1 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
3 * doc/generic.texi: Correct typos.
5 2014-03-24 Tobias Burnus <burnus@net-b.de>
7 * doc/invoke.texi (-flto): Expand section about
8 using static libraries with LTO.
10 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
12 PR rtl-optimization/60501
13 * optabs.def (addptr3_optab): New optab.
14 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
15 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
16 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
18 * lra.c (emit_add3_insn): Use the addptr pattern if available.
20 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
22 2014-03-24 Ulrich Drepper <drepper@gmail.com>
24 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
27 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
28 (_mm256_undefined_ps): Define.
29 (_mm256_undefined_pd): Define.
30 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
31 (_mm_undefined_pd): Define.
32 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
33 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
34 (_mm512_undefined_ps): Define.
35 (_mm512_undefined_pd): Define.
37 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
39 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
41 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
42 (lshr_simd): DI mode added.
43 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
44 (aarch64_ushr_simddi): Likewise.
45 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
46 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
47 (vshrd_n_u64): Likewise.
49 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
51 * Makefile.in (s-macro_list): Depend on cc1.
53 2014-03-23 Teresa Johnson <tejohnson@google.com>
55 * ipa-utils.c (ipa_print_order): Use specified dump file.
57 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
59 PR rtl-optimization/60601
60 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
62 * gcc.c (eval_spec_function): Initialize save_growing_value.
64 2014-03-22 Jakub Jelinek <jakub@redhat.com>
67 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
68 code == MINUS_EXPR, never swap op0 with op1.
70 * toplev.c (init_local_tick): Avoid signed integer multiplication
72 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
73 shift by first operand's bitsize.
75 2014-03-21 Jakub Jelinek <jakub@redhat.com>
78 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
80 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
81 TARGET_ISA_64BIT_P(x).
83 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
85 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
86 pattern for vector nor instead of subtract from splat(-1).
87 (altivec_expand_vec_perm_const_le): Likewise.
89 2014-03-21 Richard Henderson <rth@twiddle.net>
92 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
93 related insns after epilogue_completed.
95 2014-03-21 Martin Jambor <mjambor@suse.cz>
98 * cgraph.h (symtab_node): New flag body_removed.
99 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
100 when removing bodies.
101 * symtab.c (dump_symtab_base): Dump body_removed flag.
102 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
103 had their bodies removed.
105 2014-03-21 Martin Jambor <mjambor@suse.cz>
108 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
111 2014-03-21 Richard Biener <rguenther@suse.de>
113 PR tree-optimization/60577
114 * tree-core.h (struct tree_base): Document nothrow_flag use
115 in VAR_DECL_NONALIASED.
116 * tree.h (VAR_DECL_NONALIASED): New.
117 (may_be_aliased): Adjust.
118 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
120 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
122 * expr.c (expand_expr_real_1): Remove outdated comment.
124 2014-03-20 Jakub Jelinek <jakub@redhat.com>
127 * ira.c (adjust_cleared_regs): Call copy_rtx on
128 *reg_equiv[REGNO (loc)].src_p before passing it to
129 simplify_replace_fn_rtx.
132 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
133 into CONST, put pic register as first operand of PLUS. Use
134 gen_const_mem for both 32-bit and 64-bit PIC got loads.
136 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
138 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
140 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
142 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
143 around for store forwarding issue in the FPU on the UT699.
144 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
145 loads and operations if -mfix-ut699 is specified.
146 (divtf3_hq): Tweak attribute.
147 (sqrttf2_hq): Likewise.
149 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
151 * calls.c (store_one_arg): Remove incorrect const qualification on the
152 type of the temporary.
153 * cfgexpand.c (expand_return): Likewise.
154 * expr.c (expand_constructor): Likewise.
155 (expand_expr_real_1): Likewise.
157 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
159 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
162 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
165 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
167 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
169 * config/arm/aarch-common-protos.h
170 (alu_cost_table): Fix spelling of "extend".
171 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
173 2014-03-19 Richard Biener <rguenther@suse.de>
176 * tree-core.h (tree_type_common): Re-order pointer members
177 to reduce recursion depth during GC walks.
179 2014-03-19 Marek Polacek <polacek@redhat.com>
182 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
185 2014-03-19 Richard Biener <rguenther@suse.de>
188 * lto-streamer-in.c (input_function): In WPA stage do not drop
191 2014-03-19 Jakub Jelinek <jakub@redhat.com>
193 PR tree-optimization/60559
194 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
195 with build_zero_cst assignment.
197 2014-03-18 Kai Tietz <ktietz@redhat.com>
199 PR rtl-optimization/56356
200 * sdbout.c (sdbout_parms): Verify that parms'
201 incoming argument is valid.
202 (sdbout_reg_parms): Likewise.
204 2014-03-18 Richard Henderson <rth@redhat.com>
207 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
208 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
209 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
211 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
213 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
214 * doc/plugins.texi (Plugin callbacks): Mention
216 Italicize plugin event names in description. Explain that
217 PLUGIN_PRAGMAS has no sense for lto1. Explain
219 Remind that no GCC functions should be called after
221 Explain what pragmas with expansion are.
223 2014-03-18 Martin Liska <mliska@suse.cz>
225 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
226 gimple call statement is update.
227 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
228 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
230 2014-03-18 Jakub Jelinek <jakub@redhat.com>
233 * ubsan.c (ubsan_instrument_unreachable): Call
234 initialize_sanitizer_builtins.
235 (ubsan_pass): Likewise.
238 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
239 varpool_finalize_decl instead of rest_of_decl_compilation.
241 2014-03-18 Richard Biener <rguenther@suse.de>
243 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
244 by using bitmap_and_compl instead of bitmap_and_compl_into.
245 (df_rd_transfer_function): Likewise.
247 2014-03-18 Richard Biener <rguenther@suse.de>
249 * doc/lto.texi (fresolution): Fix typo.
251 2014-03-18 Richard Biener <rguenther@suse.de>
253 * doc/invoke.texi (flto): Update for changes in 4.9.
255 2014-03-18 Richard Biener <rguenther@suse.de>
257 * doc/loop.texi: Remove section on the removed lambda framework.
258 Update loop docs with recent changes in preserving loop structure.
260 2014-03-18 Richard Biener <rguenther@suse.de>
262 * doc/lto.texi (-fresolution): Document.
264 2014-03-18 Richard Biener <rguenther@suse.de>
266 * doc/contrib.texi: Adjust my name.
268 2014-03-18 Jakub Jelinek <jakub@redhat.com>
271 * internal-fn.c: Include diagnostic-core.h.
272 (expand_BUILTIN_EXPECT): New function.
273 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
274 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
275 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
276 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
278 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
279 Revert 3 argument __builtin_expect code.
280 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
281 * gimple-fold.c (gimple_fold_call): Likewise.
282 * tree.h (fold_builtin_expect): New prototype.
283 * builtins.c (build_builtin_expect_predicate): Add predictor
284 argument, if non-NULL, create 3 argument __builtin_expect.
285 (fold_builtin_expect): No longer static. Add ARG2 argument,
286 pass it through to build_builtin_expect_predicate.
287 (fold_builtin_2): Adjust caller.
288 (fold_builtin_3): Handle BUILT_IN_EXPECT.
289 * internal-fn.def (BUILTIN_EXPECT): New.
291 2014-03-18 Tobias Burnus <burnus@net-b.de>
294 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
295 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
296 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
298 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
301 * predict.c (combine_predictions_for_bb): Fix up formatting.
302 (expr_expected_value_1, expr_expected_value): Add predictor argument,
303 fill what it points to if non-NULL.
304 (tree_predict_by_opcode): Adjust caller, use the predictor.
305 * predict.def (PRED_COMPARE_AND_SWAP): Add.
307 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
309 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
310 proper constant for the store mode.
312 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
314 * symtab.c (change_decl_assembler_name): Fix transparent alias
317 2014-03-16 Renlin Li <Renlin.Li@arm.com>
319 * config/aarch64/aarch64.c: Correct the comments about the
320 aarch64 stack layout.
322 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
324 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
325 check for GF_OMP_FOR_KIND_FOR.
327 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
329 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
330 ymm and zmm register names.
332 2014-03-17 Jakub Jelinek <jakub@redhat.com>
335 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
336 note creation for the 2010-08-31 changes.
338 2014-03-17 Marek Polacek <polacek@redhat.com>
341 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
342 as -fno-tree-loop-vectorize.
343 (expand_omp_simd): Likewise.
345 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
347 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
348 (eligible_for_call_delay): New prototype.
349 * config/sparc/sparc.c (tls_call_delay): Rename into...
350 (eligible_for_call_delay): ...this. Return false if the instruction
351 cannot be put in the delay slot of a branch.
352 (eligible_for_restore_insn): Simplify.
353 (eligible_for_return_delay): Return false if the instruction cannot be
354 put in the delay slot of a branch and simplify.
355 (eligible_for_sibcall_delay): Return false if the instruction cannot be
356 put in the delay slot of a branch.
357 * config/sparc/sparc.md (fix_ut699): New attribute.
358 (tls_call_delay): Delete.
359 (in_call_delay): Reimplement.
360 (eligible_for_sibcall_delay): Rename into...
361 (in_sibcall_delay): ...this.
362 (eligible_for_return_delay): Rename into...
363 (in_return_delay): ...this.
364 (in_branch_delay): Reimplement.
365 (in_uncond_branch_delay): Delete.
366 (in_annul_branch_delay): Delete.
368 2014-03-14 Richard Henderson <rth@redhat.com>
371 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
372 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
373 (*floathi<X87MODEF>2_i387_with_temp): Remove.
374 (floathi splitters): Remove.
375 (float<SWI48x>xf2): New pattern.
376 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
377 code that tried to handle DImode for 32-bit, but which was excluded
378 by the pattern's condition. Drop allocation of stack temporary.
379 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
380 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
381 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
382 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
383 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
384 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
385 (*float<SWI48><MODEF>2_sse_interunit): Remove.
386 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
387 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
388 (*float<SWI48x><X87MODEF>2_i387): Remove.
389 (all float _with_temp splitters): Remove.
390 (*float<SWI48x><MODEF>2_i387): New pattern.
391 (*float<SWI48><MODEF>2_sse): New pattern.
392 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
393 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
395 2014-03-14 Jakub Jelinek <jakub@redhat.com>
396 Marek Polacek <polacek@redhat.com>
399 * common.opt (dump_base_name_prefixed): New Variable.
400 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
401 if x_dump_base_name_prefixed is already set, set it at the end.
403 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
405 PR rtl-optimization/60508
406 * lra-constraints.c (get_reload_reg): Add new parameter
408 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
409 Pass the new parameter values.
411 2014-03-14 Richard Biener <rguenther@suse.de>
413 * common.opt: Revert unintented changes from r205065.
416 2014-03-14 Richard Biener <rguenther@suse.de>
419 * cfghooks.c (split_block): Properly adjust all loops the
420 block was a latch of.
422 2014-03-14 Martin Jambor <mjambor@suse.cz>
425 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
428 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
431 * config/avr/avr.c (avr_set_current_function): Pass function name
432 through default_strip_name_encoding before sanity checking instead
433 of skipping the first char of the assembler name.
435 2014-03-13 Richard Henderson <rth@redhat.com>
438 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
439 (ix86_force_to_memory, ix86_free_from_memory): Remove.
440 * config/i386/i386-protos.h: Likewise.
441 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
442 in the expander instead of a splitter.
443 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
444 any possibility of requiring a memory.
445 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
446 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
447 (fp branch splitters): Update for ix86_split_fp_branch.
448 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
449 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
450 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
451 (*fop_<MODEF>_2_i387): Remove f/r alternative.
452 (*fop_<MODEF>_3_i387): Likewise.
453 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
454 (splitters for the fop_* register patterns): Remove.
455 (fscalexf4_i387): Rename from *fscalexf4_i387.
456 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
458 2014-03-13 Jakub Jelinek <jakub@redhat.com>
460 PR tree-optimization/59779
461 * tree-dfa.c (get_ref_base_and_extent): Use double_int
462 type for bitsize and maxsize instead of HOST_WIDE_INT.
464 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
466 PR rtl-optimization/57320
467 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
468 the CFG after thread_prologue_and_epilogue_insns.
470 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
472 PR rtl-optimization/57189
473 * lra-constraints.c (process_alt_operands): Disfavor spilling
476 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
478 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
481 2014-03-13 Jakub Jelinek <jakub@redhat.com>
483 PR tree-optimization/59025
485 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
486 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
488 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
491 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
492 calls of avr_out_plus_1.
494 2014-03-13 Bin Cheng <bin.cheng@arm.com>
496 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
497 BB's single pred and update the father loop's latch info later.
499 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
501 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
505 (VEC_base): Likewise.
506 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
507 registers, we need to swap double words in little endian mode.
509 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
510 to be a container mode for 128-bit integer operations added in ISA
511 2.07. Unlike TImode and PTImode, the preferred register set is
512 the Altivec/VMX registers for the 128-bit operations.
514 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
516 (rs6000_split_128bit_ok_p): Likewise.
518 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
519 macros for creating ISA 2.07 normal and overloaded builtin
520 functions with 3 arguments.
521 (BU_P8V_OVERLOAD_3): Likewise.
522 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
523 for use as overloaded functions.
524 (VPERM_1TI_UNS): Likewise.
525 (VSEL_1TI): Likewise.
526 (VSEL_1TI_UNS): Likewise.
527 (ST_INTERNAL_1ti): Likewise.
528 (LD_INTERNAL_1ti): Likewise.
529 (XXSEL_1TI): Likewise.
530 (XXSEL_1TI_UNS): Likewise.
531 (VPERM_1TI): Likewise.
532 (VPERM_1TI_UNS): Likewise.
533 (XXPERMDI_1TI): Likewise.
535 (LXVD2X_V1TI): Likewise.
536 (STXVD2X_V1TI): Likewise.
537 (VEC_INIT_V1TI): Likewise.
538 (VEC_SET_V1TI): Likewise.
539 (VEC_EXT_V1TI): Likewise.
540 (EQV_V1TI): Likewise.
541 (NAND_V1TI): Likewise.
542 (ORC_V1TI): Likewise.
543 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
544 added in ISA 2.07. Add both normal 'altivec' builtins, and the
548 (VADDEUQM): Likewise.
549 (VADDECUQ): Likewise.
550 (VSUBEUQM): Likewise.
551 (VSUBECUQ): Likewise.
553 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
554 __int128_t and __uint128_t types.
555 (__uint128_type): Likewise.
556 (altivec_categorize_keyword): Add support for vector __int128_t,
557 vector __uint128_t, vector __int128, and vector unsigned __int128
558 as a container type for TImode operations that need to be done in
559 VSX/Altivec registers.
560 (rs6000_macro_to_expand): Likewise.
561 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
562 to support 128-bit integer instructions vaddcuq, vadduqm,
563 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
564 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
566 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
567 for V1TImode, and set up preferences to use VSX/Altivec registers.
568 Setup VSX reload handlers.
569 (rs6000_debug_reg_global): Likewise.
570 (rs6000_init_hard_regno_mode_ok): Likewise.
571 (rs6000_preferred_simd_mode): Likewise.
572 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
573 (easy_altivec_constant): Likewise.
574 (output_vec_const_move): Likewise.
575 (rs6000_expand_vector_set): Convert V1TImode set and extract to
577 (rs6000_expand_vector_extract): Likewise.
578 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
580 (rs6000_const_vec): Add support for V1TImode.
581 (rs6000_emit_le_vsx_load): Swap double words when loading or
582 storing TImode/V1TImode.
583 (rs6000_emit_le_vsx_store): Likewise.
584 (rs6000_emit_le_vsx_move): Likewise.
585 (rs6000_emit_move): Add support for V1TImode.
586 (altivec_expand_ld_builtin): Likewise.
587 (altivec_expand_st_builtin): Likewise.
588 (altivec_expand_vec_init_builtin): Likewise.
589 (altivec_expand_builtin): Likewise.
590 (rs6000_init_builtins): Add support for V1TImode type. Add
591 support for ISA 2.07 128-bit integer builtins. Define type names
592 for the VSX/Altivec vector types.
593 (altivec_init_builtins): Add support for overloaded vector
594 functions with V1TImode type.
595 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
596 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
598 (rs6000_split_128bit_ok_p): Likewise.
599 (rs6000_handle_altivec_attribute): Create V1TImode from vector
600 __int128_t and vector __uint128_t.
602 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
610 (VS_scalar): Likewise.
611 (VS_double): Likewise.
612 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
614 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
615 we support the ISA 2.07 128-bit integer arithmetic instructions.
616 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
617 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
618 and TImode types for use with the builtin functions.
619 (V1TI_type_node): Likewise.
620 (unsigned_V1TI_type_node): Likewise.
621 (intTI_type_internal_node): Likewise.
622 (uintTI_type_internal_node): Likewise.
624 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
625 128-bit builtin functions.
626 (UNSPEC_VADDEUQM): Likewise.
627 (UNSPEC_VADDECUQ): Likewise.
628 (UNSPEC_VSUBCUQ): Likewise.
629 (UNSPEC_VSUBEUQM): Likewise.
