1 2014-03-24 Tobias Burnus <burnus@net-b.de>
3 * doc/invoke.texi (-flto): Expand section about
4 using static libraries with LTO.
6 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
8 PR rtl-optimization/60501
9 * optabs.def (addptr3_optab): New optab.
10 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
11 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
12 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
14 * lra.c (emit_add3_insn): Use the addptr pattern if available.
16 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
18 2014-03-24 Ulrich Drepper <drepper@gmail.com>
20 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
23 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
24 (_mm256_undefined_ps): Define.
25 (_mm256_undefined_pd): Define.
26 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
27 (_mm_undefined_pd): Define.
28 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
29 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
30 (_mm512_undefined_ps): Define.
31 (_mm512_undefined_pd): Define.
33 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
35 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
37 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
38 (lshr_simd): DI mode added.
39 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
40 (aarch64_ushr_simddi): Likewise.
41 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
42 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
43 (vshrd_n_u64): Likewise.
45 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
47 * Makefile.in (s-macro_list): Depend on cc1.
49 2014-03-23 Teresa Johnson <tejohnson@google.com>
51 * ipa-utils.c (ipa_print_order): Use specified dump file.
53 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
55 PR rtl-optimization/60601
56 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
58 * gcc.c (eval_spec_function): Initialize save_growing_value.
60 2014-03-22 Jakub Jelinek <jakub@redhat.com>
63 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
64 code == MINUS_EXPR, never swap op0 with op1.
66 * toplev.c (init_local_tick): Avoid signed integer multiplication
68 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
69 shift by first operand's bitsize.
71 2014-03-21 Jakub Jelinek <jakub@redhat.com>
74 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
76 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
77 TARGET_ISA_64BIT_P(x).
79 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
81 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
82 pattern for vector nor instead of subtract from splat(-1).
83 (altivec_expand_vec_perm_const_le): Likewise.
85 2014-03-21 Richard Henderson <rth@twiddle.net>
88 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
89 related insns after epilogue_completed.
91 2014-03-21 Martin Jambor <mjambor@suse.cz>
94 * cgraph.h (symtab_node): New flag body_removed.
95 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
97 * symtab.c (dump_symtab_base): Dump body_removed flag.
98 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
99 had their bodies removed.
101 2014-03-21 Martin Jambor <mjambor@suse.cz>
104 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
107 2014-03-21 Richard Biener <rguenther@suse.de>
109 PR tree-optimization/60577
110 * tree-core.h (struct tree_base): Document nothrow_flag use
111 in VAR_DECL_NONALIASED.
112 * tree.h (VAR_DECL_NONALIASED): New.
113 (may_be_aliased): Adjust.
114 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
116 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
118 * expr.c (expand_expr_real_1): Remove outdated comment.
120 2014-03-20 Jakub Jelinek <jakub@redhat.com>
123 * ira.c (adjust_cleared_regs): Call copy_rtx on
124 *reg_equiv[REGNO (loc)].src_p before passing it to
125 simplify_replace_fn_rtx.
128 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
129 into CONST, put pic register as first operand of PLUS. Use
130 gen_const_mem for both 32-bit and 64-bit PIC got loads.
132 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
134 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
136 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
138 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
139 around for store forwarding issue in the FPU on the UT699.
140 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
141 loads and operations if -mfix-ut699 is specified.
142 (divtf3_hq): Tweak attribute.
143 (sqrttf2_hq): Likewise.
145 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
147 * calls.c (store_one_arg): Remove incorrect const qualification on the
148 type of the temporary.
149 * cfgexpand.c (expand_return): Likewise.
150 * expr.c (expand_constructor): Likewise.
151 (expand_expr_real_1): Likewise.
153 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
155 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
158 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
161 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
163 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
165 * config/arm/aarch-common-protos.h
166 (alu_cost_table): Fix spelling of "extend".
167 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
169 2014-03-19 Richard Biener <rguenther@suse.de>
172 * tree-core.h (tree_type_common): Re-order pointer members
173 to reduce recursion depth during GC walks.
175 2014-03-19 Marek Polacek <polacek@redhat.com>
178 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
181 2014-03-19 Richard Biener <rguenther@suse.de>
184 * lto-streamer-in.c (input_function): In WPA stage do not drop
187 2014-03-19 Jakub Jelinek <jakub@redhat.com>
189 PR tree-optimization/60559
190 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
191 with build_zero_cst assignment.
193 2014-03-18 Kai Tietz <ktietz@redhat.com>
195 PR rtl-optimization/56356
196 * sdbout.c (sdbout_parms): Verify that parms'
197 incoming argument is valid.
198 (sdbout_reg_parms): Likewise.
200 2014-03-18 Richard Henderson <rth@redhat.com>
203 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
204 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
205 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
207 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
209 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
210 * doc/plugins.texi (Plugin callbacks): Mention
212 Italicize plugin event names in description. Explain that
213 PLUGIN_PRAGMAS has no sense for lto1. Explain
215 Remind that no GCC functions should be called after
217 Explain what pragmas with expansion are.
219 2014-03-18 Martin Liska <mliska@suse.cz>
221 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
222 gimple call statement is update.
223 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
224 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
226 2014-03-18 Jakub Jelinek <jakub@redhat.com>
229 * ubsan.c (ubsan_instrument_unreachable): Call
230 initialize_sanitizer_builtins.
231 (ubsan_pass): Likewise.
234 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
235 varpool_finalize_decl instead of rest_of_decl_compilation.
237 2014-03-18 Richard Biener <rguenther@suse.de>
239 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
240 by using bitmap_and_compl instead of bitmap_and_compl_into.
241 (df_rd_transfer_function): Likewise.
243 2014-03-18 Richard Biener <rguenther@suse.de>
245 * doc/lto.texi (fresolution): Fix typo.
247 2014-03-18 Richard Biener <rguenther@suse.de>
249 * doc/invoke.texi (flto): Update for changes in 4.9.
251 2014-03-18 Richard Biener <rguenther@suse.de>
253 * doc/loop.texi: Remove section on the removed lambda framework.
254 Update loop docs with recent changes in preserving loop structure.
256 2014-03-18 Richard Biener <rguenther@suse.de>
258 * doc/lto.texi (-fresolution): Document.
260 2014-03-18 Richard Biener <rguenther@suse.de>
262 * doc/contrib.texi: Adjust my name.
264 2014-03-18 Jakub Jelinek <jakub@redhat.com>
267 * internal-fn.c: Include diagnostic-core.h.
268 (expand_BUILTIN_EXPECT): New function.
269 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
270 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
271 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
272 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
274 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
275 Revert 3 argument __builtin_expect code.
276 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
277 * gimple-fold.c (gimple_fold_call): Likewise.
278 * tree.h (fold_builtin_expect): New prototype.
279 * builtins.c (build_builtin_expect_predicate): Add predictor
280 argument, if non-NULL, create 3 argument __builtin_expect.
281 (fold_builtin_expect): No longer static. Add ARG2 argument,
282 pass it through to build_builtin_expect_predicate.
283 (fold_builtin_2): Adjust caller.
284 (fold_builtin_3): Handle BUILT_IN_EXPECT.
285 * internal-fn.def (BUILTIN_EXPECT): New.
287 2014-03-18 Tobias Burnus <burnus@net-b.de>
290 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
291 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
292 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
294 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
297 * predict.c (combine_predictions_for_bb): Fix up formatting.
298 (expr_expected_value_1, expr_expected_value): Add predictor argument,
299 fill what it points to if non-NULL.
300 (tree_predict_by_opcode): Adjust caller, use the predictor.
301 * predict.def (PRED_COMPARE_AND_SWAP): Add.
303 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
305 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
306 proper constant for the store mode.
308 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
310 * symtab.c (change_decl_assembler_name): Fix transparent alias
313 2014-03-16 Renlin Li <Renlin.Li@arm.com>
315 * config/aarch64/aarch64.c: Correct the comments about the
316 aarch64 stack layout.
318 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
320 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
321 check for GF_OMP_FOR_KIND_FOR.
323 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
325 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
326 ymm and zmm register names.
328 2014-03-17 Jakub Jelinek <jakub@redhat.com>
331 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
332 note creation for the 2010-08-31 changes.
334 2014-03-17 Marek Polacek <polacek@redhat.com>
337 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
338 as -fno-tree-loop-vectorize.
339 (expand_omp_simd): Likewise.
341 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
343 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
344 (eligible_for_call_delay): New prototype.
345 * config/sparc/sparc.c (tls_call_delay): Rename into...
346 (eligible_for_call_delay): ...this. Return false if the instruction
347 cannot be put in the delay slot of a branch.
348 (eligible_for_restore_insn): Simplify.
349 (eligible_for_return_delay): Return false if the instruction cannot be
350 put in the delay slot of a branch and simplify.
351 (eligible_for_sibcall_delay): Return false if the instruction cannot be
352 put in the delay slot of a branch.
353 * config/sparc/sparc.md (fix_ut699): New attribute.
354 (tls_call_delay): Delete.
355 (in_call_delay): Reimplement.
356 (eligible_for_sibcall_delay): Rename into...
357 (in_sibcall_delay): ...this.
358 (eligible_for_return_delay): Rename into...
359 (in_return_delay): ...this.
360 (in_branch_delay): Reimplement.
361 (in_uncond_branch_delay): Delete.
362 (in_annul_branch_delay): Delete.
364 2014-03-14 Richard Henderson <rth@redhat.com>
367 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
368 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
369 (*floathi<X87MODEF>2_i387_with_temp): Remove.
370 (floathi splitters): Remove.
371 (float<SWI48x>xf2): New pattern.
372 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
373 code that tried to handle DImode for 32-bit, but which was excluded
374 by the pattern's condition. Drop allocation of stack temporary.
375 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
376 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
377 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
378 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
379 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
380 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
381 (*float<SWI48><MODEF>2_sse_interunit): Remove.
382 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
383 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
384 (*float<SWI48x><X87MODEF>2_i387): Remove.
385 (all float _with_temp splitters): Remove.
386 (*float<SWI48x><MODEF>2_i387): New pattern.
387 (*float<SWI48><MODEF>2_sse): New pattern.
388 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
389 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
391 2014-03-14 Jakub Jelinek <jakub@redhat.com>
392 Marek Polacek <polacek@redhat.com>
395 * common.opt (dump_base_name_prefixed): New Variable.
396 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
397 if x_dump_base_name_prefixed is already set, set it at the end.
399 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
401 PR rtl-optimization/60508
402 * lra-constraints.c (get_reload_reg): Add new parameter
404 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
405 Pass the new parameter values.
407 2014-03-14 Richard Biener <rguenther@suse.de>
409 * common.opt: Revert unintented changes from r205065.
412 2014-03-14 Richard Biener <rguenther@suse.de>
415 * cfghooks.c (split_block): Properly adjust all loops the
416 block was a latch of.
418 2014-03-14 Martin Jambor <mjambor@suse.cz>
421 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
424 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
427 * config/avr/avr.c (avr_set_current_function): Pass function name
428 through default_strip_name_encoding before sanity checking instead
429 of skipping the first char of the assembler name.
431 2014-03-13 Richard Henderson <rth@redhat.com>
434 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
435 (ix86_force_to_memory, ix86_free_from_memory): Remove.
436 * config/i386/i386-protos.h: Likewise.
437 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
438 in the expander instead of a splitter.
439 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
440 any possibility of requiring a memory.
441 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
442 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
443 (fp branch splitters): Update for ix86_split_fp_branch.
444 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
445 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
446 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
447 (*fop_<MODEF>_2_i387): Remove f/r alternative.
448 (*fop_<MODEF>_3_i387): Likewise.
449 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
450 (splitters for the fop_* register patterns): Remove.
451 (fscalexf4_i387): Rename from *fscalexf4_i387.
452 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
454 2014-03-13 Jakub Jelinek <jakub@redhat.com>
456 PR tree-optimization/59779
457 * tree-dfa.c (get_ref_base_and_extent): Use double_int
458 type for bitsize and maxsize instead of HOST_WIDE_INT.
460 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
462 PR rtl-optimization/57320
463 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
464 the CFG after thread_prologue_and_epilogue_insns.
466 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
468 PR rtl-optimization/57189
469 * lra-constraints.c (process_alt_operands): Disfavor spilling
472 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
474 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
477 2014-03-13 Jakub Jelinek <jakub@redhat.com>
479 PR tree-optimization/59025
481 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
482 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
484 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
487 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
488 calls of avr_out_plus_1.
490 2014-03-13 Bin Cheng <bin.cheng@arm.com>
492 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
493 BB's single pred and update the father loop's latch info later.
495 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
497 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
501 (VEC_base): Likewise.
502 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
503 registers, we need to swap double words in little endian mode.
505 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
506 to be a container mode for 128-bit integer operations added in ISA
507 2.07. Unlike TImode and PTImode, the preferred register set is
508 the Altivec/VMX registers for the 128-bit operations.
510 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
512 (rs6000_split_128bit_ok_p): Likewise.
514 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
515 macros for creating ISA 2.07 normal and overloaded builtin
516 functions with 3 arguments.
517 (BU_P8V_OVERLOAD_3): Likewise.
518 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
519 for use as overloaded functions.
520 (VPERM_1TI_UNS): Likewise.
521 (VSEL_1TI): Likewise.
522 (VSEL_1TI_UNS): Likewise.
523 (ST_INTERNAL_1ti): Likewise.
524 (LD_INTERNAL_1ti): Likewise.
525 (XXSEL_1TI): Likewise.
526 (XXSEL_1TI_UNS): Likewise.
527 (VPERM_1TI): Likewise.
528 (VPERM_1TI_UNS): Likewise.
529 (XXPERMDI_1TI): Likewise.
531 (LXVD2X_V1TI): Likewise.
532 (STXVD2X_V1TI): Likewise.
533 (VEC_INIT_V1TI): Likewise.
534 (VEC_SET_V1TI): Likewise.
535 (VEC_EXT_V1TI): Likewise.
536 (EQV_V1TI): Likewise.
537 (NAND_V1TI): Likewise.
538 (ORC_V1TI): Likewise.
539 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
540 added in ISA 2.07. Add both normal 'altivec' builtins, and the
544 (VADDEUQM): Likewise.
545 (VADDECUQ): Likewise.
546 (VSUBEUQM): Likewise.
547 (VSUBECUQ): Likewise.
549 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
550 __int128_t and __uint128_t types.
551 (__uint128_type): Likewise.
552 (altivec_categorize_keyword): Add support for vector __int128_t,
553 vector __uint128_t, vector __int128, and vector unsigned __int128
554 as a container type for TImode operations that need to be done in
555 VSX/Altivec registers.
556 (rs6000_macro_to_expand): Likewise.
557 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
558 to support 128-bit integer instructions vaddcuq, vadduqm,
559 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
560 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
562 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
563 for V1TImode, and set up preferences to use VSX/Altivec registers.
564 Setup VSX reload handlers.
565 (rs6000_debug_reg_global): Likewise.
566 (rs6000_init_hard_regno_mode_ok): Likewise.
567 (rs6000_preferred_simd_mode): Likewise.
568 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
569 (easy_altivec_constant): Likewise.
570 (output_vec_const_move): Likewise.
571 (rs6000_expand_vector_set): Convert V1TImode set and extract to
573 (rs6000_expand_vector_extract): Likewise.
574 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
576 (rs6000_const_vec): Add support for V1TImode.
577 (rs6000_emit_le_vsx_load): Swap double words when loading or
578 storing TImode/V1TImode.
579 (rs6000_emit_le_vsx_store): Likewise.
580 (rs6000_emit_le_vsx_move): Likewise.
581 (rs6000_emit_move): Add support for V1TImode.
582 (altivec_expand_ld_builtin): Likewise.
583 (altivec_expand_st_builtin): Likewise.
584 (altivec_expand_vec_init_builtin): Likewise.
585 (altivec_expand_builtin): Likewise.
586 (rs6000_init_builtins): Add support for V1TImode type. Add
587 support for ISA 2.07 128-bit integer builtins. Define type names
588 for the VSX/Altivec vector types.
589 (altivec_init_builtins): Add support for overloaded vector
590 functions with V1TImode type.
591 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
592 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
594 (rs6000_split_128bit_ok_p): Likewise.
595 (rs6000_handle_altivec_attribute): Create V1TImode from vector
596 __int128_t and vector __uint128_t.
598 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
606 (VS_scalar): Likewise.
607 (VS_double): Likewise.
608 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
610 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
611 we support the ISA 2.07 128-bit integer arithmetic instructions.
612 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
613 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
614 and TImode types for use with the builtin functions.
615 (V1TI_type_node): Likewise.
616 (unsigned_V1TI_type_node): Likewise.
617 (intTI_type_internal_node): Likewise.
618 (uintTI_type_internal_node): Likewise.
620 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
621 128-bit builtin functions.
622 (UNSPEC_VADDEUQM): Likewise.
623 (UNSPEC_VADDECUQ): Likewise.
624 (UNSPEC_VSUBCUQ): Likewise.
625 (UNSPEC_VSUBEUQM): Likewise.
626 (UNSPEC_VSUBECUQ): Likewise.
627 (VM): Add V1TImode to vector mode iterators.
