1 2014-03-25 Richard Biener <rguenther@suse.de>
3 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like
7 2014-03-25 Jakub Jelinek <jakub@redhat.com>
9 * combine.c (simplify_compare_const): Add MODE argument.
10 Handle mode_width 0 as very large mode_width.
11 (try_combine, simplify_comparison): Adjust callers.
13 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
14 type to avoid signed integer overflow.
15 * explow.c (plus_constant): Likewise.
17 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
19 * doc/generic.texi: Correct typos.
21 2014-03-24 Tobias Burnus <burnus@net-b.de>
23 * doc/invoke.texi (-flto): Expand section about
24 using static libraries with LTO.
26 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
28 PR rtl-optimization/60501
29 * optabs.def (addptr3_optab): New optab.
30 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
31 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
32 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
34 * lra.c (emit_add3_insn): Use the addptr pattern if available.
36 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
38 2014-03-24 Ulrich Drepper <drepper@gmail.com>
40 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
43 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
44 (_mm256_undefined_ps): Define.
45 (_mm256_undefined_pd): Define.
46 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
47 (_mm_undefined_pd): Define.
48 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
49 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
50 (_mm512_undefined_ps): Define.
51 (_mm512_undefined_pd): Define.
53 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
55 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
57 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
58 (lshr_simd): DI mode added.
59 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
60 (aarch64_ushr_simddi): Likewise.
61 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
62 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
63 (vshrd_n_u64): Likewise.
65 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
67 * Makefile.in (s-macro_list): Depend on cc1.
69 2014-03-23 Teresa Johnson <tejohnson@google.com>
71 * ipa-utils.c (ipa_print_order): Use specified dump file.
73 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
75 PR rtl-optimization/60601
76 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
78 * gcc.c (eval_spec_function): Initialize save_growing_value.
80 2014-03-22 Jakub Jelinek <jakub@redhat.com>
83 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
84 code == MINUS_EXPR, never swap op0 with op1.
86 * toplev.c (init_local_tick): Avoid signed integer multiplication
88 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
89 shift by first operand's bitsize.
91 2014-03-21 Jakub Jelinek <jakub@redhat.com>
94 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
96 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
97 TARGET_ISA_64BIT_P(x).
99 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
101 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
102 pattern for vector nor instead of subtract from splat(-1).
103 (altivec_expand_vec_perm_const_le): Likewise.
105 2014-03-21 Richard Henderson <rth@twiddle.net>
108 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
109 related insns after epilogue_completed.
111 2014-03-21 Martin Jambor <mjambor@suse.cz>
114 * cgraph.h (symtab_node): New flag body_removed.
115 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
116 when removing bodies.
117 * symtab.c (dump_symtab_base): Dump body_removed flag.
118 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
119 had their bodies removed.
121 2014-03-21 Martin Jambor <mjambor@suse.cz>
124 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
127 2014-03-21 Richard Biener <rguenther@suse.de>
129 PR tree-optimization/60577
130 * tree-core.h (struct tree_base): Document nothrow_flag use
131 in VAR_DECL_NONALIASED.
132 * tree.h (VAR_DECL_NONALIASED): New.
133 (may_be_aliased): Adjust.
134 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
136 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
138 * expr.c (expand_expr_real_1): Remove outdated comment.
140 2014-03-20 Jakub Jelinek <jakub@redhat.com>
143 * ira.c (adjust_cleared_regs): Call copy_rtx on
144 *reg_equiv[REGNO (loc)].src_p before passing it to
145 simplify_replace_fn_rtx.
148 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
149 into CONST, put pic register as first operand of PLUS. Use
150 gen_const_mem for both 32-bit and 64-bit PIC got loads.
152 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
154 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
156 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
158 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
159 around for store forwarding issue in the FPU on the UT699.
160 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
161 loads and operations if -mfix-ut699 is specified.
162 (divtf3_hq): Tweak attribute.
163 (sqrttf2_hq): Likewise.
165 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
167 * calls.c (store_one_arg): Remove incorrect const qualification on the
168 type of the temporary.
169 * cfgexpand.c (expand_return): Likewise.
170 * expr.c (expand_constructor): Likewise.
171 (expand_expr_real_1): Likewise.
173 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
175 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
178 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
181 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
183 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
185 * config/arm/aarch-common-protos.h
186 (alu_cost_table): Fix spelling of "extend".
187 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
189 2014-03-19 Richard Biener <rguenther@suse.de>
192 * tree-core.h (tree_type_common): Re-order pointer members
193 to reduce recursion depth during GC walks.
195 2014-03-19 Marek Polacek <polacek@redhat.com>
198 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
201 2014-03-19 Richard Biener <rguenther@suse.de>
204 * lto-streamer-in.c (input_function): In WPA stage do not drop
207 2014-03-19 Jakub Jelinek <jakub@redhat.com>
209 PR tree-optimization/60559
210 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
211 with build_zero_cst assignment.
213 2014-03-18 Kai Tietz <ktietz@redhat.com>
215 PR rtl-optimization/56356
216 * sdbout.c (sdbout_parms): Verify that parms'
217 incoming argument is valid.
218 (sdbout_reg_parms): Likewise.
220 2014-03-18 Richard Henderson <rth@redhat.com>
223 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
224 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
225 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
227 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
229 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
230 * doc/plugins.texi (Plugin callbacks): Mention
232 Italicize plugin event names in description. Explain that
233 PLUGIN_PRAGMAS has no sense for lto1. Explain
235 Remind that no GCC functions should be called after
237 Explain what pragmas with expansion are.
239 2014-03-18 Martin Liska <mliska@suse.cz>
241 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
242 gimple call statement is update.
243 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
244 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
246 2014-03-18 Jakub Jelinek <jakub@redhat.com>
249 * ubsan.c (ubsan_instrument_unreachable): Call
250 initialize_sanitizer_builtins.
251 (ubsan_pass): Likewise.
254 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
255 varpool_finalize_decl instead of rest_of_decl_compilation.
257 2014-03-18 Richard Biener <rguenther@suse.de>
259 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
260 by using bitmap_and_compl instead of bitmap_and_compl_into.
261 (df_rd_transfer_function): Likewise.
263 2014-03-18 Richard Biener <rguenther@suse.de>
265 * doc/lto.texi (fresolution): Fix typo.
267 2014-03-18 Richard Biener <rguenther@suse.de>
269 * doc/invoke.texi (flto): Update for changes in 4.9.
271 2014-03-18 Richard Biener <rguenther@suse.de>
273 * doc/loop.texi: Remove section on the removed lambda framework.
274 Update loop docs with recent changes in preserving loop structure.
276 2014-03-18 Richard Biener <rguenther@suse.de>
278 * doc/lto.texi (-fresolution): Document.
280 2014-03-18 Richard Biener <rguenther@suse.de>
282 * doc/contrib.texi: Adjust my name.
284 2014-03-18 Jakub Jelinek <jakub@redhat.com>
287 * internal-fn.c: Include diagnostic-core.h.
288 (expand_BUILTIN_EXPECT): New function.
289 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
290 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
291 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
292 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
294 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
295 Revert 3 argument __builtin_expect code.
296 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
297 * gimple-fold.c (gimple_fold_call): Likewise.
298 * tree.h (fold_builtin_expect): New prototype.
299 * builtins.c (build_builtin_expect_predicate): Add predictor
300 argument, if non-NULL, create 3 argument __builtin_expect.
301 (fold_builtin_expect): No longer static. Add ARG2 argument,
302 pass it through to build_builtin_expect_predicate.
303 (fold_builtin_2): Adjust caller.
304 (fold_builtin_3): Handle BUILT_IN_EXPECT.