630 (UNSPEC_VSUBECUQ): Likewise.
631 (VM): Add V1TImode to vector mode iterators.
634 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
635 (altivec_vaddcuq): Likewise.
636 (altivec_vsubuqm): Likewise.
637 (altivec_vsubcuq): Likewise.
638 (altivec_vaddeuqm): Likewise.
639 (altivec_vaddecuq): Likewise.
640 (altivec_vsubeuqm): Likewise.
641 (altivec_vsubecuq): Likewise.
643 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
645 (BOOL_128): Likewise.
646 (BOOL_REGS_OUTPUT): Likewise.
647 (BOOL_REGS_OP1): Likewise.
648 (BOOL_REGS_OP2): Likewise.
649 (BOOL_REGS_UNARY): Likewise.
650 (BOOL_REGS_AND_CR0): Likewise.
652 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
653 128-bit integer builtin support.
654 (vec_vadduqm): Likewise.
655 (vec_vaddecuq): Likewise.
656 (vec_vaddeuqm): Likewise.
657 (vec_vsubecuq): Likewise.
658 (vec_vsubeuqm): Likewise.
659 (vec_vsubcuq): Likewise.
660 (vec_vsubuqm): Likewise.
662 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
663 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
664 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
665 128-bit integer add/subtract to ISA 2.07.
667 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
669 * config/arc/arc.c (arc_predicate_delay_insns):
670 Fix third argument passed to conditionalize_nonjump.
672 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
674 * config/aarch64/aarch64-builtins.c
675 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
676 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
677 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
678 instead of __builtin_lfloor.
679 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
681 2014-03-12 Jakub Jelinek <jakub@redhat.com>
683 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
684 (tree_ssa_ifcombine_bb_1): New function.
685 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
686 is an empty forwarder block to then_bb or vice versa and then_bb
687 and else_bb are effectively swapped.
689 2014-03-12 Christian Bruel <christian.bruel@st.com>
692 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
693 REG_CFA_DEF_CFA note.
694 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
695 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
697 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
699 PR tree-optimization/60454
700 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
702 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
704 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
705 Do not define target_cpu_default2 to generic.
706 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
707 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
708 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
710 2014-03-12 Jakub Jelinek <jakub@redhat.com>
711 Marc Glisse <marc.glisse@inria.fr>
713 PR tree-optimization/60502
714 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
715 instead of build_low_bits_mask.
717 2014-03-12 Jakub Jelinek <jakub@redhat.com>
720 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
721 if there are multiple uses, but op doesn't live on E edge.
722 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
723 clobber stmts before __builtin_unreachable.
725 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
727 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
728 hard_frame_pointer_rtx.
729 * cse.c (cse_insn): Remove volatile check.
730 * cselib.c (cselib_process_insn): Likewise.
731 * dse.c (scan_insn): Likewise.
733 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
735 * config/arc/arc.c (conditionalize_nonjump): New function,
737 (arc_ifcvt): ... this.
738 (arc_predicate_delay_insns): Use it.
740 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
742 * config/arc/predicates.md (extend_operand): During/after reload,
743 allow const_int_operand.
744 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
745 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
746 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
748 (umulsi3_highpart_i): Likewise.
750 2014-03-11 Richard Biener <rguenther@suse.de>
752 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
753 Add asserts to guard possible wrong-code bugs.
755 2014-03-11 Richard Biener <rguenther@suse.de>
757 PR tree-optimization/60429
758 PR tree-optimization/60485
759 * tree-ssa-structalias.c (set_union_with_increment): Properly
760 take into account all fields that overlap the shifted vars.
761 (do_sd_constraint): Likewise.
762 (do_ds_constraint): Likewise.
763 (get_constraint_for_ptr_offset): Likewise.
765 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
767 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
768 (nios2_compute_frame_layout):
769 Add calculation of cfun->machine->fp_save_offset.
770 (nios2_expand_prologue): Correct setting of frame pointer register
772 (nios2_expand_epilogue): Update recovery of stack pointer from
773 frame pointer accordingly.
774 (nios2_initial_elimination_offset): Update calculation of offset
775 for eliminating to HARD_FRAME_POINTER_REGNUM.
777 2014-03-10 Jakub Jelinek <jakub@redhat.com>
780 * ipa.c (symtab_remove_unreachable_nodes): Don't call
781 cgraph_get_create_node on VAR_DECLs.
783 2014-03-10 Richard Biener <rguenther@suse.de>
786 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
788 2014-03-08 Douglas B Rupp <rupp@gnat.com>
790 * config/vms/vms.opt (vms_float_format): New variable.
792 2014-03-08 Tobias Burnus <burnus@net-b.de>
794 * doc/invoke.texi (-fcilkplus): Update implementation status.
796 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
797 Richard Biener <rguenther@suse.de>
799 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
800 consistently accross all TUs.
801 (run_gcc): Enable -fshort-double automatically at link at link-time
802 and disallow override.
804 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
807 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
808 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
809 if they can't be used.
811 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
813 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
814 for Solaris 11/x86 ld.
815 * configure: Regenerate.
817 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
819 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
820 (LIB_TLS_SPEC): Save as ld_tls_libs.
821 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
822 (HAVE_AS_IX86_TLSLDM): New test.
823 * configure, config.in: Regenerate.
824 * config/i386/i386.c (legitimize_tls_address): Fall back to
825 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
826 cannot support TLS_MODEL_LOCAL_DYNAMIC.
827 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
828 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
830 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
832 * common.opt (fira-loop-pressure): Mark as optimization.
834 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
836 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
837 an OpenMP mappable type.
839 2014-03-06 Matthias Klose <doko@ubuntu.com>
841 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
842 MULTILIB_OSDIRNAMES is not defined.
844 2014-03-06 Jakub Jelinek <jakub@redhat.com>
845 Meador Inge <meadori@codesourcery.com>
848 * config/arm/arm.c (arm_tls_symbol_p): Remove.
849 (arm_legitimize_address): Call legitimize_tls_address for any
850 arm_tls_referenced_p expression, handle constant addend. Call it
851 before testing for !TARGET_ARM.
852 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
854 2014-03-06 Richard Biener <rguenther@suse.de>
860 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
862 * tree-streamer.c (record_common_node): Assert we don't record
863 nodes with type double.
864 (preload_common_node): Skip type double, complex double and double
865 pointer since it is now frontend dependent due to fshort-double option.
867 2014-03-06 Richard Biener <rguenther@suse.de>
869 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
870 or -fno-lto is specified and the linker has full plugin support.
871 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
872 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
873 * lto-wrapper.c (merge_and_complain): Merge compile-time
875 (run_gcc): And pass it through to the link options.
877 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
881 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
883 * cselib.c (remove_useless_values): Skip to avoid quadratic
884 behavior if the condition moved from...
885 (cselib_process_insn): ... here holds.
887 2014-03-05 Jakub Jelinek <jakub@redhat.com>
890 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
891 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
894 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
895 (TM_H): Add x86-tune.def.
897 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
899 * config/aarch64/aarch64.c (generic_tunings):
900 Use cortexa57_extra_costs.
902 2014-03-05 Jakub Jelinek <jakub@redhat.com>
905 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
906 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
907 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
910 2014-03-04 Heiher <r@hev.cc>
912 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
913 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
915 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
917 * config/i386/predicates.md (const2356_operand): Change to ...
918 (const2367_operand): ... this.
919 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
921 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
922 (*avx512pf_scatterpf<mode>sf): Ditto.
923 (avx512pf_scatterpf<mode>df): Ditto.
924 (*avx512pf_scatterpf<mode>df_mask): Ditto.
925 (*avx512pf_scatterpf<mode>df): Ditto.
926 * config/i386/i386.c (ix86_expand_builtin): Update
927 incorrect hint operand error message.
929 2014-03-04 Richard Biener <rguenther@suse.de>
931 * lto-section-in.c (lto_get_section_data): Fix const cast.
933 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
935 * tree-streamer.c (record_common_node): Assert we don't record
936 nodes with type double.
937 (preload_common_node): Skip type double, complex double and double
938 pointer since it is now frontend dependent due to fshort-double option.
940 2014-03-04 Richard Biener <rguenther@suse.de>
943 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
944 (lto_input_toplevel_asms): Likewise.
945 * lto-section-in.c (lto_get_section_data): Instead do it here
948 2014-03-04 Richard Biener <rguenther@suse.de>
950 PR tree-optimization/60382
951 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
952 dead PHIs a reduction.
954 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
956 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
958 (_mm_prefetch): Move out of GCC target("sse") pragma.
959 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
960 GCC target("prfchw") pragma.
961 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
963 * config/i386/i386.c (ix86_option_override_internal): Enable
964 -mprfchw with -mprefetchwt1.
966 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
968 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
971 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
973 * opts.h (CL_PCH_IGNORE): Define.
974 * targhooks.c (option_affects_pch_p):
975 Return false for options that have CL_PCH_IGNORE set.
976 * opt-functions.awk: Process PchIgnore.
977 * doc/options.texi: Document PchIgnore.
979 * config/arc/arc.opt (misize): Add PchIgnore property.
981 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
983 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
984 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
985 constraint on constants to permit them being loaded into
986 GENERAL_REGS or BASE_REGS.
988 2014-03-03 Nick Clifton <nickc@redhat.com>
990 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
991 anti-cacnonical alternatives.
992 (negandhi3_real): New pattern.
993 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
995 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
997 * config/avr/avr-mcus.def: Remove atxmega16x1.
998 * config/avr/avr-tables.opt: Regenerate.
999 * config/avr/t-multilib: Regenerate.
1000 * doc/avr-mmcu.texi: Regenerate.
1002 2014-03-03 Tobias Grosser <tobias@grosser.es>
1003 Mircea Namolaru <mircea.namolaru@inria.fr>
1005 PR tree-optimization/58028
1006 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1009 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1011 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1012 not handled by recognizers.
1014 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1017 * function.c (expand_function_end): Don't emit
1018 clobber_return_register sequence if clobber_after is a BARRIER.
1019 * cfgexpand.c (construct_exit_block): Append instructions before
1020 return_label to prev_bb.
1022 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1024 * config/rs6000/constraints.md: Document reserved use of "wc".
1026 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1029 * ipa.c (function_and_variable_visibility): When dissolving comdat
1030 group, also set all symbols to local.
1032 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1037 2013-12-14 Jan Hubicka <jh@suse.cz>
1039 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1041 2014-03-02 Jon Beniston <jon@beniston.com>
1047 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1048 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1049 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1050 (simple_return, *simple_return): New patterns
1051 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1052 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1054 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1056 * dwarf2out.c (gen_subprogram_die): Tidy.
1058 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1061 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1062 (*mov_t_msb_neg_negc): ... this new insn.
1064 2014-02-28 Jason Merrill <jason@redhat.com>
1067 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1070 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1073 * dwarf2out.c (decltype_auto_die): New static.
1074 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1075 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1076 (is_cxx_auto): Likewise.
1078 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1080 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1081 we are not using general regs only.
1083 2014-02-28 Richard Biener <rguenther@suse.de>
1086 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1087 previous fix and only allow to remove trivial pre-headers
1088 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1089 (remove_forwarder_block): Properly update the latch of a loop.
1091 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1094 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1095 (cselib_preserved_hash_table): New.
1096 (preserve_constants_and_equivs): Move preserved vals to it.
1097 (cselib_find_slot): Look it up first.
1098 (cselib_init): Initialize it.
1099 (cselib_finish): Release it.
1100 (dump_cselib_table): Dump it.
1102 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1105 * cselib.c (remove_useless_values): Skip to avoid quadratic
1106 behavior if the condition moved from...
1107 (cselib_process_insn): ... here holds.
1109 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1112 * var-tracking.c (vt_initialize): Apply the same condition to
1113 preserve the CFA base value.
1115 2014-02-28 Joey Ye <joey.ye@arm.com>
1118 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1119 if reload in progress or completed.
1121 2014-02-28 Tobias Burnus <burnus@net-b.de>
1124 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1127 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1129 * doc/tm.texi.in (Condition Code Status): Update documention for
1130 relative locations of cc0-setter and cc0-user.
1132 2014-02-27 Jeff Law <law@redhat.com>
1134 PR rtl-optimization/52714
1135 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1136 into two independent simple sets, if I3 is a jump, ensure the
1137 pattern we place into I3 is a (set (pc) ...).
1139 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1140 Jeff Law <law@redhat.com>
1142 PR rtl-optimization/49847
1143 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1144 are in different blocks.
1145 * doc/tm.texi (Condition Code Status): Update documention for
1146 relative locations of cc0-setter and cc0-user.
1148 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1151 * lra.c (lra_emit_add): Check SUBREG too.
1153 2014-02-27 Andreas Schwab <schwab@suse.de>
1155 * config/m68k/m68k.c (m68k_option_override): Disable
1156 -flive-range-shrinkage for classic m68k.
1157 (m68k_override_options_after_change): Likewise.
1159 2014-02-27 Marek Polacek <polacek@redhat.com>
1162 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1163 -Wmaybe-uninitialized.
1165 2014-02-27 Alan Modra <amodra@gmail.com>
1168 * reload1.c (emit_input_reload_insns): When reload_override_in,
1169 set old to rl->in_reg when rl->in_reg is a subreg.
1171 2014-02-26 Richard Biener <rguenther@suse.de>
1174 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1176 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1178 * common/config/i386/predicates.md (const1256_operand): Remove.
1179 (const2356_operand): New.
1180 (const_1_to_2_operand): Remove.
1181 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1182 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1183 (*avx512pf_gatherpf<mode>sf): Ditto.
1184 (avx512pf_gatherpf<mode>df): Ditto.
1185 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1186 (*avx512pf_gatherpf<mode>df): Ditto.
1187 (avx512pf_scatterpf<mode>sf): Ditto.
1188 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1189 (*avx512pf_scatterpf<mode>sf): Ditto.
1190 (avx512pf_scatterpf<mode>df): Ditto.
1191 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1192 (*avx512pf_scatterpf<mode>df): Ditto.
1193 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1195 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1197 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1198 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1199 (_mm512_mask_testn_epi64_mask): Move to ...
1200 * config/i386/avx512cdintrin.h: Here.
1201 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1202 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1203 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1204 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1205 TARGET_AVX512F from TARGET_AVX512CD.
1207 2014-02-26 Richard Biener <rguenther@suse.de>
1210 * ipa.c (walk_polymorphic_call_targets): Properly guard
1211 call to inline_update_overall_summary.
1213 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1216 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1217 and latches only if requested. Fix latch if it is removed.
1218 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1219 LOOPS_HAVE_PREHEADERS.
1221 2014-02-25 Andrew Pinski <apinski@cavium.com>
1223 * builtins.c (expand_builtin_thread_pointer): Create a new target
1224 when the target is NULL.
1226 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1228 PR rtl-optimization/60317
1229 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1230 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1231 * lra-assigns.c: Include params.h.
1232 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1233 other reload pseudos considerations.
1235 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1237 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1238 to use canonical form for nor<mode>3.
1240 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1243 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1246 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1248 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1249 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1250 (ix86_handle_option): Handle OPT_mprefetchwt1.
1251 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1252 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1254 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1255 OPTION_MASK_ISA_PREFETCHWT1.
1256 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1257 (PTA_PREFETCHWT1): New.
1258 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1259 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1260 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1261 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1262 (*prefetch_avx512pf_<mode>_: Change into ...
1263 (*prefetch_prefetchwt1_<mode>: This.
1264 * config/i386/i386.opt (mprefetchwt1): New.
1265 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1266 (_mm_prefetch): Handle intent to write.
1267 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1269 2014-02-25 Richard Biener <rguenther@suse.de>
1272 * emit-rtl.c (mem_attrs_htab): Remove.
1273 (mem_attrs_htab_hash): Likewise.
1274 (mem_attrs_htab_eq): Likewise.
1275 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1276 (init_emit_once): Do not allocate mem_attrs_htab.
1278 2014-02-25 Richard Biener <rguenther@suse.de>
1281 * lto-opts.c (lto_write_options): Output non-explicit conservative
1282 -fwrapv, -fno-trapv and -fno-strict-overflow.
1283 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1284 (run_gcc): And pass them through.
1286 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1288 * sel-sched.c (calculate_new_fences): New parameter ptime.
1289 Calculate it as a maximum over all fence cycles.
1290 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1291 Print the final schedule timing when sched_verbose.
1293 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1295 PR rtl-optimization/60292
1296 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1297 bit fot the fence instruction.
1299 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1301 * calls.h: Fix typo in comment.
1303 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1305 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1306 adjusting offsetable addresses.
1308 2014-02-24 Guozhi Wei <carrot@google.com>
1310 * sparseset.h (sparseset_pop): Fix the wrong index.
1312 2014-02-24 Walter Lee <walt@tilera.com>
1314 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1315 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1317 * common/config/tilegx/tilegx-common.c
1318 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1319 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1321 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1322 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1323 (tilegx_gimplify_va_arg_expr): Handle big endian.
1324 (tilegx_expand_unaligned_load): Ditto.
1325 (tilegx_expand_unaligned_store): Ditto.
1326 (TARGET_RETURN_IN_MSB): New.
1327 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1328 (TARGET_ENDIAN_DEFAULT): New.
