and set *completep to false.
2814 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2815 bb with no successors.
2817 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2820 * doc/invoke.texi (-march): Clarify documentation for ARM.
2824 2014-02-05 Richard Biener <rguenther@suse.de>
2826 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2827 when not vectorizing because of too many alias checks.
2828 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2829 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2831 2014-02-05 Nick Clifton <nickc@redhat.com>
2833 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2834 accept extended registers in any mode when compiling for the MN10300.
2836 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2838 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2839 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2840 sanitization attributes.
2841 (can_inline_edge_p): Likewise.
2842 (sanitize_attrs_match_for_inline_p): New function.
2844 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2846 * ipa-prop.c (detect_type_change): Shor circuit testing of
2847 type changes on THIS pointer.
2849 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2852 * config/pa/pa.c (legitimize_tls_address): Return original address
2853 if not passed a SYMBOL_REF rtx.
2854 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2856 (pa_emit_move_sequence): Simplify TLS source operands.
2857 (pa_legitimate_constant_p): Reject all TLS constants.
2858 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2859 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2861 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2863 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2864 groups when we know they are controlled by LTO.
2865 * varasm.c (default_binds_local_p_1): If object is in other partition,
2866 it will be resolved locally.
2868 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2870 * config/host-linux.c (linux_gt_pch_use_address): Don't
2871 use SSIZE_MAX because it is not always defined.
2873 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2876 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2877 threshold for pseudo splitting.
2878 (update_ebb_live_info): Process call argument hard registers and
2879 hard registers from insn definition too.
2880 (max_small_class_regs_num): New constant.
2881 (inherit_in_ebb): Update live hard regs through EBBs. Update
2882 reloads_num only for small register classes. Don't split for
2885 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2888 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2891 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2893 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2896 2014-02-04 Marek Polacek <polacek@redhat.com>
2898 * gdbinit.in (pel): Define.
2900 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2902 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2905 2014-02-04 Richard Biener <rguenther@suse.de>
2908 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2909 in function context local.
2910 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2911 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2912 similar to LTO_imported_decl_ref.
2914 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2916 PR tree-optimization/60002
2917 * cgraphclones.c (build_function_decl_skip_args): Clear
2920 PR tree-optimization/60023
2921 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2922 false to gsi_replace.
2923 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2924 has been in some EH region and vec_stmt could throw, add
2925 vec_stmt into the same EH region.
2926 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2927 has no lhs, ignore it.
2928 * internal-fn.c (expand_MASK_LOAD): Likewise.
2931 * tree-inline.c (copy_forbidden): Fail for
2932 __attribute__((optimize (0))) functions.
2935 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2936 is set, copy one less argument.
2937 (expand_simd_clones): Don't subtract clone_info->inbranch
2938 from simd_clone_struct_alloc argument.
2940 PR rtl-optimization/57915
2941 * recog.c (simplify_while_replacing): If all unary/binary/relational
2942 operation arguments are constant, attempt to simplify those.
2945 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2946 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2948 2014-02-04 Richard Biener <rguenther@suse.de>
2950 PR tree-optimization/60012
2951 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2952 TBAA disambiguation to all DDRs.
2954 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2957 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2958 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2960 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2963 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2965 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2967 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2968 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2970 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2973 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2974 to figure out targets of polymorphic calls with known decl.
2975 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2976 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2977 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2978 (get_polymorphic_call_info): ... here.
2979 (get_polymorphic_call_info_from_invariant): New function.
2981 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2983 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2984 lookup via vtable pointer; check for type consistency
2985 and turn inconsitent facts into UNREACHABLE.
2986 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2987 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2988 type inconsistent querries; return UNREACHABLE instead.
2990 2014-02-03 Richard Henderson <rth@twiddle.net>
2993 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2994 already processed this node.
2995 (normalize_one_pred_1): Pass along mark_set.
2996 (normalize_one_pred): Create and destroy a pointer_set_t.
2997 (normalize_one_pred_chain): Likewise.
2999 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3001 PR gcov-profile/58602
3002 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3004 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3007 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3008 -fno-devirtualize; try to devirtualize by the knowledge of
3009 virtual table pointer given by aggregate propagation.
3010 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3011 (ipa_print_node_jump_functions): Dump also offset that
3012 is relevant for polymorphic calls.
3013 (determine_known_aggregate_parts): Add arg_type parameter; use it
3014 instead of determining the type from pointer type.
3015 (ipa_compute_jump_functions_for_edge): Update call of
3016 determine_known_aggregate_parts.
3017 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3018 (gimple_get_virt_method_for_binfo): ... here; simplify using
3019 vtable_pointer_value_to_vtable.
3020 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3021 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3022 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3023 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3025 (vtable_pointer_value_to_binfo): ... here.
3026 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3028 2014-02-03 Teresa Johnson <tejohnson@google.com>
3030 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3031 redef of outer loop index variable.
3033 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3037 * doc/extend.texi (Function Attributes): Typo.
3039 2014-02-03 Cong Hou <congh@google.com>
3041 PR tree-optimization/60000
3042 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3043 if the vectorized statement is a store. A store statement can only
3044 appear at the end of pattern statements.
3046 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3048 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3049 (ix86_option_override_internal): Default long double to 64-bit for
3050 32-bit Bionic and to 128-bit for 64-bit Bionic.
3052 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3053 TARGET_LONG_DOUBLE_128 is true.
3054 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3056 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3057 (mlong-double-64): Negate -mlong-double-128.
3058 (mlong-double-128): New option.
3060 * config/i386/i386-c.c (ix86_target_macros): Define
3061 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3063 * doc/invoke.texi: Document -mlong-double-128.
3065 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3067 PR rtl-optimization/60024
3068 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3070 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3072 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3074 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3076 PR rtl-optimization/57662
3077 * sel-sched.c (code_motion_path_driver): Do not mark already not
3078 existing blocks in the visiting bitmap.
3080 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3082 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3083 on the insn being emitted.
3085 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3086 Will Deacon <will.deacon@arm.com>
3088 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3090 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3092 * config/arm/arm-tables.opt: Regenerate.
3094 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3096 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3097 for vector types other than V16QImode.
3098 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3099 define_expand, and call altivec_expand_vec_perm_le when producing
3100 code with little endian element order.
3101 (*altivec_vperm_<mode>_internal): New insn having previous
3102 behavior of altivec_vperm_<mode>.
3103 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3104 altivec_expand_vec_perm_le when producing code with little endian
3106 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3107 behavior of altivec_vperm_<mode>_uns.
3109 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3111 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3112 (altivec_vsumsws): Add handling for -maltivec=be with a little
3114 (altivec_vsumsws_direct): New.
3115 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3116 gen_altivec_vsumsws.
3118 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3120 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3121 vtable_pointer_value_to_binfo): New functions.
3122 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3123 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3125 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3127 * config/nios2/nios2.md (load_got_register): Initialize GOT
3128 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3129 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3131 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3133 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3134 preserverd by passthrough, do not propagate the type.
3136 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3138 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3139 (mips_atomic_assign_expand_fenv): New function.
3140 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3142 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3144 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3145 (__builtin_mips_set_fcsr): Likewise.
3146 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3147 MIPS_USI_FTYPE_VOID.
3148 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3149 (mips16_expand_set_fcsr): Likewise.
3150 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3151 (mips16_set_fcsr_stub): Likewise.
3152 (mips16_get_fcsr_one_only_stub): New class.
3153 (mips16_set_fcsr_one_only_stub): Likewise.
3154 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3155 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3156 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3157 (hard_float): New availability predicate.
3158 (mips_builtins): Add get_fcsr and set_fcsr.
3159 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3160 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3161 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3162 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3163 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3166 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3168 * config/mips/mips.c (mips_one_only_stub): New class.
3169 (mips_need_mips16_rdhwr_p): Replace with...
3170 (mips16_rdhwr_stub): ...this new variable.
3171 (mips16_stub_call_address): New function.
3172 (mips16_rdhwr_one_only_stub): New class.
3173 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3174 (mips_output_mips16_rdhwr): Delete.
3175 (mips_finish_stub): New function.
3176 (mips_code_end): Use it to handle rdhwr stubs.
3178 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3181 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3182 when calculating size of integer atomic types.
3184 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3186 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3188 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3190 PR tree-optimization/60003
3191 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3192 * profile.c (branch_prob): Use gimple_call_builtin_p
3193 to check for BUILT_IN_SETJMP_RECEIVER.
3194 * tree-inline.c (copy_bb): Call notice_special_calls.
3196 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3199 * lra-constraints.c (process_alt_operands): Update reload_sum only
3202 2014-01-31 Richard Henderson <rth@redhat.com>
3205 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3206 until after else_eh is processed.
3208 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3210 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3211 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3212 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3213 in smmintrin.h, remove them.
3214 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3215 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3216 * config/i386/i386.md (ROUND_SAE): Fix value.
3217 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3218 (const48_operand): New.
3219 * config/i386/subst.md (round), (round_expand): Use
3220 const_4_or_8_to_11_operand.
3221 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3223 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3225 * config/i386/constraints.md (Yk): Swap meaning with k.
3226 * config/i386/i386.md (movhi_internal): Change Yk to k.