305 * internal-fn.def (BUILTIN_EXPECT): New.
307 2014-03-18 Tobias Burnus <burnus@net-b.de>
310 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
311 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
312 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
314 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
317 * predict.c (combine_predictions_for_bb): Fix up formatting.
318 (expr_expected_value_1, expr_expected_value): Add predictor argument,
319 fill what it points to if non-NULL.
320 (tree_predict_by_opcode): Adjust caller, use the predictor.
321 * predict.def (PRED_COMPARE_AND_SWAP): Add.
323 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
325 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
326 proper constant for the store mode.
328 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
330 * symtab.c (change_decl_assembler_name): Fix transparent alias
333 2014-03-16 Renlin Li <Renlin.Li@arm.com>
335 * config/aarch64/aarch64.c: Correct the comments about the
336 aarch64 stack layout.
338 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
340 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
341 check for GF_OMP_FOR_KIND_FOR.
343 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
345 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
346 ymm and zmm register names.
348 2014-03-17 Jakub Jelinek <jakub@redhat.com>
351 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
352 note creation for the 2010-08-31 changes.
354 2014-03-17 Marek Polacek <polacek@redhat.com>
357 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
358 as -fno-tree-loop-vectorize.
359 (expand_omp_simd): Likewise.
361 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
363 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
364 (eligible_for_call_delay): New prototype.
365 * config/sparc/sparc.c (tls_call_delay): Rename into...
366 (eligible_for_call_delay): ...this. Return false if the instruction
367 cannot be put in the delay slot of a branch.
368 (eligible_for_restore_insn): Simplify.
369 (eligible_for_return_delay): Return false if the instruction cannot be
370 put in the delay slot of a branch and simplify.
371 (eligible_for_sibcall_delay): Return false if the instruction cannot be
372 put in the delay slot of a branch.
373 * config/sparc/sparc.md (fix_ut699): New attribute.
374 (tls_call_delay): Delete.
375 (in_call_delay): Reimplement.
376 (eligible_for_sibcall_delay): Rename into...
377 (in_sibcall_delay): ...this.
378 (eligible_for_return_delay): Rename into...
379 (in_return_delay): ...this.
380 (in_branch_delay): Reimplement.
381 (in_uncond_branch_delay): Delete.
382 (in_annul_branch_delay): Delete.
384 2014-03-14 Richard Henderson <rth@redhat.com>
387 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
388 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
389 (*floathi<X87MODEF>2_i387_with_temp): Remove.
390 (floathi splitters): Remove.
391 (float<SWI48x>xf2): New pattern.
392 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
393 code that tried to handle DImode for 32-bit, but which was excluded
394 by the pattern's condition. Drop allocation of stack temporary.
395 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
396 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
397 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
398 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
399 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
400 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
401 (*float<SWI48><MODEF>2_sse_interunit): Remove.
402 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
403 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
404 (*float<SWI48x><X87MODEF>2_i387): Remove.
405 (all float _with_temp splitters): Remove.
406 (*float<SWI48x><MODEF>2_i387): New pattern.
407 (*float<SWI48><MODEF>2_sse): New pattern.
408 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
409 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
411 2014-03-14 Jakub Jelinek <jakub@redhat.com>
412 Marek Polacek <polacek@redhat.com>
415 * common.opt (dump_base_name_prefixed): New Variable.
416 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
417 if x_dump_base_name_prefixed is already set, set it at the end.
419 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
421 PR rtl-optimization/60508
422 * lra-constraints.c (get_reload_reg): Add new parameter
424 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
425 Pass the new parameter values.
427 2014-03-14 Richard Biener <rguenther@suse.de>
429 * common.opt: Revert unintented changes from r205065.
432 2014-03-14 Richard Biener <rguenther@suse.de>
435 * cfghooks.c (split_block): Properly adjust all loops the
436 block was a latch of.
438 2014-03-14 Martin Jambor <mjambor@suse.cz>
441 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
444 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
447 * config/avr/avr.c (avr_set_current_function): Pass function name
448 through default_strip_name_encoding before sanity checking instead
449 of skipping the first char of the assembler name.
451 2014-03-13 Richard Henderson <rth@redhat.com>
454 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
455 (ix86_force_to_memory, ix86_free_from_memory): Remove.
456 * config/i386/i386-protos.h: Likewise.
457 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
458 in the expander instead of a splitter.
459 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
460 any possibility of requiring a memory.
461 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
462 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
463 (fp branch splitters): Update for ix86_split_fp_branch.
464 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
465 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
466 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
467 (*fop_<MODEF>_2_i387): Remove f/r alternative.
468 (*fop_<MODEF>_3_i387): Likewise.
469 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
470 (splitters for the fop_* register patterns): Remove.
471 (fscalexf4_i387): Rename from *fscalexf4_i387.
472 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
474 2014-03-13 Jakub Jelinek <jakub@redhat.com>
476 PR tree-optimization/59779
477 * tree-dfa.c (get_ref_base_and_extent): Use double_int
478 type for bitsize and maxsize instead of HOST_WIDE_INT.
480 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
482 PR rtl-optimization/57320
483 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
484 the CFG after thread_prologue_and_epilogue_insns.
486 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
488 PR rtl-optimization/57189
489 * lra-constraints.c (process_alt_operands): Disfavor spilling
492 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
494 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
497 2014-03-13 Jakub Jelinek <jakub@redhat.com>
499 PR tree-optimization/59025
501 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
502 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
504 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
507 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
508 calls of avr_out_plus_1.
510 2014-03-13 Bin Cheng <bin.cheng@arm.com>
512 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
513 BB's single pred and update the father loop's latch info later.
515 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
517 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
521 (VEC_base): Likewise.
522 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
523 registers, we need to swap double words in little endian mode.
525 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
526 to be a container mode for 128-bit integer operations added in ISA
527 2.07. Unlike TImode and PTImode, the preferred register set is
528 the Altivec/VMX registers for the 128-bit operations.
530 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
532 (rs6000_split_128bit_ok_p): Likewise.
534 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
535 macros for creating ISA 2.07 normal and overloaded builtin
536 functions with 3 arguments.
537 (BU_P8V_OVERLOAD_3): Likewise.
538 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
539 for use as overloaded functions.
540 (VPERM_1TI_UNS): Likewise.
541 (VSEL_1TI): Likewise.
542 (VSEL_1TI_UNS): Likewise.
543 (ST_INTERNAL_1ti): Likewise.
544 (LD_INTERNAL_1ti): Likewise.
545 (XXSEL_1TI): Likewise.
546 (XXSEL_1TI_UNS): Likewise.
547 (VPERM_1TI): Likewise.
548 (VPERM_1TI_UNS): Likewise.
549 (XXPERMDI_1TI): Likewise.
551 (LXVD2X_V1TI): Likewise.
552 (STXVD2X_V1TI): Likewise.
553 (VEC_INIT_V1TI): Likewise.
554 (VEC_SET_V1TI): Likewise.
555 (VEC_EXT_V1TI): Likewise.
556 (EQV_V1TI): Likewise.
557 (NAND_V1TI): Likewise.
558 (ORC_V1TI): Likewise.
559 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
560 added in ISA 2.07. Add both normal 'altivec' builtins, and the
564 (VADDEUQM): Likewise.
565 (VADDECUQ): Likewise.
566 (VSUBEUQM): Likewise.
567 (VSUBECUQ): Likewise.
569 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
570 __int128_t and __uint128_t types.
571 (__uint128_type): Likewise.
572 (altivec_categorize_keyword): Add support for vector __int128_t,
573 vector __uint128_t, vector __int128, and vector unsigned __int128
574 as a container type for TImode operations that need to be done in
575 VSX/Altivec registers.
576 (rs6000_macro_to_expand): Likewise.