1329 (TARGET_BIG_ENDIAN): Handle big endian.
1330 (BYTES_BIG_ENDIAN): Ditto.
1331 (WORDS_BIG_ENDIAN): Ditto.
1332 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1335 * config/tilegx/tilegx.md (extv): Handle big endian.
1337 (insn_st<n>): Ditto.
1338 (insn_st<n>_add<bitsuffix>): Ditto.
1339 (insn_stnt<n>): Ditto.
1340 (insn_stnt<n>_add<bitsuffix>):Ditto.
1341 (vec_interleave_highv8qi): Handle big endian.
1342 (vec_interleave_highv8qi_be): New.
1343 (vec_interleave_highv8qi_le): New.
1344 (insn_v1int_h): Handle big endian.
1345 (vec_interleave_lowv8qi): Handle big endian.
1346 (vec_interleave_lowv8qi_be): New.
1347 (vec_interleave_lowv8qi_le): New.
1348 (insn_v1int_l): Handle big endian.
1349 (vec_interleave_highv4hi): Handle big endian.
1350 (vec_interleave_highv4hi_be): New.
1351 (vec_interleave_highv4hi_le): New.
1352 (insn_v2int_h): Handle big endian.
1353 (vec_interleave_lowv4hi): Handle big endian.
1354 (vec_interleave_lowv4hi_be): New.
1355 (vec_interleave_lowv4hi_le): New.
1356 (insn_v2int_l): Handle big endian.
1357 (vec_interleave_highv2si): Handle big endian.
1358 (vec_interleave_highv2si_be): New.
1359 (vec_interleave_highv2si_le): New.
1360 (insn_v4int_h): Handle big endian.
1361 (vec_interleave_lowv2si): Handle big endian.
1362 (vec_interleave_lowv2si_be): New.
1363 (vec_interleave_lowv2si_le): New.
1364 (insn_v4int_l): Handle big endian.
1365 * config/tilegx/tilegx.opt (mbig-endian): New option.
1366 (mlittle-endian): New option.
1367 * doc/install.texi: Document tilegxbe-linux.
1368 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1370 2014-02-24 Martin Jambor <mjambor@suse.cz>
1373 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1374 there are no parameter descriptors.
1376 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1378 PR rtl-optimization/60268
1379 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1380 initialization to ...
1381 (sched_rgn_init): ... here.
1382 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1384 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1386 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1389 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1391 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1394 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1396 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1397 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1399 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1401 * config/microblaze/predicates.md: Add cmp_op predicate.
1402 * config/microblaze/microblaze.md: Add branch_compare instruction
1403 which uses cmp_op predicate and emits cmp insn before branch.
1404 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1405 to microblaze_expand_conditional_branch and consolidate logic.
1406 (microblaze_expand_conditional_branch): emit branch_compare
1407 insn instead of handling cmp op separate from branch insn.
1409 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1411 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1414 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1416 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1417 define_insn with define_expand and new define_insn
1418 *altivec_lve<VI_char>x_internal.
1419 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1420 and new define_insn *altivec_stve<VI_char>x_internal.
1421 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1423 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1425 (altivec_expand_stvex_be): New function.
1427 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1429 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1430 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1431 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1432 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1434 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1437 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1438 instead of emit_move_insn.
1440 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1442 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1444 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1445 gen_altivec_vsumsws.
1447 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1449 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1450 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1451 (altivec_lvxl_<mode>): New define_expand incorporating
1452 -maltivec=be semantics where needed.
1453 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1454 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1455 semantics where needed.
1456 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1457 (altivec_stvx_<mode>): New define_expand incorporating
1458 -maltivec=be semantics where needed.
1459 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1460 VM2 iterator instead of V4SI.
1461 (altivec_stvxl_<mode>): New define_expand incorporating
1462 -maltivec=be semantics where needed.
1463 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1464 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1465 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1466 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1467 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1468 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1469 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1470 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1471 ALTIVEC_BUILTIN_STVXL.
1472 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1473 (altivec_expand_stvx_be): Likewise.
1474 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1475 (altivec_expand_lvx_be): Likewise.
1476 (altivec_expand_stvx_be): Likewise.
1477 (altivec_expand_builtin): Add cases for
1478 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1479 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1480 (altivec_init_builtins): Add definitions for
1481 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1482 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1484 2014-02-21 Catherine Moore <clm@codesourcery.com>
1486 * doc/invoke.texi (mvirt, mno-virt): Document.
1487 * config/mips/mips.opt (mvirt): New option.
1488 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1490 2014-02-21 Richard Biener <rguenther@suse.de>
1492 PR tree-optimization/60276
1493 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1494 (STMT_VINFO_MIN_NEG_DIST): New macro.
1495 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1496 STMT_VINFO_MIN_NEG_DIST.
1497 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1498 made for negative dependence distances still hold.
1500 2014-02-21 Richard Biener <rguenther@suse.de>
1503 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1504 DECL_INITIAL for globals not in the current function context.
1506 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1508 PR tree-optimization/56490
1509 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1510 * tree-ssa-uninit.c: Include params.h.
1511 (compute_control_dep_chain): Add num_calls argument, return false
1512 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1513 num_calls to recursive call.
1514 (find_predicates): Change dep_chain into normal array,
1515 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1516 variable and adjust compute_control_dep_chain caller.
1517 (find_def_preds): Likewise.
1519 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1521 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1522 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1524 2014-02-21 Nick Clifton <nickc@redhat.com>
1526 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1527 (pushhi1): Likewise.
1528 (popqi1): Add mode to pre_dec.
1531 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1533 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1534 mode for mask of V8SFmode permutation.
1536 2014-02-20 Richard Henderson <rth@redhat.com>
1539 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1540 a new pseudo for OLDVAL.
1542 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1545 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1546 gen_reg_rtx if d->testing_p.
1547 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1548 if d->testing_p and we will certainly return true.
1549 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1552 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1554 * emit-rtl.c (gen_reg_rtx): Assert that
1555 crtl->emit.regno_pointer_align_length is non-zero.
1557 2014-02-20 Richard Henderson <rth@redhat.com>
1560 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1561 on failure the store back into EXPECT.
1563 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1564 Sandra Loosemore <sandra@codesourcery.com>
1566 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1567 * config/nios2/nios2.c (nios2_function_profiler): Add
1568 -fPIC (flag_pic == 2) support.
1569 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1570 (nios2_large_offset_p): New function.
1571 (nios2_unspec_reloc_p): Move up position, update to use
1572 nios2_large_offset_p.
1573 (nios2_unspec_address): Remove function.
1574 (nios2_unspec_offset): New function.
1575 (nios2_large_got_address): New function.
1576 (nios2_got_address): Add large offset support.
1577 (nios2_legitimize_tls_address): Update usage of removed and new
1579 (nios2_symbol_binds_local_p): New function.
1580 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1581 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1582 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1583 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1584 processing for (const (unspec ...)).
1585 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1587 2014-02-20 Richard Biener <rguenther@suse.de>
1589 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1590 doing the substitution.
1591 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1593 2014-02-20 Martin Jambor <mjambor@suse.cz>
1596 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1597 info when checking whether lattices are bottom.
1599 2014-02-20 Richard Biener <rguenther@suse.de>
1602 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1605 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1608 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1609 parameter specifying the scaling.
1610 (inline_call): Update.
1611 (want_inline_recursively): Guard division by zero.
1612 (recursive_inlining): Update.
1613 * ipa-inline.h (clone_inlined_nodes): Update.
1615 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1618 * config/i386/i386.c (classify_argument): Pass structures of size
1619 64 bytes or less in register.
1621 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1622 Kirill Yukhin <kirill.yukhin@intel.com>
1624 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1625 (_mm_rcp28_round_ss): Ditto.
1626 (_mm_rsqrt28_round_sd): Ditto.
1627 (_mm_rsqrt28_round_ss): Ditto.
1628 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1629 (_mm_rcp14_round_ss): Ditto.
1630 (_mm_rsqrt14_round_sd): Ditto.
1631 (_mm_rsqrt14_round_ss): Ditto.
1632 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1633 the first input operand, get rid of match_dup.
1634 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1636 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1638 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1639 operand as the first input operand, set type attribute.
1640 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1642 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1643 operand as the first input operand, set type attribute.
1645 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1647 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1650 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1653 * config/i386/i386.c (construct_container): Remove TFmode check
1654 for X86_64_INTEGER_CLASS.
1656 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1659 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1660 only when -Wpsabi is enabled.
1662 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1665 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1666 passing arrays in registers are the same as for structs, so remove the
1667 special case for them.
1669 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1671 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1672 destination type, extract only the valid bits if the source type is not
1673 integral and has a different mode.
1675 2014-02-19 Richard Biener <rguenther@suse.de>
1678 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1681 2014-02-19 Richard Biener <rguenther@suse.de>
1684 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1685 (ipa_modify_call_arguments): Emit an argument load explicitely and
1686 preserve virtual SSA form there and for the replacement call.
1687 Do not update SSA form nor free dominance info.
1689 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1691 * ipa.c (function_and_variable_visibility): Also clear WEAK
1692 flag when disolving COMDAT_GROUP.
1694 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1696 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1697 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1699 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1700 do more sanity checks.
1701 (detect_type_change): Return true when giving up early.
1702 (compute_complex_assign_jump_func): Fix type parameter of
1703 ipa_set_ancestor_jf.
1704 (compute_complex_ancestor_jump_func): Likewise.
1705 (update_jump_functions_after_inlining): Fix updating of
1707 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1709 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1711 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1712 inline clones when edge disappears.
1714 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1717 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1718 Split 64-bit moves into 2 patterns. Do not allow the use of
1719 direct move for TDmode in little endian, since the decimal value
1720 has little endian bytes within a word, but the 64-bit pieces are
1721 ordered in a big endian fashion, and normal subreg's of TDmode are
1723 (mov<mode>_64bit_dm): Likewise.
1724 (movtd_64bit_nodm): Likewise.
1726 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1728 PR tree-optimization/60174
1729 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1730 statement of an SSA_NAME that occurs in an abnormal PHI node.
1732 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1735 * final.c (SEEN_BB): Remove.
1736 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1737 (final_scan_insn): Don't force_source_line on second
1738 NOTE_INSN_BASIC_BLOCK.
1740 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1743 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1744 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1745 (type_natural_mode): Warn ABI change when %zmm register is not
1746 available for AVX512F vector value passing.
1748 2014-02-18 Kai Tietz <ktietz@redhat.com>
1751 * config/i386/i386.c (ix86_expand_prologue): Use value in
1752 rax register as displacement when restoring %r10 or %rax.
1753 Fix wrong offset when restoring both registers.
1755 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1757 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1758 assertion with conditional return.
1760 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1761 Uros Bizjak <ubizjak@gmail.com>
1764 * config/i386/driver-i386.c (host_detect_local_cpu): If
1765 YMM state is not saved by the OS, also clear has_f16c. Move
1766 CPUID 0x80000001 handling before YMM state saving checking.
1768 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1770 PR rtl-optimization/58960
1771 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1772 factored out from ...
1773 (sched_init): ... here.
1774 (free_global_sched_pressure_data): New, factored out from ...
1775 (sched_finish): ... here.
1776 * sched-int.h (free_global_sched_pressure_data): Declare.
1777 * sched-rgn.c (nr_regions_initial): New static global.
1778 (haifa_find_rgns): Initialize it.
1779 (schedule_region): Disable sched-pressure for the newly
1782 2014-02-17 Richard Biener <rguenther@suse.de>
1784 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1785 release SSA defs of pattern stmts.
1787 2014-02-17 Richard Biener <rguenther@suse.de>
1789 * tree-inline.c (expand_call_inline): Release the virtual
1790 operand defined by the call we are about to inline.
1792 2014-02-17 Richard Biener <rguenther@suse.de>
1794 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1796 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1797 Ilya Tocar <ilya.tocar@intel.com>
1799 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1800 arguments order in builtin.
1801 (_mm512_permutexvar_epi64): Ditto.
1802 (_mm512_mask_permutexvar_epi64): Ditto
1803 (_mm512_maskz_permutexvar_epi32): Ditto
1804 (_mm512_permutexvar_epi32): Ditto
1805 (_mm512_mask_permutexvar_epi32): Ditto
1807 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1809 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1810 (p8_vmrgow): Likewise.
1812 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1814 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1817 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1820 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1821 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1822 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1823 using direct move instructions on ISA 2.07. Also adjust
1824 instruction length for 64-bit.
1825 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1826 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1828 2014-02-15 Alan Modra <amodra@gmail.com>
1832 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1833 find_replacement on parts of insn rtl that might be reloaded.
1835 2014-02-15 Richard Biener <rguenther@suse.de>
1837 PR tree-optimization/60183
1838 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1839 (tree_ssa_phiprop): Calculate and free post-dominators.
1841 2014-02-14 Jeff Law <law@redhat.com>
1843 PR rtl-optimization/60131
1844 * ree.c (get_extended_src_reg): New function.
1845 (combine_reaching_defs): Use it rather than assuming location of REG.
1846 (find_and_remove_re): Verify first operand of extension is
1847 a REG before adding the insns to the copy list.
1849 2014-02-14 Roland McGrath <mcgrathr@google.com>
1851 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1852 * configure: Regenerated.
1853 * config.in: Regenerated.
1854 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1855 instead of ASM_SHORT.
1857 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1858 Richard Earnshaw <rearnsha@arm.com>
1860 PR rtl-optimization/59535
1861 * lra-constraints.c (process_alt_operands): Encourage alternative
1862 when unassigned pseudo class is superset of the alternative class.
1863 (inherit_reload_reg): Don't inherit when optimizing for code size.
1864 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1865 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1866 modes not less than 4 for Thumb1.
1868 2014-02-14 Kyle McMartin <kyle@redhat.com>
1871 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1873 2014-02-14 Richard Biener <rguenther@suse.de>
1875 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1876 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1877 Do not strip INDIRECT_REFs.
1879 2014-02-14 Richard Biener <rguenther@suse.de>
1882 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1883 DECL_FUNCTION_SPECIFIC_TARGET.
1884 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1885 * tree-streamer-out.c (pack_ts_target_option): Remove.
1886 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1887 (write_ts_function_decl_tree_pointers): Do not stream
1888 DECL_FUNCTION_SPECIFIC_TARGET.
1889 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1890 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1891 (lto_input_ts_function_decl_tree_pointers): Do not stream
1892 DECL_FUNCTION_SPECIFIC_TARGET.
1894 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1896 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1897 (get_initial_def_for_induction, vectorizable_induction): Ignore
1898 debug stmts when looking for exit_phi.
1899 (vectorizable_live_operation): Fix up condition.
1901 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1903 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1904 nreverse() because it changes the content of original tree list.
1906 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1908 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1909 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1911 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1913 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1914 GNU coding standards.
1916 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1919 * dwarf2out.c (gen_subprogram_die): Don't call
1920 add_calling_convention_attribute if subr_die is old_die.
1922 2014-02-13 Sharad Singhai <singhai@google.com>
1924 * doc/optinfo.texi: Fix order of nodes.
1926 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1928 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1929 operands[2], not operands[3].
1931 2014-02-13 Richard Biener <rguenther@suse.de>
1934 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1935 mention the possibility of an in-tree build.
1936 (CLooG): Update recommended version to 0.18.1, mention the
1937 possibility of an in-tree build and clarify that the ISL
1938 bundled with CLooG does not work.
1940 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1943 * expr.c (compress_float_constant): If x is a hard register,
1944 extend into a pseudo and then move to x.
1946 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1948 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1949 caused by bad second argument to warning_at() with -mhotpatch and
1950 nested functions (e.g. with gfortran).
1952 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1954 * opts.c (option_name): Remove "enabled by default" rider.
1956 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1958 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1960 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1961 Uros Bizjak <ubizjak@gmail.com>
1964 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1965 * configure: Regenerated.
1967 2014-02-12 Richard Biener <rguenther@suse.de>
1969 * vec.c (vec_prefix::calculate_allocation): Move as
1970 inline variant to vec.h.
1971 (vec_prefix::calculate_allocation_1): New out-of-line version.
1972 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1973 (vec_prefix::m_has_auto_buf): Rename to ...
1974 (vec_prefix::m_using_auto_storage): ... this.
1975 (vec_prefix::calculate_allocation): Inline the easy cases
1976 and dispatch to calculate_allocation_1 which doesn't need the
1978 (va_heap::reserve): Use gcc_checking_assert.
1979 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1980 m_using_auto_storage.
1981 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1983 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1984 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1985 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1987 2014-02-12 Richard Biener <rguenther@suse.de>
1989 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1990 when we found a dependence.
1992 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1994 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1996 (maybe_fold_stmt): ... into this new function.
1997 * omp-low.c (lower_omp): Update comment.
1999 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2002 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2005 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2007 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2008 identifiers in comments.
2009 (cortexa53_extra_costs): Likewise.
2010 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2011 (cortexa7_extra_costs): Likewise.
2012 (cortexa12_extra_costs): Likewise.
2013 (cortexa15_extra_costs): Likewise.
2014 (v7m_extra_costs): Likewise.