3227 (movqi_internal): Ditto.
3228 (*k<logic><mode>): Ditto.
3231 (kandn<mode>): Ditto.
3232 (*<code>hi_1): Ditto.
3233 (*<code>qi_1): Ditto.
3234 (kxnor<mode>): Ditto.
3235 (kortestzhi): Ditto.
3236 (kortestchi): Ditto.
3238 (*one_cmplhi2_1): Ditto.
3239 (*one_cmplqi2_1): Ditto.
3240 * config/i386/sse.md (): Change k to Yk.
3241 (avx512f_load<mode>_mask): Ditto.
3242 (avx512f_blendm<mode>): Ditto.
3243 (avx512f_store<mode>_mask): Ditto.
3244 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3245 (avx512f_storedqu<mode>_mask): Ditto.
3246 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3248 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3249 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3250 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3251 (avx512f_maskcmp<mode>3): Ditto.
3252 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3253 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3254 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3255 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3256 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3257 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3258 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3259 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3260 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3261 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3262 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3263 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3264 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3265 (vec_extract_lo_<mode>_maskm): Ditto.
3266 (vec_extract_hi_<mode>_maskm): Ditto.
3267 (avx512f_vternlog<mode>_mask): Ditto.
3268 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3269 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3270 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3271 (avx512f_<code>v8div16qi2_mask): Ditto.
3272 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3273 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3274 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3275 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3276 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3277 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3278 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3279 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3280 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3281 (avx512cd_maskb_vec_dupv8di): Ditto.
3282 (avx512cd_maskw_vec_dupv16si): Ditto.
3283 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3284 (avx512f_vpermi2var<mode>3_mask): Ditto.
3285 (avx512f_vpermi2var<mode>3_mask): Ditto.
3286 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3287 (*avx512f_gathersi<mode>): Ditto.
3288 (*avx512f_gathersi<mode>_2): Ditto.
3289 (*avx512f_gatherdi<mode>): Ditto.
3290 (*avx512f_gatherdi<mode>_2): Ditto.
3291 (*avx512f_scattersi<mode>): Ditto.
3292 (*avx512f_scatterdi<mode>): Ditto.
3293 (avx512f_compress<mode>_mask): Ditto.
3294 (avx512f_compressstore<mode>_mask): Ditto.
3295 (avx512f_expand<mode>_mask): Ditto.
3296 * config/i386/subst.md (mask): Change k to Yk.
3297 (mask_scalar_merge): Ditto.
3300 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3302 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3304 2014-01-31 Richard Biener <rguenther@suse.de>
3307 * builtins.c (fold_builtin_memory_op): Make sure to not
3308 use a floating-point mode or a boolean or enumeral type for
3311 2014-01-30 DJ Delorie <dj@redhat.com>
3313 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3314 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3315 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3316 whenever main() has an epilogue.
3318 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3320 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3321 unused variable "field".
3322 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3323 (vsx_mergeh_<mode>): Likewise.
3324 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3325 (altivec_vmrghh): Likewise.
3326 (altivec_vmrghw): Likewise.
3327 (altivec_vmrglb): Likewise.
3328 (altivec_vmrglh): Likewise.
3329 (altivec_vmrglw): Likewise.
3330 (altivec_vspltb): Add missing uses.
3331 (altivec_vsplth): Likewise.
3332 (altivec_vspltw): Likewise.
3333 (altivec_vspltsf): Likewise.
3335 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3338 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3339 frame related instructions.
3341 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3343 PR rtl-optimization/59959
3344 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3345 any reload of register whose subreg is invalid.
3347 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3349 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3350 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3351 Add missing return type - void.
3353 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3355 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3356 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3357 remove element index adjustment for endian (now handled in vsx.md
3359 (altivec_expand_vec_perm_const): Use
3360 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3361 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3362 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3363 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3364 define_expand and a new define_insn *altivec_vspltb_internal;
3365 adjust for -maltivec=be on a little endian target.
3366 (altivec_vspltb_direct): New.
3367 (altivec_vsplth): Divide into a define_expand and a new
3368 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3369 little endian target.
3370 (altivec_vsplth_direct): New.
3371 (altivec_vspltw): Divide into a define_expand and a new
3372 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3373 little endian target.
3374 (altivec_vspltw_direct): New.
3375 (altivec_vspltsf): Divide into a define_expand and a new
3376 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3377 a little endian target.
3379 2014-01-30 Richard Biener <rguenther@suse.de>
3381 PR tree-optimization/59993
3382 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3383 can propagate form the earlier stmt and avoid the transform
3384 when the intermediate result is needed.
3386 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3388 * README.Portability: Fix typo.
3390 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3392 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3393 comparison_operator with ordered_comparison_operator.
3395 2014-01-30 Nick Clifton <nickc@redhat.com>
3397 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3398 Rename to mn10300_store_multiple_regs.
3399 * config/mn10300/mn10300.c: Likewise.
3400 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3401 store_multiple_regs.
3402 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3403 Call mn10300_store_multiple_regs.
3405 2014-01-30 Nick Clifton <nickc@redhat.com>
3406 DJ Delorie <dj@redhat.com>
3408 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3409 %fp 2 to keep registers after it properly word-aligned.
3410 (rl78_alloc_physical_registers_umul): Handle the case where both
3411 input operands are the same.
3413 2014-01-30 Richard Biener <rguenther@suse.de>
3415 PR tree-optimization/59903
3416 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3419 2014-01-30 Jason Merrill <jason@redhat.com>
3422 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3425 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3427 2014-01-30 Richard Biener <rguenther@suse.de>
3429 PR tree-optimization/59951
3430 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3432 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3435 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3436 SFmode to DFmode case.
3438 2014-01-29 DJ Delorie <dj@redhat.com>
3440 * config/msp430/msp430.opt (-minrt): New.
3441 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3443 (ENDFILE_SPEC): Likewise.
3445 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3447 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3448 (estimate_function_body_sizes): Use it.
3450 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3453 * dwarf2out.c (is_cxx_auto): New.
3454 (is_base_type): Use it.
3455 (gen_type_die_with_usage): Likewise.
3457 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3459 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3460 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3461 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3462 -maltivec=be with LE targets.
3463 (vsx_mergeh_<mode>): Likewise.
3464 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3465 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3466 (altivec_vmrghb): Replace with define_expand and new
3467 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3468 (altivec_vmrghb_direct): New define_insn.
3469 (altivec_vmrghh): Replace with define_expand and new
3470 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3471 (altivec_vmrghh_direct): New define_insn.
3472 (altivec_vmrghw): Replace with define_expand and new
3473 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3474 (altivec_vmrghw_direct): New define_insn.
3475 (*altivec_vmrghsf): Adjust for endianness.
3476 (altivec_vmrglb): Replace with define_expand and new
3477 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3478 (altivec_vmrglb_direct): New define_insn.
3479 (altivec_vmrglh): Replace with define_expand and new
3480 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3481 (altivec_vmrglh_direct): New define_insn.
3482 (altivec_vmrglw): Replace with define_expand and new
3483 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3484 (altivec_vmrglw_direct): New define_insn.
3485 (*altivec_vmrglsf): Adjust for endianness.
3486 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3487 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3488 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3489 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3490 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3491 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3492 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3493 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3495 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3497 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3498 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3501 2014-01-29 Richard Biener <rguenther@suse.de>
3503 PR tree-optimization/58742
3504 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3505 associate_pointerplus_align.
3506 (associate_pointerplus_diff): New function.
3507 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3508 and associate_pointerplus_diff.
3510 2014-01-29 Richard Biener <rguenther@suse.de>
3512 * lto-streamer.h (LTO_major_version): Bump to 3.
3513 (LTO_minor_version): Reset to 0.
3515 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3517 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3518 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3519 (arm_file_start): Generate correct asm header for armv7ve.
3520 * config/arm/bpabi.h: Add multilib support for armv7ve.
3521 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3522 and cortex-a15 to armv7ve.
3523 * config/arm/t-aprofile: Add multilib support for armv7ve.
3524 * doc/invoke.texi: Document -march=armv7ve.
3526 2014-01-29 Richard Biener <rguenther@suse.de>
3528 PR tree-optimization/58742
3529 * tree-ssa-forwprop.c (associate_plusminus): Return true
3530 if we changed sth, defer EH cleanup to ...
3531 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3532 (simplify_mult): New function.
3534 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3537 PR tree-optimization/59920
3538 * tree.c (build_common_builtin_nodes): Remove
3539 __builtin_setjmp_dispatcher initialization.
3540 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3541 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3542 instead of gsi_after_labels + manually skipping debug stmts.
3543 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3544 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3545 * tree-inline.c (copy_edges_for_bb): Remove
3546 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3547 argument. Ignore computed_goto_p stmts. Don't call
3548 make_abnormal_goto_edges. If a call might need abnormal edges
3549 for non-local gotos, see if it already has an edge to
3550 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3551 with true argument, don't do anything then, otherwise add
3552 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3553 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3555 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3556 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3557 (lower_stmt): Don't set data->calls_builtin_setjmp.
3558 (lower_builtin_setjmp): Adjust comment.