630 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
631 (altivec_vaddcuq): Likewise.
632 (altivec_vsubuqm): Likewise.
633 (altivec_vsubcuq): Likewise.
634 (altivec_vaddeuqm): Likewise.
635 (altivec_vaddecuq): Likewise.
636 (altivec_vsubeuqm): Likewise.
637 (altivec_vsubecuq): Likewise.
639 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
641 (BOOL_128): Likewise.
642 (BOOL_REGS_OUTPUT): Likewise.
643 (BOOL_REGS_OP1): Likewise.
644 (BOOL_REGS_OP2): Likewise.
645 (BOOL_REGS_UNARY): Likewise.
646 (BOOL_REGS_AND_CR0): Likewise.
648 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
649 128-bit integer builtin support.
650 (vec_vadduqm): Likewise.
651 (vec_vaddecuq): Likewise.
652 (vec_vaddeuqm): Likewise.
653 (vec_vsubecuq): Likewise.
654 (vec_vsubeuqm): Likewise.
655 (vec_vsubcuq): Likewise.
656 (vec_vsubuqm): Likewise.
658 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
659 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
660 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
661 128-bit integer add/subtract to ISA 2.07.
663 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
665 * config/arc/arc.c (arc_predicate_delay_insns):
666 Fix third argument passed to conditionalize_nonjump.
668 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
670 * config/aarch64/aarch64-builtins.c
671 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
672 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
673 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
674 instead of __builtin_lfloor.
675 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
677 2014-03-12 Jakub Jelinek <jakub@redhat.com>
679 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
680 (tree_ssa_ifcombine_bb_1): New function.
681 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
682 is an empty forwarder block to then_bb or vice versa and then_bb
683 and else_bb are effectively swapped.
685 2014-03-12 Christian Bruel <christian.bruel@st.com>
688 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
689 REG_CFA_DEF_CFA note.
690 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
691 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
693 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
695 PR tree-optimization/60454
696 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
698 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
700 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
701 Do not define target_cpu_default2 to generic.
702 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
703 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
704 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
706 2014-03-12 Jakub Jelinek <jakub@redhat.com>
707 Marc Glisse <marc.glisse@inria.fr>
709 PR tree-optimization/60502
710 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
711 instead of build_low_bits_mask.
713 2014-03-12 Jakub Jelinek <jakub@redhat.com>
716 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
717 if there are multiple uses, but op doesn't live on E edge.
718 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
719 clobber stmts before __builtin_unreachable.
721 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
723 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
724 hard_frame_pointer_rtx.
725 * cse.c (cse_insn): Remove volatile check.
726 * cselib.c (cselib_process_insn): Likewise.
727 * dse.c (scan_insn): Likewise.
729 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
731 * config/arc/arc.c (conditionalize_nonjump): New function,
733 (arc_ifcvt): ... this.
734 (arc_predicate_delay_insns): Use it.
736 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
738 * config/arc/predicates.md (extend_operand): During/after reload,
739 allow const_int_operand.
740 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
741 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
742 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
744 (umulsi3_highpart_i): Likewise.
746 2014-03-11 Richard Biener <rguenther@suse.de>
748 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
749 Add asserts to guard possible wrong-code bugs.
751 2014-03-11 Richard Biener <rguenther@suse.de>
753 PR tree-optimization/60429
754 PR tree-optimization/60485
755 * tree-ssa-structalias.c (set_union_with_increment): Properly
756 take into account all fields that overlap the shifted vars.
757 (do_sd_constraint): Likewise.
758 (do_ds_constraint): Likewise.
759 (get_constraint_for_ptr_offset): Likewise.
761 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
763 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
764 (nios2_compute_frame_layout):
765 Add calculation of cfun->machine->fp_save_offset.
766 (nios2_expand_prologue): Correct setting of frame pointer register
768 (nios2_expand_epilogue): Update recovery of stack pointer from
769 frame pointer accordingly.
770 (nios2_initial_elimination_offset): Update calculation of offset
771 for eliminating to HARD_FRAME_POINTER_REGNUM.
773 2014-03-10 Jakub Jelinek <jakub@redhat.com>
776 * ipa.c (symtab_remove_unreachable_nodes): Don't call
777 cgraph_get_create_node on VAR_DECLs.
779 2014-03-10 Richard Biener <rguenther@suse.de>
782 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
784 2014-03-08 Douglas B Rupp <rupp@gnat.com>
786 * config/vms/vms.opt (vms_float_format): New variable.
788 2014-03-08 Tobias Burnus <burnus@net-b.de>
790 * doc/invoke.texi (-fcilkplus): Update implementation status.
792 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
793 Richard Biener <rguenther@suse.de>
795 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
796 consistently accross all TUs.
797 (run_gcc): Enable -fshort-double automatically at link at link-time
798 and disallow override.
800 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
803 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
804 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
805 if they can't be used.
807 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
809 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
810 for Solaris 11/x86 ld.
811 * configure: Regenerate.
813 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
815 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
816 (LIB_TLS_SPEC): Save as ld_tls_libs.
817 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
818 (HAVE_AS_IX86_TLSLDM): New test.
819 * configure, config.in: Regenerate.
820 * config/i386/i386.c (legitimize_tls_address): Fall back to
821 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
822 cannot support TLS_MODEL_LOCAL_DYNAMIC.
823 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
824 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
826 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
828 * common.opt (fira-loop-pressure): Mark as optimization.
830 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
832 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
833 an OpenMP mappable type.
835 2014-03-06 Matthias Klose <doko@ubuntu.com>
837 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
838 MULTILIB_OSDIRNAMES is not defined.
840 2014-03-06 Jakub Jelinek <jakub@redhat.com>
841 Meador Inge <meadori@codesourcery.com>
844 * config/arm/arm.c (arm_tls_symbol_p): Remove.
845 (arm_legitimize_address): Call legitimize_tls_address for any
846 arm_tls_referenced_p expression, handle constant addend. Call it
847 before testing for !TARGET_ARM.
848 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
850 2014-03-06 Richard Biener <rguenther@suse.de>
856 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
858 * tree-streamer.c (record_common_node): Assert we don't record
859 nodes with type double.
860 (preload_common_node): Skip type double, complex double and double
861 pointer since it is now frontend dependent due to fshort-double option.
863 2014-03-06 Richard Biener <rguenther@suse.de>
865 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
866 or -fno-lto is specified and the linker has full plugin support.
867 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
868 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
869 * lto-wrapper.c (merge_and_complain): Merge compile-time
871 (run_gcc): And pass it through to the link options.
873 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
877 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
879 * cselib.c (remove_useless_values): Skip to avoid quadratic
880 behavior if the condition moved from...
881 (cselib_process_insn): ... here holds.
883 2014-03-05 Jakub Jelinek <jakub@redhat.com>
886 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
887 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
890 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
891 (TM_H): Add x86-tune.def.
893 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
895 * config/aarch64/aarch64.c (generic_tunings):
896 Use cortexa57_extra_costs.
898 2014-03-05 Jakub Jelinek <jakub@redhat.com>
901 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
902 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
903 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
906 2014-03-04 Heiher <r@hev.cc>
908 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
909 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
911 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
913 * config/i386/predicates.md (const2356_operand): Change to ...
914 (const2367_operand): ... this.
915 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
917 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
918 (*avx512pf_scatterpf<mode>sf): Ditto.
919 (avx512pf_scatterpf<mode>df): Ditto.
920 (*avx512pf_scatterpf<mode>df_mask): Ditto.
921 (*avx512pf_scatterpf<mode>df): Ditto.
922 * config/i386/i386.c (ix86_expand_builtin): Update
923 incorrect hint operand error message.
925 2014-03-04 Richard Biener <rguenther@suse.de>
927 * lto-section-in.c (lto_get_section_data): Fix const cast.
929 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
931 * tree-streamer.c (record_common_node): Assert we don't record
932 nodes with type double.
933 (preload_common_node): Skip type double, complex double and double
934 pointer since it is now frontend dependent due to fshort-double option.
936 2014-03-04 Richard Biener <rguenther@suse.de>
939 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
940 (lto_input_toplevel_asms): Likewise.
941 * lto-section-in.c (lto_get_section_data): Instead do it here
944 2014-03-04 Richard Biener <rguenther@suse.de>
946 PR tree-optimization/60382
947 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
948 dead PHIs a reduction.
950 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
952 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
954 (_mm_prefetch): Move out of GCC target("sse") pragma.
955 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
956 GCC target("prfchw") pragma.
957 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
959 * config/i386/i386.c (ix86_option_override_internal): Enable
960 -mprfchw with -mprefetchwt1.
962 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
964 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
967 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
969 * opts.h (CL_PCH_IGNORE): Define.
970 * targhooks.c (option_affects_pch_p):
971 Return false for options that have CL_PCH_IGNORE set.
972 * opt-functions.awk: Process PchIgnore.
973 * doc/options.texi: Document PchIgnore.
975 * config/arc/arc.opt (misize): Add PchIgnore property.
977 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
979 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
980 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
981 constraint on constants to permit them being loaded into
982 GENERAL_REGS or BASE_REGS.
984 2014-03-03 Nick Clifton <nickc@redhat.com>
986 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
987 anti-cacnonical alternatives.
988 (negandhi3_real): New pattern.
989 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
991 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
993 * config/avr/avr-mcus.def: Remove atxmega16x1.
994 * config/avr/avr-tables.opt: Regenerate.
995 * config/avr/t-multilib: Regenerate.
996 * doc/avr-mmcu.texi: Regenerate.
998 2014-03-03 Tobias Grosser <tobias@grosser.es>
999 Mircea Namolaru <mircea.namolaru@inria.fr>
1001 PR tree-optimization/58028
1002 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1005 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1007 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1008 not handled by recognizers.
1010 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1013 * function.c (expand_function_end): Don't emit
1014 clobber_return_register sequence if clobber_after is a BARRIER.
1015 * cfgexpand.c (construct_exit_block): Append instructions before
1016 return_label to prev_bb.
1018 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1020 * config/rs6000/constraints.md: Document reserved use of "wc".
1022 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1025 * ipa.c (function_and_variable_visibility): When dissolving comdat
1026 group, also set all symbols to local.
1028 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1033 2013-12-14 Jan Hubicka <jh@suse.cz>
1035 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1037 2014-03-02 Jon Beniston <jon@beniston.com>
1043 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1044 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1045 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1046 (simple_return, *simple_return): New patterns
1047 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1048 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1050 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1052 * dwarf2out.c (gen_subprogram_die): Tidy.
1054 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1057 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1058 (*mov_t_msb_neg_negc): ... this new insn.
1060 2014-02-28 Jason Merrill <jason@redhat.com>
1063 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1066 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1069 * dwarf2out.c (decltype_auto_die): New static.
1070 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1071 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1072 (is_cxx_auto): Likewise.
1074 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1076 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1077 we are not using general regs only.
1079 2014-02-28 Richard Biener <rguenther@suse.de>
1082 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1083 previous fix and only allow to remove trivial pre-headers
1084 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1085 (remove_forwarder_block): Properly update the latch of a loop.
1087 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1090 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1091 (cselib_preserved_hash_table): New.
1092 (preserve_constants_and_equivs): Move preserved vals to it.
1093 (cselib_find_slot): Look it up first.
1094 (cselib_init): Initialize it.
1095 (cselib_finish): Release it.
1096 (dump_cselib_table): Dump it.
1098 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1101 * cselib.c (remove_useless_values): Skip to avoid quadratic
1102 behavior if the condition moved from...
1103 (cselib_process_insn): ... here holds.
1105 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1108 * var-tracking.c (vt_initialize): Apply the same condition to
1109 preserve the CFA base value.
1111 2014-02-28 Joey Ye <joey.ye@arm.com>
1114 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1115 if reload in progress or completed.
1117 2014-02-28 Tobias Burnus <burnus@net-b.de>
1120 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1123 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1125 * doc/tm.texi.in (Condition Code Status): Update documention for
1126 relative locations of cc0-setter and cc0-user.
1128 2014-02-27 Jeff Law <law@redhat.com>
1130 PR rtl-optimization/52714
1131 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1132 into two independent simple sets, if I3 is a jump, ensure the
1133 pattern we place into I3 is a (set (pc) ...).
1135 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1136 Jeff Law <law@redhat.com>
1138 PR rtl-optimization/49847
1139 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1140 are in different blocks.
1141 * doc/tm.texi (Condition Code Status): Update documention for
1142 relative locations of cc0-setter and cc0-user.
1144 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1147 * lra.c (lra_emit_add): Check SUBREG too.
1149 2014-02-27 Andreas Schwab <schwab@suse.de>
1151 * config/m68k/m68k.c (m68k_option_override): Disable
1152 -flive-range-shrinkage for classic m68k.
1153 (m68k_override_options_after_change): Likewise.
1155 2014-02-27 Marek Polacek <polacek@redhat.com>
1158 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1159 -Wmaybe-uninitialized.
1161 2014-02-27 Alan Modra <amodra@gmail.com>
1164 * reload1.c (emit_input_reload_insns): When reload_override_in,
1165 set old to rl->in_reg when rl->in_reg is a subreg.
1167 2014-02-26 Richard Biener <rguenther@suse.de>
1170 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1172 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1174 * common/config/i386/predicates.md (const1256_operand): Remove.
1175 (const2356_operand): New.
1176 (const_1_to_2_operand): Remove.
1177 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1178 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1179 (*avx512pf_gatherpf<mode>sf): Ditto.
1180 (avx512pf_gatherpf<mode>df): Ditto.
1181 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1182 (*avx512pf_gatherpf<mode>df): Ditto.
1183 (avx512pf_scatterpf<mode>sf): Ditto.
1184 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1185 (*avx512pf_scatterpf<mode>sf): Ditto.
1186 (avx512pf_scatterpf<mode>df): Ditto.
1187 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1188 (*avx512pf_scatterpf<mode>df): Ditto.
1189 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1191 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1193 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1194 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1195 (_mm512_mask_testn_epi64_mask): Move to ...
1196 * config/i386/avx512cdintrin.h: Here.
1197 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1198 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1199 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1200 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1201 TARGET_AVX512F from TARGET_AVX512CD.
1203 2014-02-26 Richard Biener <rguenther@suse.de>
1206 * ipa.c (walk_polymorphic_call_targets): Properly guard
1207 call to inline_update_overall_summary.
1209 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1212 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1213 and latches only if requested. Fix latch if it is removed.
1214 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1215 LOOPS_HAVE_PREHEADERS.
1217 2014-02-25 Andrew Pinski <apinski@cavium.com>
1219 * builtins.c (expand_builtin_thread_pointer): Create a new target
1220 when the target is NULL.
1222 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1224 PR rtl-optimization/60317
1225 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1226 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1227 * lra-assigns.c: Include params.h.
1228 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1229 other reload pseudos considerations.
1231 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1233 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1234 to use canonical form for nor<mode>3.
1236 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1239 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1242 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1244 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1245 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1246 (ix86_handle_option): Handle OPT_mprefetchwt1.
1247 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1248 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1250 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1251 OPTION_MASK_ISA_PREFETCHWT1.
1252 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1253 (PTA_PREFETCHWT1): New.
1254 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1255 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1256 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1257 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1258 (*prefetch_avx512pf_<mode>_: Change into ...
1259 (*prefetch_prefetchwt1_<mode>: This.
1260 * config/i386/i386.opt (mprefetchwt1): New.
1261 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1262 (_mm_prefetch): Handle intent to write.
1263 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1265 2014-02-25 Richard Biener <rguenther@suse.de>
1268 * emit-rtl.c (mem_attrs_htab): Remove.
1269 (mem_attrs_htab_hash): Likewise.
1270 (mem_attrs_htab_eq): Likewise.
1271 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1272 (init_emit_once): Do not allocate mem_attrs_htab.
1274 2014-02-25 Richard Biener <rguenther@suse.de>
1277 * lto-opts.c (lto_write_options): Output non-explicit conservative
1278 -fwrapv, -fno-trapv and -fno-strict-overflow.
1279 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1280 (run_gcc): And pass them through.
1282 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1284 * sel-sched.c (calculate_new_fences): New parameter ptime.
1285 Calculate it as a maximum over all fence cycles.
1286 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1287 Print the final schedule timing when sched_verbose.
1289 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1291 PR rtl-optimization/60292
1292 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1293 bit fot the fence instruction.
1295 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1297 * calls.h: Fix typo in comment.
1299 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1301 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1302 adjusting offsetable addresses.
1304 2014-02-24 Guozhi Wei <carrot@google.com>
1306 * sparseset.h (sparseset_pop): Fix the wrong index.
1308 2014-02-24 Walter Lee <walt@tilera.com>
1310 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1311 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1313 * common/config/tilegx/tilegx-common.c
1314 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1315 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1317 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1318 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1319 (tilegx_gimplify_va_arg_expr): Handle big endian.
1320 (tilegx_expand_unaligned_load): Ditto.
1321 (tilegx_expand_unaligned_store): Ditto.
1322 (TARGET_RETURN_IN_MSB): New.
1323 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1324 (TARGET_ENDIAN_DEFAULT): New.