2016 2014-02-12 Richard Biener <rguenther@suse.de>
2019 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2020 of posix_memalign being successful.
2021 (lower_stmt): Restrict lowering of posix_memalign to when
2022 -ftree-bit-ccp is enabled.
2024 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2026 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2028 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2030 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2032 PR rtl-optimization/60116
2033 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2034 other_insn once the combination has been validated.
2036 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2039 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2041 * ipa-devirt.c: Include demangle.h
2042 (odr_violation_reported): New static variable.
2043 (add_type_duplicate): Update odr_violations.
2044 (maybe_record_node): Add completep parameter; update it.
2045 (record_target_from_binfo): Add COMPLETEP parameter;
2046 update it as needed.
2047 (possible_polymorphic_call_targets_1): Likewise.
2048 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2049 rename FINAL to COMPLETE.
2050 (record_targets_from_bases): Sanity check we found the binfo;
2051 fix COMPLETEP updating.
2052 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2053 parameter, fix computing of COMPLETEP.
2054 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2055 at LTO time do demangling.
2056 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2057 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2059 (gimple_get_virt_method_for_binfo): Likewise.
2060 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2061 gimple_get_virt_method_for_vtable): Update prototypes.
2063 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2066 * genautomata.c (add_presence_absence): Fix typo with
2067 {final_}presence_list.
2069 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2072 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2073 for VSX/Altivec vectors that land in GPR registers.
2075 2014-02-11 Richard Henderson <rth@redhat.com>
2076 Jakub Jelinek <jakub@redhat.com>
2079 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2080 around drhs if type conversion to lacc->type is not useless.
2082 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2084 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2086 (cortex-a57.cortex-a53): Likewise.
2087 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2089 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2091 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2094 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2096 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2097 add_options_for_arm_vfp3.
2099 2014-02-11 Jeff Law <law@redhat.com>
2102 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2103 object with an undesirable mode.
2105 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2108 * config/i386/sol2-9.h: New file.
2109 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2110 *-*-solaris2.9*): Use it.
2112 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2114 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2115 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2117 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2119 * config/microblaze/microblaze.c: Extend mcpu version format
2121 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2123 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2125 2014-02-10 Richard Henderson <rth@redhat.com>
2128 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2129 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2130 ms-abi vs -mno-accumulate-outgoing-args.
2131 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2132 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2135 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2138 * cfgexpand.c (expand_asm_operands): Attach source location to
2139 ASM_INPUT rtx objects.
2140 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2142 2014-02-10 Nick Clifton <nickc@redhat.com>
2144 * config/mn10300/mn10300.c (popcount): New function.
2145 (mn10300_expand_prologue): Include saved registers in stack usage
2148 2014-02-10 Jeff Law <law@redhat.com>
2151 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2152 when changing the SET_DEST of a prior insn to avoid an input reload.
2154 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2156 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2157 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2158 -mcall-openbsd, or -mcall-linux.
2159 (CC1_ENDIAN_BIG_SPEC): Remove.
2160 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2161 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2162 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2163 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2164 and %cc1_endian_default.
2165 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2167 2014-02-10 Richard Biener <rguenther@suse.de>
2169 PR tree-optimization/60115
2170 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2171 MEM_REF handling. Properly verify that the accesses are not
2172 out of the objects bound.
2174 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2176 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2179 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2181 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2182 proper constants and fix formatting.
2183 (possible_polymorphic_call_targets): Fix formatting.
2185 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2186 Ilya Tocar <ilya.tocar@intel.com>
2188 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2189 (_mm512_loadu_epi32): Renamed into...
2190 (_mm512_loadu_si512): This.
2191 (_mm512_storeu_epi32): Renamed into...
2192 (_mm512_storeu_si512): This.
2193 (_mm512_maskz_ceil_ps): Removed.
2194 (_mm512_maskz_ceil_pd): Ditto.
2195 (_mm512_maskz_floor_ps): Ditto.
2196 (_mm512_maskz_floor_pd): Ditto.
2197 (_mm512_floor_round_ps): Ditto.
2198 (_mm512_floor_round_pd): Ditto.
2199 (_mm512_ceil_round_ps): Ditto.
2200 (_mm512_ceil_round_pd): Ditto.
2201 (_mm512_mask_floor_round_ps): Ditto.
2202 (_mm512_mask_floor_round_pd): Ditto.
2203 (_mm512_mask_ceil_round_ps): Ditto.
2204 (_mm512_mask_ceil_round_pd): Ditto.
2205 (_mm512_maskz_floor_round_ps): Ditto.
2206 (_mm512_maskz_floor_round_pd): Ditto.
2207 (_mm512_maskz_ceil_round_ps): Ditto.
2208 (_mm512_maskz_ceil_round_pd): Ditto.
2209 (_mm512_expand_pd): Ditto.
2210 (_mm512_expand_ps): Ditto.
2211 * config/i386/i386.c (ix86_builtins): Remove
2212 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2213 (bdesc_args): Ditto.
2214 * config/i386/predicates.md (const1256_operand): New.
2215 (const_1_to_2_operand): Ditto.
2216 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2217 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2218 (*avx512pf_gatherpf<mode>sf): Ditto.
2219 (avx512pf_gatherpf<mode>df): Ditto.
2220 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2221 (*avx512pf_gatherpf<mode>df): Ditto.
2222 (avx512pf_scatterpf<mode>sf): Ditto.
2223 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2224 (*avx512pf_scatterpf<mode>sf): Ditto.
2225 (avx512pf_scatterpf<mode>df): Ditto.
2226 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2227 (*avx512pf_scatterpf<mode>df): Ditto.
2228 (avx512f_expand<mode>): Removed.
2229 (<shift_insn><mode>3<mask_name>): Change predicate type.
2231 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2233 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2234 not at the end of datarefs vector use ordered_remove to avoid
2235 reordering datarefs vector.
2238 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2239 mark local addressable non-static vars as GOVD_PRIVATE
2240 instead of GOVD_LOCAL.
2241 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2242 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2246 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2247 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2248 assume_aligned or alloc_align attributes.
2249 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2250 arguments. Handle also assume_aligned and alloc_align attributes.
2251 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2252 calls to functions with assume_aligned or alloc_align attributes.
2253 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2255 2014-02-08 Terry Guo <terry.guo@arm.com>
2257 * doc/invoke.texi: Document ARM -march=armv7e-m.
2259 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2261 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2262 flag on __cilkrts_rethrow builtin.
2265 * ipa-cp.c (determine_versionability): Fail at -O0
2266 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2267 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2270 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2273 * tree-inline.c (copy_forbidden): Fail for
2274 __attribute__((optimize (0))) functions.
2276 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2278 * varpool.c: Include pointer-set.h.
2279 (varpool_remove_unreferenced_decls): Variables in other partitions
2280 will not be output; be however careful to not lose information
2283 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2285 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2286 lookup in the vtable constructor.
2288 2014-02-07 Jeff Law <law@redhat.com>
2291 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2292 define_insn_and_split.
2294 * ipa-inline.c (inline_small_functions): Fix typos.
2296 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2298 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2299 (s390_can_use_return_insn): Declare.
2300 * config/s390/s390.h (EPILOGUE_USES): Define.
2301 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2303 (s390_chunkify_start): Handle return JUMP_LABELs.
2304 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2305 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2306 (s390_can_use_return_insn): New functions.
2307 (s390_fix_long_loop_prediction): Handle conditional returns.
2308 (TARGET_SET_UP_BY_PROLOGUE): Define.
2309 * config/s390/s390.md (ANY_RETURN): New code iterator.
2310 (*creturn, *csimple_return, return, simple_return): New patterns.
2312 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2314 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2315 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2316 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2317 REG_CFA_RESTORE list when deciding not to restore a register.
2319 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2321 * config/s390/s390.c: Include tree-pass.h and context.h.
2322 (s390_early_mach): New function, split out from...
2323 (s390_emit_prologue): ...here.
2324 (pass_data_s390_early_mach): New pass structure.
2325 (pass_s390_early_mach): New class.
2326 (s390_option_override): Create and register early_mach pass.
2327 Move to end of file.
2329 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2331 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2332 to match for the exit block.
2334 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2336 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2337 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2338 Reject misaligned operands.
2340 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2342 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2344 2014-02-07 Richard Biener <rguenther@suse.de>
2347 * gimple-low.c (lower_builtin_posix_memalign): New function.
2348 (lower_stmt): Call it to lower posix_memalign in a way
2349 to make alignment info accessible.
2351 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2354 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2355 __builtin_setjmp_receiver.
2357 2014-02-07 Richard Biener <rguenther@suse.de>
2360 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2361 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2362 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2363 Handle BUILT_IN_POSIX_MEMALIGN.
2364 (find_func_clobbers): Likewise.
2365 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2366 (call_may_clobber_ref_p_1): Likewise.
2368 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2371 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2374 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2377 * lto-cgraph.c (lto_output_node): Use
2378 symtab_get_symbol_partitioning_class.
2379 (lto_output_varpool_node): likewise.
2380 (symtab_get_symbol_partitioning_class): Move here from
2382 * cgraph.h (symbol_partitioning_class): Likewise.
2383 (symtab_get_symbol_partitioning_class): Declare.
2385 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2387 * ggc.h (ggc_internal_cleared_alloc): New macro.
2388 * vec.h (vec_safe_copy): Handle memory stats.
2389 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2390 * target-globals.c (save_target_globals): Likewise.
2392 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2395 * expr.c (emit_move_resolve_push): Export; be bit more selective
2396 on when to clear alias set.
2397 * expr.h (emit_move_resolve_push): Declare.
2398 * function.h (struct function): Add tail_call_marked.
2399 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2400 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2401 * config/i386/i386.md (TImode move expander): De not call
2403 (FP push expanders): Preserve memory attributes.
2404 * config/i386/sse.md (push<mode>1): Remove.
2405 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2406 (ix86_expand_push): Remove.
2407 * config/i386/mmx.md (push<mode>1): Remove.
2409 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2411 PR rtl-optimization/60030
2412 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2413 lopart with paradoxical subreg before shifting it up by hprec.
2415 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2417 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2418 Remove extra newline at end of file.
2419 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2420 (arm_issue_rate): Handle cortexa57.
2421 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2422 (cortex-a57.cortex-a53): Likewise.
2424 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2427 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2428 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2430 (arm_expand_prologue): Adjust all callers.
2431 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2432 info, registers also at the lowest numbered registers side. Use
2433 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2437 * var-tracking.c (adjust_mems): Before adding a SET to
2438 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2440 2014-02-06 Alan Modra <amodra@gmail.com>
2443 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2444 change SDmode to DDmode when lra_in_progress.
2446 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2449 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2450 free_data_ref on the dr first, and before goto again also set dr
2451 to the next dr. For simd_lane_access, free old datarefs[i] before
2452 overwriting it. For get_vectype_for_scalar_type failure, don't
2453 free_data_ref if simd_lane_access.
2455 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2458 * tree.h (opts_for_fn): New inline function.
2459 (opt_for_fn): Define.
2460 * config/i386/i386.c (ix86_function_regparm): Use
2461 opt_for_fn (decl, optimize) instead of optimize.
2463 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2465 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2466 for SYMBOL_REF in large memory model.
2468 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2470 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2472 (cortex-a57): Likewise.
2473 (cortex-a57.cortex-a53): Likewise.
2475 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2476 Kugan Vivekanandarajah <kuganv@linaro.org>
2478 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2480 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2482 2014-02-06 Richard Biener <rguenther@suse.de>
2484 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2485 set_loop_copy and initialize_original_copy_tables.
2487 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2489 * config/aarch64/aarch64-simd.md
2490 (aarch64_ashr_simddi): Change QI to SI.
2492 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2493 Jakub Jelinek <jakub@redhat.com>
2496 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2499 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2501 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2502 CODE_FOR_altivec_vpku[hw]um to
2503 CODE_FOR_altivec_vpku[hw]um_direct.
2504 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2505 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2506 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2507 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2509 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2511 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2512 generation for -maltivec=be.
2513 (altivec_vsumsws): Simplify redundant test.
2515 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2517 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2518 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2519 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2520 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2521 gen_altivec_vpkuwum.
2522 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2524 (altivec_vpks<VI_char>ss): Likewise.
2525 (altivec_vpks<VI_char>us): Likewise.
2526 (altivec_vpku<VI_char>us): Likewise.
2527 (altivec_vpku<VI_char>um): Likewise.
2528 (altivec_vpku<VI_char>um_direct): New (copy of
2529 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2531 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2532 target is little endian and -maltivec=be is not specified.
2533 (*altivec_vupkhs<VU_char>_direct): New (copy of
2534 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2535 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2536 target is little endian and -maltivec=be is not specified.
2537 (*altivec_vupkls<VU_char>_direct): New (copy of
2538 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2539 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2540 little endian and -maltivec=be is not specified.
2541 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2542 little endian and -maltivec=be is not specified.
2544 2014-02-05 Richard Henderson <rth@redhat.com>
2547 * combine-stack-adj.c: Revert r206943.
2548 * sched-int.h (struct deps_desc): Add last_args_size.
2549 * sched-deps.c (init_deps): Initialize it.
2550 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2551 contain REG_ARGS_SIZE notes.
2553 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2555 * lto-cgraph.c (asm_nodes_output): Make global.
2556 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2557 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2558 (driver_handle_option): Handle OPT_fwpa.
2560 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2563 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2564 a comment typo and formatting issue. If odr_hash hasn't been
2565 created, return vNULL and set *completep to false.
2568 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2569 bb with no successors.
2571 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2574 * doc/invoke.texi (-march): Clarify documentation for ARM.
2578 2014-02-05 Richard Biener <rguenther@suse.de>
2580 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2581 when not vectorizing because of too many alias checks.
2582 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2583 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2585 2014-02-05 Nick Clifton <nickc@redhat.com>
2587 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2588 accept extended registers in any mode when compiling for the MN10300.
2590 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2592 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2593 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2594 sanitization attributes.
2595 (can_inline_edge_p): Likewise.
2596 (sanitize_attrs_match_for_inline_p): New function.
2598 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2600 * ipa-prop.c (detect_type_change): Shor circuit testing of
2601 type changes on THIS pointer.
2603 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2606 * config/pa/pa.c (legitimize_tls_address): Return original address
2607 if not passed a SYMBOL_REF rtx.
2608 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2610 (pa_emit_move_sequence): Simplify TLS source operands.
2611 (pa_legitimate_constant_p): Reject all TLS constants.
2612 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2613 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2615 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2617 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2618 groups when we know they are controlled by LTO.
2619 * varasm.c (default_binds_local_p_1): If object is in other partition,
2620 it will be resolved locally.
2622 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2624 * config/host-linux.c (linux_gt_pch_use_address): Don't
2625 use SSIZE_MAX because it is not always defined.
2627 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2630 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2631 threshold for pseudo splitting.
2632 (update_ebb_live_info): Process call argument hard registers and
2633 hard registers from insn definition too.
2634 (max_small_class_regs_num): New constant.
2635 (inherit_in_ebb): Update live hard regs through EBBs. Update
2636 reloads_num only for small register classes. Don't split for
2639 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2642 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2645 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2647 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2650 2014-02-04 Marek Polacek <polacek@redhat.com>
2652 * gdbinit.in (pel): Define.
2654 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2656 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2659 2014-02-04 Richard Biener <rguenther@suse.de>
2662 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2663 in function context local.
2664 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2665 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2666 similar to LTO_imported_decl_ref.
2668 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2670 PR tree-optimization/60002
2671 * cgraphclones.c (build_function_decl_skip_args): Clear
2674 PR tree-optimization/60023
2675 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2676 false to gsi_replace.
2677 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2678 has been in some EH region and vec_stmt could throw, add
2679 vec_stmt into the same EH region.
2680 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2681 has no lhs, ignore it.
2682 * internal-fn.c (expand_MASK_LOAD): Likewise.
2685 * tree-inline.c (copy_forbidden): Fail for
2686 __attribute__((optimize (0))) functions.
2689 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2690 is set, copy one less argument.
2691 (expand_simd_clones): Don't subtract clone_info->inbranch
2692 from simd_clone_struct_alloc argument.
2694 PR rtl-optimization/57915
2695 * recog.c (simplify_while_replacing): If all unary/binary/relational
2696 operation arguments are constant, attempt to simplify those.
2699 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2700 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2702 2014-02-04 Richard Biener <rguenther@suse.de>
2704 PR tree-optimization/60012
2705 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2706 TBAA disambiguation to all DDRs.
2708 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2711 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2712 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2714 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2717 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2719 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2721 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2722 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2724 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2727 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2728 to figure out targets of polymorphic calls with known decl.
2729 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2730 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2731 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2732 (get_polymorphic_call_info): ... here.
2733 (get_polymorphic_call_info_from_invariant): New function.
2735 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2737 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2738 lookup via vtable pointer; check for type consistency
2739 and turn inconsitent facts into UNREACHABLE.
2740 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2741 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2742 type inconsistent querries; return UNREACHABLE instead.
2744 2014-02-03 Richard Henderson <rth@twiddle.net>
2747 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2748 already processed this node.
2749 (normalize_one_pred_1): Pass along mark_set.
2750 (normalize_one_pred): Create and destroy a pointer_set_t.