3559 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3560 * tree-cfg.c (found_computed_goto): Remove.
3561 (factor_computed_gotos): Remove.
3562 (make_goto_expr_edges): Return bool, true for computed gotos.
3563 Don't call make_abnormal_goto_edges.
3564 (build_gimple_cfg): Don't set found_computed_goto, don't call
3565 factor_computed_gotos.
3566 (computed_goto_p): No longer static.
3567 (make_blocks): Don't set found_computed_goto.
3568 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3569 (make_edges): If make_goto_expr_edges returns true, push bb
3570 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3571 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3572 vector. Record mapping between bbs and OpenMP regions if there
3573 are any, adjust make_gimple_omp_edges caller. Call
3574 handle_abnormal_edges.
3575 (make_abnormal_goto_edges): Remove.
3576 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3577 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3578 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3579 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3580 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3581 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3582 filling *region also set *region_idx to (*region)->entry->index.
3585 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3586 For REGs set ORIGINAL_REGNO.
3588 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3590 * doc/md.texi: Mention that a target shouldn't implement
3591 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3594 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3596 PR tree-optimization/59594
3597 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3598 a copy of the datarefs vector rather than the vector itself.
3600 2014-01-28 Jason Merrill <jason@redhat.com>
3603 * dwarf2out.c (auto_die): New static.
3604 (gen_type_die_with_usage): Handle C++1y 'auto'.
3605 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3608 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3611 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3612 (SPEC_X32): Likewise.
3613 (SPEC_64): Likewise.
3614 * config/i386/i386.c (ix86_option_override_internal): Turn off
3615 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3617 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3618 * config/i386/i386.h (TARGET_16BIT): New macro.
3619 (TARGET_16BIT_P): Likewise.
3620 * config/i386/i386.opt: Add m16.
3621 * doc/invoke.texi: Document -m16.
3623 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3625 PR preprocessor/59935
3626 * input.c (location_get_source_line): Bail out on when line number
3627 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3629 2014-01-28 Richard Biener <rguenther@suse.de>
3631 PR tree-optimization/58742
3632 * tree-ssa-forwprop.c (associate_plusminus): Handle
3633 pointer subtraction of the form (T)(P + A) - (T)P.
3635 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3637 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3640 2014-01-28 Richard Biener <rguenther@suse.de>
3643 2014-01-28 Richard Biener <rguenther@suse.de>
3645 PR rtl-optimization/45364
3646 PR rtl-optimization/59890
3647 * var-tracking.c (local_get_addr_clear_given_value): Handle
3648 already cleared slot.
3649 (val_reset): Handle not allocated local_get_addr_cache.
3650 (vt_find_locations): Use post-order on the inverted CFG.
3652 2014-01-28 Richard Biener <rguenther@suse.de>
3654 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3656 2014-01-28 Richard Biener <rguenther@suse.de>
3658 PR rtl-optimization/45364
3659 PR rtl-optimization/59890
3660 * var-tracking.c (local_get_addr_clear_given_value): Handle
3661 already cleared slot.
3662 (val_reset): Handle not allocated local_get_addr_cache.
3663 (vt_find_locations): Use post-order on the inverted CFG.
3665 2014-01-28 Alan Modra <amodra@gmail.com>
3667 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3668 * configure.ac <recursive call for build != host>: Define
3669 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3670 and LD_FOR_BUILD too.
3671 * configure: Regenerate.
3673 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3675 * config/i386/i386.c (get_builtin_code_for_version): Separate
3676 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3677 Broadwell from Haswell.
3679 2014-01-27 Steve Ellcey <sellcey@mips.com>
3681 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3682 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3683 * config/mips/mips.c (mips_option_override): Change setting
3685 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3686 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3687 Change from Mask to Var.
3689 2014-01-27 Jeff Law <law@redhat.com>
3691 * ipa-inline.c (inline_small_functions): Fix typo.
3693 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3695 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3696 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3697 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3698 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3699 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3700 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3701 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3702 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3703 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3704 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3705 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3706 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3707 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3708 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3709 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3710 (_mm512_storeu_epi64): Ditto.
3711 (_mm512_cmpge_epi32_mask): Ditto.
3712 (_mm512_cmpge_epu32_mask): Ditto.
3713 (_mm512_cmpge_epi64_mask): Ditto.
3714 (_mm512_cmpge_epu64_mask): Ditto.
3715 (_mm512_cmple_epi32_mask): Ditto.
3716 (_mm512_cmple_epu32_mask): Ditto.
3717 (_mm512_cmple_epi64_mask): Ditto.
3718 (_mm512_cmple_epu64_mask): Ditto.
3719 (_mm512_cmplt_epi32_mask): Ditto.
3720 (_mm512_cmplt_epu32_mask): Ditto.
3721 (_mm512_cmplt_epi64_mask): Ditto.
3722 (_mm512_cmplt_epu64_mask): Ditto.
3723 (_mm512_cmpneq_epi32_mask): Ditto.
3724 (_mm512_cmpneq_epu32_mask): Ditto.
3725 (_mm512_cmpneq_epi64_mask): Ditto.
3726 (_mm512_cmpneq_epu64_mask): Ditto.
3727 (_mm512_expand_pd): Ditto.
3728 (_mm512_expand_ps): Ditto.
3729 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3730 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3731 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3732 * config/i386/i386.c (ix86_builtins): Add
3733 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3734 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3735 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3736 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3737 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3738 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3739 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3740 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3741 IX86_BUILTIN_PMOVUSQW512_MEM.
3742 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3743 __builtin_ia32_pmovsqd512mem_mask,
3744 __builtin_ia32_pmovqd512mem_mask,
3745 __builtin_ia32_pmovusqw512mem_mask,
3746 __builtin_ia32_pmovsqw512mem_mask,
3747 __builtin_ia32_pmovqw512mem_mask,
3748 __builtin_ia32_pmovusdw512mem_mask,
3749 __builtin_ia32_pmovsdw512mem_mask,
3750 __builtin_ia32_pmovdw512mem_mask,
3751 __builtin_ia32_pmovqb512mem_mask,
3752 __builtin_ia32_pmovusqb512mem_mask,
3753 __builtin_ia32_pmovsqb512mem_mask,
3754 __builtin_ia32_pmovusdb512mem_mask,
3755 __builtin_ia32_pmovsdb512mem_mask,
3756 __builtin_ia32_pmovdb512mem_mask.
3757 (bdesc_args): Add __builtin_ia32_expanddf512,
3758 __builtin_ia32_expandsf512.
3759 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3760 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3761 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3762 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3763 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3764 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3765 (avx512f_<code>v8div16qi2_mask_store): This.
3766 (avx512f_expand<mode>): New.
3768 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3770 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3772 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3773 (_mm512_prefetch_i32scatter_pd): Ditto.
3774 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3775 (_mm512_prefetch_i64scatter_pd): Ditto.
3776 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3777 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3778 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3779 (_mm512_prefetch_i32scatter_ps): Ditto.
3780 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3781 (_mm512_prefetch_i64scatter_ps): Ditto.
3782 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3783 * config/i386/i386-builtin-types.def: Define
3784 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3785 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3786 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3787 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3788 IX86_BUILTIN_SCATTERPFQPD.
3789 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3790 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3791 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3792 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3793 __builtin_ia32_scatterpfqps.
3794 (ix86_expand_builtin): Expand new built-ins.
3795 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3796 fix memory access data type.
3797 (*avx512pf_gatherpf<mode>_mask): Ditto.
3798 (*avx512pf_gatherpf<mode>): Ditto.
3799 (avx512pf_scatterpf<mode>): Ditto.
3800 (*avx512pf_scatterpf<mode>_mask): Ditto.
3801 (*avx512pf_scatterpf<mode>): Ditto.
3802 (GATHER_SCATTER_SF_MEM_MODE): New.
3803 (avx512pf_gatherpf<mode>df): Ditto.
3804 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3805 (*avx512pf_scatterpf<mode>df): Ditto.
3807 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3810 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3811 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3814 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3816 * common/config/arm/arm-common.c
3817 (arm_rewrite_mcpu): Handle multiple names.
3819 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3821 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3823 * gimple-builder.h (create_gimple_tmp): Delete.
3825 2014-01-27 Christian Bruel <christian.bruel@st.com>
3827 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3830 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3832 * config/pa/pa.md (call): Generate indirect long calls to non-local
3833 functions when outputing 32-bit code.
3834 (call_value): Likewise except for special call to buggy powf function.
3836 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3837 portable runtime and PIC indirect calls.
3838 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3839 and PIC call sequences. Use ldo instead of blr to set return register
3840 in PIC call sequence.
3842 2014-01-25 Walter Lee <walt@tilera.com>
3844 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3845 avoid clobbering a live register.
3847 2014-01-25 Walter Lee <walt@tilera.com>
3849 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3850 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3851 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3852 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3854 2014-01-25 Walter Lee <walt@tilera.com>
3856 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3857 arguments on even registers.
3858 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3860 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3861 (BIGGEST_ALIGNMENT): Ditto.
3862 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3864 2014-01-25 Walter Lee <walt@tilera.com>
3866 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3867 insns before bundling.