577 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
578 to support 128-bit integer instructions vaddcuq, vadduqm,
579 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
580 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
582 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
583 for V1TImode, and set up preferences to use VSX/Altivec registers.
584 Setup VSX reload handlers.
585 (rs6000_debug_reg_global): Likewise.
586 (rs6000_init_hard_regno_mode_ok): Likewise.
587 (rs6000_preferred_simd_mode): Likewise.
588 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
589 (easy_altivec_constant): Likewise.
590 (output_vec_const_move): Likewise.
591 (rs6000_expand_vector_set): Convert V1TImode set and extract to
593 (rs6000_expand_vector_extract): Likewise.
594 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
596 (rs6000_const_vec): Add support for V1TImode.
597 (rs6000_emit_le_vsx_load): Swap double words when loading or
598 storing TImode/V1TImode.
599 (rs6000_emit_le_vsx_store): Likewise.
600 (rs6000_emit_le_vsx_move): Likewise.
601 (rs6000_emit_move): Add support for V1TImode.
602 (altivec_expand_ld_builtin): Likewise.
603 (altivec_expand_st_builtin): Likewise.
604 (altivec_expand_vec_init_builtin): Likewise.
605 (altivec_expand_builtin): Likewise.
606 (rs6000_init_builtins): Add support for V1TImode type. Add
607 support for ISA 2.07 128-bit integer builtins. Define type names
608 for the VSX/Altivec vector types.
609 (altivec_init_builtins): Add support for overloaded vector
610 functions with V1TImode type.
611 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
612 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
614 (rs6000_split_128bit_ok_p): Likewise.
615 (rs6000_handle_altivec_attribute): Create V1TImode from vector
616 __int128_t and vector __uint128_t.
618 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
626 (VS_scalar): Likewise.
627 (VS_double): Likewise.
628 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
630 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
631 we support the ISA 2.07 128-bit integer arithmetic instructions.
632 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
633 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
634 and TImode types for use with the builtin functions.
635 (V1TI_type_node): Likewise.
636 (unsigned_V1TI_type_node): Likewise.
637 (intTI_type_internal_node): Likewise.
638 (uintTI_type_internal_node): Likewise.
640 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
641 128-bit builtin functions.
642 (UNSPEC_VADDEUQM): Likewise.
643 (UNSPEC_VADDECUQ): Likewise.
644 (UNSPEC_VSUBCUQ): Likewise.
645 (UNSPEC_VSUBEUQM): Likewise.
646 (UNSPEC_VSUBECUQ): Likewise.
647 (VM): Add V1TImode to vector mode iterators.
650 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
651 (altivec_vaddcuq): Likewise.
652 (altivec_vsubuqm): Likewise.
653 (altivec_vsubcuq): Likewise.
654 (altivec_vaddeuqm): Likewise.
655 (altivec_vaddecuq): Likewise.
656 (altivec_vsubeuqm): Likewise.
657 (altivec_vsubecuq): Likewise.
659 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
661 (BOOL_128): Likewise.
662 (BOOL_REGS_OUTPUT): Likewise.
663 (BOOL_REGS_OP1): Likewise.
664 (BOOL_REGS_OP2): Likewise.
665 (BOOL_REGS_UNARY): Likewise.
666 (BOOL_REGS_AND_CR0): Likewise.
668 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
669 128-bit integer builtin support.
670 (vec_vadduqm): Likewise.
671 (vec_vaddecuq): Likewise.
672 (vec_vaddeuqm): Likewise.
673 (vec_vsubecuq): Likewise.
674 (vec_vsubeuqm): Likewise.
675 (vec_vsubcuq): Likewise.
676 (vec_vsubuqm): Likewise.
678 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
679 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
680 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
681 128-bit integer add/subtract to ISA 2.07.
683 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
685 * config/arc/arc.c (arc_predicate_delay_insns):
686 Fix third argument passed to conditionalize_nonjump.
688 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
690 * config/aarch64/aarch64-builtins.c
691 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
692 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
693 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
694 instead of __builtin_lfloor.
695 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
697 2014-03-12 Jakub Jelinek <jakub@redhat.com>
699 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
700 (tree_ssa_ifcombine_bb_1): New function.
701 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
702 is an empty forwarder block to then_bb or vice versa and then_bb
703 and else_bb are effectively swapped.
705 2014-03-12 Christian Bruel <christian.bruel@st.com>
708 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
709 REG_CFA_DEF_CFA note.
710 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
711 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
713 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
715 PR tree-optimization/60454
716 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
718 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
720 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
721 Do not define target_cpu_default2 to generic.
722 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
723 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
724 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
726 2014-03-12 Jakub Jelinek <jakub@redhat.com>
727 Marc Glisse <marc.glisse@inria.fr>
729 PR tree-optimization/60502
730 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
731 instead of build_low_bits_mask.
733 2014-03-12 Jakub Jelinek <jakub@redhat.com>
736 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
737 if there are multiple uses, but op doesn't live on E edge.
738 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
739 clobber stmts before __builtin_unreachable.
741 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
743 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
744 hard_frame_pointer_rtx.
745 * cse.c (cse_insn): Remove volatile check.
746 * cselib.c (cselib_process_insn): Likewise.
747 * dse.c (scan_insn): Likewise.
749 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
751 * config/arc/arc.c (conditionalize_nonjump): New function,
753 (arc_ifcvt): ... this.
754 (arc_predicate_delay_insns): Use it.
756 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
758 * config/arc/predicates.md (extend_operand): During/after reload,
759 allow const_int_operand.
760 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
761 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
762 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
764 (umulsi3_highpart_i): Likewise.
766 2014-03-11 Richard Biener <rguenther@suse.de>
768 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
769 Add asserts to guard possible wrong-code bugs.
771 2014-03-11 Richard Biener <rguenther@suse.de>
773 PR tree-optimization/60429
774 PR tree-optimization/60485
775 * tree-ssa-structalias.c (set_union_with_increment): Properly
776 take into account all fields that overlap the shifted vars.
777 (do_sd_constraint): Likewise.
778 (do_ds_constraint): Likewise.
779 (get_constraint_for_ptr_offset): Likewise.
781 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
783 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
784 (nios2_compute_frame_layout):
785 Add calculation of cfun->machine->fp_save_offset.
786 (nios2_expand_prologue): Correct setting of frame pointer register
788 (nios2_expand_epilogue): Update recovery of stack pointer from
789 frame pointer accordingly.
790 (nios2_initial_elimination_offset): Update calculation of offset
791 for eliminating to HARD_FRAME_POINTER_REGNUM.
793 2014-03-10 Jakub Jelinek <jakub@redhat.com>
796 * ipa.c (symtab_remove_unreachable_nodes): Don't call
797 cgraph_get_create_node on VAR_DECLs.
799 2014-03-10 Richard Biener <rguenther@suse.de>
802 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
804 2014-03-08 Douglas B Rupp <rupp@gnat.com>
806 * config/vms/vms.opt (vms_float_format): New variable.
808 2014-03-08 Tobias Burnus <burnus@net-b.de>
810 * doc/invoke.texi (-fcilkplus): Update implementation status.
812 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
813 Richard Biener <rguenther@suse.de>
815 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
816 consistently accross all TUs.
817 (run_gcc): Enable -fshort-double automatically at link at link-time
818 and disallow override.
820 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
823 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
824 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
825 if they can't be used.
827 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
829 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
830 for Solaris 11/x86 ld.
831 * configure: Regenerate.
833 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
835 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
836 (LIB_TLS_SPEC): Save as ld_tls_libs.
837 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
838 (HAVE_AS_IX86_TLSLDM): New test.