2751 (normalize_one_pred_chain): Likewise.
2753 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2755 PR gcov-profile/58602
2756 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2758 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2761 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2762 -fno-devirtualize; try to devirtualize by the knowledge of
2763 virtual table pointer given by aggregate propagation.
2764 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2765 (ipa_print_node_jump_functions): Dump also offset that
2766 is relevant for polymorphic calls.
2767 (determine_known_aggregate_parts): Add arg_type parameter; use it
2768 instead of determining the type from pointer type.
2769 (ipa_compute_jump_functions_for_edge): Update call of
2770 determine_known_aggregate_parts.
2771 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2772 (gimple_get_virt_method_for_binfo): ... here; simplify using
2773 vtable_pointer_value_to_vtable.
2774 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2775 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2776 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2777 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2779 (vtable_pointer_value_to_binfo): ... here.
2780 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2782 2014-02-03 Teresa Johnson <tejohnson@google.com>
2784 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2785 redef of outer loop index variable.
2787 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2791 * doc/extend.texi (Function Attributes): Typo.
2793 2014-02-03 Cong Hou <congh@google.com>
2795 PR tree-optimization/60000
2796 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2797 if the vectorized statement is a store. A store statement can only
2798 appear at the end of pattern statements.
2800 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2802 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2803 (ix86_option_override_internal): Default long double to 64-bit for
2804 32-bit Bionic and to 128-bit for 64-bit Bionic.
2806 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2807 TARGET_LONG_DOUBLE_128 is true.
2808 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2810 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2811 (mlong-double-64): Negate -mlong-double-128.
2812 (mlong-double-128): New option.
2814 * config/i386/i386-c.c (ix86_target_macros): Define
2815 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2817 * doc/invoke.texi: Document -mlong-double-128.
2819 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2821 PR rtl-optimization/60024
2822 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2824 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2826 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2828 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2830 PR rtl-optimization/57662
2831 * sel-sched.c (code_motion_path_driver): Do not mark already not
2832 existing blocks in the visiting bitmap.
2834 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2836 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2837 on the insn being emitted.
2839 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2840 Will Deacon <will.deacon@arm.com>
2842 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2844 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2846 * config/arm/arm-tables.opt: Regenerate.
2848 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2850 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2851 for vector types other than V16QImode.
2852 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2853 define_expand, and call altivec_expand_vec_perm_le when producing
2854 code with little endian element order.
2855 (*altivec_vperm_<mode>_internal): New insn having previous
2856 behavior of altivec_vperm_<mode>.
2857 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2858 altivec_expand_vec_perm_le when producing code with little endian
2860 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2861 behavior of altivec_vperm_<mode>_uns.
2863 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2865 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2866 (altivec_vsumsws): Add handling for -maltivec=be with a little
2868 (altivec_vsumsws_direct): New.
2869 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2870 gen_altivec_vsumsws.
2872 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2874 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2875 vtable_pointer_value_to_binfo): New functions.
2876 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2877 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2879 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2881 * config/nios2/nios2.md (load_got_register): Initialize GOT
2882 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2883 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2885 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2887 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2888 preserverd by passthrough, do not propagate the type.
2890 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2892 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2893 (mips_atomic_assign_expand_fenv): New function.
2894 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2896 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2898 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2899 (__builtin_mips_set_fcsr): Likewise.
2900 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2901 MIPS_USI_FTYPE_VOID.
2902 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2903 (mips16_expand_set_fcsr): Likewise.
2904 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2905 (mips16_set_fcsr_stub): Likewise.
2906 (mips16_get_fcsr_one_only_stub): New class.
2907 (mips16_set_fcsr_one_only_stub): Likewise.
2908 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2909 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2910 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2911 (hard_float): New availability predicate.
2912 (mips_builtins): Add get_fcsr and set_fcsr.
2913 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2914 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2915 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2916 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2917 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2920 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2922 * config/mips/mips.c (mips_one_only_stub): New class.
2923 (mips_need_mips16_rdhwr_p): Replace with...
2924 (mips16_rdhwr_stub): ...this new variable.
2925 (mips16_stub_call_address): New function.
2926 (mips16_rdhwr_one_only_stub): New class.
2927 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2928 (mips_output_mips16_rdhwr): Delete.
2929 (mips_finish_stub): New function.
2930 (mips_code_end): Use it to handle rdhwr stubs.
2932 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2935 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2936 when calculating size of integer atomic types.
2938 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2940 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2942 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2944 PR tree-optimization/60003
2945 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2946 * profile.c (branch_prob): Use gimple_call_builtin_p
2947 to check for BUILT_IN_SETJMP_RECEIVER.
2948 * tree-inline.c (copy_bb): Call notice_special_calls.
2950 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2953 * lra-constraints.c (process_alt_operands): Update reload_sum only
2956 2014-01-31 Richard Henderson <rth@redhat.com>
2959 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2960 until after else_eh is processed.
2962 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2964 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2965 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2966 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2967 in smmintrin.h, remove them.
2968 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2969 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2970 * config/i386/i386.md (ROUND_SAE): Fix value.
2971 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2972 (const48_operand): New.
2973 * config/i386/subst.md (round), (round_expand): Use
2974 const_4_or_8_to_11_operand.
2975 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2977 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2979 * config/i386/constraints.md (Yk): Swap meaning with k.
2980 * config/i386/i386.md (movhi_internal): Change Yk to k.
2981 (movqi_internal): Ditto.
2982 (*k<logic><mode>): Ditto.
2985 (kandn<mode>): Ditto.
2986 (*<code>hi_1): Ditto.
2987 (*<code>qi_1): Ditto.
2988 (kxnor<mode>): Ditto.
2989 (kortestzhi): Ditto.
2990 (kortestchi): Ditto.
2992 (*one_cmplhi2_1): Ditto.
2993 (*one_cmplqi2_1): Ditto.
2994 * config/i386/sse.md (): Change k to Yk.
2995 (avx512f_load<mode>_mask): Ditto.
2996 (avx512f_blendm<mode>): Ditto.
2997 (avx512f_store<mode>_mask): Ditto.
2998 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2999 (avx512f_storedqu<mode>_mask): Ditto.
3000 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3002 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3003 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3004 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3005 (avx512f_maskcmp<mode>3): Ditto.
3006 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3007 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3008 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3009 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3010 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3011 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3012 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3013 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3014 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3015 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3016 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3017 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3018 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3019 (vec_extract_lo_<mode>_maskm): Ditto.
3020 (vec_extract_hi_<mode>_maskm): Ditto.
3021 (avx512f_vternlog<mode>_mask): Ditto.
3022 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3023 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3024 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3025 (avx512f_<code>v8div16qi2_mask): Ditto.
3026 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3027 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3028 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3029 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3030 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3031 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3032 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3033 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3034 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3035 (avx512cd_maskb_vec_dupv8di): Ditto.
3036 (avx512cd_maskw_vec_dupv16si): Ditto.
3037 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3038 (avx512f_vpermi2var<mode>3_mask): Ditto.
3039 (avx512f_vpermi2var<mode>3_mask): Ditto.
3040 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3041 (*avx512f_gathersi<mode>): Ditto.
3042 (*avx512f_gathersi<mode>_2): Ditto.
3043 (*avx512f_gatherdi<mode>): Ditto.
3044 (*avx512f_gatherdi<mode>_2): Ditto.
3045 (*avx512f_scattersi<mode>): Ditto.
3046 (*avx512f_scatterdi<mode>): Ditto.
3047 (avx512f_compress<mode>_mask): Ditto.
3048 (avx512f_compressstore<mode>_mask): Ditto.
3049 (avx512f_expand<mode>_mask): Ditto.
3050 * config/i386/subst.md (mask): Change k to Yk.
3051 (mask_scalar_merge): Ditto.
3054 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3056 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3058 2014-01-31 Richard Biener <rguenther@suse.de>
3061 * builtins.c (fold_builtin_memory_op): Make sure to not
3062 use a floating-point mode or a boolean or enumeral type for
3065 2014-01-30 DJ Delorie <dj@redhat.com>
3067 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3068 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3069 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3070 whenever main() has an epilogue.
3072 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3074 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3075 unused variable "field".
3076 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3077 (vsx_mergeh_<mode>): Likewise.
3078 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3079 (altivec_vmrghh): Likewise.
3080 (altivec_vmrghw): Likewise.
3081 (altivec_vmrglb): Likewise.
3082 (altivec_vmrglh): Likewise.
3083 (altivec_vmrglw): Likewise.
3084 (altivec_vspltb): Add missing uses.
3085 (altivec_vsplth): Likewise.
3086 (altivec_vspltw): Likewise.
3087 (altivec_vspltsf): Likewise.
3089 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3092 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3093 frame related instructions.
3095 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3097 PR rtl-optimization/59959
3098 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3099 any reload of register whose subreg is invalid.
3101 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3103 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3104 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3105 Add missing return type - void.
3107 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3109 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3110 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3111 remove element index adjustment for endian (now handled in vsx.md
3113 (altivec_expand_vec_perm_const): Use
3114 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3115 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3116 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3117 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3118 define_expand and a new define_insn *altivec_vspltb_internal;
3119 adjust for -maltivec=be on a little endian target.
3120 (altivec_vspltb_direct): New.
3121 (altivec_vsplth): Divide into a define_expand and a new
3122 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3123 little endian target.
3124 (altivec_vsplth_direct): New.
3125 (altivec_vspltw): Divide into a define_expand and a new
3126 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3127 little endian target.
3128 (altivec_vspltw_direct): New.
3129 (altivec_vspltsf): Divide into a define_expand and a new
3130 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3131 a little endian target.
3133 2014-01-30 Richard Biener <rguenther@suse.de>
3135 PR tree-optimization/59993
3136 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3137 can propagate form the earlier stmt and avoid the transform
3138 when the intermediate result is needed.
3140 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3142 * README.Portability: Fix typo.
3144 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3146 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3147 comparison_operator with ordered_comparison_operator.
3149 2014-01-30 Nick Clifton <nickc@redhat.com>
3151 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3152 Rename to mn10300_store_multiple_regs.
3153 * config/mn10300/mn10300.c: Likewise.
3154 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3155 store_multiple_regs.
3156 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3157 Call mn10300_store_multiple_regs.
3159 2014-01-30 Nick Clifton <nickc@redhat.com>
3160 DJ Delorie <dj@redhat.com>
3162 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3163 %fp 2 to keep registers after it properly word-aligned.
3164 (rl78_alloc_physical_registers_umul): Handle the case where both
3165 input operands are the same.
3167 2014-01-30 Richard Biener <rguenther@suse.de>
3169 PR tree-optimization/59903
3170 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3173 2014-01-30 Jason Merrill <jason@redhat.com>
3176 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3179 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3181 2014-01-30 Richard Biener <rguenther@suse.de>
3183 PR tree-optimization/59951
3184 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3186 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3189 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3190 SFmode to DFmode case.
3192 2014-01-29 DJ Delorie <dj@redhat.com>
3194 * config/msp430/msp430.opt (-minrt): New.
3195 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3197 (ENDFILE_SPEC): Likewise.
3199 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3201 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3202 (estimate_function_body_sizes): Use it.
3204 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3207 * dwarf2out.c (is_cxx_auto): New.
3208 (is_base_type): Use it.
3209 (gen_type_die_with_usage): Likewise.
3211 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3213 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3214 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3215 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3216 -maltivec=be with LE targets.
3217 (vsx_mergeh_<mode>): Likewise.
3218 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3219 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3220 (altivec_vmrghb): Replace with define_expand and new
3221 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3222 (altivec_vmrghb_direct): New define_insn.
3223 (altivec_vmrghh): Replace with define_expand and new
3224 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3225 (altivec_vmrghh_direct): New define_insn.
3226 (altivec_vmrghw): Replace with define_expand and new
3227 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3228 (altivec_vmrghw_direct): New define_insn.
3229 (*altivec_vmrghsf): Adjust for endianness.
3230 (altivec_vmrglb): Replace with define_expand and new
3231 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3232 (altivec_vmrglb_direct): New define_insn.
3233 (altivec_vmrglh): Replace with define_expand and new
3234 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3235 (altivec_vmrglh_direct): New define_insn.
3236 (altivec_vmrglw): Replace with define_expand and new
3237 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3238 (altivec_vmrglw_direct): New define_insn.
3239 (*altivec_vmrglsf): Adjust for endianness.
3240 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3241 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3242 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3243 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3244 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3245 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3246 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3247 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3249 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3251 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3252 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3255 2014-01-29 Richard Biener <rguenther@suse.de>
3257 PR tree-optimization/58742
3258 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3259 associate_pointerplus_align.
3260 (associate_pointerplus_diff): New function.
3261 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3262 and associate_pointerplus_diff.
3264 2014-01-29 Richard Biener <rguenther@suse.de>
3266 * lto-streamer.h (LTO_major_version): Bump to 3.
3267 (LTO_minor_version): Reset to 0.
3269 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3271 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3272 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3273 (arm_file_start): Generate correct asm header for armv7ve.
3274 * config/arm/bpabi.h: Add multilib support for armv7ve.
3275 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3276 and cortex-a15 to armv7ve.
3277 * config/arm/t-aprofile: Add multilib support for armv7ve.
3278 * doc/invoke.texi: Document -march=armv7ve.
3280 2014-01-29 Richard Biener <rguenther@suse.de>
3282 PR tree-optimization/58742
3283 * tree-ssa-forwprop.c (associate_plusminus): Return true
3284 if we changed sth, defer EH cleanup to ...
3285 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3286 (simplify_mult): New function.
3288 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3291 PR tree-optimization/59920
3292 * tree.c (build_common_builtin_nodes): Remove
3293 __builtin_setjmp_dispatcher initialization.
3294 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3295 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3296 instead of gsi_after_labels + manually skipping debug stmts.
3297 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3298 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3299 * tree-inline.c (copy_edges_for_bb): Remove
3300 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3301 argument. Ignore computed_goto_p stmts. Don't call
3302 make_abnormal_goto_edges. If a call might need abnormal edges
3303 for non-local gotos, see if it already has an edge to
3304 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3305 with true argument, don't do anything then, otherwise add
3306 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3307 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3309 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3310 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3311 (lower_stmt): Don't set data->calls_builtin_setjmp.
3312 (lower_builtin_setjmp): Adjust comment.
3313 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3314 * tree-cfg.c (found_computed_goto): Remove.
3315 (factor_computed_gotos): Remove.
3316 (make_goto_expr_edges): Return bool, true for computed gotos.
3317 Don't call make_abnormal_goto_edges.
3318 (build_gimple_cfg): Don't set found_computed_goto, don't call
3319 factor_computed_gotos.
3320 (computed_goto_p): No longer static.
3321 (make_blocks): Don't set found_computed_goto.
3322 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3323 (make_edges): If make_goto_expr_edges returns true, push bb
3324 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3325 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3326 vector. Record mapping between bbs and OpenMP regions if there
3327 are any, adjust make_gimple_omp_edges caller. Call
3328 handle_abnormal_edges.
3329 (make_abnormal_goto_edges): Remove.
3330 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3331 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3332 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3333 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3334 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3335 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3336 filling *region also set *region_idx to (*region)->entry->index.
3339 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3340 For REGs set ORIGINAL_REGNO.
3342 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3344 * doc/md.texi: Mention that a target shouldn't implement
3345 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3348 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3350 PR tree-optimization/59594
3351 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3352 a copy of the datarefs vector rather than the vector itself.
3354 2014-01-28 Jason Merrill <jason@redhat.com>
3357 * dwarf2out.c (auto_die): New static.
3358 (gen_type_die_with_usage): Handle C++1y 'auto'.
3359 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3362 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3365 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3366 (SPEC_X32): Likewise.
3367 (SPEC_64): Likewise.
3368 * config/i386/i386.c (ix86_option_override_internal): Turn off
3369 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3371 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3372 * config/i386/i386.h (TARGET_16BIT): New macro.
3373 (TARGET_16BIT_P): Likewise.
3374 * config/i386/i386.opt: Add m16.
3375 * doc/invoke.texi: Document -m16.
3377 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3379 PR preprocessor/59935
3380 * input.c (location_get_source_line): Bail out on when line number
3381 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3383 2014-01-28 Richard Biener <rguenther@suse.de>
3385 PR tree-optimization/58742
3386 * tree-ssa-forwprop.c (associate_plusminus): Handle
3387 pointer subtraction of the form (T)(P + A) - (T)P.
3389 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3391 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3394 2014-01-28 Richard Biener <rguenther@suse.de>
3397 2014-01-28 Richard Biener <rguenther@suse.de>
3399 PR rtl-optimization/45364
3400 PR rtl-optimization/59890
3401 * var-tracking.c (local_get_addr_clear_given_value): Handle
3402 already cleared slot.
3403 (val_reset): Handle not allocated local_get_addr_cache.
3404 (vt_find_locations): Use post-order on the inverted CFG.
3406 2014-01-28 Richard Biener <rguenther@suse.de>
3408 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3410 2014-01-28 Richard Biener <rguenther@suse.de>
3412 PR rtl-optimization/45364
3413 PR rtl-optimization/59890
3414 * var-tracking.c (local_get_addr_clear_given_value): Handle
3415 already cleared slot.