3868 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3870 2014-01-25 Walter Lee <walt@tilera.com>
3872 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3873 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3874 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3876 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3878 * config/mips/constraints.md (kl): Delete.
3879 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3880 define expands, using...
3881 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3882 instructions for MIPS16.
3883 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3884 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3886 2014-01-25 Walter Lee <walt@tilera.com>
3888 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3892 2014-01-25 Walter Lee <walt@tilera.com>
3894 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3895 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3897 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3899 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3902 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3904 * print-rtl.c (in_call_function_usage): New var.
3905 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3906 EXPR_LIST mode as mode and not as reg note name.
3909 * cfgloopmanip.c (copy_loop_info): If
3910 loop->warned_aggressive_loop_optimizations, make sure
3911 the flag is set in target loop too.
3913 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3915 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3917 * builtins.def: Likewise.
3918 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3919 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3920 * ira.c (ira_setup_eliminable_regset): Likewise.
3921 * omp-low.c (gate_expand_omp): Likewise.
3922 (execute_lower_omp): Likewise.
3923 (diagnose_sb_0): Likewise.
3924 (gate_diagnose_omp_blocks): Likewise.
3925 (simd_clone_clauses_extract): Likewise.
3928 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3930 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3931 correction for little endian...
3932 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3935 2014-01-24 Jeff Law <law@redhat.com>
3937 PR tree-optimization/59919
3938 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3939 for non-returning calls.
3941 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3943 * common/config/aarch64/aarch64-common.c
3944 (aarch64_rewrite_mcpu): Handle multiple names.
3945 * config/aarch64/aarch64.h
3946 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3948 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3950 * input.c (add_file_to_cache_tab): Handle the case where fopen
3953 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3956 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3957 from push operand if code of push isn't PRE_DEC.
3959 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3962 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3963 -mquad-memory-atomic. Update -mquad-memory documentation to say
3964 it is only used for non-atomic loads/stores.
3966 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3967 -mquad-memory or -mquad-memory-atomic switches.
3969 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3970 -mquad-memory-atomic to ISA 2.07 support.
3972 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3973 to separate support of normal quad word memory operations (ldq, stq)
3974 from the atomic quad word memory operations.
3976 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3977 support to separate non-atomic quad word operations from atomic
3978 quad word operations. Disable non-atomic quad word operations in
3979 little endian mode so that we don't have to swap words after the
3980 load and before the store.
3981 (quad_load_store_p): Add comment about atomic quad word support.
3982 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3983 options printed with -mdebug=reg.
3985 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3986 -mquad-memory-atomic as the test for whether we have quad word
3987 atomic instructions.
3988 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3989 or -mp8-vector are used, allow byte/half-word atomic operations.
3991 * config/rs6000/sync.md (load_lockedti): Insure that the address
3992 is a proper indexed or indirect address for the lqarx instruction.
3993 On little endian systems, swap the hi/lo registers after the lqarx
3995 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3996 insure the address is valid for the lqarx instruction.
3997 (store_conditionalti): Insure that the address is a proper indexed
3998 or indirect address for the stqcrx. instruction. On little endian
3999 systems, swap the hi/lo registers before doing the stqcrx.
4001 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4002 insure the address is valid for the stqcrx. instruction.
4004 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4005 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4006 type of quad memory support is available.
4008 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4011 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4012 there is a danger of looping.
4014 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4017 force flag_ira_loop_pressure if set via command line.
4019 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4021 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4022 (ashr_simd): New builtin handling DI mode.
4023 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4024 (aarch64_sshr_simddi): New match pattern.
4025 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4026 (vshrd_n_s64): Likewise.
4027 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4029 2014-01-23 Nick Clifton <nickc@redhat.com>
4031 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4032 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4033 favour of mcu specific scripts.
4034 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4037 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4038 Alex Velenko <Alex.Velenko@arm.com>
4040 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4041 (vaddv_s16): Likewise.
4042 (vaddv_s32): Likewise.
4043 (vaddv_u8): Likewise.
4044 (vaddv_u16): Likewise.
4045 (vaddv_u32): Likewise.
4046 (vaddvq_s8): Likewise.
4047 (vaddvq_s16): Likewise.
4048 (vaddvq_s32): Likewise.
4049 (vaddvq_s64): Likewise.
4050 (vaddvq_u8): Likewise.
4051 (vaddvq_u16): Likewise.
4052 (vaddvq_u32): Likewise.
4053 (vaddvq_u64): Likewise.
4054 (vaddv_f32): Likewise.
4055 (vaddvq_f32): Likewise.
4056 (vaddvq_f64): Likewise.
4057 (vmaxv_f32): Likewise.
4058 (vmaxv_s8): Likewise.
4059 (vmaxv_s16): Likewise.
4060 (vmaxv_s32): Likewise.
4061 (vmaxv_u8): Likewise.
4062 (vmaxv_u16): Likewise.
4063 (vmaxv_u32): Likewise.
4064 (vmaxvq_f32): Likewise.
4065 (vmaxvq_f64): Likewise.
4066 (vmaxvq_s8): Likewise.
4067 (vmaxvq_s16): Likewise.
4068 (vmaxvq_s32): Likewise.
4069 (vmaxvq_u8): Likewise.
4070 (vmaxvq_u16): Likewise.
4071 (vmaxvq_u32): Likewise.
4072 (vmaxnmv_f32): Likewise.
4073 (vmaxnmvq_f32): Likewise.
4074 (vmaxnmvq_f64): Likewise.
4075 (vminv_f32): Likewise.
4076 (vminv_s8): Likewise.
4077 (vminv_s16): Likewise.
4078 (vminv_s32): Likewise.
4079 (vminv_u8): Likewise.
4080 (vminv_u16): Likewise.
4081 (vminv_u32): Likewise.
4082 (vminvq_f32): Likewise.
4083 (vminvq_f64): Likewise.
4084 (vminvq_s8): Likewise.
4085 (vminvq_s16): Likewise.
4086 (vminvq_s32): Likewise.
4087 (vminvq_u8): Likewise.
4088 (vminvq_u16): Likewise.
4089 (vminvq_u32): Likewise.
4090 (vminnmv_f32): Likewise.
4091 (vminnmvq_f32): Likewise.
4092 (vminnmvq_f64): Likewise.
4094 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4096 * config/aarch64/aarch64-simd.md
4097 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4098 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4099 (*aarch64_mul3_elt<mode>): Likewise.
4100 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4101 (*aarch64_mul3_elt_to_64v2df): Likewise.
4102 (*aarch64_mla_elt<mode>): Likewise.
4103 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4104 (*aarch64_mls_elt<mode>): Likewise.
4105 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4106 (*aarch64_fma4_elt<mode>): Likewise.
4107 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4108 (*aarch64_fma4_elt_to_64v2df): Likewise.
4109 (*aarch64_fnma4_elt<mode>): Likewise.
4110 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4111 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4112 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4113 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4114 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4115 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4116 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4117 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4118 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4120 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4122 * config/aarch64/aarch64-simd.md
4123 (aarch64_be_checked_get_lane<mode>): New define_expand.
4124 * config/aarch64/aarch64-simd-builtins.def
4125 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4126 New builtin definition.
4127 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4128 Use new safe be builtin.
4130 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4132 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4134 (aarch64_be_st1<mode>): Likewise.
4135 (aarch_ld1<VALL:mode>): Define_expand modified.
4136 (aarch_st1<VALL:mode>): Likewise.
4137 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4138 (UNSPEC_ST1): Likewise.
4140 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4142 * config/microblaze/microblaze.md: Add trap insn and attribute
4144 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4146 PR preprocessor/58580
4147 * input.h (location_get_source_line): Take an additional line_size
4149 (void diagnostics_file_cache_fini): Declare new function.
4150 * input.c (struct fcache): New type.
4151 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4152 New static constants.
4153 (diagnostic_file_cache_init, total_lines_num)
4154 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4155 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4156 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4157 (get_next_line, read_next_line, goto_next_line, read_line_num):
4158 New static function definitions.
4159 (diagnostic_file_cache_fini): New function.
4160 (location_get_source_line): Take an additional output line_len
4161 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4163 * diagnostic.c (diagnostic_finish): Call
4164 diagnostic_file_cache_fini.
4165 (adjust_line): Take an additional input parameter for the length
4166 of the line, rather than calculating it with strlen.
4167 (diagnostic_show_locus): Adjust the use of
4168 location_get_source_line and adjust_line with respect to their new
4169 signature. While displaying a line now, do not stop at the first
4170 null byte. Rather, display the zero byte as a space and keep
4171 going until we reach the size of the line.
4172 * Makefile.in: Add vec.o to OBJS-libcommon
4174 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4175 Ilya Tocar <ilya.tocar@intel.com>
4177 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4178 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4179 (__builtin_ia32_kmov16): Ditto.
4180 * config/i386/i386.md (UNSPEC_KMOV): New.
4183 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4185 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4186 (_mm512_storeu_si512): Ditto.
4188 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4191 * rtl.h (get_referenced_operands): Declare.
4192 * recog.c (get_referenced_operands): New function.
4193 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4194 operands have been referenced when recording LO_SUM references.
4196 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4198 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4200 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4202 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4203 Enable for generic and recent AMD targets.