1325 (TARGET_BIG_ENDIAN): Handle big endian.
1326 (BYTES_BIG_ENDIAN): Ditto.
1327 (WORDS_BIG_ENDIAN): Ditto.
1328 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1331 * config/tilegx/tilegx.md (extv): Handle big endian.
1333 (insn_st<n>): Ditto.
1334 (insn_st<n>_add<bitsuffix>): Ditto.
1335 (insn_stnt<n>): Ditto.
1336 (insn_stnt<n>_add<bitsuffix>):Ditto.
1337 (vec_interleave_highv8qi): Handle big endian.
1338 (vec_interleave_highv8qi_be): New.
1339 (vec_interleave_highv8qi_le): New.
1340 (insn_v1int_h): Handle big endian.
1341 (vec_interleave_lowv8qi): Handle big endian.
1342 (vec_interleave_lowv8qi_be): New.
1343 (vec_interleave_lowv8qi_le): New.
1344 (insn_v1int_l): Handle big endian.
1345 (vec_interleave_highv4hi): Handle big endian.
1346 (vec_interleave_highv4hi_be): New.
1347 (vec_interleave_highv4hi_le): New.
1348 (insn_v2int_h): Handle big endian.
1349 (vec_interleave_lowv4hi): Handle big endian.
1350 (vec_interleave_lowv4hi_be): New.
1351 (vec_interleave_lowv4hi_le): New.
1352 (insn_v2int_l): Handle big endian.
1353 (vec_interleave_highv2si): Handle big endian.
1354 (vec_interleave_highv2si_be): New.
1355 (vec_interleave_highv2si_le): New.
1356 (insn_v4int_h): Handle big endian.
1357 (vec_interleave_lowv2si): Handle big endian.
1358 (vec_interleave_lowv2si_be): New.
1359 (vec_interleave_lowv2si_le): New.
1360 (insn_v4int_l): Handle big endian.
1361 * config/tilegx/tilegx.opt (mbig-endian): New option.
1362 (mlittle-endian): New option.
1363 * doc/install.texi: Document tilegxbe-linux.
1364 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1366 2014-02-24 Martin Jambor <mjambor@suse.cz>
1369 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1370 there are no parameter descriptors.
1372 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1374 PR rtl-optimization/60268
1375 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1376 initialization to ...
1377 (sched_rgn_init): ... here.
1378 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1380 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1382 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1385 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1387 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1390 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1392 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1393 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1395 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1397 * config/microblaze/predicates.md: Add cmp_op predicate.
1398 * config/microblaze/microblaze.md: Add branch_compare instruction
1399 which uses cmp_op predicate and emits cmp insn before branch.
1400 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1401 to microblaze_expand_conditional_branch and consolidate logic.
1402 (microblaze_expand_conditional_branch): emit branch_compare
1403 insn instead of handling cmp op separate from branch insn.
1405 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1407 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1410 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1412 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1413 define_insn with define_expand and new define_insn
1414 *altivec_lve<VI_char>x_internal.
1415 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1416 and new define_insn *altivec_stve<VI_char>x_internal.
1417 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1419 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1421 (altivec_expand_stvex_be): New function.
1423 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1425 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1426 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1427 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1428 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1430 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1433 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1434 instead of emit_move_insn.
1436 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1438 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1440 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1441 gen_altivec_vsumsws.
1443 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1445 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1446 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1447 (altivec_lvxl_<mode>): New define_expand incorporating
1448 -maltivec=be semantics where needed.
1449 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1450 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1451 semantics where needed.
1452 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1453 (altivec_stvx_<mode>): New define_expand incorporating
1454 -maltivec=be semantics where needed.
1455 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1456 VM2 iterator instead of V4SI.
1457 (altivec_stvxl_<mode>): New define_expand incorporating
1458 -maltivec=be semantics where needed.
1459 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1460 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1461 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1462 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1463 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1464 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1465 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1466 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1467 ALTIVEC_BUILTIN_STVXL.
1468 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1469 (altivec_expand_stvx_be): Likewise.
1470 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1471 (altivec_expand_lvx_be): Likewise.
1472 (altivec_expand_stvx_be): Likewise.
1473 (altivec_expand_builtin): Add cases for
1474 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1475 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1476 (altivec_init_builtins): Add definitions for
1477 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1478 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1480 2014-02-21 Catherine Moore <clm@codesourcery.com>
1482 * doc/invoke.texi (mvirt, mno-virt): Document.
1483 * config/mips/mips.opt (mvirt): New option.
1484 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1486 2014-02-21 Richard Biener <rguenther@suse.de>
1488 PR tree-optimization/60276
1489 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1490 (STMT_VINFO_MIN_NEG_DIST): New macro.
1491 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1492 STMT_VINFO_MIN_NEG_DIST.
1493 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1494 made for negative dependence distances still hold.
1496 2014-02-21 Richard Biener <rguenther@suse.de>
1499 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1500 DECL_INITIAL for globals not in the current function context.
1502 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1504 PR tree-optimization/56490
1505 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1506 * tree-ssa-uninit.c: Include params.h.
1507 (compute_control_dep_chain): Add num_calls argument, return false
1508 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1509 num_calls to recursive call.
1510 (find_predicates): Change dep_chain into normal array,
1511 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1512 variable and adjust compute_control_dep_chain caller.
1513 (find_def_preds): Likewise.
1515 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1517 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1518 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1520 2014-02-21 Nick Clifton <nickc@redhat.com>
1522 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1523 (pushhi1): Likewise.
1524 (popqi1): Add mode to pre_dec.
1527 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1529 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1530 mode for mask of V8SFmode permutation.
1532 2014-02-20 Richard Henderson <rth@redhat.com>
1535 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1536 a new pseudo for OLDVAL.
1538 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1541 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1542 gen_reg_rtx if d->testing_p.
1543 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1544 if d->testing_p and we will certainly return true.
1545 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1548 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1550 * emit-rtl.c (gen_reg_rtx): Assert that
1551 crtl->emit.regno_pointer_align_length is non-zero.
1553 2014-02-20 Richard Henderson <rth@redhat.com>
1556 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1557 on failure the store back into EXPECT.
1559 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1560 Sandra Loosemore <sandra@codesourcery.com>
1562 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1563 * config/nios2/nios2.c (nios2_function_profiler): Add
1564 -fPIC (flag_pic == 2) support.
1565 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1566 (nios2_large_offset_p): New function.
1567 (nios2_unspec_reloc_p): Move up position, update to use
1568 nios2_large_offset_p.
1569 (nios2_unspec_address): Remove function.
1570 (nios2_unspec_offset): New function.
1571 (nios2_large_got_address): New function.
1572 (nios2_got_address): Add large offset support.
1573 (nios2_legitimize_tls_address): Update usage of removed and new
1575 (nios2_symbol_binds_local_p): New function.
1576 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1577 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1578 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1579 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1580 processing for (const (unspec ...)).
1581 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1583 2014-02-20 Richard Biener <rguenther@suse.de>
1585 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1586 doing the substitution.
1587 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1589 2014-02-20 Martin Jambor <mjambor@suse.cz>
1592 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1593 info when checking whether lattices are bottom.
1595 2014-02-20 Richard Biener <rguenther@suse.de>
1598 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1601 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1604 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1605 parameter specifying the scaling.
1606 (inline_call): Update.
1607 (want_inline_recursively): Guard division by zero.
1608 (recursive_inlining): Update.
1609 * ipa-inline.h (clone_inlined_nodes): Update.
1611 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1614 * config/i386/i386.c (classify_argument): Pass structures of size
1615 64 bytes or less in register.
1617 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1618 Kirill Yukhin <kirill.yukhin@intel.com>
1620 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1621 (_mm_rcp28_round_ss): Ditto.
1622 (_mm_rsqrt28_round_sd): Ditto.
1623 (_mm_rsqrt28_round_ss): Ditto.
1624 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1625 (_mm_rcp14_round_ss): Ditto.
1626 (_mm_rsqrt14_round_sd): Ditto.
1627 (_mm_rsqrt14_round_ss): Ditto.
1628 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1629 the first input operand, get rid of match_dup.
1630 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1632 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1634 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1635 operand as the first input operand, set type attribute.
1636 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1638 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1639 operand as the first input operand, set type attribute.
1641 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1643 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1646 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1649 * config/i386/i386.c (construct_container): Remove TFmode check
1650 for X86_64_INTEGER_CLASS.
1652 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1655 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1656 only when -Wpsabi is enabled.
1658 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1661 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1662 passing arrays in registers are the same as for structs, so remove the
1663 special case for them.
1665 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1667 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1668 destination type, extract only the valid bits if the source type is not
1669 integral and has a different mode.
1671 2014-02-19 Richard Biener <rguenther@suse.de>
1674 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1677 2014-02-19 Richard Biener <rguenther@suse.de>
1680 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1681 (ipa_modify_call_arguments): Emit an argument load explicitely and
1682 preserve virtual SSA form there and for the replacement call.
1683 Do not update SSA form nor free dominance info.
1685 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1687 * ipa.c (function_and_variable_visibility): Also clear WEAK
1688 flag when disolving COMDAT_GROUP.
1690 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1692 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1693 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1695 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1696 do more sanity checks.
1697 (detect_type_change): Return true when giving up early.
1698 (compute_complex_assign_jump_func): Fix type parameter of
1699 ipa_set_ancestor_jf.
1700 (compute_complex_ancestor_jump_func): Likewise.
1701 (update_jump_functions_after_inlining): Fix updating of
1703 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1705 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1707 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1708 inline clones when edge disappears.
1710 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1713 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1714 Split 64-bit moves into 2 patterns. Do not allow the use of
1715 direct move for TDmode in little endian, since the decimal value
1716 has little endian bytes within a word, but the 64-bit pieces are
1717 ordered in a big endian fashion, and normal subreg's of TDmode are
1719 (mov<mode>_64bit_dm): Likewise.
1720 (movtd_64bit_nodm): Likewise.
1722 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1724 PR tree-optimization/60174
1725 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1726 statement of an SSA_NAME that occurs in an abnormal PHI node.
1728 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1731 * final.c (SEEN_BB): Remove.
1732 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1733 (final_scan_insn): Don't force_source_line on second
1734 NOTE_INSN_BASIC_BLOCK.
1736 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1739 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1740 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1741 (type_natural_mode): Warn ABI change when %zmm register is not
1742 available for AVX512F vector value passing.
1744 2014-02-18 Kai Tietz <ktietz@redhat.com>
1747 * config/i386/i386.c (ix86_expand_prologue): Use value in
1748 rax register as displacement when restoring %r10 or %rax.
1749 Fix wrong offset when restoring both registers.
1751 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1753 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1754 assertion with conditional return.
1756 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1757 Uros Bizjak <ubizjak@gmail.com>
1760 * config/i386/driver-i386.c (host_detect_local_cpu): If
1761 YMM state is not saved by the OS, also clear has_f16c. Move
1762 CPUID 0x80000001 handling before YMM state saving checking.
1764 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1766 PR rtl-optimization/58960
1767 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1768 factored out from ...
1769 (sched_init): ... here.
1770 (free_global_sched_pressure_data): New, factored out from ...
1771 (sched_finish): ... here.
1772 * sched-int.h (free_global_sched_pressure_data): Declare.
1773 * sched-rgn.c (nr_regions_initial): New static global.
1774 (haifa_find_rgns): Initialize it.
1775 (schedule_region): Disable sched-pressure for the newly
1778 2014-02-17 Richard Biener <rguenther@suse.de>
1780 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1781 release SSA defs of pattern stmts.
1783 2014-02-17 Richard Biener <rguenther@suse.de>
1785 * tree-inline.c (expand_call_inline): Release the virtual
1786 operand defined by the call we are about to inline.
1788 2014-02-17 Richard Biener <rguenther@suse.de>
1790 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1792 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1793 Ilya Tocar <ilya.tocar@intel.com>
1795 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1796 arguments order in builtin.
1797 (_mm512_permutexvar_epi64): Ditto.
1798 (_mm512_mask_permutexvar_epi64): Ditto
1799 (_mm512_maskz_permutexvar_epi32): Ditto
1800 (_mm512_permutexvar_epi32): Ditto
1801 (_mm512_mask_permutexvar_epi32): Ditto
1803 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1805 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1806 (p8_vmrgow): Likewise.
1808 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1810 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1813 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1816 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1817 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1818 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1819 using direct move instructions on ISA 2.07. Also adjust
1820 instruction length for 64-bit.
1821 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1822 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1824 2014-02-15 Alan Modra <amodra@gmail.com>
1828 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1829 find_replacement on parts of insn rtl that might be reloaded.
1831 2014-02-15 Richard Biener <rguenther@suse.de>
1833 PR tree-optimization/60183
1834 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1835 (tree_ssa_phiprop): Calculate and free post-dominators.
1837 2014-02-14 Jeff Law <law@redhat.com>
1839 PR rtl-optimization/60131
1840 * ree.c (get_extended_src_reg): New function.
1841 (combine_reaching_defs): Use it rather than assuming location of REG.
1842 (find_and_remove_re): Verify first operand of extension is
1843 a REG before adding the insns to the copy list.
1845 2014-02-14 Roland McGrath <mcgrathr@google.com>
1847 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1848 * configure: Regenerated.
1849 * config.in: Regenerated.
1850 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1851 instead of ASM_SHORT.
1853 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1854 Richard Earnshaw <rearnsha@arm.com>
1856 PR rtl-optimization/59535
1857 * lra-constraints.c (process_alt_operands): Encourage alternative
1858 when unassigned pseudo class is superset of the alternative class.
1859 (inherit_reload_reg): Don't inherit when optimizing for code size.
1860 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1861 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1862 modes not less than 4 for Thumb1.
1864 2014-02-14 Kyle McMartin <kyle@redhat.com>
1867 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1869 2014-02-14 Richard Biener <rguenther@suse.de>
1871 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1872 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1873 Do not strip INDIRECT_REFs.
1875 2014-02-14 Richard Biener <rguenther@suse.de>
1878 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1879 DECL_FUNCTION_SPECIFIC_TARGET.
1880 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1881 * tree-streamer-out.c (pack_ts_target_option): Remove.
1882 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1883 (write_ts_function_decl_tree_pointers): Do not stream
1884 DECL_FUNCTION_SPECIFIC_TARGET.
1885 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1886 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1887 (lto_input_ts_function_decl_tree_pointers): Do not stream
1888 DECL_FUNCTION_SPECIFIC_TARGET.
1890 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1892 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1893 (get_initial_def_for_induction, vectorizable_induction): Ignore
1894 debug stmts when looking for exit_phi.
1895 (vectorizable_live_operation): Fix up condition.
1897 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1899 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1900 nreverse() because it changes the content of original tree list.
1902 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1904 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1905 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1907 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1909 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1910 GNU coding standards.
1912 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1915 * dwarf2out.c (gen_subprogram_die): Don't call
1916 add_calling_convention_attribute if subr_die is old_die.
1918 2014-02-13 Sharad Singhai <singhai@google.com>
1920 * doc/optinfo.texi: Fix order of nodes.
1922 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1924 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1925 operands[2], not operands[3].
1927 2014-02-13 Richard Biener <rguenther@suse.de>
1930 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1931 mention the possibility of an in-tree build.
1932 (CLooG): Update recommended version to 0.18.1, mention the
1933 possibility of an in-tree build and clarify that the ISL
1934 bundled with CLooG does not work.
1936 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1939 * expr.c (compress_float_constant): If x is a hard register,
1940 extend into a pseudo and then move to x.
1942 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1944 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1945 caused by bad second argument to warning_at() with -mhotpatch and
1946 nested functions (e.g. with gfortran).
1948 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1950 * opts.c (option_name): Remove "enabled by default" rider.
1952 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1954 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1956 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1957 Uros Bizjak <ubizjak@gmail.com>
1960 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1961 * configure: Regenerated.
1963 2014-02-12 Richard Biener <rguenther@suse.de>
1965 * vec.c (vec_prefix::calculate_allocation): Move as
1966 inline variant to vec.h.
1967 (vec_prefix::calculate_allocation_1): New out-of-line version.
1968 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1969 (vec_prefix::m_has_auto_buf): Rename to ...
1970 (vec_prefix::m_using_auto_storage): ... this.
1971 (vec_prefix::calculate_allocation): Inline the easy cases
1972 and dispatch to calculate_allocation_1 which doesn't need the
1974 (va_heap::reserve): Use gcc_checking_assert.
1975 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1976 m_using_auto_storage.
1977 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1979 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1980 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1981 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1983 2014-02-12 Richard Biener <rguenther@suse.de>
1985 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1986 when we found a dependence.
1988 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1990 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1992 (maybe_fold_stmt): ... into this new function.
1993 * omp-low.c (lower_omp): Update comment.
1995 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1998 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2001 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2003 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2004 identifiers in comments.
2005 (cortexa53_extra_costs): Likewise.
2006 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2007 (cortexa7_extra_costs): Likewise.
2008 (cortexa12_extra_costs): Likewise.
2009 (cortexa15_extra_costs): Likewise.
2010 (v7m_extra_costs): Likewise.
2012 2014-02-12 Richard Biener <rguenther@suse.de>
2015 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2016 of posix_memalign being successful.