3416 (val_reset): Handle not allocated local_get_addr_cache.
3417 (vt_find_locations): Use post-order on the inverted CFG.
3419 2014-01-28 Alan Modra <amodra@gmail.com>
3421 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3422 * configure.ac <recursive call for build != host>: Define
3423 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3424 and LD_FOR_BUILD too.
3425 * configure: Regenerate.
3427 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3429 * config/i386/i386.c (get_builtin_code_for_version): Separate
3430 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3431 Broadwell from Haswell.
3433 2014-01-27 Steve Ellcey <sellcey@mips.com>
3435 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3436 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3437 * config/mips/mips.c (mips_option_override): Change setting
3439 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3440 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3441 Change from Mask to Var.
3443 2014-01-27 Jeff Law <law@redhat.com>
3445 * ipa-inline.c (inline_small_functions): Fix typo.
3447 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3449 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3450 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3451 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3452 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3453 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3454 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3455 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3456 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3457 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3458 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3459 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3460 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3461 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3462 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3463 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3464 (_mm512_storeu_epi64): Ditto.
3465 (_mm512_cmpge_epi32_mask): Ditto.
3466 (_mm512_cmpge_epu32_mask): Ditto.
3467 (_mm512_cmpge_epi64_mask): Ditto.
3468 (_mm512_cmpge_epu64_mask): Ditto.
3469 (_mm512_cmple_epi32_mask): Ditto.
3470 (_mm512_cmple_epu32_mask): Ditto.
3471 (_mm512_cmple_epi64_mask): Ditto.
3472 (_mm512_cmple_epu64_mask): Ditto.
3473 (_mm512_cmplt_epi32_mask): Ditto.
3474 (_mm512_cmplt_epu32_mask): Ditto.
3475 (_mm512_cmplt_epi64_mask): Ditto.
3476 (_mm512_cmplt_epu64_mask): Ditto.
3477 (_mm512_cmpneq_epi32_mask): Ditto.
3478 (_mm512_cmpneq_epu32_mask): Ditto.
3479 (_mm512_cmpneq_epi64_mask): Ditto.
3480 (_mm512_cmpneq_epu64_mask): Ditto.
3481 (_mm512_expand_pd): Ditto.
3482 (_mm512_expand_ps): Ditto.
3483 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3484 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3485 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3486 * config/i386/i386.c (ix86_builtins): Add
3487 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3488 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3489 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3490 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3491 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3492 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3493 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3494 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3495 IX86_BUILTIN_PMOVUSQW512_MEM.
3496 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3497 __builtin_ia32_pmovsqd512mem_mask,
3498 __builtin_ia32_pmovqd512mem_mask,
3499 __builtin_ia32_pmovusqw512mem_mask,
3500 __builtin_ia32_pmovsqw512mem_mask,
3501 __builtin_ia32_pmovqw512mem_mask,
3502 __builtin_ia32_pmovusdw512mem_mask,
3503 __builtin_ia32_pmovsdw512mem_mask,
3504 __builtin_ia32_pmovdw512mem_mask,
3505 __builtin_ia32_pmovqb512mem_mask,
3506 __builtin_ia32_pmovusqb512mem_mask,
3507 __builtin_ia32_pmovsqb512mem_mask,
3508 __builtin_ia32_pmovusdb512mem_mask,
3509 __builtin_ia32_pmovsdb512mem_mask,
3510 __builtin_ia32_pmovdb512mem_mask.
3511 (bdesc_args): Add __builtin_ia32_expanddf512,
3512 __builtin_ia32_expandsf512.
3513 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3514 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3515 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3516 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3517 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3518 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3519 (avx512f_<code>v8div16qi2_mask_store): This.
3520 (avx512f_expand<mode>): New.
3522 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3524 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3526 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3527 (_mm512_prefetch_i32scatter_pd): Ditto.
3528 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3529 (_mm512_prefetch_i64scatter_pd): Ditto.
3530 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3531 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3532 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3533 (_mm512_prefetch_i32scatter_ps): Ditto.
3534 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3535 (_mm512_prefetch_i64scatter_ps): Ditto.
3536 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3537 * config/i386/i386-builtin-types.def: Define
3538 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3539 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3540 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3541 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3542 IX86_BUILTIN_SCATTERPFQPD.
3543 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3544 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3545 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3546 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3547 __builtin_ia32_scatterpfqps.
3548 (ix86_expand_builtin): Expand new built-ins.
3549 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3550 fix memory access data type.
3551 (*avx512pf_gatherpf<mode>_mask): Ditto.
3552 (*avx512pf_gatherpf<mode>): Ditto.
3553 (avx512pf_scatterpf<mode>): Ditto.
3554 (*avx512pf_scatterpf<mode>_mask): Ditto.
3555 (*avx512pf_scatterpf<mode>): Ditto.
3556 (GATHER_SCATTER_SF_MEM_MODE): New.
3557 (avx512pf_gatherpf<mode>df): Ditto.
3558 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3559 (*avx512pf_scatterpf<mode>df): Ditto.
3561 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3564 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3565 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3568 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3570 * common/config/arm/arm-common.c
3571 (arm_rewrite_mcpu): Handle multiple names.
3573 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3575 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3577 * gimple-builder.h (create_gimple_tmp): Delete.
3579 2014-01-27 Christian Bruel <christian.bruel@st.com>
3581 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3584 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3586 * config/pa/pa.md (call): Generate indirect long calls to non-local
3587 functions when outputing 32-bit code.
3588 (call_value): Likewise except for special call to buggy powf function.
3590 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3591 portable runtime and PIC indirect calls.
3592 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3593 and PIC call sequences. Use ldo instead of blr to set return register
3594 in PIC call sequence.
3596 2014-01-25 Walter Lee <walt@tilera.com>
3598 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3599 avoid clobbering a live register.
3601 2014-01-25 Walter Lee <walt@tilera.com>
3603 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3604 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3605 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3606 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3608 2014-01-25 Walter Lee <walt@tilera.com>
3610 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3611 arguments on even registers.
3612 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3614 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3615 (BIGGEST_ALIGNMENT): Ditto.
3616 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3618 2014-01-25 Walter Lee <walt@tilera.com>
3620 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3621 insns before bundling.
3622 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3624 2014-01-25 Walter Lee <walt@tilera.com>
3626 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3627 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3628 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3630 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3632 * config/mips/constraints.md (kl): Delete.
3633 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3634 define expands, using...
3635 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3636 instructions for MIPS16.
3637 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3638 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3640 2014-01-25 Walter Lee <walt@tilera.com>
3642 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3646 2014-01-25 Walter Lee <walt@tilera.com>
3648 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3649 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3651 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3653 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3656 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3658 * print-rtl.c (in_call_function_usage): New var.
3659 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3660 EXPR_LIST mode as mode and not as reg note name.
3663 * cfgloopmanip.c (copy_loop_info): If
3664 loop->warned_aggressive_loop_optimizations, make sure
3665 the flag is set in target loop too.
3667 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3669 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3671 * builtins.def: Likewise.
3672 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3673 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3674 * ira.c (ira_setup_eliminable_regset): Likewise.
3675 * omp-low.c (gate_expand_omp): Likewise.
3676 (execute_lower_omp): Likewise.
3677 (diagnose_sb_0): Likewise.
3678 (gate_diagnose_omp_blocks): Likewise.
3679 (simd_clone_clauses_extract): Likewise.
3682 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3684 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3685 correction for little endian...
3686 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3689 2014-01-24 Jeff Law <law@redhat.com>
3691 PR tree-optimization/59919
3692 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3693 for non-returning calls.
3695 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3697 * common/config/aarch64/aarch64-common.c
3698 (aarch64_rewrite_mcpu): Handle multiple names.
3699 * config/aarch64/aarch64.h
3700 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3702 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3704 * input.c (add_file_to_cache_tab): Handle the case where fopen
3707 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3710 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3711 from push operand if code of push isn't PRE_DEC.
3713 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3716 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3717 -mquad-memory-atomic. Update -mquad-memory documentation to say
3718 it is only used for non-atomic loads/stores.
3720 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3721 -mquad-memory or -mquad-memory-atomic switches.
3723 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3724 -mquad-memory-atomic to ISA 2.07 support.
3726 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3727 to separate support of normal quad word memory operations (ldq, stq)
3728 from the atomic quad word memory operations.
3730 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3731 support to separate non-atomic quad word operations from atomic
3732 quad word operations. Disable non-atomic quad word operations in
3733 little endian mode so that we don't have to swap words after the
3734 load and before the store.
3735 (quad_load_store_p): Add comment about atomic quad word support.
3736 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3737 options printed with -mdebug=reg.
3739 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3740 -mquad-memory-atomic as the test for whether we have quad word
3741 atomic instructions.
3742 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3743 or -mp8-vector are used, allow byte/half-word atomic operations.
3745 * config/rs6000/sync.md (load_lockedti): Insure that the address
3746 is a proper indexed or indirect address for the lqarx instruction.
3747 On little endian systems, swap the hi/lo registers after the lqarx
3749 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3750 insure the address is valid for the lqarx instruction.
3751 (store_conditionalti): Insure that the address is a proper indexed
3752 or indirect address for the stqcrx. instruction. On little endian
3753 systems, swap the hi/lo registers before doing the stqcrx.
3755 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3756 insure the address is valid for the stqcrx. instruction.
3758 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3759 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3760 type of quad memory support is available.
3762 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3765 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3766 there is a danger of looping.
3768 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3770 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3771 force flag_ira_loop_pressure if set via command line.
3773 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3775 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3776 (ashr_simd): New builtin handling DI mode.
3777 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3778 (aarch64_sshr_simddi): New match pattern.
3779 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3780 (vshrd_n_s64): Likewise.
3781 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3783 2014-01-23 Nick Clifton <nickc@redhat.com>
3785 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3786 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3787 favour of mcu specific scripts.
3788 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3791 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3792 Alex Velenko <Alex.Velenko@arm.com>
3794 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3795 (vaddv_s16): Likewise.
3796 (vaddv_s32): Likewise.
3797 (vaddv_u8): Likewise.
3798 (vaddv_u16): Likewise.
3799 (vaddv_u32): Likewise.
3800 (vaddvq_s8): Likewise.
3801 (vaddvq_s16): Likewise.
3802 (vaddvq_s32): Likewise.
3803 (vaddvq_s64): Likewise.
3804 (vaddvq_u8): Likewise.
3805 (vaddvq_u16): Likewise.
3806 (vaddvq_u32): Likewise.
3807 (vaddvq_u64): Likewise.
3808 (vaddv_f32): Likewise.
3809 (vaddvq_f32): Likewise.
3810 (vaddvq_f64): Likewise.
3811 (vmaxv_f32): Likewise.
3812 (vmaxv_s8): Likewise.
3813 (vmaxv_s16): Likewise.
3814 (vmaxv_s32): Likewise.
3815 (vmaxv_u8): Likewise.
3816 (vmaxv_u16): Likewise.
3817 (vmaxv_u32): Likewise.
3818 (vmaxvq_f32): Likewise.
3819 (vmaxvq_f64): Likewise.
3820 (vmaxvq_s8): Likewise.
3821 (vmaxvq_s16): Likewise.
3822 (vmaxvq_s32): Likewise.
3823 (vmaxvq_u8): Likewise.
3824 (vmaxvq_u16): Likewise.
3825 (vmaxvq_u32): Likewise.
3826 (vmaxnmv_f32): Likewise.
3827 (vmaxnmvq_f32): Likewise.
3828 (vmaxnmvq_f64): Likewise.
3829 (vminv_f32): Likewise.
3830 (vminv_s8): Likewise.
3831 (vminv_s16): Likewise.
3832 (vminv_s32): Likewise.
3833 (vminv_u8): Likewise.
3834 (vminv_u16): Likewise.
3835 (vminv_u32): Likewise.
3836 (vminvq_f32): Likewise.
3837 (vminvq_f64): Likewise.
3838 (vminvq_s8): Likewise.
3839 (vminvq_s16): Likewise.
3840 (vminvq_s32): Likewise.
3841 (vminvq_u8): Likewise.
3842 (vminvq_u16): Likewise.
3843 (vminvq_u32): Likewise.
3844 (vminnmv_f32): Likewise.
3845 (vminnmvq_f32): Likewise.
3846 (vminnmvq_f64): Likewise.
3848 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3850 * config/aarch64/aarch64-simd.md
3851 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3852 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3853 (*aarch64_mul3_elt<mode>): Likewise.
3854 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3855 (*aarch64_mul3_elt_to_64v2df): Likewise.
3856 (*aarch64_mla_elt<mode>): Likewise.
3857 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3858 (*aarch64_mls_elt<mode>): Likewise.
3859 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3860 (*aarch64_fma4_elt<mode>): Likewise.
3861 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3862 (*aarch64_fma4_elt_to_64v2df): Likewise.
3863 (*aarch64_fnma4_elt<mode>): Likewise.
3864 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3865 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3866 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3867 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3868 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3869 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3870 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3871 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3872 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3874 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3876 * config/aarch64/aarch64-simd.md
3877 (aarch64_be_checked_get_lane<mode>): New define_expand.
3878 * config/aarch64/aarch64-simd-builtins.def
3879 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3880 New builtin definition.
3881 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3882 Use new safe be builtin.
3884 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3886 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3888 (aarch64_be_st1<mode>): Likewise.
3889 (aarch_ld1<VALL:mode>): Define_expand modified.
3890 (aarch_st1<VALL:mode>): Likewise.
3891 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3892 (UNSPEC_ST1): Likewise.
3894 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3896 * config/microblaze/microblaze.md: Add trap insn and attribute
3898 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3900 PR preprocessor/58580
3901 * input.h (location_get_source_line): Take an additional line_size
3903 (void diagnostics_file_cache_fini): Declare new function.
3904 * input.c (struct fcache): New type.
3905 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3906 New static constants.
3907 (diagnostic_file_cache_init, total_lines_num)
3908 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3909 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3910 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3911 (get_next_line, read_next_line, goto_next_line, read_line_num):
3912 New static function definitions.
3913 (diagnostic_file_cache_fini): New function.
3914 (location_get_source_line): Take an additional output line_len
3915 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3917 * diagnostic.c (diagnostic_finish): Call
3918 diagnostic_file_cache_fini.
3919 (adjust_line): Take an additional input parameter for the length
3920 of the line, rather than calculating it with strlen.
3921 (diagnostic_show_locus): Adjust the use of
3922 location_get_source_line and adjust_line with respect to their new
3923 signature. While displaying a line now, do not stop at the first
3924 null byte. Rather, display the zero byte as a space and keep
3925 going until we reach the size of the line.
3926 * Makefile.in: Add vec.o to OBJS-libcommon
3928 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3929 Ilya Tocar <ilya.tocar@intel.com>
3931 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3932 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3933 (__builtin_ia32_kmov16): Ditto.
3934 * config/i386/i386.md (UNSPEC_KMOV): New.
3937 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3939 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3940 (_mm512_storeu_si512): Ditto.
3942 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3945 * rtl.h (get_referenced_operands): Declare.
3946 * recog.c (get_referenced_operands): New function.
3947 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3948 operands have been referenced when recording LO_SUM references.
3950 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3952 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3954 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3956 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3957 Enable for generic and recent AMD targets.
3959 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3961 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3962 ARG_SIZE note when adjustment was eliminated.
3964 2014-01-22 Jeff Law <law@redhat.com>
3966 PR tree-optimization/59597
3967 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3968 in file. Accept new argument REGISTERING and use it to modify
3969 dump output appropriately.
3970 (register_jump_thread): Corresponding changes.
3971 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3972 thread paths involving joiner blocks. Add code to dump cancelled
3973 jump threading paths.
3975 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3977 PR rtl-optimization/59477
3978 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3979 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3981 2014-01-22 Tom Tromey <tromey@redhat.com>
3983 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3985 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3987 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3989 PR rtl-optimization/59896
3990 * lra-constraints.c (process_alt_operands): Check unused note for
3991 matched operands of insn with no output reloads.
3993 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3995 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3996 (mips_move_from_gpr_cost): Likewise.
3998 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4000 PR rtl-optimization/59858
4001 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4002 ira_class_hard_regs_num.
4003 (process_alt_operands): Increase reject for dying matched operand.
4005 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4008 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4009 smaller than size, perform several stores or loads and stores
4010 at dst + count - size to store or copy all of size bytes, rather
4011 than just last modesize bytes.
4013 2014-01-20 DJ Delorie <dj@redhat.com>
4015 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4016 that CLOBBERs are REGs before propogating their values.
4018 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4021 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4022 (cgraph_inline_failed_type): New function.
4023 * cgraph.h (DEFCIFCODE): Add type.
4024 (cgraph_inline_failed_type_t): New enum.
4025 (cgraph_inline_failed_type): New prototype.
4026 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4027 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4028 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4029 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4030 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4031 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4032 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4033 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4034 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4035 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4036 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4037 OPTIMIZATION_MISMATCH.
4038 * tree-inline.c (expand_call_inline): Emit errors during
4039 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4041 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4044 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4045 mode attribute in insn output.