4205 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4207 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4208 ARG_SIZE note when adjustment was eliminated.
4210 2014-01-22 Jeff Law <law@redhat.com>
4212 PR tree-optimization/59597
4213 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4214 in file. Accept new argument REGISTERING and use it to modify
4215 dump output appropriately.
4216 (register_jump_thread): Corresponding changes.
4217 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4218 thread paths involving joiner blocks. Add code to dump cancelled
4219 jump threading paths.
4221 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4223 PR rtl-optimization/59477
4224 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4225 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4227 2014-01-22 Tom Tromey <tromey@redhat.com>
4229 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4231 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4233 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4235 PR rtl-optimization/59896
4236 * lra-constraints.c (process_alt_operands): Check unused note for
4237 matched operands of insn with no output reloads.
4239 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4241 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4242 (mips_move_from_gpr_cost): Likewise.
4244 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4246 PR rtl-optimization/59858
4247 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4248 ira_class_hard_regs_num.
4249 (process_alt_operands): Increase reject for dying matched operand.
4251 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4254 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4255 smaller than size, perform several stores or loads and stores
4256 at dst + count - size to store or copy all of size bytes, rather
4257 than just last modesize bytes.
4259 2014-01-20 DJ Delorie <dj@redhat.com>
4261 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4262 that CLOBBERs are REGs before propogating their values.
4264 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4267 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4268 (cgraph_inline_failed_type): New function.
4269 * cgraph.h (DEFCIFCODE): Add type.
4270 (cgraph_inline_failed_type_t): New enum.
4271 (cgraph_inline_failed_type): New prototype.
4272 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4273 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4274 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4275 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4276 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4277 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4278 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4279 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4280 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4281 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4282 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4283 OPTIMIZATION_MISMATCH.
4284 * tree-inline.c (expand_call_inline): Emit errors during
4285 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4287 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4290 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4291 mode attribute in insn output.
4293 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4295 * output.h (output_constant): Delete.
4296 * varasm.c (output_constant): Make private.
4298 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4300 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4302 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4305 * tree.h (fold_builtin_strcat): New prototype.
4306 * builtins.c (fold_builtin_strcat): No longer static. Add len
4307 argument, if non-NULL, don't call c_strlen. Optimize
4308 directly into __builtin_memcpy instead of __builtin_strcpy.
4309 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4310 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4312 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4314 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4315 for SImode_address_operand operands, having only a REG argument.
4317 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4319 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4320 loader name using mbig-endian.
4321 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4323 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4325 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4329 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4331 * config/aarch64/aarch64-protos.h
4332 (aarch64_cannot_change_mode_class_ptr): Declare.
4333 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4334 aarch64_cannot_change_mode_class_ptr): New.
4335 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4336 backend hook aarch64_cannot_change_mode_class.
4338 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4340 * common/config/aarch64/aarch64-common.c
4341 (aarch64_handle_option): Don't handle any option order logic here.
4342 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4343 selected_cpu, warn on architecture version mismatch.
4344 (aarch64_override_options): Fix parsing order for option strings.
4346 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4347 Iain Sandoe <iain@codesourcery.com>
4350 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4351 warning. Amend comment to reflect current functionality.
4353 2014-01-20 Richard Biener <rguenther@suse.de>
4356 * builtins.c (fold_builtin_strcat): Remove case better handled
4357 by tree-ssa-strlen.c.
4359 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4361 * config/aarch64/aarch64.opt
4362 (mcpu, march, mtune): Make case-insensitive.
4364 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4367 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4368 if operands[1] is a REG or ZERO_EXTEND of a REG.
4370 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4372 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4374 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4376 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4377 long non-pic millicode calls.
4379 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4381 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4383 2014-01-19 Kito Cheng <kito@0xlab.org>
4385 * builtins.c (expand_movstr): Check movstr expand done or fail.
4387 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4388 H.J. Lu <hongjiu.lu@intel.com>
4391 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4392 to DImode for zero-extended addresses.
4394 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4396 PR rtl-optimization/57763
4397 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4398 on the new indirect jump_insn and increment LABEL_NUSES (label).
4400 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4404 * config.gcc (x86_archs): New variable.
4405 (x86_64_archs): Likewise.
4406 (x86_cpus): Likewise.
4407 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4408 --with-arch/--with-cpu= options.
4409 Support --with-arch=/--with-cpu={nehalem,westmere,
4410 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4412 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4414 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4415 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4417 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4419 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4421 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4424 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4425 clear cpp_get_options (parse_in)->warn_unused_macros for
4426 ix86_target_macros_internal with cpp_define.
4428 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4430 * jump.c (delete_related_insns): Keep (use (insn))s.
4431 * reorg.c (redundant_insn): Check for barriers too.
4433 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4435 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4437 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4439 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4440 call to $$dyncall when TARGET_LONG_CALLS is true.
4442 2014-01-17 Jeff Law <law@redhat.com>
4444 * ree.c (combine_set_extension): Temporarily disable test for
4445 changing number of hard registers.
4447 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4450 * ipa-inline-analysis.c (inline_free_summary):
4451 Do not free summary of aliases.
4453 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4456 * gimplify.c (gimplify_expr): Use create_tmp_var
4457 instead of create_tmp_var_raw. If cond doesn't have
4458 integral type, don't add the IFN_ANNOTATE builtin at all.
4460 2014-01-17 Martin Jambor <mjambor@suse.cz>
4463 * ipa-cp.c (prev_edge_clone): New variable.
4464 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4465 Also resize prev_edge_clone vector.
4466 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4467 (ipcp_edge_removal_hook): New function.
4468 (ipcp_driver): Register ipcp_edge_removal_hook.
4470 2014-01-17 Andrew Pinski <apinski@cavium.com>
4471 Steve Ellcey <sellcey@mips.com>
4474 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4477 2014-01-17 Jeff Law <law@redhat.com>
4480 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4481 so that pass_ccp runs first.
4483 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4485 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4486 (ix86_adjust_cost): Use !TARGET_XXX.
4487 (do_reorder_for_imul): Likewise.
4488 (swap_top_of_ready_list): Likewise.
4489 (ix86_sched_reorder): Likewise.
4491 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4493 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4494 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4495 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4496 (intel_memset): New. Duplicate slm_memset.
4497 (intel_cost): New. Duplicate slm_cost.
4498 (m_INTEL): New macro.
4499 (processor_target_table): Add "intel".
4500 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4501 with PROCESSOR_INTEL for "intel".
4502 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4503 PROCESSOR_SILVERMONT.
4504 (ix86_issue_rate): Likewise.
4505 (ix86_adjust_cost): Likewise.
4506 (ia32_multipass_dfa_lookahead): Likewise.
4507 (swap_top_of_ready_list): Likewise.
4508 (ix86_sched_reorder): Likewise.
4509 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4510 instead of TARGET_OPT_AGU.
4511 * config/i386/i386.h (TARGET_INTEL): New.
4512 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4513 (processor_type): Add PROCESSOR_INTEL.
4514 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4515 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4517 2014-01-17 Marek Polacek <polacek@redhat.com>
4520 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4523 2014-01-17 Richard Biener <rguenther@suse.de>
4525 PR tree-optimization/46590
4526 * opts.c (default_options_table): Add entries for
4527 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4528 all enabled at -O1 but not for -Og.
4529 * common.opt (fbranch-count-reg): Remove Init(1).
4530 (fmove-loop-invariants): Likewise.
4531 (ftree-pta): Likewise.
4533 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4535 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4536 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4537 decls to at least the GCC 4.8 used alignments.
4540 * tree-nested.c (convert_nonlocal_reference_stmt,
4541 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4542 of GIMPLE_BIND stmts, adjust associated decls.
4544 2014-01-17 Richard Biener <rguenther@suse.de>
4546 PR tree-optimization/46590
4547 * vec.h (vec<>::bseach): New member function implementing
4548 binary search according to C89 bsearch.
4549 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4550 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4551 bitmap pointer again. Make accesses_in_loop a flat array.
4552 (mem_ref_obstack): New global.
4553 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4554 (mark_ref_stored): Likewise.
4555 (ref_indep_loop_p_2): Likewise.
4556 (set_ref_stored_in_loop): New helper function.
4557 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4558 (memref_free): Adjust.
4559 (record_mem_ref_loc): Simplify.
4560 (gather_mem_refs_stmt): Adjust.
4561 (sort_locs_in_loop_postorder_cmp): New function.
4562 (analyze_memory_references): Sort accesses_in_loop after
4563 loop postorder number.
4564 (find_ref_loc_in_loop_cmp): New function.
4565 (for_all_locs_in_loop): Find relevant cluster of locs in
4566 accesses_in_loop and iterate without recursion.
4567 (execute_sm): Avoid uninit warning.
4568 (struct ref_always_accessed): Simplify.
4569 (ref_always_accessed::operator ()): Likewise.
4570 (ref_always_accessed_p): Likewise.
4571 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4572 loop postorder numbers here.
4573 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4576 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4579 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4580 on decls for which assemble_alias has been called.
4582 2014-01-17 Nick Clifton <nickc@redhat.com>
4584 * config/msp430/msp430.opt: (mcpu): New option.