839 * configure, config.in: Regenerate.
840 * config/i386/i386.c (legitimize_tls_address): Fall back to
841 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
842 cannot support TLS_MODEL_LOCAL_DYNAMIC.
843 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
844 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
846 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
848 * common.opt (fira-loop-pressure): Mark as optimization.
850 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
852 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
853 an OpenMP mappable type.
855 2014-03-06 Matthias Klose <doko@ubuntu.com>
857 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
858 MULTILIB_OSDIRNAMES is not defined.
860 2014-03-06 Jakub Jelinek <jakub@redhat.com>
861 Meador Inge <meadori@codesourcery.com>
864 * config/arm/arm.c (arm_tls_symbol_p): Remove.
865 (arm_legitimize_address): Call legitimize_tls_address for any
866 arm_tls_referenced_p expression, handle constant addend. Call it
867 before testing for !TARGET_ARM.
868 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
870 2014-03-06 Richard Biener <rguenther@suse.de>
876 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
878 * tree-streamer.c (record_common_node): Assert we don't record
879 nodes with type double.
880 (preload_common_node): Skip type double, complex double and double
881 pointer since it is now frontend dependent due to fshort-double option.
883 2014-03-06 Richard Biener <rguenther@suse.de>
885 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
886 or -fno-lto is specified and the linker has full plugin support.
887 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
888 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
889 * lto-wrapper.c (merge_and_complain): Merge compile-time
891 (run_gcc): And pass it through to the link options.
893 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
897 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
899 * cselib.c (remove_useless_values): Skip to avoid quadratic
900 behavior if the condition moved from...
901 (cselib_process_insn): ... here holds.
903 2014-03-05 Jakub Jelinek <jakub@redhat.com>
906 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
907 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
910 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
911 (TM_H): Add x86-tune.def.
913 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
915 * config/aarch64/aarch64.c (generic_tunings):
916 Use cortexa57_extra_costs.
918 2014-03-05 Jakub Jelinek <jakub@redhat.com>
921 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
922 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
923 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
926 2014-03-04 Heiher <r@hev.cc>
928 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
929 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
931 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
933 * config/i386/predicates.md (const2356_operand): Change to ...
934 (const2367_operand): ... this.
935 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
937 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
938 (*avx512pf_scatterpf<mode>sf): Ditto.
939 (avx512pf_scatterpf<mode>df): Ditto.
940 (*avx512pf_scatterpf<mode>df_mask): Ditto.
941 (*avx512pf_scatterpf<mode>df): Ditto.
942 * config/i386/i386.c (ix86_expand_builtin): Update
943 incorrect hint operand error message.
945 2014-03-04 Richard Biener <rguenther@suse.de>
947 * lto-section-in.c (lto_get_section_data): Fix const cast.
949 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
951 * tree-streamer.c (record_common_node): Assert we don't record
952 nodes with type double.
953 (preload_common_node): Skip type double, complex double and double
954 pointer since it is now frontend dependent due to fshort-double option.
956 2014-03-04 Richard Biener <rguenther@suse.de>
959 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
960 (lto_input_toplevel_asms): Likewise.
961 * lto-section-in.c (lto_get_section_data): Instead do it here
964 2014-03-04 Richard Biener <rguenther@suse.de>
966 PR tree-optimization/60382
967 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
968 dead PHIs a reduction.
970 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
972 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
974 (_mm_prefetch): Move out of GCC target("sse") pragma.
975 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
976 GCC target("prfchw") pragma.
977 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
979 * config/i386/i386.c (ix86_option_override_internal): Enable
980 -mprfchw with -mprefetchwt1.
982 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
984 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
987 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
989 * opts.h (CL_PCH_IGNORE): Define.
990 * targhooks.c (option_affects_pch_p):
991 Return false for options that have CL_PCH_IGNORE set.
992 * opt-functions.awk: Process PchIgnore.
993 * doc/options.texi: Document PchIgnore.
995 * config/arc/arc.opt (misize): Add PchIgnore property.
997 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
999 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1000 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1001 constraint on constants to permit them being loaded into
1002 GENERAL_REGS or BASE_REGS.
1004 2014-03-03 Nick Clifton <nickc@redhat.com>
1006 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1007 anti-cacnonical alternatives.
1008 (negandhi3_real): New pattern.
1009 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1011 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1013 * config/avr/avr-mcus.def: Remove atxmega16x1.
1014 * config/avr/avr-tables.opt: Regenerate.
1015 * config/avr/t-multilib: Regenerate.
1016 * doc/avr-mmcu.texi: Regenerate.
1018 2014-03-03 Tobias Grosser <tobias@grosser.es>
1019 Mircea Namolaru <mircea.namolaru@inria.fr>
1021 PR tree-optimization/58028
1022 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1025 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1027 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1028 not handled by recognizers.
1030 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1033 * function.c (expand_function_end): Don't emit
1034 clobber_return_register sequence if clobber_after is a BARRIER.
1035 * cfgexpand.c (construct_exit_block): Append instructions before
1036 return_label to prev_bb.
1038 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1040 * config/rs6000/constraints.md: Document reserved use of "wc".
1042 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1045 * ipa.c (function_and_variable_visibility): When dissolving comdat
1046 group, also set all symbols to local.
1048 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1053 2013-12-14 Jan Hubicka <jh@suse.cz>
1055 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1057 2014-03-02 Jon Beniston <jon@beniston.com>
1063 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1064 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1065 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1066 (simple_return, *simple_return): New patterns
1067 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1068 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1070 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1072 * dwarf2out.c (gen_subprogram_die): Tidy.
1074 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1077 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1078 (*mov_t_msb_neg_negc): ... this new insn.
1080 2014-02-28 Jason Merrill <jason@redhat.com>
1083 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1086 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1089 * dwarf2out.c (decltype_auto_die): New static.
1090 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1091 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1092 (is_cxx_auto): Likewise.
1094 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1096 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1097 we are not using general regs only.
1099 2014-02-28 Richard Biener <rguenther@suse.de>
1102 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1103 previous fix and only allow to remove trivial pre-headers
1104 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1105 (remove_forwarder_block): Properly update the latch of a loop.
1107 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1110 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1111 (cselib_preserved_hash_table): New.
1112 (preserve_constants_and_equivs): Move preserved vals to it.
1113 (cselib_find_slot): Look it up first.
1114 (cselib_init): Initialize it.
1115 (cselib_finish): Release it.
1116 (dump_cselib_table): Dump it.
1118 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1121 * cselib.c (remove_useless_values): Skip to avoid quadratic
1122 behavior if the condition moved from...
1123 (cselib_process_insn): ... here holds.
1125 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1128 * var-tracking.c (vt_initialize): Apply the same condition to
1129 preserve the CFA base value.
1131 2014-02-28 Joey Ye <joey.ye@arm.com>
1134 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1135 if reload in progress or completed.
1137 2014-02-28 Tobias Burnus <burnus@net-b.de>
1140 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1143 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1145 * doc/tm.texi.in (Condition Code Status): Update documention for
1146 relative locations of cc0-setter and cc0-user.
1148 2014-02-27 Jeff Law <law@redhat.com>
1150 PR rtl-optimization/52714
1151 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1152 into two independent simple sets, if I3 is a jump, ensure the
1153 pattern we place into I3 is a (set (pc) ...).
1155 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1156 Jeff Law <law@redhat.com>
1158 PR rtl-optimization/49847
1159 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1160 are in different blocks.
1161 * doc/tm.texi (Condition Code Status): Update documention for
1162 relative locations of cc0-setter and cc0-user.
1164 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1167 * lra.c (lra_emit_add): Check SUBREG too.