4047 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4049 * output.h (output_constant): Delete.
4050 * varasm.c (output_constant): Make private.
4052 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4054 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4056 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4059 * tree.h (fold_builtin_strcat): New prototype.
4060 * builtins.c (fold_builtin_strcat): No longer static. Add len
4061 argument, if non-NULL, don't call c_strlen. Optimize
4062 directly into __builtin_memcpy instead of __builtin_strcpy.
4063 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4064 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4066 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4068 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4069 for SImode_address_operand operands, having only a REG argument.
4071 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4073 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4074 loader name using mbig-endian.
4075 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4077 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4079 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4083 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4085 * config/aarch64/aarch64-protos.h
4086 (aarch64_cannot_change_mode_class_ptr): Declare.
4087 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4088 aarch64_cannot_change_mode_class_ptr): New.
4089 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4090 backend hook aarch64_cannot_change_mode_class.
4092 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4094 * common/config/aarch64/aarch64-common.c
4095 (aarch64_handle_option): Don't handle any option order logic here.
4096 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4097 selected_cpu, warn on architecture version mismatch.
4098 (aarch64_override_options): Fix parsing order for option strings.
4100 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4101 Iain Sandoe <iain@codesourcery.com>
4104 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4105 warning. Amend comment to reflect current functionality.
4107 2014-01-20 Richard Biener <rguenther@suse.de>
4110 * builtins.c (fold_builtin_strcat): Remove case better handled
4111 by tree-ssa-strlen.c.
4113 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4115 * config/aarch64/aarch64.opt
4116 (mcpu, march, mtune): Make case-insensitive.
4118 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4121 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4122 if operands[1] is a REG or ZERO_EXTEND of a REG.
4124 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4126 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4128 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4130 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4131 long non-pic millicode calls.
4133 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4135 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4137 2014-01-19 Kito Cheng <kito@0xlab.org>
4139 * builtins.c (expand_movstr): Check movstr expand done or fail.
4141 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4142 H.J. Lu <hongjiu.lu@intel.com>
4145 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4146 to DImode for zero-extended addresses.
4148 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4150 PR rtl-optimization/57763
4151 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4152 on the new indirect jump_insn and increment LABEL_NUSES (label).
4154 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4158 * config.gcc (x86_archs): New variable.
4159 (x86_64_archs): Likewise.
4160 (x86_cpus): Likewise.
4161 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4162 --with-arch/--with-cpu= options.
4163 Support --with-arch=/--with-cpu={nehalem,westmere,
4164 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4166 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4168 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4169 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4171 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4173 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4175 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4178 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4179 clear cpp_get_options (parse_in)->warn_unused_macros for
4180 ix86_target_macros_internal with cpp_define.
4182 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4184 * jump.c (delete_related_insns): Keep (use (insn))s.
4185 * reorg.c (redundant_insn): Check for barriers too.
4187 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4189 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4191 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4193 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4194 call to $$dyncall when TARGET_LONG_CALLS is true.
4196 2014-01-17 Jeff Law <law@redhat.com>
4198 * ree.c (combine_set_extension): Temporarily disable test for
4199 changing number of hard registers.
4201 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4204 * ipa-inline-analysis.c (inline_free_summary):
4205 Do not free summary of aliases.
4207 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4210 * gimplify.c (gimplify_expr): Use create_tmp_var
4211 instead of create_tmp_var_raw. If cond doesn't have
4212 integral type, don't add the IFN_ANNOTATE builtin at all.
4214 2014-01-17 Martin Jambor <mjambor@suse.cz>
4217 * ipa-cp.c (prev_edge_clone): New variable.
4218 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4219 Also resize prev_edge_clone vector.
4220 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4221 (ipcp_edge_removal_hook): New function.
4222 (ipcp_driver): Register ipcp_edge_removal_hook.
4224 2014-01-17 Andrew Pinski <apinski@cavium.com>
4225 Steve Ellcey <sellcey@mips.com>
4228 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4231 2014-01-17 Jeff Law <law@redhat.com>
4234 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4235 so that pass_ccp runs first.
4237 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4239 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4240 (ix86_adjust_cost): Use !TARGET_XXX.
4241 (do_reorder_for_imul): Likewise.
4242 (swap_top_of_ready_list): Likewise.
4243 (ix86_sched_reorder): Likewise.
4245 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4247 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4248 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4249 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4250 (intel_memset): New. Duplicate slm_memset.
4251 (intel_cost): New. Duplicate slm_cost.
4252 (m_INTEL): New macro.
4253 (processor_target_table): Add "intel".
4254 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4255 with PROCESSOR_INTEL for "intel".
4256 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4257 PROCESSOR_SILVERMONT.
4258 (ix86_issue_rate): Likewise.
4259 (ix86_adjust_cost): Likewise.
4260 (ia32_multipass_dfa_lookahead): Likewise.
4261 (swap_top_of_ready_list): Likewise.
4262 (ix86_sched_reorder): Likewise.
4263 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4264 instead of TARGET_OPT_AGU.
4265 * config/i386/i386.h (TARGET_INTEL): New.
4266 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4267 (processor_type): Add PROCESSOR_INTEL.
4268 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4269 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4271 2014-01-17 Marek Polacek <polacek@redhat.com>
4274 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4277 2014-01-17 Richard Biener <rguenther@suse.de>
4279 PR tree-optimization/46590
4280 * opts.c (default_options_table): Add entries for
4281 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4282 all enabled at -O1 but not for -Og.
4283 * common.opt (fbranch-count-reg): Remove Init(1).
4284 (fmove-loop-invariants): Likewise.
4285 (ftree-pta): Likewise.
4287 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4289 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4290 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4291 decls to at least the GCC 4.8 used alignments.
4294 * tree-nested.c (convert_nonlocal_reference_stmt,
4295 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4296 of GIMPLE_BIND stmts, adjust associated decls.
4298 2014-01-17 Richard Biener <rguenther@suse.de>
4300 PR tree-optimization/46590
4301 * vec.h (vec<>::bseach): New member function implementing
4302 binary search according to C89 bsearch.
4303 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4304 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4305 bitmap pointer again. Make accesses_in_loop a flat array.
4306 (mem_ref_obstack): New global.
4307 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4308 (mark_ref_stored): Likewise.
4309 (ref_indep_loop_p_2): Likewise.
4310 (set_ref_stored_in_loop): New helper function.
4311 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4312 (memref_free): Adjust.
4313 (record_mem_ref_loc): Simplify.
4314 (gather_mem_refs_stmt): Adjust.
4315 (sort_locs_in_loop_postorder_cmp): New function.
4316 (analyze_memory_references): Sort accesses_in_loop after
4317 loop postorder number.
4318 (find_ref_loc_in_loop_cmp): New function.
4319 (for_all_locs_in_loop): Find relevant cluster of locs in
4320 accesses_in_loop and iterate without recursion.
4321 (execute_sm): Avoid uninit warning.
4322 (struct ref_always_accessed): Simplify.
4323 (ref_always_accessed::operator ()): Likewise.
4324 (ref_always_accessed_p): Likewise.
4325 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4326 loop postorder numbers here.
4327 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4330 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4333 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4334 on decls for which assemble_alias has been called.
4336 2014-01-17 Nick Clifton <nickc@redhat.com>
4338 * config/msp430/msp430.opt: (mcpu): New option.
4339 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4340 (msp430_option_override): Parse target_cpu. If the MCU name
4341 matches a generic string, clear target_mcu.
4342 (msp430_attr): Allow numeric interrupt values up to 63.
4343 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4344 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4346 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4348 * config/msp430/msp430.md (popm): Use %J rather than %I.
4349 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4350 (addhi_cy_i): Use immediate_operand for operand 2.
4351 * doc/invoke.texi: Document -mcpu option.
4353 2014-01-17 Richard Biener <rguenther@suse.de>
4355 PR rtl-optimization/38518
4356 * df.h (df_analyze_loop): Declare.
4357 * df-core.c: Include cfgloop.h.
4358 (df_analyze_1): Split out main part of df_analyze.
4359 (df_analyze): Adjust.
4360 (loop_inverted_post_order_compute): New function.
4361 (loop_post_order_compute): Likewise.
4362 (df_analyze_loop): New function avoiding whole-function
4364 * loop-invariant.c (find_defs): Use df_analyze_loop.
4365 (find_invariants): Adjust.
4366 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4368 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4370 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4371 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4373 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4375 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4376 traversal when removing references.
4378 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4381 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4383 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4386 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4387 pushing a reload for an autoinc when we had previously reloaded an
4388 inner part of the address.
4390 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4392 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4394 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4395 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4396 when not giving up or versioning for alias only because of
4398 (vect_analyze_data_ref_dependences): Set to true.
4399 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4401 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4402 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4406 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4409 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4410 operand 0 predicate for gathers, use a new pseudo as subtarget.
4412 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4415 * lra-constraints.c (process_alt_operands): Add printing debug info.
4416 Check absence of input/output reloads for matched operands too.
4418 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4420 PR rtl-optimization/59835
4421 * ira.c (ira_init_register_move_cost): Increase cost for
4424 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4426 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4428 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4431 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4432 non-register objects. Use gen_(high/low)part more consistently.
4435 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4438 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4439 endian support, remove tests for WORDS_BIG_ENDIAN.
4440 (p8_mfvsrd_3_<mode>): Likewise.
4441 (reload_gpr_from_vsx<mode>): Likewise.
4442 (reload_gpr_from_vsxsf): Likewise.
4443 (p8_mfvsrd_4_disf): Likewise.
4445 2014-01-16 Richard Biener <rguenther@suse.de>
4447 PR rtl-optimization/46590
4448 * lcm.c (compute_antinout_edge): Use postorder iteration.
4449 (compute_laterin): Use inverted postorder iteration.
4451 2014-01-16 Nick Clifton <nickc@redhat.com>
4454 * varasm.c (output_constant): Return the number of bytes actually
4456 (output_constructor_array_range): Update the field size with the
4457 number of bytes emitted by output_constant.
4458 (output_constructor_regular_field): Likewise. Also do not
4459 complain if the total number of bytes emitted is now greater
4460 than the expected fieldpos.
4461 * output.h (output_constant): Update prototype and descriptive comment.
4463 2014-01-16 Marek Polacek <polacek@redhat.com>
4466 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4467 it is error_mark_node.
4469 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4471 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4472 VALID_AVX256_REG_OR_OI_MODE.
4474 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4476 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4477 current procedure should be profiled.
4479 2014-01-15 Andrew Pinski <apinski@cavium.com>
4481 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4482 of moving from/to the STACK_REG register class.
4484 2014-01-15 Richard Henderson <rth@redhat.com>
4487 * reginfo.c (global_regs_decl): Globalize.
4488 * rtl.h (global_regs_decl): Declare.
4489 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4490 reserved via global_regs.
4492 2014-01-15 Teresa Johnson <tejohnson@google.com>
4494 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4496 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4498 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4499 and vmulosh rather than call gen_vec_widen_smult_*.
4500 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4501 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4502 (vec_widen_smult_even_v16qi): Likewise.
4503 (vec_widen_umult_even_v8hi): Likewise.
4504 (vec_widen_smult_even_v8hi): Likewise.
4505 (vec_widen_umult_odd_v16qi): Likewise.
4506 (vec_widen_smult_odd_v16qi): Likewise.
4507 (vec_widen_umult_odd_v8hi): Likewise.
4508 (vec_widen_smult_odd_v8hi): Likewise.
4509 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4510 vmuloub rather than call gen_vec_widen_umult_*.
4511 (vec_widen_umult_lo_v16qi): Likewise.
4512 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4513 vmulosb rather than call gen_vec_widen_smult_*.
4514 (vec_widen_smult_lo_v16qi): Likewise.
4515 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4516 rather than call gen_vec_widen_umult_*.
4517 (vec_widen_umult_lo_v8hi): Likewise.
4518 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4519 rather than call gen_vec_widen_smult_*.
4520 (vec_widen_smult_lo_v8hi): Likewise.
4522 2014-01-15 Jeff Law <law@redhat.com>
4524 PR tree-optimization/59747
4525 * ree.c (find_and_remove_re): Properly handle case where a second
4526 eliminated extension requires widening a copy created for elimination
4527 of a prior extension.
4528 (combine_set_extension): Ensure that the number of hard regs needed
4529 for a destination register does not change when we widen it.
4531 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4533 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4534 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4535 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4536 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4537 (avr-*-rtems*): Likewise.
4538 (bfin*-rtems*): Likewise.
4539 (moxie-*-rtems*): Likewise.
4540 (h8300-*-rtems*): Likewise.
4541 (i[34567]86-*-rtems*): Likewise.
4542 (lm32-*-rtems*): Likewise.
4543 (m32r-*-rtems*): Likewise.
4544 (m68k-*-rtems*): Likewise.
4545 (microblaze*-*-rtems*): Likewise.
4546 (mips*-*-rtems*): Likewise.
4547 (powerpc-*-rtems*): Likewise.
4548 (sh-*-rtems*): Likewise.
4549 (sparc-*-rtems*): Likewise.
4550 (sparc64-*-rtems*): Likewise.
4551 (v850-*-rtems*): Likewise.
4552 (m32c-*-rtems*): Likewise.
4554 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4556 PR rtl-optimization/59511
4557 * ira.c (ira_init_register_move_cost): Use memory costs for some
4558 cases of register move cost calculations.
4559 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4560 instead of BB frequency.
4561 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4562 * lra-assigns.c (find_hard_regno_for): Ditto.
4564 2014-01-15 Richard Biener <rguenther@suse.de>
4566 PR tree-optimization/59822
4567 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4568 (vectorizable_load): Use it to hoist defs of uses of invariant
4569 loads out of the loop.
4571 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4572 Kugan Vivekanandarajah <kuganv@linaro.org>
4575 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4578 2014-01-15 Richard Biener <rguenther@suse.de>
4580 PR rtl-optimization/59802
4581 * lcm.c (compute_available): Use inverted postorder to seed
4582 the initial worklist.
4584 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4587 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4588 ADDR_REGS for invalid symrefs in non-PIC code.
4590 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4593 * builtins.c (determine_block_size): Initialize *probable_max_size
4594 even if len_rtx is CONST_INT.
4596 2014-01-14 Andrew Pinski <apinski@cavium.com>
4598 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4599 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4600 (cortexa53_tunings): Likewise.
4601 (aarch64_sched_issue_rate): New function.
4602 (TARGET_SCHED_ISSUE_RATE): Define.
4604 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4606 * ira-costs.c (find_costs_and_classes): Add missed
4607 ira_init_register_move_cost_if_necessary.
4609 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4612 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4614 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4617 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4618 to indicate if type is used for function return value. Warn ABI
4619 change if the vector mode isn't available for function return value.
4620 (ix86_function_arg_advance): Pass false to type_natural_mode.
4621 (ix86_function_arg): Likewise.
4622 (ix86_gimplify_va_arg): Likewise.
4623 (function_arg_32): Don't warn ABI change.
4624 (ix86_function_value): Pass true to type_natural_mode.
4625 (ix86_return_in_memory): Likewise.
4626 (ix86_struct_value_rtx): Removed.
4627 (TARGET_STRUCT_VALUE_RTX): Likewise.
4629 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4631 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4632 converting a conditional jump into a conditional return.
4634 2014-01-14 Richard Biener <rguenther@suse.de>
4636 PR tree-optimization/58921
4637 PR tree-optimization/59006
4638 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4639 hoisting invariant stmts.
4640 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4641 invariant loads on the preheader edge if possible.
4643 2014-01-14 Joey Ye <joey.ye@arm.com>
4645 * doc/plugin.texi (Building GCC plugins): Update to C++.
4647 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4649 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4650 (_mm_rcp28_round_ss): Ditto.
4651 (_mm_rsqrt28_round_sd): Ditto.
4652 (_mm_rsqrt28_round_ss): Ditto.
4653 (_mm_rcp28_sd): Ditto.
4654 (_mm_rcp28_ss): Ditto.
4655 (_mm_rsqrt28_sd): Ditto.
4656 (_mm_rsqrt28_ss): Ditto.
4657 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4658 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4659 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4660 (IX86_BUILTIN_RCP28SD): Ditto.
4661 (IX86_BUILTIN_RCP28SS): Ditto.
4662 (IX86_BUILTIN_RSQRT28SD): Ditto.
4663 (IX86_BUILTIN_RSQRT28SS): Ditto.
4664 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4665 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4666 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4667 (ix86_expand_special_args_builtin): Expand new FTYPE.
4668 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4669 (srcp14<mode>): Make insn unary.
4670 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4671 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4672 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4673 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4674 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4675 Fix rounding: make it SAE only.
4676 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4678 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4680 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4681 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4682 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4683 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4684 (round_saeonly_mask_scalar_operand4): Ditto.
4685 (round_saeonly_mask_scalar_op3): Ditto.
4686 (round_saeonly_mask_scalar_op4): Ditto.
4688 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4690 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4691 Implement -maltivec=be for vec_insert and vec_extract.
4693 2014-01-10 DJ Delorie <dj@redhat.com>
4695 * config/msp430/msp430.md (call_internal): Don't allow memory
4696 references with SP as the base register.