4585 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4586 (msp430_option_override): Parse target_cpu. If the MCU name
4587 matches a generic string, clear target_mcu.
4588 (msp430_attr): Allow numeric interrupt values up to 63.
4589 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4590 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4592 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4594 * config/msp430/msp430.md (popm): Use %J rather than %I.
4595 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4596 (addhi_cy_i): Use immediate_operand for operand 2.
4597 * doc/invoke.texi: Document -mcpu option.
4599 2014-01-17 Richard Biener <rguenther@suse.de>
4601 PR rtl-optimization/38518
4602 * df.h (df_analyze_loop): Declare.
4603 * df-core.c: Include cfgloop.h.
4604 (df_analyze_1): Split out main part of df_analyze.
4605 (df_analyze): Adjust.
4606 (loop_inverted_post_order_compute): New function.
4607 (loop_post_order_compute): Likewise.
4608 (df_analyze_loop): New function avoiding whole-function
4610 * loop-invariant.c (find_defs): Use df_analyze_loop.
4611 (find_invariants): Adjust.
4612 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4614 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4616 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4617 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4619 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4621 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4622 traversal when removing references.
4624 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4627 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4629 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4632 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4633 pushing a reload for an autoinc when we had previously reloaded an
4634 inner part of the address.
4636 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4638 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4640 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4641 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4642 when not giving up or versioning for alias only because of
4644 (vect_analyze_data_ref_dependences): Set to true.
4645 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4647 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4648 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4652 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4655 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4656 operand 0 predicate for gathers, use a new pseudo as subtarget.
4658 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4661 * lra-constraints.c (process_alt_operands): Add printing debug info.
4662 Check absence of input/output reloads for matched operands too.
4664 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4666 PR rtl-optimization/59835
4667 * ira.c (ira_init_register_move_cost): Increase cost for
4670 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4672 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4674 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4677 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4678 non-register objects. Use gen_(high/low)part more consistently.
4681 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4684 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4685 endian support, remove tests for WORDS_BIG_ENDIAN.
4686 (p8_mfvsrd_3_<mode>): Likewise.
4687 (reload_gpr_from_vsx<mode>): Likewise.
4688 (reload_gpr_from_vsxsf): Likewise.
4689 (p8_mfvsrd_4_disf): Likewise.
4691 2014-01-16 Richard Biener <rguenther@suse.de>
4693 PR rtl-optimization/46590
4694 * lcm.c (compute_antinout_edge): Use postorder iteration.
4695 (compute_laterin): Use inverted postorder iteration.
4697 2014-01-16 Nick Clifton <nickc@redhat.com>
4700 * varasm.c (output_constant): Return the number of bytes actually
4702 (output_constructor_array_range): Update the field size with the
4703 number of bytes emitted by output_constant.
4704 (output_constructor_regular_field): Likewise. Also do not
4705 complain if the total number of bytes emitted is now greater
4706 than the expected fieldpos.
4707 * output.h (output_constant): Update prototype and descriptive comment.
4709 2014-01-16 Marek Polacek <polacek@redhat.com>
4712 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4713 it is error_mark_node.
4715 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4717 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4718 VALID_AVX256_REG_OR_OI_MODE.
4720 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4722 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4723 current procedure should be profiled.
4725 2014-01-15 Andrew Pinski <apinski@cavium.com>
4727 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4728 of moving from/to the STACK_REG register class.
4730 2014-01-15 Richard Henderson <rth@redhat.com>
4733 * reginfo.c (global_regs_decl): Globalize.
4734 * rtl.h (global_regs_decl): Declare.
4735 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4736 reserved via global_regs.
4738 2014-01-15 Teresa Johnson <tejohnson@google.com>
4740 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4742 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4744 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4745 and vmulosh rather than call gen_vec_widen_smult_*.
4746 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4747 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4748 (vec_widen_smult_even_v16qi): Likewise.
4749 (vec_widen_umult_even_v8hi): Likewise.
4750 (vec_widen_smult_even_v8hi): Likewise.
4751 (vec_widen_umult_odd_v16qi): Likewise.
4752 (vec_widen_smult_odd_v16qi): Likewise.
4753 (vec_widen_umult_odd_v8hi): Likewise.
4754 (vec_widen_smult_odd_v8hi): Likewise.
4755 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4756 vmuloub rather than call gen_vec_widen_umult_*.
4757 (vec_widen_umult_lo_v16qi): Likewise.
4758 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4759 vmulosb rather than call gen_vec_widen_smult_*.
4760 (vec_widen_smult_lo_v16qi): Likewise.
4761 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4762 rather than call gen_vec_widen_umult_*.
4763 (vec_widen_umult_lo_v8hi): Likewise.
4764 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4765 rather than call gen_vec_widen_smult_*.
4766 (vec_widen_smult_lo_v8hi): Likewise.
4768 2014-01-15 Jeff Law <law@redhat.com>
4770 PR tree-optimization/59747
4771 * ree.c (find_and_remove_re): Properly handle case where a second
4772 eliminated extension requires widening a copy created for elimination
4773 of a prior extension.
4774 (combine_set_extension): Ensure that the number of hard regs needed
4775 for a destination register does not change when we widen it.
4777 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4779 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4780 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4781 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4782 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4783 (avr-*-rtems*): Likewise.
4784 (bfin*-rtems*): Likewise.
4785 (moxie-*-rtems*): Likewise.
4786 (h8300-*-rtems*): Likewise.
4787 (i[34567]86-*-rtems*): Likewise.
4788 (lm32-*-rtems*): Likewise.
4789 (m32r-*-rtems*): Likewise.
4790 (m68k-*-rtems*): Likewise.
4791 (microblaze*-*-rtems*): Likewise.
4792 (mips*-*-rtems*): Likewise.
4793 (powerpc-*-rtems*): Likewise.
4794 (sh-*-rtems*): Likewise.
4795 (sparc-*-rtems*): Likewise.
4796 (sparc64-*-rtems*): Likewise.
4797 (v850-*-rtems*): Likewise.
4798 (m32c-*-rtems*): Likewise.
4800 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4802 PR rtl-optimization/59511
4803 * ira.c (ira_init_register_move_cost): Use memory costs for some
4804 cases of register move cost calculations.
4805 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4806 instead of BB frequency.
4807 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4808 * lra-assigns.c (find_hard_regno_for): Ditto.
4810 2014-01-15 Richard Biener <rguenther@suse.de>
4812 PR tree-optimization/59822
4813 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4814 (vectorizable_load): Use it to hoist defs of uses of invariant
4815 loads out of the loop.
4817 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4818 Kugan Vivekanandarajah <kuganv@linaro.org>
4821 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4824 2014-01-15 Richard Biener <rguenther@suse.de>
4826 PR rtl-optimization/59802
4827 * lcm.c (compute_available): Use inverted postorder to seed
4828 the initial worklist.
4830 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4833 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4834 ADDR_REGS for invalid symrefs in non-PIC code.
4836 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4839 * builtins.c (determine_block_size): Initialize *probable_max_size
4840 even if len_rtx is CONST_INT.
4842 2014-01-14 Andrew Pinski <apinski@cavium.com>
4844 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4845 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4846 (cortexa53_tunings): Likewise.
4847 (aarch64_sched_issue_rate): New function.
4848 (TARGET_SCHED_ISSUE_RATE): Define.
4850 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4852 * ira-costs.c (find_costs_and_classes): Add missed
4853 ira_init_register_move_cost_if_necessary.
4855 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4858 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4860 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4863 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4864 to indicate if type is used for function return value. Warn ABI
4865 change if the vector mode isn't available for function return value.
4866 (ix86_function_arg_advance): Pass false to type_natural_mode.
4867 (ix86_function_arg): Likewise.
4868 (ix86_gimplify_va_arg): Likewise.
4869 (function_arg_32): Don't warn ABI change.
4870 (ix86_function_value): Pass true to type_natural_mode.
4871 (ix86_return_in_memory): Likewise.
4872 (ix86_struct_value_rtx): Removed.
4873 (TARGET_STRUCT_VALUE_RTX): Likewise.
4875 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4877 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4878 converting a conditional jump into a conditional return.
4880 2014-01-14 Richard Biener <rguenther@suse.de>
4882 PR tree-optimization/58921
4883 PR tree-optimization/59006
4884 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4885 hoisting invariant stmts.
4886 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4887 invariant loads on the preheader edge if possible.
4889 2014-01-14 Joey Ye <joey.ye@arm.com>
4891 * doc/plugin.texi (Building GCC plugins): Update to C++.
4893 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4895 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4896 (_mm_rcp28_round_ss): Ditto.
4897 (_mm_rsqrt28_round_sd): Ditto.
4898 (_mm_rsqrt28_round_ss): Ditto.
4899 (_mm_rcp28_sd): Ditto.
4900 (_mm_rcp28_ss): Ditto.
4901 (_mm_rsqrt28_sd): Ditto.
4902 (_mm_rsqrt28_ss): Ditto.
4903 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4904 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4905 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4906 (IX86_BUILTIN_RCP28SD): Ditto.
4907 (IX86_BUILTIN_RCP28SS): Ditto.