2017 (lower_stmt): Restrict lowering of posix_memalign to when
2018 -ftree-bit-ccp is enabled.
2020 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2022 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2024 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2026 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2028 PR rtl-optimization/60116
2029 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2030 other_insn once the combination has been validated.
2032 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2035 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2037 * ipa-devirt.c: Include demangle.h
2038 (odr_violation_reported): New static variable.
2039 (add_type_duplicate): Update odr_violations.
2040 (maybe_record_node): Add completep parameter; update it.
2041 (record_target_from_binfo): Add COMPLETEP parameter;
2042 update it as needed.
2043 (possible_polymorphic_call_targets_1): Likewise.
2044 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2045 rename FINAL to COMPLETE.
2046 (record_targets_from_bases): Sanity check we found the binfo;
2047 fix COMPLETEP updating.
2048 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2049 parameter, fix computing of COMPLETEP.
2050 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2051 at LTO time do demangling.
2052 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2053 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2055 (gimple_get_virt_method_for_binfo): Likewise.
2056 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2057 gimple_get_virt_method_for_vtable): Update prototypes.
2059 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2062 * genautomata.c (add_presence_absence): Fix typo with
2063 {final_}presence_list.
2065 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2068 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2069 for VSX/Altivec vectors that land in GPR registers.
2071 2014-02-11 Richard Henderson <rth@redhat.com>
2072 Jakub Jelinek <jakub@redhat.com>
2075 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2076 around drhs if type conversion to lacc->type is not useless.
2078 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2080 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2082 (cortex-a57.cortex-a53): Likewise.
2083 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2085 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2087 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2090 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2092 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2093 add_options_for_arm_vfp3.
2095 2014-02-11 Jeff Law <law@redhat.com>
2098 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2099 object with an undesirable mode.
2101 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2104 * config/i386/sol2-9.h: New file.
2105 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2106 *-*-solaris2.9*): Use it.
2108 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2110 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2111 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2113 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2115 * config/microblaze/microblaze.c: Extend mcpu version format
2117 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2119 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2121 2014-02-10 Richard Henderson <rth@redhat.com>
2124 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2125 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2126 ms-abi vs -mno-accumulate-outgoing-args.
2127 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2128 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2131 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2134 * cfgexpand.c (expand_asm_operands): Attach source location to
2135 ASM_INPUT rtx objects.
2136 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2138 2014-02-10 Nick Clifton <nickc@redhat.com>
2140 * config/mn10300/mn10300.c (popcount): New function.
2141 (mn10300_expand_prologue): Include saved registers in stack usage
2144 2014-02-10 Jeff Law <law@redhat.com>
2147 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2148 when changing the SET_DEST of a prior insn to avoid an input reload.
2150 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2152 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2153 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2154 -mcall-openbsd, or -mcall-linux.
2155 (CC1_ENDIAN_BIG_SPEC): Remove.
2156 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2157 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2158 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2159 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2160 and %cc1_endian_default.
2161 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2163 2014-02-10 Richard Biener <rguenther@suse.de>
2165 PR tree-optimization/60115
2166 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2167 MEM_REF handling. Properly verify that the accesses are not
2168 out of the objects bound.
2170 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2172 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2175 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2177 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2178 proper constants and fix formatting.
2179 (possible_polymorphic_call_targets): Fix formatting.
2181 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2182 Ilya Tocar <ilya.tocar@intel.com>
2184 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2185 (_mm512_loadu_epi32): Renamed into...
2186 (_mm512_loadu_si512): This.
2187 (_mm512_storeu_epi32): Renamed into...
2188 (_mm512_storeu_si512): This.
2189 (_mm512_maskz_ceil_ps): Removed.
2190 (_mm512_maskz_ceil_pd): Ditto.
2191 (_mm512_maskz_floor_ps): Ditto.
2192 (_mm512_maskz_floor_pd): Ditto.
2193 (_mm512_floor_round_ps): Ditto.
2194 (_mm512_floor_round_pd): Ditto.
2195 (_mm512_ceil_round_ps): Ditto.
2196 (_mm512_ceil_round_pd): Ditto.
2197 (_mm512_mask_floor_round_ps): Ditto.
2198 (_mm512_mask_floor_round_pd): Ditto.
2199 (_mm512_mask_ceil_round_ps): Ditto.
2200 (_mm512_mask_ceil_round_pd): Ditto.
2201 (_mm512_maskz_floor_round_ps): Ditto.
2202 (_mm512_maskz_floor_round_pd): Ditto.
2203 (_mm512_maskz_ceil_round_ps): Ditto.
2204 (_mm512_maskz_ceil_round_pd): Ditto.
2205 (_mm512_expand_pd): Ditto.
2206 (_mm512_expand_ps): Ditto.
2207 * config/i386/i386.c (ix86_builtins): Remove
2208 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2209 (bdesc_args): Ditto.
2210 * config/i386/predicates.md (const1256_operand): New.
2211 (const_1_to_2_operand): Ditto.
2212 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2213 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2214 (*avx512pf_gatherpf<mode>sf): Ditto.
2215 (avx512pf_gatherpf<mode>df): Ditto.
2216 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2217 (*avx512pf_gatherpf<mode>df): Ditto.
2218 (avx512pf_scatterpf<mode>sf): Ditto.
2219 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2220 (*avx512pf_scatterpf<mode>sf): Ditto.
2221 (avx512pf_scatterpf<mode>df): Ditto.
2222 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2223 (*avx512pf_scatterpf<mode>df): Ditto.
2224 (avx512f_expand<mode>): Removed.
2225 (<shift_insn><mode>3<mask_name>): Change predicate type.
2227 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2229 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2230 not at the end of datarefs vector use ordered_remove to avoid
2231 reordering datarefs vector.
2234 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2235 mark local addressable non-static vars as GOVD_PRIVATE
2236 instead of GOVD_LOCAL.
2237 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2238 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2242 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2243 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2244 assume_aligned or alloc_align attributes.
2245 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2246 arguments. Handle also assume_aligned and alloc_align attributes.
2247 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2248 calls to functions with assume_aligned or alloc_align attributes.
2249 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2251 2014-02-08 Terry Guo <terry.guo@arm.com>
2253 * doc/invoke.texi: Document ARM -march=armv7e-m.
2255 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2257 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2258 flag on __cilkrts_rethrow builtin.
2261 * ipa-cp.c (determine_versionability): Fail at -O0
2262 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2263 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2266 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2269 * tree-inline.c (copy_forbidden): Fail for
2270 __attribute__((optimize (0))) functions.
2272 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2274 * varpool.c: Include pointer-set.h.
2275 (varpool_remove_unreferenced_decls): Variables in other partitions
2276 will not be output; be however careful to not lose information
2279 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2281 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2282 lookup in the vtable constructor.
2284 2014-02-07 Jeff Law <law@redhat.com>
2287 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2288 define_insn_and_split.
2290 * ipa-inline.c (inline_small_functions): Fix typos.
2292 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2294 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2295 (s390_can_use_return_insn): Declare.
2296 * config/s390/s390.h (EPILOGUE_USES): Define.
2297 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2299 (s390_chunkify_start): Handle return JUMP_LABELs.
2300 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2301 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2302 (s390_can_use_return_insn): New functions.
2303 (s390_fix_long_loop_prediction): Handle conditional returns.
2304 (TARGET_SET_UP_BY_PROLOGUE): Define.
2305 * config/s390/s390.md (ANY_RETURN): New code iterator.
2306 (*creturn, *csimple_return, return, simple_return): New patterns.
2308 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2310 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2311 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2312 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2313 REG_CFA_RESTORE list when deciding not to restore a register.
2315 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2317 * config/s390/s390.c: Include tree-pass.h and context.h.
2318 (s390_early_mach): New function, split out from...
2319 (s390_emit_prologue): ...here.
2320 (pass_data_s390_early_mach): New pass structure.
2321 (pass_s390_early_mach): New class.
2322 (s390_option_override): Create and register early_mach pass.
2323 Move to end of file.
2325 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2327 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2328 to match for the exit block.
2330 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2332 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2333 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2334 Reject misaligned operands.
2336 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2338 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2340 2014-02-07 Richard Biener <rguenther@suse.de>
2343 * gimple-low.c (lower_builtin_posix_memalign): New function.
2344 (lower_stmt): Call it to lower posix_memalign in a way
2345 to make alignment info accessible.
2347 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2350 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2351 __builtin_setjmp_receiver.
2353 2014-02-07 Richard Biener <rguenther@suse.de>
2356 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2357 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2358 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2359 Handle BUILT_IN_POSIX_MEMALIGN.
2360 (find_func_clobbers): Likewise.
2361 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2362 (call_may_clobber_ref_p_1): Likewise.
2364 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2367 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2370 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2373 * lto-cgraph.c (lto_output_node): Use
2374 symtab_get_symbol_partitioning_class.
2375 (lto_output_varpool_node): likewise.
2376 (symtab_get_symbol_partitioning_class): Move here from
2378 * cgraph.h (symbol_partitioning_class): Likewise.
2379 (symtab_get_symbol_partitioning_class): Declare.
2381 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2383 * ggc.h (ggc_internal_cleared_alloc): New macro.
2384 * vec.h (vec_safe_copy): Handle memory stats.
2385 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2386 * target-globals.c (save_target_globals): Likewise.
2388 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2391 * expr.c (emit_move_resolve_push): Export; be bit more selective
2392 on when to clear alias set.
2393 * expr.h (emit_move_resolve_push): Declare.
2394 * function.h (struct function): Add tail_call_marked.
2395 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2396 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2397 * config/i386/i386.md (TImode move expander): De not call
2399 (FP push expanders): Preserve memory attributes.
2400 * config/i386/sse.md (push<mode>1): Remove.
2401 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2402 (ix86_expand_push): Remove.
2403 * config/i386/mmx.md (push<mode>1): Remove.
2405 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2407 PR rtl-optimization/60030
2408 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2409 lopart with paradoxical subreg before shifting it up by hprec.
2411 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2413 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2414 Remove extra newline at end of file.
2415 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2416 (arm_issue_rate): Handle cortexa57.
2417 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2418 (cortex-a57.cortex-a53): Likewise.
2420 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2423 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2424 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2426 (arm_expand_prologue): Adjust all callers.
2427 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2428 info, registers also at the lowest numbered registers side. Use
2429 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2433 * var-tracking.c (adjust_mems): Before adding a SET to
2434 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2436 2014-02-06 Alan Modra <amodra@gmail.com>
2439 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2440 change SDmode to DDmode when lra_in_progress.
2442 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2445 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2446 free_data_ref on the dr first, and before goto again also set dr
2447 to the next dr. For simd_lane_access, free old datarefs[i] before
2448 overwriting it. For get_vectype_for_scalar_type failure, don't
2449 free_data_ref if simd_lane_access.
2451 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2454 * tree.h (opts_for_fn): New inline function.
2455 (opt_for_fn): Define.
2456 * config/i386/i386.c (ix86_function_regparm): Use
2457 opt_for_fn (decl, optimize) instead of optimize.
2459 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2461 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2462 for SYMBOL_REF in large memory model.
2464 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2466 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2468 (cortex-a57): Likewise.
2469 (cortex-a57.cortex-a53): Likewise.
2471 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2472 Kugan Vivekanandarajah <kuganv@linaro.org>
2474 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2476 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2478 2014-02-06 Richard Biener <rguenther@suse.de>
2480 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2481 set_loop_copy and initialize_original_copy_tables.
2483 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2485 * config/aarch64/aarch64-simd.md
2486 (aarch64_ashr_simddi): Change QI to SI.
2488 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2489 Jakub Jelinek <jakub@redhat.com>
2492 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2495 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2497 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2498 CODE_FOR_altivec_vpku[hw]um to
2499 CODE_FOR_altivec_vpku[hw]um_direct.
2500 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2501 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2502 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2503 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2505 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2507 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2508 generation for -maltivec=be.
2509 (altivec_vsumsws): Simplify redundant test.
2511 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2513 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2514 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2515 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2516 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2517 gen_altivec_vpkuwum.
2518 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2520 (altivec_vpks<VI_char>ss): Likewise.
2521 (altivec_vpks<VI_char>us): Likewise.
2522 (altivec_vpku<VI_char>us): Likewise.
2523 (altivec_vpku<VI_char>um): Likewise.
2524 (altivec_vpku<VI_char>um_direct): New (copy of
2525 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2527 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2528 target is little endian and -maltivec=be is not specified.
2529 (*altivec_vupkhs<VU_char>_direct): New (copy of
2530 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2531 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2532 target is little endian and -maltivec=be is not specified.
2533 (*altivec_vupkls<VU_char>_direct): New (copy of
2534 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2535 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2536 little endian and -maltivec=be is not specified.
2537 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2538 little endian and -maltivec=be is not specified.
2540 2014-02-05 Richard Henderson <rth@redhat.com>
2543 * combine-stack-adj.c: Revert r206943.
2544 * sched-int.h (struct deps_desc): Add last_args_size.
2545 * sched-deps.c (init_deps): Initialize it.
2546 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2547 contain REG_ARGS_SIZE notes.
2549 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2551 * lto-cgraph.c (asm_nodes_output): Make global.
2552 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2553 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2554 (driver_handle_option): Handle OPT_fwpa.
2556 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2559 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2560 a comment typo and formatting issue. If odr_hash hasn't been
2561 created, return vNULL and set *completep to false.
2564 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2565 bb with no successors.
2567 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2570 * doc/invoke.texi (-march): Clarify documentation for ARM.
2574 2014-02-05 Richard Biener <rguenther@suse.de>
2576 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2577 when not vectorizing because of too many alias checks.
2578 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2579 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2581 2014-02-05 Nick Clifton <nickc@redhat.com>
2583 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2584 accept extended registers in any mode when compiling for the MN10300.
2586 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2588 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2589 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2590 sanitization attributes.
2591 (can_inline_edge_p): Likewise.
2592 (sanitize_attrs_match_for_inline_p): New function.
2594 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2596 * ipa-prop.c (detect_type_change): Shor circuit testing of
2597 type changes on THIS pointer.
2599 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2602 * config/pa/pa.c (legitimize_tls_address): Return original address
2603 if not passed a SYMBOL_REF rtx.
2604 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2606 (pa_emit_move_sequence): Simplify TLS source operands.
2607 (pa_legitimate_constant_p): Reject all TLS constants.
2608 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2609 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2611 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2613 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2614 groups when we know they are controlled by LTO.
2615 * varasm.c (default_binds_local_p_1): If object is in other partition,
2616 it will be resolved locally.
2618 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2620 * config/host-linux.c (linux_gt_pch_use_address): Don't
2621 use SSIZE_MAX because it is not always defined.
2623 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2626 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2627 threshold for pseudo splitting.
2628 (update_ebb_live_info): Process call argument hard registers and
2629 hard registers from insn definition too.
2630 (max_small_class_regs_num): New constant.
2631 (inherit_in_ebb): Update live hard regs through EBBs. Update
2632 reloads_num only for small register classes. Don't split for
2635 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2638 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2641 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2643 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2646 2014-02-04 Marek Polacek <polacek@redhat.com>
2648 * gdbinit.in (pel): Define.
2650 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2652 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2655 2014-02-04 Richard Biener <rguenther@suse.de>
2658 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2659 in function context local.
2660 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2661 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2662 similar to LTO_imported_decl_ref.
2664 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2666 PR tree-optimization/60002
2667 * cgraphclones.c (build_function_decl_skip_args): Clear
2670 PR tree-optimization/60023
2671 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2672 false to gsi_replace.
2673 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2674 has been in some EH region and vec_stmt could throw, add
2675 vec_stmt into the same EH region.
2676 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2677 has no lhs, ignore it.
2678 * internal-fn.c (expand_MASK_LOAD): Likewise.
2681 * tree-inline.c (copy_forbidden): Fail for
2682 __attribute__((optimize (0))) functions.
2685 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2686 is set, copy one less argument.
2687 (expand_simd_clones): Don't subtract clone_info->inbranch
2688 from simd_clone_struct_alloc argument.
2690 PR rtl-optimization/57915
2691 * recog.c (simplify_while_replacing): If all unary/binary/relational
2692 operation arguments are constant, attempt to simplify those.
2695 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2696 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2698 2014-02-04 Richard Biener <rguenther@suse.de>
2700 PR tree-optimization/60012
2701 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2702 TBAA disambiguation to all DDRs.
2704 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2707 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2708 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2710 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2713 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2715 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2717 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2718 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2720 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2723 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2724 to figure out targets of polymorphic calls with known decl.
2725 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2726 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2727 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2728 (get_polymorphic_call_info): ... here.
2729 (get_polymorphic_call_info_from_invariant): New function.
2731 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2733 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2734 lookup via vtable pointer; check for type consistency
2735 and turn inconsitent facts into UNREACHABLE.
2736 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2737 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2738 type inconsistent querries; return UNREACHABLE instead.
2740 2014-02-03 Richard Henderson <rth@twiddle.net>
2743 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2744 already processed this node.
2745 (normalize_one_pred_1): Pass along mark_set.
2746 (normalize_one_pred): Create and destroy a pointer_set_t.