1169 2014-02-27 Andreas Schwab <schwab@suse.de>
1171 * config/m68k/m68k.c (m68k_option_override): Disable
1172 -flive-range-shrinkage for classic m68k.
1173 (m68k_override_options_after_change): Likewise.
1175 2014-02-27 Marek Polacek <polacek@redhat.com>
1178 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1179 -Wmaybe-uninitialized.
1181 2014-02-27 Alan Modra <amodra@gmail.com>
1184 * reload1.c (emit_input_reload_insns): When reload_override_in,
1185 set old to rl->in_reg when rl->in_reg is a subreg.
1187 2014-02-26 Richard Biener <rguenther@suse.de>
1190 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1192 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1194 * common/config/i386/predicates.md (const1256_operand): Remove.
1195 (const2356_operand): New.
1196 (const_1_to_2_operand): Remove.
1197 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1198 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1199 (*avx512pf_gatherpf<mode>sf): Ditto.
1200 (avx512pf_gatherpf<mode>df): Ditto.
1201 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1202 (*avx512pf_gatherpf<mode>df): Ditto.
1203 (avx512pf_scatterpf<mode>sf): Ditto.
1204 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1205 (*avx512pf_scatterpf<mode>sf): Ditto.
1206 (avx512pf_scatterpf<mode>df): Ditto.
1207 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1208 (*avx512pf_scatterpf<mode>df): Ditto.
1209 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1211 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1213 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1214 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1215 (_mm512_mask_testn_epi64_mask): Move to ...
1216 * config/i386/avx512cdintrin.h: Here.
1217 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1218 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1219 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1220 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1221 TARGET_AVX512F from TARGET_AVX512CD.
1223 2014-02-26 Richard Biener <rguenther@suse.de>
1226 * ipa.c (walk_polymorphic_call_targets): Properly guard
1227 call to inline_update_overall_summary.
1229 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1232 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1233 and latches only if requested. Fix latch if it is removed.
1234 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1235 LOOPS_HAVE_PREHEADERS.
1237 2014-02-25 Andrew Pinski <apinski@cavium.com>
1239 * builtins.c (expand_builtin_thread_pointer): Create a new target
1240 when the target is NULL.
1242 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1244 PR rtl-optimization/60317
1245 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1246 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1247 * lra-assigns.c: Include params.h.
1248 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1249 other reload pseudos considerations.
1251 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1253 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1254 to use canonical form for nor<mode>3.
1256 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1259 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1262 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1264 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1265 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1266 (ix86_handle_option): Handle OPT_mprefetchwt1.
1267 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1268 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1270 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1271 OPTION_MASK_ISA_PREFETCHWT1.
1272 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1273 (PTA_PREFETCHWT1): New.
1274 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1275 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1276 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1277 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1278 (*prefetch_avx512pf_<mode>_: Change into ...
1279 (*prefetch_prefetchwt1_<mode>: This.
1280 * config/i386/i386.opt (mprefetchwt1): New.
1281 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1282 (_mm_prefetch): Handle intent to write.
1283 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1285 2014-02-25 Richard Biener <rguenther@suse.de>
1288 * emit-rtl.c (mem_attrs_htab): Remove.
1289 (mem_attrs_htab_hash): Likewise.
1290 (mem_attrs_htab_eq): Likewise.
1291 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1292 (init_emit_once): Do not allocate mem_attrs_htab.
1294 2014-02-25 Richard Biener <rguenther@suse.de>
1297 * lto-opts.c (lto_write_options): Output non-explicit conservative
1298 -fwrapv, -fno-trapv and -fno-strict-overflow.
1299 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1300 (run_gcc): And pass them through.
1302 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1304 * sel-sched.c (calculate_new_fences): New parameter ptime.
1305 Calculate it as a maximum over all fence cycles.
1306 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1307 Print the final schedule timing when sched_verbose.
1309 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1311 PR rtl-optimization/60292
1312 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1313 bit fot the fence instruction.
1315 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1317 * calls.h: Fix typo in comment.
1319 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1321 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1322 adjusting offsetable addresses.
1324 2014-02-24 Guozhi Wei <carrot@google.com>
1326 * sparseset.h (sparseset_pop): Fix the wrong index.
1328 2014-02-24 Walter Lee <walt@tilera.com>
1330 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1331 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1333 * common/config/tilegx/tilegx-common.c
1334 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1335 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1337 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1338 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1339 (tilegx_gimplify_va_arg_expr): Handle big endian.
1340 (tilegx_expand_unaligned_load): Ditto.
1341 (tilegx_expand_unaligned_store): Ditto.
1342 (TARGET_RETURN_IN_MSB): New.
1343 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1344 (TARGET_ENDIAN_DEFAULT): New.
1345 (TARGET_BIG_ENDIAN): Handle big endian.
1346 (BYTES_BIG_ENDIAN): Ditto.
1347 (WORDS_BIG_ENDIAN): Ditto.
1348 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1351 * config/tilegx/tilegx.md (extv): Handle big endian.
1353 (insn_st<n>): Ditto.
1354 (insn_st<n>_add<bitsuffix>): Ditto.
1355 (insn_stnt<n>): Ditto.
1356 (insn_stnt<n>_add<bitsuffix>):Ditto.
1357 (vec_interleave_highv8qi): Handle big endian.
1358 (vec_interleave_highv8qi_be): New.
1359 (vec_interleave_highv8qi_le): New.
1360 (insn_v1int_h): Handle big endian.
1361 (vec_interleave_lowv8qi): Handle big endian.
1362 (vec_interleave_lowv8qi_be): New.
1363 (vec_interleave_lowv8qi_le): New.
1364 (insn_v1int_l): Handle big endian.
1365 (vec_interleave_highv4hi): Handle big endian.
1366 (vec_interleave_highv4hi_be): New.
1367 (vec_interleave_highv4hi_le): New.
1368 (insn_v2int_h): Handle big endian.
1369 (vec_interleave_lowv4hi): Handle big endian.
1370 (vec_interleave_lowv4hi_be): New.
1371 (vec_interleave_lowv4hi_le): New.
1372 (insn_v2int_l): Handle big endian.
1373 (vec_interleave_highv2si): Handle big endian.
1374 (vec_interleave_highv2si_be): New.
1375 (vec_interleave_highv2si_le): New.
1376 (insn_v4int_h): Handle big endian.
1377 (vec_interleave_lowv2si): Handle big endian.
1378 (vec_interleave_lowv2si_be): New.
1379 (vec_interleave_lowv2si_le): New.
1380 (insn_v4int_l): Handle big endian.
1381 * config/tilegx/tilegx.opt (mbig-endian): New option.
1382 (mlittle-endian): New option.
1383 * doc/install.texi: Document tilegxbe-linux.
1384 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1386 2014-02-24 Martin Jambor <mjambor@suse.cz>
1389 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1390 there are no parameter descriptors.
1392 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1394 PR rtl-optimization/60268
1395 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1396 initialization to ...
1397 (sched_rgn_init): ... here.
1398 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1400 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1402 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1405 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1407 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1410 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1412 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1413 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1415 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1417 * config/microblaze/predicates.md: Add cmp_op predicate.
1418 * config/microblaze/microblaze.md: Add branch_compare instruction
1419 which uses cmp_op predicate and emits cmp insn before branch.
1420 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1421 to microblaze_expand_conditional_branch and consolidate logic.
1422 (microblaze_expand_conditional_branch): emit branch_compare
1423 insn instead of handling cmp op separate from branch insn.
1425 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1427 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1430 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1432 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1433 define_insn with define_expand and new define_insn
1434 *altivec_lve<VI_char>x_internal.
1435 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1436 and new define_insn *altivec_stve<VI_char>x_internal.