4697 (call_value_internal): Likewise.
4698 * config/msp430/constraints.md (Yc): New. For memory references
4699 that don't use SP as a base register.
4701 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4702 "an integer without a # prefix"
4703 * config/msp430/msp430.md (epilogue_helper): Use it.
4705 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4708 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4709 AVX512F gather builtins.
4710 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4711 on gather decls with INTEGER_TYPE masktype.
4712 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4713 directly into the builtin rather than hoisting it before loop.
4715 PR tree-optimization/59387
4716 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4717 (scev_const_prop): If folded_casts and type has undefined overflow,
4718 use force_gimple_operand instead of force_gimple_operand_gsi and
4719 for each added stmt if it is assign with
4720 arith_code_with_undefined_signed_overflow, call
4721 rewrite_to_defined_overflow.
4722 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4723 gimple-fold.h instead.
4724 (arith_code_with_undefined_signed_overflow,
4725 rewrite_to_defined_overflow): Moved to ...
4726 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4727 rewrite_to_defined_overflow): ... here. No longer static.
4728 Include gimplify-me.h.
4729 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4730 rewrite_to_defined_overflow): New prototypes.
4732 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4734 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4736 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4738 * builtins.c (get_object_alignment_2): Minor tweak.
4739 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4741 2014-01-13 Christian Bruel <christian.bruel@st.com>
4743 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4744 optimized non constant lengths.
4746 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4749 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4750 load as __atomic_load_N if possible.
4752 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4754 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4756 (rs6000_expand_builtin): Adjust call.
4758 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4761 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4762 * config/rs6000/rs6000.c: Include target-globals.h.
4763 (rs6000_set_current_function): Instead of doing target_reinit
4764 unconditionally, use save_target_globals_default_opts and
4765 restore_target_globals.
4767 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4769 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4770 (rs6000_expand_builtin): Handle mffs and mtfsf.
4771 (rs6000_init_builtins): Define mffs and mtfsf.
4772 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4773 (rs6000_mffs): New pattern.
4774 (rs6000_mtfsf): New pattern.
4776 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4778 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4779 Start narrowing with START. Apply candidate-use pair
4780 and check overall cost in narrowing.
4781 (iv_ca_prune): Pass new argument.
4783 2014-01-10 Jeff Law <law@redhat.com>
4786 * ree.c (combine_reaching_defs): Ensure the defining statement
4787 occurs before the extension when optimizing extensions with
4788 different source and destination hard registers.
4790 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4793 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4794 vtables into the type inheritance graph.
4796 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4798 PR rtl-optimization/59754
4799 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4800 modes in the REGNO != REGNO case.
4802 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4804 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4806 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4808 PR tree-optimization/59745
4809 * tree-predcom.c (tree_predictive_commoning_loop): Call
4810 free_affine_expand_cache if giving up because components is NULL.
4812 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4813 GC in payload of target_globals struct instead of allocating them on
4814 the heap and the larger structs separately using GC.
4815 * target-globals.h (struct target_globals): Make regs, hard_regs,
4816 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4817 of GTY((skip)) and change type to void *.
4818 (reset_target_globals): Cast loads from those fields to corresponding
4821 2014-01-10 Steve Ellcey <sellcey@mips.com>
4824 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4825 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4826 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4828 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4831 * aarch64-modes.def (CC_Zmode): New flags mode.
4832 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4833 represents an equality.
4834 (aarch64_get_condition_code): Handle CC_Zmode.
4835 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4837 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4839 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4840 extraction in good case.
4842 2014-01-10 Richard Biener <rguenther@suse.de>
4844 PR tree-optimization/59374
4845 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4846 checking after SLP discovery. Mark stmts not participating
4847 in any SLP instance properly.
4849 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4851 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4852 when handling a SET rtx.
4854 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4856 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4857 (cortex-a57): Likewise.
4858 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4860 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4862 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4863 non-iwmmxt builtins.
4865 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4869 * ipa-devirt.c record_target_from_binfo): Take as argument
4870 stack of binfos and lookup matching one for virtual inheritance.
4871 (possible_polymorphic_call_targets_1): Update.
4873 2014-01-10 Huacai Chen <chenhc@lemote.com>
4875 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4876 kernel strings for Loongson-2E/2F/3A.
4878 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4881 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4882 is_gimple_call before calling gimple_call_internal_p.
4884 2014-01-09 Steve Ellcey <sellcey@mips.com>
4886 * Makefile.in (TREE_FLOW_H): Remove.
4887 (TREE_SSA_H): Add file names from tree-flow.h.
4888 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4889 * tree.h: Remove tree-flow.h reference.
4890 * hash-table.h: Remove tree-flow.h reference.
4891 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4892 reference with tree-ssa-loop.h.
4894 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4896 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4897 default element-order behavior for -maltivec.
4898 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4899 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4900 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4901 when targeting big endian, at least for now.
4902 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4904 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4907 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4908 var satisfies use_register_for_decl, just take into account type
4909 alignment, rather than decl alignment.
4911 PR tree-optimization/59622
4912 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4913 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4914 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4915 Don't devirtualize for inplace at all. For targets.length () == 1,
4916 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4918 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4920 * config/i386/i386.md (cpu): Remove the unused btver1.
4922 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4924 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4926 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4929 * tree-core.h (struct target_globals): New forward declaration.
4930 (struct tree_target_option): Add globals field.
4931 * tree.h (TREE_TARGET_GLOBALS): Define.
4932 (prepare_target_option_nodes_for_pch): New prototype.
4933 * target-globals.h (struct target_globals): Define even if
4935 * tree.c (prepare_target_option_node_for_pch,
4936 prepare_target_option_nodes_for_pch): New functions.
4937 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4938 * config/i386/i386.c: Include target-globals.h.
4939 (ix86_set_current_function): Instead of doing target_reinit
4940 unconditionally, use save_target_globals_default_opts and
4941 restore_target_globals.
4943 2014-01-09 Richard Biener <rguenther@suse.de>
4945 PR tree-optimization/59715
4946 * tree-cfg.h (split_critical_edges): Declare.
4947 * tree-cfg.c (split_critical_edges): Export.
4948 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4950 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4952 * cfgexpand.c (expand_stack_vars): Optionally disable
4953 asan stack protection.
4954 (expand_used_vars): Likewise.
4955 (partition_stack_vars): Likewise.
4956 * asan.c (asan_emit_stack_protection): Optionally disable
4957 after return stack usage.
4958 (instrument_derefs): Optionally disable memory access instrumentation.
4959 (instrument_builtin_call): Likewise.
4960 (instrument_strlen_call): Likewise.
4961 (asan_protect_global): Optionally disable global variables protection.
4962 * doc/invoke.texi: Added doc for new options.
4963 * params.def: Added new options.
4964 * params.h: Likewise.
4966 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4968 PR rtl-optimization/59724
4969 * ifcvt.c (cond_exec_process_if_block): Don't call
4970 flow_find_head_matching_sequence with 0 longest_match.
4971 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4972 non-active insns if !stop_after.
4973 (try_head_merge_bb): Revert 2014-01-07 changes.
4975 2014-01-08 Jeff Law <law@redhat.com>
4977 * ree.c (get_sub_rtx): New function, extracted from...
4978 (merge_def_and_ext): Here.
4979 (combine_reaching_defs): Use get_sub_rtx.
4981 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4983 * cgraph.h (varpool_variable_node): Do not choke on null node.
4985 2014-01-08 Catherine Moore <clm@codesourcery.com>
4987 * config/mips/mips.md (simple_return): Attempt to use JRC
4989 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4991 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4993 PR rtl-optimization/59137
4994 * reorg.c (steal_delay_list_from_target): Call update_block for
4996 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4998 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5000 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5001 two duplicate entries.
5003 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5006 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5008 * config/mips/mips.c (mips_truncated_op_cost): New function.
5009 (mips_rtx_costs): Adjust test for BADDU.
5010 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5012 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5014 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5015 (*baddu_si): ...this new pattern.
5017 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5020 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5022 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5025 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5027 (expand_expr, expand_normal): Adjust.
5028 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5029 inner_reference_p. Use inner_reference_p to expand inner references.
5030 (store_expr): Adjust.
5031 * cfgexpand.c (expand_call_stmt): Adjust.
5033 2014-01-08 Rong Xu <xur@google.com>
5035 * gcov-io.c (gcov_var): Move from gcov-io.h.
5036 (gcov_position): Ditto.
5037 (gcov_is_error): Ditto.
5038 (gcov_rewrite): Ditto.
5039 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5040 only part to libgcc/libgcov.h.
5042 2014-01-08 Marek Polacek <polacek@redhat.com>
5045 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5047 2014-01-08 Marek Polacek <polacek@redhat.com>
5050 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5052 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5054 PR rtl-optimization/59649
5055 * stor-layout.c (get_mode_bounds): For BImode return
5056 0 and STORE_FLAG_VALUE.
5058 2014-01-08 Richard Biener <rguenther@suse.de>
5061 * gimple.h (is_gimple_builtin_call): Remove.
5062 (gimple_builtin_call_types_compatible_p): New.
5063 (gimple_call_builtin_p): New overload.
5064 * gimple.c (is_gimple_builtin_call): Remove.
5065 (validate_call): Rename to ...
5066 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5068 (validate_type): New static function.
5069 (gimple_call_builtin_p): New overload and adjust.
5070 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5071 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5072 (gimple_fold_stmt_to_constant_1): Likewise.
5073 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5075 2014-01-08 Richard Biener <rguenther@suse.de>
5078 * gimplify.c (gimplify_expr): Gimplify register-register type
5079 VIEW_CONVERT_EXPRs to separate stmts.
5081 2014-01-07 Jeff Law <law@redhat.com>
5084 * ree.c (combine_set_extension): Handle case where source
5085 and destination registers in an extension insn are different.
5086 (combine_reaching_defs): Allow source and destination registers
5087 in extension to be different under limited circumstances.
5088 (add_removable_extension): Remove restriction that the
5089 source and destination registers in the extension are the same.
5090 (find_and_remove_re): Emit a copy from the extension's
5091 destination to its source after the defining insn if
5092 the source and destination registers are different.
5095 * ifcvt.c (merge_if_block): If we are merging a block with more than
5096 one successor with a block with no successors, remove any BARRIER
5097 after the second block.
5099 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5101 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5103 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5106 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5107 for 14-bit register offsets when INT14_OK_STRICT is false.
5109 2014-01-07 Roland Stigge <stigge@antcom.de>
5110 Michael Meissner <meissner@linux.vnet.ibm.com>
5113 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5114 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5116 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5118 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5121 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5123 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5124 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5125 rtx is const0_rtx or not.
5127 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5130 * target-globals.c (save_target_globals): Remove this_fn_optab
5132 * toplev.c: Include optabs.h.
5133 (target_reinit): Temporarily restore the global options if another
5134 set of options are in force.
5136 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5138 PR rtl-optimization/58668
5139 * cfgcleanup.c (flow_find_cross_jump): Don't count
5140 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5141 to determine what is counted.
5142 (flow_find_head_matching_sequence): Use active_insn_p to determine
5144 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5146 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5147 determine what is counted.
5149 PR tree-optimization/59643
5150 * tree-predcom.c (split_data_refs_to_components): If one dr is
5151 read and one write, determine_offset fails and the write isn't
5152 in the bad component, just put the read into the bad component.
5154 2014-01-07 Mike Stump <mikestump@comcast.net>
5155 Jakub Jelinek <jakub@redhat.com>
5158 * tree-core.h (struct tree_optimization_option): Change optabs
5159 type from unsigned char * to void *.
5160 * optabs.c (init_tree_optimization_optabs): Adjust
5161 TREE_OPTIMIZATION_OPTABS initialization.
5163 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5166 * config/i386/i386.h (struct machine_function): Add
5167 no_drap_save_restore field.
5168 * config/i386/i386.c (ix86_save_reg): Use
5169 !cfun->machine->no_drap_save_restore instead of
5170 crtl->stack_realign_needed.
5171 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5172 this function clears frame_pointer_needed. Set
5173 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5174 and DRAP reg is needed.
5176 2014-01-06 Marek Polacek <polacek@redhat.com>
5179 * doc/implement-c.texi: Mention that other integer types are
5180 permitted as bit-field types in strictly conforming mode.
5182 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5184 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5187 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5189 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5191 2014-01-06 Martin Jambor <mjambor@suse.cz>
5194 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5196 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5198 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5202 * var-tracking.c (add_stores): Preserve the value of the source even if
5203 we don't record the store.
5205 2014-01-06 Terry Guo <terry.guo@arm.com>
5207 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5209 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5212 * config/darwin.c (darwin_function_section): Adjust return values to
5213 correspond to optimisation changes made in r206070.
5215 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5217 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5218 from prefetch_block tune setting.
5219 (nocona_cost): Correct size of prefetch block to 64.
5221 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5223 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5224 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5225 used to save the static chain register in the computation of the offset
5226 from which the FP registers need to be restored.
5228 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5230 PR tree-optimization/59519
5231 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5232 ICE if get_current_def (current_new_name) is already non-NULL, as long
5233 as it is a phi result of some other phi in *new_exit_bb that has
5236 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5237 or vmovdqu* for misaligned_operand.
5238 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5239 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5240 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5241 aligned_mem for AVX512F masked aligned load and store builtins and for
5244 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5246 PR tree-optimization/59651
5247 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5248 Address range for negative step should be added by TYPE_SIZE_UNIT.
5250 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5252 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5253 overlapping registers also for registers other than the stack pointer.
5255 2014-01-03 Marek Polacek <polacek@redhat.com>
5258 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5261 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5264 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5267 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5268 (push splitter): Use <P:MODE_SIZE> instead of
5269 GET_MODE_SIZE (<P:MODE>mode).
5270 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5271 (mov -1, reg peephole2): Likewise.
5272 * config/i386/sse.md (*mov<mode>_internal,
5273 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5274 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5275 *<code><mode>3, *andnot<mode>3<mask_name>,
5276 <mask_codefor><code><mode>3<mask_name>): Likewise.
5277 * config/i386/subst.md (mask_mode512bit_condition,
5278 sd_mask_mode512bit_condition): Likewise.
5280 2014-01-02 Xinliang David Li <davidxl@google.com>
5282 PR tree-optimization/59303
5283 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5284 (dump_predicates): Better output format.
5285 (pred_equal_p): New function.
5286 (is_neq_relop_p): Ditto.
5287 (is_neq_zero_form_p): Ditto.
5288 (pred_expr_equal_p): Ditto.
5289 (pred_neg_p): Ditto.
5290 (simplify_pred): Ditto.
5291 (simplify_preds_2): Ditto.
5292 (simplify_preds_3): Ditto.
5293 (simplify_preds_4): Ditto.
5294 (simplify_preds): Ditto.
5296 (push_to_worklist): Ditto.
5297 (get_pred_info_from_cmp): Ditto.
5298 (is_degenerated_phi): Ditto.
5299 (normalize_one_pred_1): Ditto.
5300 (normalize_one_pred): Ditto.
5301 (normalize_one_pred_chain): Ditto.
5302 (normalize_preds): Ditto.
5303 (normalize_cond_1): Remove function.
5304 (normalize_cond): Ditto.
5305 (is_gcond_subset_of): Ditto.
5306 (is_subset_of_any): Ditto.
5307 (is_or_set_subset_of): Ditto.
5308 (is_and_set_subset_of): Ditto.
5309 (is_norm_cond_subset_of): Ditto.
5310 (pred_chain_length_cmp): Ditto.
5311 (convert_control_dep_chain_into_preds): Type change.
5312 (find_predicates): Ditto.
5313 (find_def_preds): Ditto.
5314 (destroy_predicates_vecs): Ditto.
5315 (find_matching_predicates_in_rest_chains): Ditto.
5316 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5317 (is_pred_expr_subset): Ditto.
5318 (is_pred_chain_subset_of): Ditto.
5319 (is_included_in): Ditto.
5320 (is_superset_of): Ditto.
5322 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5324 Update copyright years.
5326 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5328 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5329 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5330 config/arc/arc.md, config/arc/arc.opt,
5331 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5332 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5333 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5334 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5335 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5336 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5337 vtable-verify.c, vtable-verify.h: Use the standard form for the
5340 2014-01-02 Tobias Burnus <burnus@net-b.de>
5342 * gcc.c (process_command): Update copyright notice dates.
5343 * gcov-dump.c: Ditto.
5345 * doc/cpp.texi: Bump @copying's copyright year.
5346 * doc/cppinternals.texi: Ditto.
5347 * doc/gcc.texi: Ditto.
5348 * doc/gccint.texi: Ditto.
5349 * doc/gcov.texi: Ditto.
5350 * doc/install.texi: Ditto.
5351 * doc/invoke.texi: Ditto.
5353 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5355 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5357 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5359 * config/i386/sse.md (*mov<mode>_internal): Guard
5360 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5362 PR rtl-optimization/59647
5363 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5364 new_rtx into UNSIGNED_FLOAT rtxes.
5366 Copyright (C) 2014 Free Software Foundation, Inc.
5368 Copying and distribution of this file, with or without modification,
5369 are permitted in any medium without royalty provided the copyright
5370 notice and this notice are preserved.