2747 (normalize_one_pred_chain): Likewise.
2749 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2751 PR gcov-profile/58602
2752 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2754 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2757 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2758 -fno-devirtualize; try to devirtualize by the knowledge of
2759 virtual table pointer given by aggregate propagation.
2760 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2761 (ipa_print_node_jump_functions): Dump also offset that
2762 is relevant for polymorphic calls.
2763 (determine_known_aggregate_parts): Add arg_type parameter; use it
2764 instead of determining the type from pointer type.
2765 (ipa_compute_jump_functions_for_edge): Update call of
2766 determine_known_aggregate_parts.
2767 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2768 (gimple_get_virt_method_for_binfo): ... here; simplify using
2769 vtable_pointer_value_to_vtable.
2770 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2771 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2772 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2773 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2775 (vtable_pointer_value_to_binfo): ... here.
2776 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2778 2014-02-03 Teresa Johnson <tejohnson@google.com>
2780 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2781 redef of outer loop index variable.
2783 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2787 * doc/extend.texi (Function Attributes): Typo.
2789 2014-02-03 Cong Hou <congh@google.com>
2791 PR tree-optimization/60000
2792 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2793 if the vectorized statement is a store. A store statement can only
2794 appear at the end of pattern statements.
2796 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2798 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2799 (ix86_option_override_internal): Default long double to 64-bit for
2800 32-bit Bionic and to 128-bit for 64-bit Bionic.
2802 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2803 TARGET_LONG_DOUBLE_128 is true.
2804 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2806 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2807 (mlong-double-64): Negate -mlong-double-128.
2808 (mlong-double-128): New option.
2810 * config/i386/i386-c.c (ix86_target_macros): Define
2811 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2813 * doc/invoke.texi: Document -mlong-double-128.
2815 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2817 PR rtl-optimization/60024
2818 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2820 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2822 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2824 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2826 PR rtl-optimization/57662
2827 * sel-sched.c (code_motion_path_driver): Do not mark already not
2828 existing blocks in the visiting bitmap.
2830 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2832 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2833 on the insn being emitted.
2835 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2836 Will Deacon <will.deacon@arm.com>
2838 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2840 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2842 * config/arm/arm-tables.opt: Regenerate.
2844 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2846 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2847 for vector types other than V16QImode.
2848 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2849 define_expand, and call altivec_expand_vec_perm_le when producing
2850 code with little endian element order.
2851 (*altivec_vperm_<mode>_internal): New insn having previous
2852 behavior of altivec_vperm_<mode>.
2853 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2854 altivec_expand_vec_perm_le when producing code with little endian
2856 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2857 behavior of altivec_vperm_<mode>_uns.
2859 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2861 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2862 (altivec_vsumsws): Add handling for -maltivec=be with a little
2864 (altivec_vsumsws_direct): New.
2865 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2866 gen_altivec_vsumsws.
2868 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2870 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2871 vtable_pointer_value_to_binfo): New functions.
2872 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2873 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2875 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2877 * config/nios2/nios2.md (load_got_register): Initialize GOT
2878 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2879 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2881 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2883 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2884 preserverd by passthrough, do not propagate the type.
2886 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2888 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2889 (mips_atomic_assign_expand_fenv): New function.
2890 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2892 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2894 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2895 (__builtin_mips_set_fcsr): Likewise.
2896 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2897 MIPS_USI_FTYPE_VOID.
2898 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2899 (mips16_expand_set_fcsr): Likewise.
2900 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2901 (mips16_set_fcsr_stub): Likewise.
2902 (mips16_get_fcsr_one_only_stub): New class.
2903 (mips16_set_fcsr_one_only_stub): Likewise.
2904 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2905 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2906 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2907 (hard_float): New availability predicate.
2908 (mips_builtins): Add get_fcsr and set_fcsr.
2909 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2910 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2911 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2912 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2913 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2916 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2918 * config/mips/mips.c (mips_one_only_stub): New class.
2919 (mips_need_mips16_rdhwr_p): Replace with...
2920 (mips16_rdhwr_stub): ...this new variable.
2921 (mips16_stub_call_address): New function.
2922 (mips16_rdhwr_one_only_stub): New class.
2923 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2924 (mips_output_mips16_rdhwr): Delete.
2925 (mips_finish_stub): New function.
2926 (mips_code_end): Use it to handle rdhwr stubs.
2928 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2931 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2932 when calculating size of integer atomic types.
2934 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2936 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2938 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2940 PR tree-optimization/60003
2941 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2942 * profile.c (branch_prob): Use gimple_call_builtin_p
2943 to check for BUILT_IN_SETJMP_RECEIVER.
2944 * tree-inline.c (copy_bb): Call notice_special_calls.
2946 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2949 * lra-constraints.c (process_alt_operands): Update reload_sum only
2952 2014-01-31 Richard Henderson <rth@redhat.com>
2955 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2956 until after else_eh is processed.
2958 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2960 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2961 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2962 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2963 in smmintrin.h, remove them.
2964 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2965 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2966 * config/i386/i386.md (ROUND_SAE): Fix value.
2967 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2968 (const48_operand): New.
2969 * config/i386/subst.md (round), (round_expand): Use
2970 const_4_or_8_to_11_operand.
2971 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2973 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2975 * config/i386/constraints.md (Yk): Swap meaning with k.
2976 * config/i386/i386.md (movhi_internal): Change Yk to k.
2977 (movqi_internal): Ditto.
2978 (*k<logic><mode>): Ditto.
2981 (kandn<mode>): Ditto.
2982 (*<code>hi_1): Ditto.
2983 (*<code>qi_1): Ditto.
2984 (kxnor<mode>): Ditto.
2985 (kortestzhi): Ditto.
2986 (kortestchi): Ditto.
2988 (*one_cmplhi2_1): Ditto.
2989 (*one_cmplqi2_1): Ditto.
2990 * config/i386/sse.md (): Change k to Yk.
2991 (avx512f_load<mode>_mask): Ditto.
2992 (avx512f_blendm<mode>): Ditto.
2993 (avx512f_store<mode>_mask): Ditto.
2994 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2995 (avx512f_storedqu<mode>_mask): Ditto.
2996 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2998 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2999 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3000 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3001 (avx512f_maskcmp<mode>3): Ditto.
3002 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3003 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3004 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3005 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3006 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3007 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3008 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3009 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3010 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3011 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3012 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3013 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3014 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3015 (vec_extract_lo_<mode>_maskm): Ditto.
3016 (vec_extract_hi_<mode>_maskm): Ditto.
3017 (avx512f_vternlog<mode>_mask): Ditto.
3018 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3019 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3020 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3021 (avx512f_<code>v8div16qi2_mask): Ditto.
3022 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3023 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3024 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3025 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3026 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3027 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3028 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3029 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3030 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3031 (avx512cd_maskb_vec_dupv8di): Ditto.
3032 (avx512cd_maskw_vec_dupv16si): Ditto.
3033 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3034 (avx512f_vpermi2var<mode>3_mask): Ditto.
3035 (avx512f_vpermi2var<mode>3_mask): Ditto.
3036 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3037 (*avx512f_gathersi<mode>): Ditto.
3038 (*avx512f_gathersi<mode>_2): Ditto.
3039 (*avx512f_gatherdi<mode>): Ditto.
3040 (*avx512f_gatherdi<mode>_2): Ditto.
3041 (*avx512f_scattersi<mode>): Ditto.
3042 (*avx512f_scatterdi<mode>): Ditto.
3043 (avx512f_compress<mode>_mask): Ditto.
3044 (avx512f_compressstore<mode>_mask): Ditto.
3045 (avx512f_expand<mode>_mask): Ditto.
3046 * config/i386/subst.md (mask): Change k to Yk.
3047 (mask_scalar_merge): Ditto.
3050 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3052 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3054 2014-01-31 Richard Biener <rguenther@suse.de>
3057 * builtins.c (fold_builtin_memory_op): Make sure to not
3058 use a floating-point mode or a boolean or enumeral type for
3061 2014-01-30 DJ Delorie <dj@redhat.com>
3063 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3064 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3065 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3066 whenever main() has an epilogue.
3068 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3070 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3071 unused variable "field".
3072 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3073 (vsx_mergeh_<mode>): Likewise.
3074 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3075 (altivec_vmrghh): Likewise.
3076 (altivec_vmrghw): Likewise.
3077 (altivec_vmrglb): Likewise.
3078 (altivec_vmrglh): Likewise.
3079 (altivec_vmrglw): Likewise.
3080 (altivec_vspltb): Add missing uses.
3081 (altivec_vsplth): Likewise.
3082 (altivec_vspltw): Likewise.
3083 (altivec_vspltsf): Likewise.
3085 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3088 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3089 frame related instructions.
3091 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3093 PR rtl-optimization/59959
3094 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3095 any reload of register whose subreg is invalid.
3097 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3099 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3100 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3101 Add missing return type - void.
3103 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3105 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3106 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3107 remove element index adjustment for endian (now handled in vsx.md
3109 (altivec_expand_vec_perm_const): Use
3110 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3111 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3112 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3113 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3114 define_expand and a new define_insn *altivec_vspltb_internal;
3115 adjust for -maltivec=be on a little endian target.
3116 (altivec_vspltb_direct): New.
3117 (altivec_vsplth): Divide into a define_expand and a new
3118 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3119 little endian target.
3120 (altivec_vsplth_direct): New.
3121 (altivec_vspltw): Divide into a define_expand and a new
3122 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3123 little endian target.
3124 (altivec_vspltw_direct): New.
3125 (altivec_vspltsf): Divide into a define_expand and a new
3126 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3127 a little endian target.
3129 2014-01-30 Richard Biener <rguenther@suse.de>
3131 PR tree-optimization/59993
3132 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3133 can propagate form the earlier stmt and avoid the transform
3134 when the intermediate result is needed.
3136 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3138 * README.Portability: Fix typo.
3140 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3142 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3143 comparison_operator with ordered_comparison_operator.
3145 2014-01-30 Nick Clifton <nickc@redhat.com>
3147 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3148 Rename to mn10300_store_multiple_regs.
3149 * config/mn10300/mn10300.c: Likewise.
3150 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3151 store_multiple_regs.
3152 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3153 Call mn10300_store_multiple_regs.
3155 2014-01-30 Nick Clifton <nickc@redhat.com>
3156 DJ Delorie <dj@redhat.com>
3158 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3159 %fp 2 to keep registers after it properly word-aligned.
3160 (rl78_alloc_physical_registers_umul): Handle the case where both
3161 input operands are the same.
3163 2014-01-30 Richard Biener <rguenther@suse.de>
3165 PR tree-optimization/59903
3166 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3169 2014-01-30 Jason Merrill <jason@redhat.com>
3172 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3175 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3177 2014-01-30 Richard Biener <rguenther@suse.de>
3179 PR tree-optimization/59951
3180 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3182 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3185 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3186 SFmode to DFmode case.
3188 2014-01-29 DJ Delorie <dj@redhat.com>
3190 * config/msp430/msp430.opt (-minrt): New.
3191 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3193 (ENDFILE_SPEC): Likewise.
3195 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3197 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3198 (estimate_function_body_sizes): Use it.
3200 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3203 * dwarf2out.c (is_cxx_auto): New.
3204 (is_base_type): Use it.
3205 (gen_type_die_with_usage): Likewise.
3207 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3209 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3210 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3211 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3212 -maltivec=be with LE targets.
3213 (vsx_mergeh_<mode>): Likewise.
3214 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3215 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3216 (altivec_vmrghb): Replace with define_expand and new
3217 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3218 (altivec_vmrghb_direct): New define_insn.
3219 (altivec_vmrghh): Replace with define_expand and new
3220 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3221 (altivec_vmrghh_direct): New define_insn.
3222 (altivec_vmrghw): Replace with define_expand and new
3223 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3224 (altivec_vmrghw_direct): New define_insn.
3225 (*altivec_vmrghsf): Adjust for endianness.
3226 (altivec_vmrglb): Replace with define_expand and new
3227 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3228 (altivec_vmrglb_direct): New define_insn.
3229 (altivec_vmrglh): Replace with define_expand and new
3230 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3231 (altivec_vmrglh_direct): New define_insn.
3232 (altivec_vmrglw): Replace with define_expand and new
3233 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3234 (altivec_vmrglw_direct): New define_insn.
3235 (*altivec_vmrglsf): Adjust for endianness.
3236 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3237 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3238 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3239 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3240 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3241 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3242 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3243 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3245 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3247 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3248 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3251 2014-01-29 Richard Biener <rguenther@suse.de>
3253 PR tree-optimization/58742
3254 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3255 associate_pointerplus_align.
3256 (associate_pointerplus_diff): New function.
3257 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3258 and associate_pointerplus_diff.
3260 2014-01-29 Richard Biener <rguenther@suse.de>
3262 * lto-streamer.h (LTO_major_version): Bump to 3.
3263 (LTO_minor_version): Reset to 0.
3265 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3267 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3268 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3269 (arm_file_start): Generate correct asm header for armv7ve.
3270 * config/arm/bpabi.h: Add multilib support for armv7ve.
3271 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3272 and cortex-a15 to armv7ve.
3273 * config/arm/t-aprofile: Add multilib support for armv7ve.
3274 * doc/invoke.texi: Document -march=armv7ve.
3276 2014-01-29 Richard Biener <rguenther@suse.de>
3278 PR tree-optimization/58742
3279 * tree-ssa-forwprop.c (associate_plusminus): Return true
3280 if we changed sth, defer EH cleanup to ...
3281 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3282 (simplify_mult): New function.
3284 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3287 PR tree-optimization/59920
3288 * tree.c (build_common_builtin_nodes): Remove
3289 __builtin_setjmp_dispatcher initialization.
3290 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3291 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3292 instead of gsi_after_labels + manually skipping debug stmts.
3293 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3294 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3295 * tree-inline.c (copy_edges_for_bb): Remove
3296 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3297 argument. Ignore computed_goto_p stmts. Don't call
3298 make_abnormal_goto_edges. If a call might need abnormal edges
3299 for non-local gotos, see if it already has an edge to
3300 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3301 with true argument, don't do anything then, otherwise add
3302 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3303 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3305 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3306 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3307 (lower_stmt): Don't set data->calls_builtin_setjmp.
3308 (lower_builtin_setjmp): Adjust comment.
3309 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3310 * tree-cfg.c (found_computed_goto): Remove.
3311 (factor_computed_gotos): Remove.
3312 (make_goto_expr_edges): Return bool, true for computed gotos.
3313 Don't call make_abnormal_goto_edges.
3314 (build_gimple_cfg): Don't set found_computed_goto, don't call
3315 factor_computed_gotos.
3316 (computed_goto_p): No longer static.
3317 (make_blocks): Don't set found_computed_goto.
3318 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3319 (make_edges): If make_goto_expr_edges returns true, push bb
3320 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3321 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3322 vector. Record mapping between bbs and OpenMP regions if there
3323 are any, adjust make_gimple_omp_edges caller. Call
3324 handle_abnormal_edges.
3325 (make_abnormal_goto_edges): Remove.
3326 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3327 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3328 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3329 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3330 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3331 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3332 filling *region also set *region_idx to (*region)->entry->index.
3335 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3336 For REGs set ORIGINAL_REGNO.
3338 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3340 * doc/md.texi: Mention that a target shouldn't implement
3341 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3344 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3346 PR tree-optimization/59594
3347 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3348 a copy of the datarefs vector rather than the vector itself.
3350 2014-01-28 Jason Merrill <jason@redhat.com>
3353 * dwarf2out.c (auto_die): New static.
3354 (gen_type_die_with_usage): Handle C++1y 'auto'.
3355 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3358 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3361 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3362 (SPEC_X32): Likewise.
3363 (SPEC_64): Likewise.
3364 * config/i386/i386.c (ix86_option_override_internal): Turn off
3365 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3367 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3368 * config/i386/i386.h (TARGET_16BIT): New macro.
3369 (TARGET_16BIT_P): Likewise.
3370 * config/i386/i386.opt: Add m16.
3371 * doc/invoke.texi: Document -m16.
3373 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3375 PR preprocessor/59935
3376 * input.c (location_get_source_line): Bail out on when line number
3377 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3379 2014-01-28 Richard Biener <rguenther@suse.de>
3381 PR tree-optimization/58742
3382 * tree-ssa-forwprop.c (associate_plusminus): Handle
3383 pointer subtraction of the form (T)(P + A) - (T)P.
3385 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3387 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3390 2014-01-28 Richard Biener <rguenther@suse.de>
3393 2014-01-28 Richard Biener <rguenther@suse.de>
3395 PR rtl-optimization/45364
3396 PR rtl-optimization/59890
3397 * var-tracking.c (local_get_addr_clear_given_value): Handle
3398 already cleared slot.
3399 (val_reset): Handle not allocated local_get_addr_cache.
3400 (vt_find_locations): Use post-order on the inverted CFG.
3402 2014-01-28 Richard Biener <rguenther@suse.de>
3404 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3406 2014-01-28 Richard Biener <rguenther@suse.de>
3408 PR rtl-optimization/45364
3409 PR rtl-optimization/59890
3410 * var-tracking.c (local_get_addr_clear_given_value): Handle
3411 already cleared slot.