4908 (IX86_BUILTIN_RSQRT28SD): Ditto.
4909 (IX86_BUILTIN_RSQRT28SS): Ditto.
4910 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4911 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4912 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4913 (ix86_expand_special_args_builtin): Expand new FTYPE.
4914 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4915 (srcp14<mode>): Make insn unary.
4916 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4917 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4918 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4919 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4920 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4921 Fix rounding: make it SAE only.
4922 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4924 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4926 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4927 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4928 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4929 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4930 (round_saeonly_mask_scalar_operand4): Ditto.
4931 (round_saeonly_mask_scalar_op3): Ditto.
4932 (round_saeonly_mask_scalar_op4): Ditto.
4934 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4936 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4937 Implement -maltivec=be for vec_insert and vec_extract.
4939 2014-01-10 DJ Delorie <dj@redhat.com>
4941 * config/msp430/msp430.md (call_internal): Don't allow memory
4942 references with SP as the base register.
4943 (call_value_internal): Likewise.
4944 * config/msp430/constraints.md (Yc): New. For memory references
4945 that don't use SP as a base register.
4947 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4948 "an integer without a # prefix"
4949 * config/msp430/msp430.md (epilogue_helper): Use it.
4951 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4954 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4955 AVX512F gather builtins.
4956 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4957 on gather decls with INTEGER_TYPE masktype.
4958 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4959 directly into the builtin rather than hoisting it before loop.
4961 PR tree-optimization/59387
4962 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4963 (scev_const_prop): If folded_casts and type has undefined overflow,
4964 use force_gimple_operand instead of force_gimple_operand_gsi and
4965 for each added stmt if it is assign with
4966 arith_code_with_undefined_signed_overflow, call
4967 rewrite_to_defined_overflow.
4968 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4969 gimple-fold.h instead.
4970 (arith_code_with_undefined_signed_overflow,
4971 rewrite_to_defined_overflow): Moved to ...
4972 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4973 rewrite_to_defined_overflow): ... here. No longer static.
4974 Include gimplify-me.h.
4975 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4976 rewrite_to_defined_overflow): New prototypes.
4978 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4980 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4982 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4984 * builtins.c (get_object_alignment_2): Minor tweak.
4985 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4987 2014-01-13 Christian Bruel <christian.bruel@st.com>
4989 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4990 optimized non constant lengths.
4992 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4995 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4996 load as __atomic_load_N if possible.
4998 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5000 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5002 (rs6000_expand_builtin): Adjust call.
5004 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5007 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5008 * config/rs6000/rs6000.c: Include target-globals.h.
5009 (rs6000_set_current_function): Instead of doing target_reinit
5010 unconditionally, use save_target_globals_default_opts and
5011 restore_target_globals.
5013 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5015 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5016 (rs6000_expand_builtin): Handle mffs and mtfsf.
5017 (rs6000_init_builtins): Define mffs and mtfsf.
5018 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5019 (rs6000_mffs): New pattern.
5020 (rs6000_mtfsf): New pattern.
5022 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5024 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5025 Start narrowing with START. Apply candidate-use pair
5026 and check overall cost in narrowing.
5027 (iv_ca_prune): Pass new argument.
5029 2014-01-10 Jeff Law <law@redhat.com>
5032 * ree.c (combine_reaching_defs): Ensure the defining statement
5033 occurs before the extension when optimizing extensions with
5034 different source and destination hard registers.
5036 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5039 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5040 vtables into the type inheritance graph.
5042 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5044 PR rtl-optimization/59754
5045 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5046 modes in the REGNO != REGNO case.
5048 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5050 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5052 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5054 PR tree-optimization/59745
5055 * tree-predcom.c (tree_predictive_commoning_loop): Call
5056 free_affine_expand_cache if giving up because components is NULL.
5058 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5059 GC in payload of target_globals struct instead of allocating them on
5060 the heap and the larger structs separately using GC.
5061 * target-globals.h (struct target_globals): Make regs, hard_regs,
5062 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5063 of GTY((skip)) and change type to void *.
5064 (reset_target_globals): Cast loads from those fields to corresponding
5067 2014-01-10 Steve Ellcey <sellcey@mips.com>
5070 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5071 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5072 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5074 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5077 * aarch64-modes.def (CC_Zmode): New flags mode.
5078 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5079 represents an equality.
5080 (aarch64_get_condition_code): Handle CC_Zmode.
5081 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5083 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5085 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5086 extraction in good case.
5088 2014-01-10 Richard Biener <rguenther@suse.de>
5090 PR tree-optimization/59374
5091 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5092 checking after SLP discovery. Mark stmts not participating
5093 in any SLP instance properly.
5095 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5097 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5098 when handling a SET rtx.
5100 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5102 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5103 (cortex-a57): Likewise.
5104 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5106 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5108 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5109 non-iwmmxt builtins.
5111 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5115 * ipa-devirt.c record_target_from_binfo): Take as argument
5116 stack of binfos and lookup matching one for virtual inheritance.
5117 (possible_polymorphic_call_targets_1): Update.
5119 2014-01-10 Huacai Chen <chenhc@lemote.com>
5121 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5122 kernel strings for Loongson-2E/2F/3A.
5124 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5127 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5128 is_gimple_call before calling gimple_call_internal_p.
5130 2014-01-09 Steve Ellcey <sellcey@mips.com>
5132 * Makefile.in (TREE_FLOW_H): Remove.
5133 (TREE_SSA_H): Add file names from tree-flow.h.
5134 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5135 * tree.h: Remove tree-flow.h reference.
5136 * hash-table.h: Remove tree-flow.h reference.
5137 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5138 reference with tree-ssa-loop.h.
5140 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5142 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5143 default element-order behavior for -maltivec.
5144 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5145 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5146 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5147 when targeting big endian, at least for now.
5148 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5150 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5153 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5154 var satisfies use_register_for_decl, just take into account type
5155 alignment, rather than decl alignment.
5157 PR tree-optimization/59622
5158 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5159 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5160 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5161 Don't devirtualize for inplace at all. For targets.length () == 1,
5162 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5164 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5166 * config/i386/i386.md (cpu): Remove the unused btver1.
5168 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5170 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5172 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5175 * tree-core.h (struct target_globals): New forward declaration.
5176 (struct tree_target_option): Add globals field.
5177 * tree.h (TREE_TARGET_GLOBALS): Define.
5178 (prepare_target_option_nodes_for_pch): New prototype.
5179 * target-globals.h (struct target_globals): Define even if
5181 * tree.c (prepare_target_option_node_for_pch,
5182 prepare_target_option_nodes_for_pch): New functions.
5183 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5184 * config/i386/i386.c: Include target-globals.h.
5185 (ix86_set_current_function): Instead of doing target_reinit
5186 unconditionally, use save_target_globals_default_opts and
5187 restore_target_globals.
5189 2014-01-09 Richard Biener <rguenther@suse.de>
5191 PR tree-optimization/59715
5192 * tree-cfg.h (split_critical_edges): Declare.
5193 * tree-cfg.c (split_critical_edges): Export.
5194 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5196 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5198 * cfgexpand.c (expand_stack_vars): Optionally disable
5199 asan stack protection.
5200 (expand_used_vars): Likewise.
5201 (partition_stack_vars): Likewise.
5202 * asan.c (asan_emit_stack_protection): Optionally disable
5203 after return stack usage.
5204 (instrument_derefs): Optionally disable memory access instrumentation.
5205 (instrument_builtin_call): Likewise.
5206 (instrument_strlen_call): Likewise.
5207 (asan_protect_global): Optionally disable global variables protection.
5208 * doc/invoke.texi: Added doc for new options.
5209 * params.def: Added new options.
5210 * params.h: Likewise.
5212 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5214 PR rtl-optimization/59724
5215 * ifcvt.c (cond_exec_process_if_block): Don't call
5216 flow_find_head_matching_sequence with 0 longest_match.
5217 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5218 non-active insns if !stop_after.
5219 (try_head_merge_bb): Revert 2014-01-07 changes.
5221 2014-01-08 Jeff Law <law@redhat.com>
5223 * ree.c (get_sub_rtx): New function, extracted from...
5224 (merge_def_and_ext): Here.
5225 (combine_reaching_defs): Use get_sub_rtx.
5227 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5229 * cgraph.h (varpool_variable_node): Do not choke on null node.
5231 2014-01-08 Catherine Moore <clm@codesourcery.com>
5233 * config/mips/mips.md (simple_return): Attempt to use JRC
5235 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5237 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5239 PR rtl-optimization/59137
5240 * reorg.c (steal_delay_list_from_target): Call update_block for
5242 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5244 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5246 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5247 two duplicate entries.
5249 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5252 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5254 * config/mips/mips.c (mips_truncated_op_cost): New function.
5255 (mips_rtx_costs): Adjust test for BADDU.
5256 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5258 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5260 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5261 (*baddu_si): ...this new pattern.
5263 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5266 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5268 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5271 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5273 (expand_expr, expand_normal): Adjust.
5274 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5275 inner_reference_p. Use inner_reference_p to expand inner references.
5276 (store_expr): Adjust.
5277 * cfgexpand.c (expand_call_stmt): Adjust.