1437 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1439 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1441 (altivec_expand_stvex_be): New function.
1443 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1445 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1446 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1447 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1448 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1450 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1453 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1454 instead of emit_move_insn.
1456 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1458 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1460 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1461 gen_altivec_vsumsws.
1463 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1465 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1466 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1467 (altivec_lvxl_<mode>): New define_expand incorporating
1468 -maltivec=be semantics where needed.
1469 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1470 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1471 semantics where needed.
1472 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1473 (altivec_stvx_<mode>): New define_expand incorporating
1474 -maltivec=be semantics where needed.
1475 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1476 VM2 iterator instead of V4SI.
1477 (altivec_stvxl_<mode>): New define_expand incorporating
1478 -maltivec=be semantics where needed.
1479 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1480 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1481 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1482 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1483 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1484 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1485 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1486 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1487 ALTIVEC_BUILTIN_STVXL.
1488 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1489 (altivec_expand_stvx_be): Likewise.
1490 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1491 (altivec_expand_lvx_be): Likewise.
1492 (altivec_expand_stvx_be): Likewise.
1493 (altivec_expand_builtin): Add cases for
1494 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1495 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1496 (altivec_init_builtins): Add definitions for
1497 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1498 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1500 2014-02-21 Catherine Moore <clm@codesourcery.com>
1502 * doc/invoke.texi (mvirt, mno-virt): Document.
1503 * config/mips/mips.opt (mvirt): New option.
1504 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1506 2014-02-21 Richard Biener <rguenther@suse.de>
1508 PR tree-optimization/60276
1509 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1510 (STMT_VINFO_MIN_NEG_DIST): New macro.
1511 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1512 STMT_VINFO_MIN_NEG_DIST.
1513 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1514 made for negative dependence distances still hold.
1516 2014-02-21 Richard Biener <rguenther@suse.de>
1519 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1520 DECL_INITIAL for globals not in the current function context.
1522 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1524 PR tree-optimization/56490
1525 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1526 * tree-ssa-uninit.c: Include params.h.
1527 (compute_control_dep_chain): Add num_calls argument, return false
1528 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1529 num_calls to recursive call.
1530 (find_predicates): Change dep_chain into normal array,
1531 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1532 variable and adjust compute_control_dep_chain caller.
1533 (find_def_preds): Likewise.
1535 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1537 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1538 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1540 2014-02-21 Nick Clifton <nickc@redhat.com>
1542 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1543 (pushhi1): Likewise.
1544 (popqi1): Add mode to pre_dec.
1547 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1549 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1550 mode for mask of V8SFmode permutation.
1552 2014-02-20 Richard Henderson <rth@redhat.com>
1555 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1556 a new pseudo for OLDVAL.
1558 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1561 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1562 gen_reg_rtx if d->testing_p.
1563 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1564 if d->testing_p and we will certainly return true.
1565 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1568 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1570 * emit-rtl.c (gen_reg_rtx): Assert that
1571 crtl->emit.regno_pointer_align_length is non-zero.
1573 2014-02-20 Richard Henderson <rth@redhat.com>
1576 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1577 on failure the store back into EXPECT.
1579 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1580 Sandra Loosemore <sandra@codesourcery.com>
1582 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1583 * config/nios2/nios2.c (nios2_function_profiler): Add
1584 -fPIC (flag_pic == 2) support.
1585 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1586 (nios2_large_offset_p): New function.
1587 (nios2_unspec_reloc_p): Move up position, update to use
1588 nios2_large_offset_p.
1589 (nios2_unspec_address): Remove function.
1590 (nios2_unspec_offset): New function.
1591 (nios2_large_got_address): New function.
1592 (nios2_got_address): Add large offset support.
1593 (nios2_legitimize_tls_address): Update usage of removed and new
1595 (nios2_symbol_binds_local_p): New function.
1596 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1597 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1598 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1599 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1600 processing for (const (unspec ...)).
1601 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1603 2014-02-20 Richard Biener <rguenther@suse.de>
1605 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1606 doing the substitution.
1607 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1609 2014-02-20 Martin Jambor <mjambor@suse.cz>
1612 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1613 info when checking whether lattices are bottom.
1615 2014-02-20 Richard Biener <rguenther@suse.de>
1618 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1621 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1624 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1625 parameter specifying the scaling.
1626 (inline_call): Update.
1627 (want_inline_recursively): Guard division by zero.
1628 (recursive_inlining): Update.
1629 * ipa-inline.h (clone_inlined_nodes): Update.
1631 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1634 * config/i386/i386.c (classify_argument): Pass structures of size
1635 64 bytes or less in register.
1637 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1638 Kirill Yukhin <kirill.yukhin@intel.com>
1640 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1641 (_mm_rcp28_round_ss): Ditto.
1642 (_mm_rsqrt28_round_sd): Ditto.
1643 (_mm_rsqrt28_round_ss): Ditto.
1644 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1645 (_mm_rcp14_round_ss): Ditto.
1646 (_mm_rsqrt14_round_sd): Ditto.
1647 (_mm_rsqrt14_round_ss): Ditto.
1648 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1649 the first input operand, get rid of match_dup.
1650 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1652 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1654 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1655 operand as the first input operand, set type attribute.
1656 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1658 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1659 operand as the first input operand, set type attribute.
1661 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1663 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1666 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1669 * config/i386/i386.c (construct_container): Remove TFmode check
1670 for X86_64_INTEGER_CLASS.
1672 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1675 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1676 only when -Wpsabi is enabled.
1678 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1681 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1682 passing arrays in registers are the same as for structs, so remove the
1683 special case for them.
1685 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1687 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1688 destination type, extract only the valid bits if the source type is not
1689 integral and has a different mode.
1691 2014-02-19 Richard Biener <rguenther@suse.de>
1694 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1697 2014-02-19 Richard Biener <rguenther@suse.de>
1700 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1701 (ipa_modify_call_arguments): Emit an argument load explicitely and
1702 preserve virtual SSA form there and for the replacement call.
1703 Do not update SSA form nor free dominance info.
1705 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1707 * ipa.c (function_and_variable_visibility): Also clear WEAK
1708 flag when disolving COMDAT_GROUP.
1710 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1712 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1713 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1715 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1716 do more sanity checks.
1717 (detect_type_change): Return true when giving up early.
1718 (compute_complex_assign_jump_func): Fix type parameter of
1719 ipa_set_ancestor_jf.
1720 (compute_complex_ancestor_jump_func): Likewise.
1721 (update_jump_functions_after_inlining): Fix updating of
1723 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1725 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1727 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1728 inline clones when edge disappears.
1730 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1733 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1734 Split 64-bit moves into 2 patterns. Do not allow the use of
1735 direct move for TDmode in little endian, since the decimal value
1736 has little endian bytes within a word, but the 64-bit pieces are
1737 ordered in a big endian fashion, and normal subreg's of TDmode are
1739 (mov<mode>_64bit_dm): Likewise.
1740 (movtd_64bit_nodm): Likewise.
1742 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1744 PR tree-optimization/60174
1745 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1746 statement of an SSA_NAME that occurs in an abnormal PHI node.
1748 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1751 * final.c (SEEN_BB): Remove.
1752 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1753 (final_scan_insn): Don't force_source_line on second
1754 NOTE_INSN_BASIC_BLOCK.
1756 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1759 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1760 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1761 (type_natural_mode): Warn ABI change when %zmm register is not
1762 available for AVX512F vector value passing.
1764 2014-02-18 Kai Tietz <ktietz@redhat.com>
1767 * config/i386/i386.c (ix86_expand_prologue): Use value in
1768 rax register as displacement when restoring %r10 or %rax.