3412 (val_reset): Handle not allocated local_get_addr_cache.
3413 (vt_find_locations): Use post-order on the inverted CFG.
3415 2014-01-28 Alan Modra <amodra@gmail.com>
3417 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3418 * configure.ac <recursive call for build != host>: Define
3419 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3420 and LD_FOR_BUILD too.
3421 * configure: Regenerate.
3423 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3425 * config/i386/i386.c (get_builtin_code_for_version): Separate
3426 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3427 Broadwell from Haswell.
3429 2014-01-27 Steve Ellcey <sellcey@mips.com>
3431 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3432 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3433 * config/mips/mips.c (mips_option_override): Change setting
3435 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3436 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3437 Change from Mask to Var.
3439 2014-01-27 Jeff Law <law@redhat.com>
3441 * ipa-inline.c (inline_small_functions): Fix typo.
3443 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3445 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3446 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3447 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3448 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3449 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3450 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3451 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3452 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3453 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3454 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3455 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3456 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3457 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3458 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3459 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3460 (_mm512_storeu_epi64): Ditto.
3461 (_mm512_cmpge_epi32_mask): Ditto.
3462 (_mm512_cmpge_epu32_mask): Ditto.
3463 (_mm512_cmpge_epi64_mask): Ditto.
3464 (_mm512_cmpge_epu64_mask): Ditto.
3465 (_mm512_cmple_epi32_mask): Ditto.
3466 (_mm512_cmple_epu32_mask): Ditto.
3467 (_mm512_cmple_epi64_mask): Ditto.
3468 (_mm512_cmple_epu64_mask): Ditto.
3469 (_mm512_cmplt_epi32_mask): Ditto.
3470 (_mm512_cmplt_epu32_mask): Ditto.
3471 (_mm512_cmplt_epi64_mask): Ditto.
3472 (_mm512_cmplt_epu64_mask): Ditto.
3473 (_mm512_cmpneq_epi32_mask): Ditto.
3474 (_mm512_cmpneq_epu32_mask): Ditto.
3475 (_mm512_cmpneq_epi64_mask): Ditto.
3476 (_mm512_cmpneq_epu64_mask): Ditto.
3477 (_mm512_expand_pd): Ditto.
3478 (_mm512_expand_ps): Ditto.
3479 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3480 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3481 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3482 * config/i386/i386.c (ix86_builtins): Add
3483 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3484 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3485 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3486 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3487 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3488 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3489 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3490 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3491 IX86_BUILTIN_PMOVUSQW512_MEM.
3492 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3493 __builtin_ia32_pmovsqd512mem_mask,
3494 __builtin_ia32_pmovqd512mem_mask,
3495 __builtin_ia32_pmovusqw512mem_mask,
3496 __builtin_ia32_pmovsqw512mem_mask,
3497 __builtin_ia32_pmovqw512mem_mask,
3498 __builtin_ia32_pmovusdw512mem_mask,
3499 __builtin_ia32_pmovsdw512mem_mask,
3500 __builtin_ia32_pmovdw512mem_mask,
3501 __builtin_ia32_pmovqb512mem_mask,
3502 __builtin_ia32_pmovusqb512mem_mask,
3503 __builtin_ia32_pmovsqb512mem_mask,
3504 __builtin_ia32_pmovusdb512mem_mask,
3505 __builtin_ia32_pmovsdb512mem_mask,
3506 __builtin_ia32_pmovdb512mem_mask.
3507 (bdesc_args): Add __builtin_ia32_expanddf512,
3508 __builtin_ia32_expandsf512.
3509 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3510 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3511 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3512 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3513 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3514 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3515 (avx512f_<code>v8div16qi2_mask_store): This.
3516 (avx512f_expand<mode>): New.
3518 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3520 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3522 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3523 (_mm512_prefetch_i32scatter_pd): Ditto.
3524 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3525 (_mm512_prefetch_i64scatter_pd): Ditto.
3526 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3527 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3528 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3529 (_mm512_prefetch_i32scatter_ps): Ditto.
3530 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3531 (_mm512_prefetch_i64scatter_ps): Ditto.
3532 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3533 * config/i386/i386-builtin-types.def: Define
3534 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3535 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3536 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3537 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3538 IX86_BUILTIN_SCATTERPFQPD.
3539 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3540 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3541 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3542 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3543 __builtin_ia32_scatterpfqps.
3544 (ix86_expand_builtin): Expand new built-ins.
3545 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3546 fix memory access data type.
3547 (*avx512pf_gatherpf<mode>_mask): Ditto.
3548 (*avx512pf_gatherpf<mode>): Ditto.
3549 (avx512pf_scatterpf<mode>): Ditto.
3550 (*avx512pf_scatterpf<mode>_mask): Ditto.
3551 (*avx512pf_scatterpf<mode>): Ditto.
3552 (GATHER_SCATTER_SF_MEM_MODE): New.
3553 (avx512pf_gatherpf<mode>df): Ditto.
3554 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3555 (*avx512pf_scatterpf<mode>df): Ditto.
3557 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3560 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3561 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3564 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3566 * common/config/arm/arm-common.c
3567 (arm_rewrite_mcpu): Handle multiple names.
3569 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3571 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3573 * gimple-builder.h (create_gimple_tmp): Delete.
3575 2014-01-27 Christian Bruel <christian.bruel@st.com>
3577 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3580 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3582 * config/pa/pa.md (call): Generate indirect long calls to non-local
3583 functions when outputing 32-bit code.
3584 (call_value): Likewise except for special call to buggy powf function.
3586 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3587 portable runtime and PIC indirect calls.
3588 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3589 and PIC call sequences. Use ldo instead of blr to set return register
3590 in PIC call sequence.
3592 2014-01-25 Walter Lee <walt@tilera.com>
3594 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3595 avoid clobbering a live register.
3597 2014-01-25 Walter Lee <walt@tilera.com>
3599 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3600 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3601 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3602 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3604 2014-01-25 Walter Lee <walt@tilera.com>
3606 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3607 arguments on even registers.
3608 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3610 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3611 (BIGGEST_ALIGNMENT): Ditto.
3612 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3614 2014-01-25 Walter Lee <walt@tilera.com>
3616 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3617 insns before bundling.
3618 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3620 2014-01-25 Walter Lee <walt@tilera.com>
3622 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3623 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3624 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3626 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3628 * config/mips/constraints.md (kl): Delete.
3629 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3630 define expands, using...
3631 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3632 instructions for MIPS16.
3633 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3634 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3636 2014-01-25 Walter Lee <walt@tilera.com>
3638 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3642 2014-01-25 Walter Lee <walt@tilera.com>
3644 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3645 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3647 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3649 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3652 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3654 * print-rtl.c (in_call_function_usage): New var.
3655 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3656 EXPR_LIST mode as mode and not as reg note name.
3659 * cfgloopmanip.c (copy_loop_info): If
3660 loop->warned_aggressive_loop_optimizations, make sure
3661 the flag is set in target loop too.
3663 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3665 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3667 * builtins.def: Likewise.
3668 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3669 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3670 * ira.c (ira_setup_eliminable_regset): Likewise.
3671 * omp-low.c (gate_expand_omp): Likewise.
3672 (execute_lower_omp): Likewise.
3673 (diagnose_sb_0): Likewise.
3674 (gate_diagnose_omp_blocks): Likewise.
3675 (simd_clone_clauses_extract): Likewise.
3678 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3680 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3681 correction for little endian...
3682 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3685 2014-01-24 Jeff Law <law@redhat.com>
3687 PR tree-optimization/59919
3688 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3689 for non-returning calls.
3691 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3693 * common/config/aarch64/aarch64-common.c
3694 (aarch64_rewrite_mcpu): Handle multiple names.
3695 * config/aarch64/aarch64.h
3696 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3698 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3700 * input.c (add_file_to_cache_tab): Handle the case where fopen
3703 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3706 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3707 from push operand if code of push isn't PRE_DEC.
3709 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3712 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3713 -mquad-memory-atomic. Update -mquad-memory documentation to say
3714 it is only used for non-atomic loads/stores.
3716 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3717 -mquad-memory or -mquad-memory-atomic switches.
3719 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3720 -mquad-memory-atomic to ISA 2.07 support.
3722 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3723 to separate support of normal quad word memory operations (ldq, stq)
3724 from the atomic quad word memory operations.
3726 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3727 support to separate non-atomic quad word operations from atomic
3728 quad word operations. Disable non-atomic quad word operations in
3729 little endian mode so that we don't have to swap words after the
3730 load and before the store.
3731 (quad_load_store_p): Add comment about atomic quad word support.
3732 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3733 options printed with -mdebug=reg.
3735 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3736 -mquad-memory-atomic as the test for whether we have quad word
3737 atomic instructions.
3738 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3739 or -mp8-vector are used, allow byte/half-word atomic operations.
3741 * config/rs6000/sync.md (load_lockedti): Insure that the address
3742 is a proper indexed or indirect address for the lqarx instruction.
3743 On little endian systems, swap the hi/lo registers after the lqarx
3745 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3746 insure the address is valid for the lqarx instruction.
3747 (store_conditionalti): Insure that the address is a proper indexed
3748 or indirect address for the stqcrx. instruction. On little endian
3749 systems, swap the hi/lo registers before doing the stqcrx.
3751 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3752 insure the address is valid for the stqcrx. instruction.
3754 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3755 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3756 type of quad memory support is available.
3758 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3761 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3762 there is a danger of looping.
3764 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3767 force flag_ira_loop_pressure if set via command line.
3769 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3771 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3772 (ashr_simd): New builtin handling DI mode.
3773 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3774 (aarch64_sshr_simddi): New match pattern.
3775 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3776 (vshrd_n_s64): Likewise.
3777 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3779 2014-01-23 Nick Clifton <nickc@redhat.com>
3781 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3782 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3783 favour of mcu specific scripts.
3784 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3787 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3788 Alex Velenko <Alex.Velenko@arm.com>
3790 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3791 (vaddv_s16): Likewise.
3792 (vaddv_s32): Likewise.
3793 (vaddv_u8): Likewise.
3794 (vaddv_u16): Likewise.
3795 (vaddv_u32): Likewise.
3796 (vaddvq_s8): Likewise.
3797 (vaddvq_s16): Likewise.
3798 (vaddvq_s32): Likewise.
3799 (vaddvq_s64): Likewise.
3800 (vaddvq_u8): Likewise.
3801 (vaddvq_u16): Likewise.
3802 (vaddvq_u32): Likewise.
3803 (vaddvq_u64): Likewise.
3804 (vaddv_f32): Likewise.
3805 (vaddvq_f32): Likewise.
3806 (vaddvq_f64): Likewise.
3807 (vmaxv_f32): Likewise.
3808 (vmaxv_s8): Likewise.
3809 (vmaxv_s16): Likewise.
3810 (vmaxv_s32): Likewise.
3811 (vmaxv_u8): Likewise.
3812 (vmaxv_u16): Likewise.
3813 (vmaxv_u32): Likewise.
3814 (vmaxvq_f32): Likewise.
3815 (vmaxvq_f64): Likewise.
3816 (vmaxvq_s8): Likewise.
3817 (vmaxvq_s16): Likewise.
3818 (vmaxvq_s32): Likewise.
3819 (vmaxvq_u8): Likewise.
3820 (vmaxvq_u16): Likewise.
3821 (vmaxvq_u32): Likewise.
3822 (vmaxnmv_f32): Likewise.
3823 (vmaxnmvq_f32): Likewise.
3824 (vmaxnmvq_f64): Likewise.
3825 (vminv_f32): Likewise.
3826 (vminv_s8): Likewise.
3827 (vminv_s16): Likewise.
3828 (vminv_s32): Likewise.
3829 (vminv_u8): Likewise.
3830 (vminv_u16): Likewise.
3831 (vminv_u32): Likewise.
3832 (vminvq_f32): Likewise.
3833 (vminvq_f64): Likewise.
3834 (vminvq_s8): Likewise.
3835 (vminvq_s16): Likewise.
3836 (vminvq_s32): Likewise.
3837 (vminvq_u8): Likewise.
3838 (vminvq_u16): Likewise.
3839 (vminvq_u32): Likewise.
3840 (vminnmv_f32): Likewise.
3841 (vminnmvq_f32): Likewise.
3842 (vminnmvq_f64): Likewise.
3844 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3846 * config/aarch64/aarch64-simd.md
3847 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3848 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3849 (*aarch64_mul3_elt<mode>): Likewise.
3850 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3851 (*aarch64_mul3_elt_to_64v2df): Likewise.
3852 (*aarch64_mla_elt<mode>): Likewise.
3853 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3854 (*aarch64_mls_elt<mode>): Likewise.
3855 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3856 (*aarch64_fma4_elt<mode>): Likewise.
3857 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3858 (*aarch64_fma4_elt_to_64v2df): Likewise.
3859 (*aarch64_fnma4_elt<mode>): Likewise.
3860 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3861 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3862 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3863 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3864 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3865 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3866 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3867 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3868 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3870 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3872 * config/aarch64/aarch64-simd.md
3873 (aarch64_be_checked_get_lane<mode>): New define_expand.
3874 * config/aarch64/aarch64-simd-builtins.def
3875 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3876 New builtin definition.
3877 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3878 Use new safe be builtin.
3880 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3882 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3884 (aarch64_be_st1<mode>): Likewise.
3885 (aarch_ld1<VALL:mode>): Define_expand modified.
3886 (aarch_st1<VALL:mode>): Likewise.
3887 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3888 (UNSPEC_ST1): Likewise.
3890 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3892 * config/microblaze/microblaze.md: Add trap insn and attribute
3894 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3896 PR preprocessor/58580
3897 * input.h (location_get_source_line): Take an additional line_size
3899 (void diagnostics_file_cache_fini): Declare new function.
3900 * input.c (struct fcache): New type.
3901 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3902 New static constants.
3903 (diagnostic_file_cache_init, total_lines_num)
3904 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3905 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3906 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3907 (get_next_line, read_next_line, goto_next_line, read_line_num):
3908 New static function definitions.
3909 (diagnostic_file_cache_fini): New function.
3910 (location_get_source_line): Take an additional output line_len
3911 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3913 * diagnostic.c (diagnostic_finish): Call
3914 diagnostic_file_cache_fini.
3915 (adjust_line): Take an additional input parameter for the length
3916 of the line, rather than calculating it with strlen.
3917 (diagnostic_show_locus): Adjust the use of
3918 location_get_source_line and adjust_line with respect to their new
3919 signature. While displaying a line now, do not stop at the first
3920 null byte. Rather, display the zero byte as a space and keep
3921 going until we reach the size of the line.
3922 * Makefile.in: Add vec.o to OBJS-libcommon
3924 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3925 Ilya Tocar <ilya.tocar@intel.com>
3927 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3928 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3929 (__builtin_ia32_kmov16): Ditto.
3930 * config/i386/i386.md (UNSPEC_KMOV): New.
3933 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3935 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3936 (_mm512_storeu_si512): Ditto.
3938 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3941 * rtl.h (get_referenced_operands): Declare.
3942 * recog.c (get_referenced_operands): New function.
3943 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3944 operands have been referenced when recording LO_SUM references.
3946 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3948 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3950 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3952 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3953 Enable for generic and recent AMD targets.
3955 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3957 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3958 ARG_SIZE note when adjustment was eliminated.
3960 2014-01-22 Jeff Law <law@redhat.com>
3962 PR tree-optimization/59597
3963 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3964 in file. Accept new argument REGISTERING and use it to modify
3965 dump output appropriately.
3966 (register_jump_thread): Corresponding changes.
3967 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3968 thread paths involving joiner blocks. Add code to dump cancelled
3969 jump threading paths.
3971 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3973 PR rtl-optimization/59477
3974 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3975 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3977 2014-01-22 Tom Tromey <tromey@redhat.com>
3979 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3981 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3983 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3985 PR rtl-optimization/59896
3986 * lra-constraints.c (process_alt_operands): Check unused note for
3987 matched operands of insn with no output reloads.
3989 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3991 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3992 (mips_move_from_gpr_cost): Likewise.
3994 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3996 PR rtl-optimization/59858
3997 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3998 ira_class_hard_regs_num.
3999 (process_alt_operands): Increase reject for dying matched operand.
4001 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4004 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4005 smaller than size, perform several stores or loads and stores
4006 at dst + count - size to store or copy all of size bytes, rather
4007 than just last modesize bytes.
4009 2014-01-20 DJ Delorie <dj@redhat.com>
4011 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4012 that CLOBBERs are REGs before propogating their values.
4014 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4017 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4018 (cgraph_inline_failed_type): New function.
4019 * cgraph.h (DEFCIFCODE): Add type.
4020 (cgraph_inline_failed_type_t): New enum.
4021 (cgraph_inline_failed_type): New prototype.
4022 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4023 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4024 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4025 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4026 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4027 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4028 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4029 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4030 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4031 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4032 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4033 OPTIMIZATION_MISMATCH.
4034 * tree-inline.c (expand_call_inline): Emit errors during
4035 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4037 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4040 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4041 mode attribute in insn output.