5279 2014-01-08 Rong Xu <xur@google.com>
5281 * gcov-io.c (gcov_var): Move from gcov-io.h.
5282 (gcov_position): Ditto.
5283 (gcov_is_error): Ditto.
5284 (gcov_rewrite): Ditto.
5285 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5286 only part to libgcc/libgcov.h.
5288 2014-01-08 Marek Polacek <polacek@redhat.com>
5291 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5293 2014-01-08 Marek Polacek <polacek@redhat.com>
5296 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5298 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5300 PR rtl-optimization/59649
5301 * stor-layout.c (get_mode_bounds): For BImode return
5302 0 and STORE_FLAG_VALUE.
5304 2014-01-08 Richard Biener <rguenther@suse.de>
5307 * gimple.h (is_gimple_builtin_call): Remove.
5308 (gimple_builtin_call_types_compatible_p): New.
5309 (gimple_call_builtin_p): New overload.
5310 * gimple.c (is_gimple_builtin_call): Remove.
5311 (validate_call): Rename to ...
5312 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5314 (validate_type): New static function.
5315 (gimple_call_builtin_p): New overload and adjust.
5316 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5317 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5318 (gimple_fold_stmt_to_constant_1): Likewise.
5319 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5321 2014-01-08 Richard Biener <rguenther@suse.de>
5324 * gimplify.c (gimplify_expr): Gimplify register-register type
5325 VIEW_CONVERT_EXPRs to separate stmts.
5327 2014-01-07 Jeff Law <law@redhat.com>
5330 * ree.c (combine_set_extension): Handle case where source
5331 and destination registers in an extension insn are different.
5332 (combine_reaching_defs): Allow source and destination registers
5333 in extension to be different under limited circumstances.
5334 (add_removable_extension): Remove restriction that the
5335 source and destination registers in the extension are the same.
5336 (find_and_remove_re): Emit a copy from the extension's
5337 destination to its source after the defining insn if
5338 the source and destination registers are different.
5341 * ifcvt.c (merge_if_block): If we are merging a block with more than
5342 one successor with a block with no successors, remove any BARRIER
5343 after the second block.
5345 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5347 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5349 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5352 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5353 for 14-bit register offsets when INT14_OK_STRICT is false.
5355 2014-01-07 Roland Stigge <stigge@antcom.de>
5356 Michael Meissner <meissner@linux.vnet.ibm.com>
5359 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5360 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5362 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5364 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5367 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5369 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5370 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5371 rtx is const0_rtx or not.
5373 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5376 * target-globals.c (save_target_globals): Remove this_fn_optab
5378 * toplev.c: Include optabs.h.
5379 (target_reinit): Temporarily restore the global options if another
5380 set of options are in force.
5382 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5384 PR rtl-optimization/58668
5385 * cfgcleanup.c (flow_find_cross_jump): Don't count
5386 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5387 to determine what is counted.
5388 (flow_find_head_matching_sequence): Use active_insn_p to determine
5390 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5392 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5393 determine what is counted.
5395 PR tree-optimization/59643
5396 * tree-predcom.c (split_data_refs_to_components): If one dr is
5397 read and one write, determine_offset fails and the write isn't
5398 in the bad component, just put the read into the bad component.
5400 2014-01-07 Mike Stump <mikestump@comcast.net>
5401 Jakub Jelinek <jakub@redhat.com>
5404 * tree-core.h (struct tree_optimization_option): Change optabs
5405 type from unsigned char * to void *.
5406 * optabs.c (init_tree_optimization_optabs): Adjust
5407 TREE_OPTIMIZATION_OPTABS initialization.
5409 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5412 * config/i386/i386.h (struct machine_function): Add
5413 no_drap_save_restore field.
5414 * config/i386/i386.c (ix86_save_reg): Use
5415 !cfun->machine->no_drap_save_restore instead of
5416 crtl->stack_realign_needed.
5417 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5418 this function clears frame_pointer_needed. Set
5419 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5420 and DRAP reg is needed.
5422 2014-01-06 Marek Polacek <polacek@redhat.com>
5425 * doc/implement-c.texi: Mention that other integer types are
5426 permitted as bit-field types in strictly conforming mode.
5428 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5430 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5433 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5435 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5437 2014-01-06 Martin Jambor <mjambor@suse.cz>
5440 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5442 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5444 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5448 * var-tracking.c (add_stores): Preserve the value of the source even if
5449 we don't record the store.
5451 2014-01-06 Terry Guo <terry.guo@arm.com>
5453 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5455 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5458 * config/darwin.c (darwin_function_section): Adjust return values to
5459 correspond to optimisation changes made in r206070.
5461 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5463 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5464 from prefetch_block tune setting.
5465 (nocona_cost): Correct size of prefetch block to 64.
5467 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5469 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5470 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5471 used to save the static chain register in the computation of the offset
5472 from which the FP registers need to be restored.
5474 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5476 PR tree-optimization/59519
5477 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5478 ICE if get_current_def (current_new_name) is already non-NULL, as long
5479 as it is a phi result of some other phi in *new_exit_bb that has
5482 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5483 or vmovdqu* for misaligned_operand.
5484 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5485 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5486 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5487 aligned_mem for AVX512F masked aligned load and store builtins and for
5490 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5492 PR tree-optimization/59651
5493 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5494 Address range for negative step should be added by TYPE_SIZE_UNIT.
5496 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5498 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5499 overlapping registers also for registers other than the stack pointer.
5501 2014-01-03 Marek Polacek <polacek@redhat.com>
5504 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5507 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5510 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5513 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5514 (push splitter): Use <P:MODE_SIZE> instead of
5515 GET_MODE_SIZE (<P:MODE>mode).
5516 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5517 (mov -1, reg peephole2): Likewise.
5518 * config/i386/sse.md (*mov<mode>_internal,
5519 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5520 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5521 *<code><mode>3, *andnot<mode>3<mask_name>,
5522 <mask_codefor><code><mode>3<mask_name>): Likewise.
5523 * config/i386/subst.md (mask_mode512bit_condition,
5524 sd_mask_mode512bit_condition): Likewise.
5526 2014-01-02 Xinliang David Li <davidxl@google.com>
5528 PR tree-optimization/59303
5529 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5530 (dump_predicates): Better output format.
5531 (pred_equal_p): New function.
5532 (is_neq_relop_p): Ditto.
5533 (is_neq_zero_form_p): Ditto.
5534 (pred_expr_equal_p): Ditto.
5535 (pred_neg_p): Ditto.
5536 (simplify_pred): Ditto.
5537 (simplify_preds_2): Ditto.
5538 (simplify_preds_3): Ditto.
5539 (simplify_preds_4): Ditto.
5540 (simplify_preds): Ditto.
5542 (push_to_worklist): Ditto.
5543 (get_pred_info_from_cmp): Ditto.
5544 (is_degenerated_phi): Ditto.
5545 (normalize_one_pred_1): Ditto.
5546 (normalize_one_pred): Ditto.
5547 (normalize_one_pred_chain): Ditto.
5548 (normalize_preds): Ditto.
5549 (normalize_cond_1): Remove function.
5550 (normalize_cond): Ditto.
5551 (is_gcond_subset_of): Ditto.
5552 (is_subset_of_any): Ditto.
5553 (is_or_set_subset_of): Ditto.
5554 (is_and_set_subset_of): Ditto.
5555 (is_norm_cond_subset_of): Ditto.
5556 (pred_chain_length_cmp): Ditto.
5557 (convert_control_dep_chain_into_preds): Type change.
5558 (find_predicates): Ditto.
5559 (find_def_preds): Ditto.
5560 (destroy_predicates_vecs): Ditto.
5561 (find_matching_predicates_in_rest_chains): Ditto.
5562 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5563 (is_pred_expr_subset): Ditto.
5564 (is_pred_chain_subset_of): Ditto.
5565 (is_included_in): Ditto.
5566 (is_superset_of): Ditto.
5568 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5570 Update copyright years.
5572 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5574 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5575 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5576 config/arc/arc.md, config/arc/arc.opt,
5577 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5578 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5579 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5580 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5581 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5582 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5583 vtable-verify.c, vtable-verify.h: Use the standard form for the
5586 2014-01-02 Tobias Burnus <burnus@net-b.de>
5588 * gcc.c (process_command): Update copyright notice dates.
5589 * gcov-dump.c: Ditto.
5591 * doc/cpp.texi: Bump @copying's copyright year.
5592 * doc/cppinternals.texi: Ditto.
5593 * doc/gcc.texi: Ditto.
5594 * doc/gccint.texi: Ditto.
5595 * doc/gcov.texi: Ditto.
5596 * doc/install.texi: Ditto.
5597 * doc/invoke.texi: Ditto.
5599 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5601 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5603 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5605 * config/i386/sse.md (*mov<mode>_internal): Guard
5606 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5608 PR rtl-optimization/59647
5609 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5610 new_rtx into UNSIGNED_FLOAT rtxes.
5612 Copyright (C) 2014 Free Software Foundation, Inc.
5614 Copying and distribution of this file, with or without modification,
5615 are permitted in any medium without royalty provided the copyright
5616 notice and this notice are preserved.