1769 Fix wrong offset when restoring both registers.
1771 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1773 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1774 assertion with conditional return.
1776 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1777 Uros Bizjak <ubizjak@gmail.com>
1780 * config/i386/driver-i386.c (host_detect_local_cpu): If
1781 YMM state is not saved by the OS, also clear has_f16c. Move
1782 CPUID 0x80000001 handling before YMM state saving checking.
1784 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1786 PR rtl-optimization/58960
1787 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1788 factored out from ...
1789 (sched_init): ... here.
1790 (free_global_sched_pressure_data): New, factored out from ...
1791 (sched_finish): ... here.
1792 * sched-int.h (free_global_sched_pressure_data): Declare.
1793 * sched-rgn.c (nr_regions_initial): New static global.
1794 (haifa_find_rgns): Initialize it.
1795 (schedule_region): Disable sched-pressure for the newly
1798 2014-02-17 Richard Biener <rguenther@suse.de>
1800 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1801 release SSA defs of pattern stmts.
1803 2014-02-17 Richard Biener <rguenther@suse.de>
1805 * tree-inline.c (expand_call_inline): Release the virtual
1806 operand defined by the call we are about to inline.
1808 2014-02-17 Richard Biener <rguenther@suse.de>
1810 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1812 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1813 Ilya Tocar <ilya.tocar@intel.com>
1815 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1816 arguments order in builtin.
1817 (_mm512_permutexvar_epi64): Ditto.
1818 (_mm512_mask_permutexvar_epi64): Ditto
1819 (_mm512_maskz_permutexvar_epi32): Ditto
1820 (_mm512_permutexvar_epi32): Ditto
1821 (_mm512_mask_permutexvar_epi32): Ditto
1823 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1825 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1826 (p8_vmrgow): Likewise.
1828 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1830 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1833 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1836 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1837 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1838 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1839 using direct move instructions on ISA 2.07. Also adjust
1840 instruction length for 64-bit.
1841 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1842 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1844 2014-02-15 Alan Modra <amodra@gmail.com>
1848 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1849 find_replacement on parts of insn rtl that might be reloaded.
1851 2014-02-15 Richard Biener <rguenther@suse.de>
1853 PR tree-optimization/60183
1854 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1855 (tree_ssa_phiprop): Calculate and free post-dominators.
1857 2014-02-14 Jeff Law <law@redhat.com>
1859 PR rtl-optimization/60131
1860 * ree.c (get_extended_src_reg): New function.
1861 (combine_reaching_defs): Use it rather than assuming location of REG.
1862 (find_and_remove_re): Verify first operand of extension is
1863 a REG before adding the insns to the copy list.
1865 2014-02-14 Roland McGrath <mcgrathr@google.com>
1867 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1868 * configure: Regenerated.
1869 * config.in: Regenerated.
1870 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1871 instead of ASM_SHORT.
1873 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1874 Richard Earnshaw <rearnsha@arm.com>
1876 PR rtl-optimization/59535
1877 * lra-constraints.c (process_alt_operands): Encourage alternative
1878 when unassigned pseudo class is superset of the alternative class.
1879 (inherit_reload_reg): Don't inherit when optimizing for code size.
1880 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1881 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1882 modes not less than 4 for Thumb1.
1884 2014-02-14 Kyle McMartin <kyle@redhat.com>
1887 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1889 2014-02-14 Richard Biener <rguenther@suse.de>
1891 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1892 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1893 Do not strip INDIRECT_REFs.
1895 2014-02-14 Richard Biener <rguenther@suse.de>
1898 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1899 DECL_FUNCTION_SPECIFIC_TARGET.
1900 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1901 * tree-streamer-out.c (pack_ts_target_option): Remove.
1902 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1903 (write_ts_function_decl_tree_pointers): Do not stream
1904 DECL_FUNCTION_SPECIFIC_TARGET.
1905 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1906 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1907 (lto_input_ts_function_decl_tree_pointers): Do not stream
1908 DECL_FUNCTION_SPECIFIC_TARGET.
1910 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1912 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1913 (get_initial_def_for_induction, vectorizable_induction): Ignore
1914 debug stmts when looking for exit_phi.
1915 (vectorizable_live_operation): Fix up condition.
1917 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1919 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1920 nreverse() because it changes the content of original tree list.
1922 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1924 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1925 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1927 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1929 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1930 GNU coding standards.
1932 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1935 * dwarf2out.c (gen_subprogram_die): Don't call
1936 add_calling_convention_attribute if subr_die is old_die.
1938 2014-02-13 Sharad Singhai <singhai@google.com>
1940 * doc/optinfo.texi: Fix order of nodes.
1942 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1944 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1945 operands[2], not operands[3].
1947 2014-02-13 Richard Biener <rguenther@suse.de>
1950 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1951 mention the possibility of an in-tree build.
1952 (CLooG): Update recommended version to 0.18.1, mention the
1953 possibility of an in-tree build and clarify that the ISL
1954 bundled with CLooG does not work.
1956 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1959 * expr.c (compress_float_constant): If x is a hard register,
1960 extend into a pseudo and then move to x.
1962 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1964 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1965 caused by bad second argument to warning_at() with -mhotpatch and
1966 nested functions (e.g. with gfortran).
1968 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1970 * opts.c (option_name): Remove "enabled by default" rider.
1972 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1974 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1976 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1977 Uros Bizjak <ubizjak@gmail.com>
1980 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1981 * configure: Regenerated.
1983 2014-02-12 Richard Biener <rguenther@suse.de>
1985 * vec.c (vec_prefix::calculate_allocation): Move as
1986 inline variant to vec.h.
1987 (vec_prefix::calculate_allocation_1): New out-of-line version.
1988 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1989 (vec_prefix::m_has_auto_buf): Rename to ...
1990 (vec_prefix::m_using_auto_storage): ... this.
1991 (vec_prefix::calculate_allocation): Inline the easy cases
1992 and dispatch to calculate_allocation_1 which doesn't need the
1994 (va_heap::reserve): Use gcc_checking_assert.
1995 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1996 m_using_auto_storage.
1997 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1999 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2000 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2001 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2003 2014-02-12 Richard Biener <rguenther@suse.de>
2005 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2006 when we found a dependence.
2008 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2010 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2012 (maybe_fold_stmt): ... into this new function.
2013 * omp-low.c (lower_omp): Update comment.
2015 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2018 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2021 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2023 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2024 identifiers in comments.
2025 (cortexa53_extra_costs): Likewise.
2026 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2027 (cortexa7_extra_costs): Likewise.
2028 (cortexa12_extra_costs): Likewise.
2029 (cortexa15_extra_costs): Likewise.
2030 (v7m_extra_costs): Likewise.
2032 2014-02-12 Richard Biener <rguenther@suse.de>
2035 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2036 of posix_memalign being successful.
2037 (lower_stmt): Restrict lowering of posix_memalign to when
2038 -ftree-bit-ccp is enabled.
2040 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2042 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2044 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2046 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2048 PR rtl-optimization/60116
2049 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2050 other_insn once the combination has been validated.
2052 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2055 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2057 * ipa-devirt.c: Include demangle.h
2058 (odr_violation_reported): New static variable.
2059 (add_type_duplicate): Update odr_violations.
2060 (maybe_record_node): Add completep parameter; update it.
2061 (record_target_from_binfo): Add COMPLETEP parameter;
2062 update it as needed.
2063 (possible_polymorphic_call_targets_1): Likewise.
2064 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2065 rename FINAL to COMPLETE.
2066 (record_targets_from_bases): Sanity check we found the binfo;
2067 fix COMPLETEP updating.
2068 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2069 parameter, fix computing of COMPLETEP.