4043 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4045 * output.h (output_constant): Delete.
4046 * varasm.c (output_constant): Make private.
4048 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4050 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4052 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4055 * tree.h (fold_builtin_strcat): New prototype.
4056 * builtins.c (fold_builtin_strcat): No longer static. Add len
4057 argument, if non-NULL, don't call c_strlen. Optimize
4058 directly into __builtin_memcpy instead of __builtin_strcpy.
4059 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4060 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4062 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4064 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4065 for SImode_address_operand operands, having only a REG argument.
4067 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4069 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4070 loader name using mbig-endian.
4071 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4073 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4075 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4079 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4081 * config/aarch64/aarch64-protos.h
4082 (aarch64_cannot_change_mode_class_ptr): Declare.
4083 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4084 aarch64_cannot_change_mode_class_ptr): New.
4085 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4086 backend hook aarch64_cannot_change_mode_class.
4088 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4090 * common/config/aarch64/aarch64-common.c
4091 (aarch64_handle_option): Don't handle any option order logic here.
4092 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4093 selected_cpu, warn on architecture version mismatch.
4094 (aarch64_override_options): Fix parsing order for option strings.
4096 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4097 Iain Sandoe <iain@codesourcery.com>
4100 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4101 warning. Amend comment to reflect current functionality.
4103 2014-01-20 Richard Biener <rguenther@suse.de>
4106 * builtins.c (fold_builtin_strcat): Remove case better handled
4107 by tree-ssa-strlen.c.
4109 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4111 * config/aarch64/aarch64.opt
4112 (mcpu, march, mtune): Make case-insensitive.
4114 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4117 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4118 if operands[1] is a REG or ZERO_EXTEND of a REG.
4120 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4122 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4124 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4126 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4127 long non-pic millicode calls.
4129 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4131 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4133 2014-01-19 Kito Cheng <kito@0xlab.org>
4135 * builtins.c (expand_movstr): Check movstr expand done or fail.
4137 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4138 H.J. Lu <hongjiu.lu@intel.com>
4141 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4142 to DImode for zero-extended addresses.
4144 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4146 PR rtl-optimization/57763
4147 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4148 on the new indirect jump_insn and increment LABEL_NUSES (label).
4150 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4154 * config.gcc (x86_archs): New variable.
4155 (x86_64_archs): Likewise.
4156 (x86_cpus): Likewise.
4157 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4158 --with-arch/--with-cpu= options.
4159 Support --with-arch=/--with-cpu={nehalem,westmere,
4160 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4162 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4164 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4165 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4167 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4169 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4171 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4174 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4175 clear cpp_get_options (parse_in)->warn_unused_macros for
4176 ix86_target_macros_internal with cpp_define.
4178 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4180 * jump.c (delete_related_insns): Keep (use (insn))s.
4181 * reorg.c (redundant_insn): Check for barriers too.
4183 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4185 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4187 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4189 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4190 call to $$dyncall when TARGET_LONG_CALLS is true.
4192 2014-01-17 Jeff Law <law@redhat.com>
4194 * ree.c (combine_set_extension): Temporarily disable test for
4195 changing number of hard registers.
4197 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4200 * ipa-inline-analysis.c (inline_free_summary):
4201 Do not free summary of aliases.
4203 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4206 * gimplify.c (gimplify_expr): Use create_tmp_var
4207 instead of create_tmp_var_raw. If cond doesn't have
4208 integral type, don't add the IFN_ANNOTATE builtin at all.
4210 2014-01-17 Martin Jambor <mjambor@suse.cz>
4213 * ipa-cp.c (prev_edge_clone): New variable.
4214 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4215 Also resize prev_edge_clone vector.
4216 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4217 (ipcp_edge_removal_hook): New function.
4218 (ipcp_driver): Register ipcp_edge_removal_hook.
4220 2014-01-17 Andrew Pinski <apinski@cavium.com>
4221 Steve Ellcey <sellcey@mips.com>
4224 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4227 2014-01-17 Jeff Law <law@redhat.com>
4230 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4231 so that pass_ccp runs first.
4233 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4235 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4236 (ix86_adjust_cost): Use !TARGET_XXX.
4237 (do_reorder_for_imul): Likewise.
4238 (swap_top_of_ready_list): Likewise.
4239 (ix86_sched_reorder): Likewise.
4241 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4243 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4244 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4245 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4246 (intel_memset): New. Duplicate slm_memset.
4247 (intel_cost): New. Duplicate slm_cost.
4248 (m_INTEL): New macro.
4249 (processor_target_table): Add "intel".
4250 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4251 with PROCESSOR_INTEL for "intel".
4252 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4253 PROCESSOR_SILVERMONT.
4254 (ix86_issue_rate): Likewise.
4255 (ix86_adjust_cost): Likewise.
4256 (ia32_multipass_dfa_lookahead): Likewise.
4257 (swap_top_of_ready_list): Likewise.
4258 (ix86_sched_reorder): Likewise.
4259 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4260 instead of TARGET_OPT_AGU.
4261 * config/i386/i386.h (TARGET_INTEL): New.
4262 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4263 (processor_type): Add PROCESSOR_INTEL.
4264 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4265 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4267 2014-01-17 Marek Polacek <polacek@redhat.com>
4270 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4273 2014-01-17 Richard Biener <rguenther@suse.de>
4275 PR tree-optimization/46590
4276 * opts.c (default_options_table): Add entries for
4277 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4278 all enabled at -O1 but not for -Og.
4279 * common.opt (fbranch-count-reg): Remove Init(1).
4280 (fmove-loop-invariants): Likewise.
4281 (ftree-pta): Likewise.
4283 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4285 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4286 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4287 decls to at least the GCC 4.8 used alignments.
4290 * tree-nested.c (convert_nonlocal_reference_stmt,
4291 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4292 of GIMPLE_BIND stmts, adjust associated decls.
4294 2014-01-17 Richard Biener <rguenther@suse.de>
4296 PR tree-optimization/46590
4297 * vec.h (vec<>::bseach): New member function implementing
4298 binary search according to C89 bsearch.
4299 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4300 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4301 bitmap pointer again. Make accesses_in_loop a flat array.
4302 (mem_ref_obstack): New global.
4303 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4304 (mark_ref_stored): Likewise.
4305 (ref_indep_loop_p_2): Likewise.
4306 (set_ref_stored_in_loop): New helper function.
4307 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4308 (memref_free): Adjust.
4309 (record_mem_ref_loc): Simplify.
4310 (gather_mem_refs_stmt): Adjust.
4311 (sort_locs_in_loop_postorder_cmp): New function.
4312 (analyze_memory_references): Sort accesses_in_loop after
4313 loop postorder number.
4314 (find_ref_loc_in_loop_cmp): New function.
4315 (for_all_locs_in_loop): Find relevant cluster of locs in
4316 accesses_in_loop and iterate without recursion.
4317 (execute_sm): Avoid uninit warning.
4318 (struct ref_always_accessed): Simplify.
4319 (ref_always_accessed::operator ()): Likewise.
4320 (ref_always_accessed_p): Likewise.
4321 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4322 loop postorder numbers here.
4323 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4326 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4329 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4330 on decls for which assemble_alias has been called.
4332 2014-01-17 Nick Clifton <nickc@redhat.com>
4334 * config/msp430/msp430.opt: (mcpu): New option.
4335 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4336 (msp430_option_override): Parse target_cpu. If the MCU name
4337 matches a generic string, clear target_mcu.
4338 (msp430_attr): Allow numeric interrupt values up to 63.
4339 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4340 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4342 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4344 * config/msp430/msp430.md (popm): Use %J rather than %I.
4345 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4346 (addhi_cy_i): Use immediate_operand for operand 2.
4347 * doc/invoke.texi: Document -mcpu option.
4349 2014-01-17 Richard Biener <rguenther@suse.de>
4351 PR rtl-optimization/38518
4352 * df.h (df_analyze_loop): Declare.
4353 * df-core.c: Include cfgloop.h.
4354 (df_analyze_1): Split out main part of df_analyze.
4355 (df_analyze): Adjust.
4356 (loop_inverted_post_order_compute): New function.
4357 (loop_post_order_compute): Likewise.
4358 (df_analyze_loop): New function avoiding whole-function
4360 * loop-invariant.c (find_defs): Use df_analyze_loop.
4361 (find_invariants): Adjust.
4362 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4364 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4366 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4367 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4369 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4371 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4372 traversal when removing references.
4374 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4377 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4379 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4382 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4383 pushing a reload for an autoinc when we had previously reloaded an
4384 inner part of the address.
4386 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4388 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4390 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4391 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4392 when not giving up or versioning for alias only because of
4394 (vect_analyze_data_ref_dependences): Set to true.
4395 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4397 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4398 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4402 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4405 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4406 operand 0 predicate for gathers, use a new pseudo as subtarget.
4408 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4411 * lra-constraints.c (process_alt_operands): Add printing debug info.
4412 Check absence of input/output reloads for matched operands too.
4414 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4416 PR rtl-optimization/59835
4417 * ira.c (ira_init_register_move_cost): Increase cost for
4420 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4422 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4424 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4427 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4428 non-register objects. Use gen_(high/low)part more consistently.
4431 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4434 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4435 endian support, remove tests for WORDS_BIG_ENDIAN.
4436 (p8_mfvsrd_3_<mode>): Likewise.
4437 (reload_gpr_from_vsx<mode>): Likewise.
4438 (reload_gpr_from_vsxsf): Likewise.
4439 (p8_mfvsrd_4_disf): Likewise.
4441 2014-01-16 Richard Biener <rguenther@suse.de>
4443 PR rtl-optimization/46590
4444 * lcm.c (compute_antinout_edge): Use postorder iteration.
4445 (compute_laterin): Use inverted postorder iteration.
4447 2014-01-16 Nick Clifton <nickc@redhat.com>
4450 * varasm.c (output_constant): Return the number of bytes actually
4452 (output_constructor_array_range): Update the field size with the
4453 number of bytes emitted by output_constant.
4454 (output_constructor_regular_field): Likewise. Also do not
4455 complain if the total number of bytes emitted is now greater
4456 than the expected fieldpos.
4457 * output.h (output_constant): Update prototype and descriptive comment.
4459 2014-01-16 Marek Polacek <polacek@redhat.com>
4462 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4463 it is error_mark_node.
4465 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4467 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4468 VALID_AVX256_REG_OR_OI_MODE.
4470 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4472 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4473 current procedure should be profiled.
4475 2014-01-15 Andrew Pinski <apinski@cavium.com>
4477 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4478 of moving from/to the STACK_REG register class.
4480 2014-01-15 Richard Henderson <rth@redhat.com>
4483 * reginfo.c (global_regs_decl): Globalize.
4484 * rtl.h (global_regs_decl): Declare.
4485 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4486 reserved via global_regs.
4488 2014-01-15 Teresa Johnson <tejohnson@google.com>
4490 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4492 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4494 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4495 and vmulosh rather than call gen_vec_widen_smult_*.
4496 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4497 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4498 (vec_widen_smult_even_v16qi): Likewise.
4499 (vec_widen_umult_even_v8hi): Likewise.
4500 (vec_widen_smult_even_v8hi): Likewise.
4501 (vec_widen_umult_odd_v16qi): Likewise.
4502 (vec_widen_smult_odd_v16qi): Likewise.
4503 (vec_widen_umult_odd_v8hi): Likewise.
4504 (vec_widen_smult_odd_v8hi): Likewise.
4505 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4506 vmuloub rather than call gen_vec_widen_umult_*.
4507 (vec_widen_umult_lo_v16qi): Likewise.
4508 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4509 vmulosb rather than call gen_vec_widen_smult_*.
4510 (vec_widen_smult_lo_v16qi): Likewise.
4511 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4512 rather than call gen_vec_widen_umult_*.
4513 (vec_widen_umult_lo_v8hi): Likewise.
4514 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4515 rather than call gen_vec_widen_smult_*.
4516 (vec_widen_smult_lo_v8hi): Likewise.
4518 2014-01-15 Jeff Law <law@redhat.com>
4520 PR tree-optimization/59747
4521 * ree.c (find_and_remove_re): Properly handle case where a second
4522 eliminated extension requires widening a copy created for elimination
4523 of a prior extension.
4524 (combine_set_extension): Ensure that the number of hard regs needed
4525 for a destination register does not change when we widen it.
4527 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4529 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4530 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4531 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4532 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4533 (avr-*-rtems*): Likewise.
4534 (bfin*-rtems*): Likewise.
4535 (moxie-*-rtems*): Likewise.
4536 (h8300-*-rtems*): Likewise.
4537 (i[34567]86-*-rtems*): Likewise.
4538 (lm32-*-rtems*): Likewise.
4539 (m32r-*-rtems*): Likewise.
4540 (m68k-*-rtems*): Likewise.
4541 (microblaze*-*-rtems*): Likewise.
4542 (mips*-*-rtems*): Likewise.
4543 (powerpc-*-rtems*): Likewise.
4544 (sh-*-rtems*): Likewise.
4545 (sparc-*-rtems*): Likewise.
4546 (sparc64-*-rtems*): Likewise.
4547 (v850-*-rtems*): Likewise.
4548 (m32c-*-rtems*): Likewise.
4550 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4552 PR rtl-optimization/59511
4553 * ira.c (ira_init_register_move_cost): Use memory costs for some
4554 cases of register move cost calculations.
4555 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4556 instead of BB frequency.
4557 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4558 * lra-assigns.c (find_hard_regno_for): Ditto.
4560 2014-01-15 Richard Biener <rguenther@suse.de>
4562 PR tree-optimization/59822
4563 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4564 (vectorizable_load): Use it to hoist defs of uses of invariant
4565 loads out of the loop.
4567 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4568 Kugan Vivekanandarajah <kuganv@linaro.org>
4571 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4574 2014-01-15 Richard Biener <rguenther@suse.de>
4576 PR rtl-optimization/59802
4577 * lcm.c (compute_available): Use inverted postorder to seed
4578 the initial worklist.
4580 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4583 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4584 ADDR_REGS for invalid symrefs in non-PIC code.
4586 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4589 * builtins.c (determine_block_size): Initialize *probable_max_size
4590 even if len_rtx is CONST_INT.
4592 2014-01-14 Andrew Pinski <apinski@cavium.com>
4594 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4595 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4596 (cortexa53_tunings): Likewise.
4597 (aarch64_sched_issue_rate): New function.
4598 (TARGET_SCHED_ISSUE_RATE): Define.
4600 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4602 * ira-costs.c (find_costs_and_classes): Add missed
4603 ira_init_register_move_cost_if_necessary.
4605 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4608 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4610 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4613 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4614 to indicate if type is used for function return value. Warn ABI
4615 change if the vector mode isn't available for function return value.
4616 (ix86_function_arg_advance): Pass false to type_natural_mode.
4617 (ix86_function_arg): Likewise.
4618 (ix86_gimplify_va_arg): Likewise.
4619 (function_arg_32): Don't warn ABI change.
4620 (ix86_function_value): Pass true to type_natural_mode.
4621 (ix86_return_in_memory): Likewise.
4622 (ix86_struct_value_rtx): Removed.
4623 (TARGET_STRUCT_VALUE_RTX): Likewise.
4625 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4627 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4628 converting a conditional jump into a conditional return.
4630 2014-01-14 Richard Biener <rguenther@suse.de>
4632 PR tree-optimization/58921
4633 PR tree-optimization/59006
4634 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4635 hoisting invariant stmts.
4636 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4637 invariant loads on the preheader edge if possible.
4639 2014-01-14 Joey Ye <joey.ye@arm.com>
4641 * doc/plugin.texi (Building GCC plugins): Update to C++.
4643 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4645 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4646 (_mm_rcp28_round_ss): Ditto.
4647 (_mm_rsqrt28_round_sd): Ditto.
4648 (_mm_rsqrt28_round_ss): Ditto.
4649 (_mm_rcp28_sd): Ditto.
4650 (_mm_rcp28_ss): Ditto.
4651 (_mm_rsqrt28_sd): Ditto.
4652 (_mm_rsqrt28_ss): Ditto.
4653 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4654 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4655 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4656 (IX86_BUILTIN_RCP28SD): Ditto.
4657 (IX86_BUILTIN_RCP28SS): Ditto.
4658 (IX86_BUILTIN_RSQRT28SD): Ditto.
4659 (IX86_BUILTIN_RSQRT28SS): Ditto.
4660 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4661 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4662 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4663 (ix86_expand_special_args_builtin): Expand new FTYPE.
4664 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4665 (srcp14<mode>): Make insn unary.
4666 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4667 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4668 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4669 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4670 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4671 Fix rounding: make it SAE only.
4672 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4674 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4676 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4677 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4678 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4679 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4680 (round_saeonly_mask_scalar_operand4): Ditto.
4681 (round_saeonly_mask_scalar_op3): Ditto.
4682 (round_saeonly_mask_scalar_op4): Ditto.
4684 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4686 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4687 Implement -maltivec=be for vec_insert and vec_extract.
4689 2014-01-10 DJ Delorie <dj@redhat.com>
4691 * config/msp430/msp430.md (call_internal): Don't allow memory
4692 references with SP as the base register.
4693 (call_value_internal): Likewise.
4694 * config/msp430/constraints.md (Yc): New. For memory references
4695 that don't use SP as a base register.
4697 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4698 "an integer without a # prefix"
4699 * config/msp430/msp430.md (epilogue_helper): Use it.
4701 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4704 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4705 AVX512F gather builtins.
4706 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4707 on gather decls with INTEGER_TYPE masktype.
4708 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4709 directly into the builtin rather than hoisting it before loop.
4711 PR tree-optimization/59387
4712 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4713 (scev_const_prop): If folded_casts and type has undefined overflow,
4714 use force_gimple_operand instead of force_gimple_operand_gsi and
4715 for each added stmt if it is assign with
4716 arith_code_with_undefined_signed_overflow, call
4717 rewrite_to_defined_overflow.
4718 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4719 gimple-fold.h instead.
4720 (arith_code_with_undefined_signed_overflow,
4721 rewrite_to_defined_overflow): Moved to ...
4722 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4723 rewrite_to_defined_overflow): ... here. No longer static.
4724 Include gimplify-me.h.
4725 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4726 rewrite_to_defined_overflow): New prototypes.
4728 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4730 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4732 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4734 * builtins.c (get_object_alignment_2): Minor tweak.
4735 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4737 2014-01-13 Christian Bruel <christian.bruel@st.com>
4739 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4740 optimized non constant lengths.
4742 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4745 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4746 load as __atomic_load_N if possible.
4748 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4750 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4752 (rs6000_expand_builtin): Adjust call.
4754 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4757 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4758 * config/rs6000/rs6000.c: Include target-globals.h.
4759 (rs6000_set_current_function): Instead of doing target_reinit
4760 unconditionally, use save_target_globals_default_opts and
4761 restore_target_globals.
4763 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4765 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4766 (rs6000_expand_builtin): Handle mffs and mtfsf.
4767 (rs6000_init_builtins): Define mffs and mtfsf.
4768 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4769 (rs6000_mffs): New pattern.
4770 (rs6000_mtfsf): New pattern.
4772 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4774 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4775 Start narrowing with START. Apply candidate-use pair
4776 and check overall cost in narrowing.
4777 (iv_ca_prune): Pass new argument.
4779 2014-01-10 Jeff Law <law@redhat.com>
4782 * ree.c (combine_reaching_defs): Ensure the defining statement
4783 occurs before the extension when optimizing extensions with
4784 different source and destination hard registers.
4786 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4789 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4790 vtables into the type inheritance graph.
4792 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4794 PR rtl-optimization/59754
4795 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4796 modes in the REGNO != REGNO case.
4798 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4800 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4802 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4804 PR tree-optimization/59745
4805 * tree-predcom.c (tree_predictive_commoning_loop): Call
4806 free_affine_expand_cache if giving up because components is NULL.
4808 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4809 GC in payload of target_globals struct instead of allocating them on
4810 the heap and the larger structs separately using GC.
4811 * target-globals.h (struct target_globals): Make regs, hard_regs,
4812 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4813 of GTY((skip)) and change type to void *.
4814 (reset_target_globals): Cast loads from those fields to corresponding
4817 2014-01-10 Steve Ellcey <sellcey@mips.com>
4820 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4821 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4822 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4824 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4827 * aarch64-modes.def (CC_Zmode): New flags mode.
4828 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4829 represents an equality.
4830 (aarch64_get_condition_code): Handle CC_Zmode.
4831 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4833 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4835 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4836 extraction in good case.
4838 2014-01-10 Richard Biener <rguenther@suse.de>
4840 PR tree-optimization/59374
4841 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4842 checking after SLP discovery. Mark stmts not participating
4843 in any SLP instance properly.
4845 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4847 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4848 when handling a SET rtx.
4850 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4852 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4853 (cortex-a57): Likewise.
4854 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4856 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4858 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4859 non-iwmmxt builtins.
4861 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4865 * ipa-devirt.c record_target_from_binfo): Take as argument
4866 stack of binfos and lookup matching one for virtual inheritance.
4867 (possible_polymorphic_call_targets_1): Update.
4869 2014-01-10 Huacai Chen <chenhc@lemote.com>
4871 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4872 kernel strings for Loongson-2E/2F/3A.
4874 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4877 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4878 is_gimple_call before calling gimple_call_internal_p.
4880 2014-01-09 Steve Ellcey <sellcey@mips.com>
4882 * Makefile.in (TREE_FLOW_H): Remove.
4883 (TREE_SSA_H): Add file names from tree-flow.h.
4884 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4885 * tree.h: Remove tree-flow.h reference.
4886 * hash-table.h: Remove tree-flow.h reference.
4887 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4888 reference with tree-ssa-loop.h.
4890 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4892 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4893 default element-order behavior for -maltivec.
4894 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4896 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4897 when targeting big endian, at least for now.
4898 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4900 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4903 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4904 var satisfies use_register_for_decl, just take into account type
4905 alignment, rather than decl alignment.
4907 PR tree-optimization/59622
4908 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4909 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4910 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4911 Don't devirtualize for inplace at all. For targets.length () == 1,
4912 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4914 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4916 * config/i386/i386.md (cpu): Remove the unused btver1.
4918 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4920 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4922 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4925 * tree-core.h (struct target_globals): New forward declaration.
4926 (struct tree_target_option): Add globals field.
4927 * tree.h (TREE_TARGET_GLOBALS): Define.
4928 (prepare_target_option_nodes_for_pch): New prototype.
4929 * target-globals.h (struct target_globals): Define even if
4931 * tree.c (prepare_target_option_node_for_pch,
4932 prepare_target_option_nodes_for_pch): New functions.
4933 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4934 * config/i386/i386.c: Include target-globals.h.
4935 (ix86_set_current_function): Instead of doing target_reinit
4936 unconditionally, use save_target_globals_default_opts and
4937 restore_target_globals.
4939 2014-01-09 Richard Biener <rguenther@suse.de>
4941 PR tree-optimization/59715
4942 * tree-cfg.h (split_critical_edges): Declare.
4943 * tree-cfg.c (split_critical_edges): Export.
4944 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4946 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4948 * cfgexpand.c (expand_stack_vars): Optionally disable
4949 asan stack protection.
4950 (expand_used_vars): Likewise.
4951 (partition_stack_vars): Likewise.
4952 * asan.c (asan_emit_stack_protection): Optionally disable
4953 after return stack usage.
4954 (instrument_derefs): Optionally disable memory access instrumentation.
4955 (instrument_builtin_call): Likewise.
4956 (instrument_strlen_call): Likewise.
4957 (asan_protect_global): Optionally disable global variables protection.
4958 * doc/invoke.texi: Added doc for new options.
4959 * params.def: Added new options.
4960 * params.h: Likewise.
4962 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4964 PR rtl-optimization/59724
4965 * ifcvt.c (cond_exec_process_if_block): Don't call
4966 flow_find_head_matching_sequence with 0 longest_match.
4967 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4968 non-active insns if !stop_after.
4969 (try_head_merge_bb): Revert 2014-01-07 changes.
4971 2014-01-08 Jeff Law <law@redhat.com>
4973 * ree.c (get_sub_rtx): New function, extracted from...
4974 (merge_def_and_ext): Here.
4975 (combine_reaching_defs): Use get_sub_rtx.
4977 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4979 * cgraph.h (varpool_variable_node): Do not choke on null node.
4981 2014-01-08 Catherine Moore <clm@codesourcery.com>
4983 * config/mips/mips.md (simple_return): Attempt to use JRC
4985 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4987 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4989 PR rtl-optimization/59137
4990 * reorg.c (steal_delay_list_from_target): Call update_block for
4992 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4994 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4997 two duplicate entries.
4999 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5002 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5004 * config/mips/mips.c (mips_truncated_op_cost): New function.
5005 (mips_rtx_costs): Adjust test for BADDU.
5006 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5008 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5010 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5011 (*baddu_si): ...this new pattern.
5013 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5016 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5018 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5021 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5023 (expand_expr, expand_normal): Adjust.
5024 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5025 inner_reference_p. Use inner_reference_p to expand inner references.
5026 (store_expr): Adjust.
5027 * cfgexpand.c (expand_call_stmt): Adjust.
5029 2014-01-08 Rong Xu <xur@google.com>
5031 * gcov-io.c (gcov_var): Move from gcov-io.h.
5032 (gcov_position): Ditto.
5033 (gcov_is_error): Ditto.
5034 (gcov_rewrite): Ditto.
5035 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5036 only part to libgcc/libgcov.h.
5038 2014-01-08 Marek Polacek <polacek@redhat.com>
5041 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5043 2014-01-08 Marek Polacek <polacek@redhat.com>
5046 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5048 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5050 PR rtl-optimization/59649
5051 * stor-layout.c (get_mode_bounds): For BImode return
5052 0 and STORE_FLAG_VALUE.
5054 2014-01-08 Richard Biener <rguenther@suse.de>
5057 * gimple.h (is_gimple_builtin_call): Remove.
5058 (gimple_builtin_call_types_compatible_p): New.
5059 (gimple_call_builtin_p): New overload.
5060 * gimple.c (is_gimple_builtin_call): Remove.
5061 (validate_call): Rename to ...
5062 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5064 (validate_type): New static function.
5065 (gimple_call_builtin_p): New overload and adjust.
5066 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5067 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5068 (gimple_fold_stmt_to_constant_1): Likewise.
5069 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5071 2014-01-08 Richard Biener <rguenther@suse.de>
5074 * gimplify.c (gimplify_expr): Gimplify register-register type
5075 VIEW_CONVERT_EXPRs to separate stmts.
5077 2014-01-07 Jeff Law <law@redhat.com>
5080 * ree.c (combine_set_extension): Handle case where source
5081 and destination registers in an extension insn are different.
5082 (combine_reaching_defs): Allow source and destination registers
5083 in extension to be different under limited circumstances.
5084 (add_removable_extension): Remove restriction that the
5085 source and destination registers in the extension are the same.
5086 (find_and_remove_re): Emit a copy from the extension's
5087 destination to its source after the defining insn if
5088 the source and destination registers are different.
5091 * ifcvt.c (merge_if_block): If we are merging a block with more than
5092 one successor with a block with no successors, remove any BARRIER
5093 after the second block.
5095 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5097 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5099 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5102 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5103 for 14-bit register offsets when INT14_OK_STRICT is false.
5105 2014-01-07 Roland Stigge <stigge@antcom.de>
5106 Michael Meissner <meissner@linux.vnet.ibm.com>
5109 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5110 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5112 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5114 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5117 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5119 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5120 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5121 rtx is const0_rtx or not.
5123 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5126 * target-globals.c (save_target_globals): Remove this_fn_optab
5128 * toplev.c: Include optabs.h.
5129 (target_reinit): Temporarily restore the global options if another
5130 set of options are in force.
5132 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5134 PR rtl-optimization/58668
5135 * cfgcleanup.c (flow_find_cross_jump): Don't count
5136 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5137 to determine what is counted.
5138 (flow_find_head_matching_sequence): Use active_insn_p to determine
5140 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5142 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5143 determine what is counted.
5145 PR tree-optimization/59643
5146 * tree-predcom.c (split_data_refs_to_components): If one dr is
5147 read and one write, determine_offset fails and the write isn't
5148 in the bad component, just put the read into the bad component.
5150 2014-01-07 Mike Stump <mikestump@comcast.net>
5151 Jakub Jelinek <jakub@redhat.com>
5154 * tree-core.h (struct tree_optimization_option): Change optabs
5155 type from unsigned char * to void *.
5156 * optabs.c (init_tree_optimization_optabs): Adjust
5157 TREE_OPTIMIZATION_OPTABS initialization.
5159 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5162 * config/i386/i386.h (struct machine_function): Add
5163 no_drap_save_restore field.
5164 * config/i386/i386.c (ix86_save_reg): Use
5165 !cfun->machine->no_drap_save_restore instead of
5166 crtl->stack_realign_needed.
5167 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5168 this function clears frame_pointer_needed. Set
5169 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5170 and DRAP reg is needed.
5172 2014-01-06 Marek Polacek <polacek@redhat.com>
5175 * doc/implement-c.texi: Mention that other integer types are
5176 permitted as bit-field types in strictly conforming mode.
5178 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5180 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5183 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5185 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5187 2014-01-06 Martin Jambor <mjambor@suse.cz>
5190 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5192 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5194 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5198 * var-tracking.c (add_stores): Preserve the value of the source even if
5199 we don't record the store.
5201 2014-01-06 Terry Guo <terry.guo@arm.com>
5203 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5205 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5208 * config/darwin.c (darwin_function_section): Adjust return values to
5209 correspond to optimisation changes made in r206070.
5211 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5213 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5214 from prefetch_block tune setting.
5215 (nocona_cost): Correct size of prefetch block to 64.
5217 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5219 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5220 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5221 used to save the static chain register in the computation of the offset
5222 from which the FP registers need to be restored.
5224 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5226 PR tree-optimization/59519
5227 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5228 ICE if get_current_def (current_new_name) is already non-NULL, as long
5229 as it is a phi result of some other phi in *new_exit_bb that has
5232 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5233 or vmovdqu* for misaligned_operand.
5234 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5235 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5236 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5237 aligned_mem for AVX512F masked aligned load and store builtins and for
5240 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5242 PR tree-optimization/59651
5243 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5244 Address range for negative step should be added by TYPE_SIZE_UNIT.
5246 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5248 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5249 overlapping registers also for registers other than the stack pointer.
5251 2014-01-03 Marek Polacek <polacek@redhat.com>
5254 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5257 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5260 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5263 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5264 (push splitter): Use <P:MODE_SIZE> instead of
5265 GET_MODE_SIZE (<P:MODE>mode).
5266 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5267 (mov -1, reg peephole2): Likewise.
5268 * config/i386/sse.md (*mov<mode>_internal,
5269 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5270 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5271 *<code><mode>3, *andnot<mode>3<mask_name>,
5272 <mask_codefor><code><mode>3<mask_name>): Likewise.
5273 * config/i386/subst.md (mask_mode512bit_condition,
5274 sd_mask_mode512bit_condition): Likewise.
5276 2014-01-02 Xinliang David Li <davidxl@google.com>
5278 PR tree-optimization/59303
5279 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5280 (dump_predicates): Better output format.
5281 (pred_equal_p): New function.
5282 (is_neq_relop_p): Ditto.
5283 (is_neq_zero_form_p): Ditto.
5284 (pred_expr_equal_p): Ditto.
5285 (pred_neg_p): Ditto.
5286 (simplify_pred): Ditto.
5287 (simplify_preds_2): Ditto.
5288 (simplify_preds_3): Ditto.
5289 (simplify_preds_4): Ditto.
5290 (simplify_preds): Ditto.
5292 (push_to_worklist): Ditto.
5293 (get_pred_info_from_cmp): Ditto.
5294 (is_degenerated_phi): Ditto.
5295 (normalize_one_pred_1): Ditto.
5296 (normalize_one_pred): Ditto.
5297 (normalize_one_pred_chain): Ditto.
5298 (normalize_preds): Ditto.
5299 (normalize_cond_1): Remove function.
5300 (normalize_cond): Ditto.
5301 (is_gcond_subset_of): Ditto.
5302 (is_subset_of_any): Ditto.
5303 (is_or_set_subset_of): Ditto.
5304 (is_and_set_subset_of): Ditto.
5305 (is_norm_cond_subset_of): Ditto.
5306 (pred_chain_length_cmp): Ditto.
5307 (convert_control_dep_chain_into_preds): Type change.
5308 (find_predicates): Ditto.
5309 (find_def_preds): Ditto.
5310 (destroy_predicates_vecs): Ditto.
5311 (find_matching_predicates_in_rest_chains): Ditto.
5312 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5313 (is_pred_expr_subset): Ditto.
5314 (is_pred_chain_subset_of): Ditto.
5315 (is_included_in): Ditto.
5316 (is_superset_of): Ditto.
5318 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5320 Update copyright years.
5322 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5324 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5325 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5326 config/arc/arc.md, config/arc/arc.opt,
5327 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5328 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5329 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5330 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5331 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5332 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5333 vtable-verify.c, vtable-verify.h: Use the standard form for the
5336 2014-01-02 Tobias Burnus <burnus@net-b.de>
5338 * gcc.c (process_command): Update copyright notice dates.
5339 * gcov-dump.c: Ditto.
5341 * doc/cpp.texi: Bump @copying's copyright year.
5342 * doc/cppinternals.texi: Ditto.
5343 * doc/gcc.texi: Ditto.
5344 * doc/gccint.texi: Ditto.
5345 * doc/gcov.texi: Ditto.
5346 * doc/install.texi: Ditto.
5347 * doc/invoke.texi: Ditto.
5349 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5351 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5353 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5355 * config/i386/sse.md (*mov<mode>_internal): Guard
5356 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5358 PR rtl-optimization/59647
5359 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5360 new_rtx into UNSIGNED_FLOAT rtxes.
5362 Copyright (C) 2014 Free Software Foundation, Inc.
5364 Copying and distribution of this file, with or without modification,
5365 are permitted in any medium without royalty provided the copyright
5366 notice and this notice are preserved.