2070 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2071 at LTO time do demangling.
2072 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2073 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2075 (gimple_get_virt_method_for_binfo): Likewise.
2076 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2077 gimple_get_virt_method_for_vtable): Update prototypes.
2079 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2082 * genautomata.c (add_presence_absence): Fix typo with
2083 {final_}presence_list.
2085 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2088 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2089 for VSX/Altivec vectors that land in GPR registers.
2091 2014-02-11 Richard Henderson <rth@redhat.com>
2092 Jakub Jelinek <jakub@redhat.com>
2095 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2096 around drhs if type conversion to lacc->type is not useless.
2098 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2100 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2102 (cortex-a57.cortex-a53): Likewise.
2103 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2105 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2107 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2110 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2112 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2113 add_options_for_arm_vfp3.
2115 2014-02-11 Jeff Law <law@redhat.com>
2118 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2119 object with an undesirable mode.
2121 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2124 * config/i386/sol2-9.h: New file.
2125 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2126 *-*-solaris2.9*): Use it.
2128 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2130 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2131 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2133 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2135 * config/microblaze/microblaze.c: Extend mcpu version format
2137 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2139 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2141 2014-02-10 Richard Henderson <rth@redhat.com>
2144 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2145 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2146 ms-abi vs -mno-accumulate-outgoing-args.
2147 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2148 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2151 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2154 * cfgexpand.c (expand_asm_operands): Attach source location to
2155 ASM_INPUT rtx objects.
2156 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2158 2014-02-10 Nick Clifton <nickc@redhat.com>
2160 * config/mn10300/mn10300.c (popcount): New function.
2161 (mn10300_expand_prologue): Include saved registers in stack usage
2164 2014-02-10 Jeff Law <law@redhat.com>
2167 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2168 when changing the SET_DEST of a prior insn to avoid an input reload.
2170 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2172 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2173 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2174 -mcall-openbsd, or -mcall-linux.
2175 (CC1_ENDIAN_BIG_SPEC): Remove.
2176 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2177 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2178 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2179 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2180 and %cc1_endian_default.
2181 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2183 2014-02-10 Richard Biener <rguenther@suse.de>
2185 PR tree-optimization/60115
2186 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2187 MEM_REF handling. Properly verify that the accesses are not
2188 out of the objects bound.
2190 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2192 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2195 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2197 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2198 proper constants and fix formatting.
2199 (possible_polymorphic_call_targets): Fix formatting.
2201 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2202 Ilya Tocar <ilya.tocar@intel.com>
2204 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2205 (_mm512_loadu_epi32): Renamed into...
2206 (_mm512_loadu_si512): This.
2207 (_mm512_storeu_epi32): Renamed into...
2208 (_mm512_storeu_si512): This.
2209 (_mm512_maskz_ceil_ps): Removed.
2210 (_mm512_maskz_ceil_pd): Ditto.
2211 (_mm512_maskz_floor_ps): Ditto.
2212 (_mm512_maskz_floor_pd): Ditto.
2213 (_mm512_floor_round_ps): Ditto.
2214 (_mm512_floor_round_pd): Ditto.
2215 (_mm512_ceil_round_ps): Ditto.
2216 (_mm512_ceil_round_pd): Ditto.
2217 (_mm512_mask_floor_round_ps): Ditto.
2218 (_mm512_mask_floor_round_pd): Ditto.
2219 (_mm512_mask_ceil_round_ps): Ditto.
2220 (_mm512_mask_ceil_round_pd): Ditto.
2221 (_mm512_maskz_floor_round_ps): Ditto.
2222 (_mm512_maskz_floor_round_pd): Ditto.
2223 (_mm512_maskz_ceil_round_ps): Ditto.
2224 (_mm512_maskz_ceil_round_pd): Ditto.
2225 (_mm512_expand_pd): Ditto.
2226 (_mm512_expand_ps): Ditto.
2227 * config/i386/i386.c (ix86_builtins): Remove
2228 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2229 (bdesc_args): Ditto.
2230 * config/i386/predicates.md (const1256_operand): New.
2231 (const_1_to_2_operand): Ditto.
2232 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2233 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2234 (*avx512pf_gatherpf<mode>sf): Ditto.
2235 (avx512pf_gatherpf<mode>df): Ditto.
2236 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2237 (*avx512pf_gatherpf<mode>df): Ditto.
2238 (avx512pf_scatterpf<mode>sf): Ditto.
2239 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2240 (*avx512pf_scatterpf<mode>sf): Ditto.
2241 (avx512pf_scatterpf<mode>df): Ditto.
2242 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2243 (*avx512pf_scatterpf<mode>df): Ditto.
2244 (avx512f_expand<mode>): Removed.
2245 (<shift_insn><mode>3<mask_name>): Change predicate type.
2247 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2249 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2250 not at the end of datarefs vector use ordered_remove to avoid
2251 reordering datarefs vector.
2254 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2255 mark local addressable non-static vars as GOVD_PRIVATE
2256 instead of GOVD_LOCAL.
2257 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2258 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2262 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2263 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2264 assume_aligned or alloc_align attributes.
2265 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2266 arguments. Handle also assume_aligned and alloc_align attributes.
2267 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2268 calls to functions with assume_aligned or alloc_align attributes.
2269 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2271 2014-02-08 Terry Guo <terry.guo@arm.com>
2273 * doc/invoke.texi: Document ARM -march=armv7e-m.
2275 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2277 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2278 flag on __cilkrts_rethrow builtin.
2281 * ipa-cp.c (determine_versionability): Fail at -O0
2282 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2283 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2286 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2289 * tree-inline.c (copy_forbidden): Fail for
2290 __attribute__((optimize (0))) functions.
2292 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2294 * varpool.c: Include pointer-set.h.
2295 (varpool_remove_unreferenced_decls): Variables in other partitions
2296 will not be output; be however careful to not lose information
2299 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2301 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2302 lookup in the vtable constructor.
2304 2014-02-07 Jeff Law <law@redhat.com>
2307 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2308 define_insn_and_split.
2310 * ipa-inline.c (inline_small_functions): Fix typos.
2312 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2314 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2315 (s390_can_use_return_insn): Declare.
2316 * config/s390/s390.h (EPILOGUE_USES): Define.
2317 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2319 (s390_chunkify_start): Handle return JUMP_LABELs.
2320 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2321 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2322 (s390_can_use_return_insn): New functions.
2323 (s390_fix_long_loop_prediction): Handle conditional returns.
2324 (TARGET_SET_UP_BY_PROLOGUE): Define.
2325 * config/s390/s390.md (ANY_RETURN): New code iterator.
2326 (*creturn, *csimple_return, return, simple_return): New patterns.
2328 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2330 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2331 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2332 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2333 REG_CFA_RESTORE list when deciding not to restore a register.
2335 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2337 * config/s390/s390.c: Include tree-pass.h and context.h.
2338 (s390_early_mach): New function, split out from...
2339 (s390_emit_prologue): ...here.
2340 (pass_data_s390_early_mach): New pass structure.
2341 (pass_s390_early_mach): New class.
2342 (s390_option_override): Create and register early_mach pass.
2343 Move to end of file.
2345 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2347 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2348 to match for the exit block.
2350 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2352 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2353 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2354 Reject misaligned operands.
2356 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2358 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2360 2014-02-07 Richard Biener <rguenther@suse.de>
2363 * gimple-low.c (lower_builtin_posix_memalign): New function.
2364 (lower_stmt): Call it to lower posix_memalign in a way
2365 to make alignment info accessible.
2367 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2370 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2371 __builtin_setjmp_receiver.
2373 2014-02-07 Richard Biener <rguenther@suse.de>
2376 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2377 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2378 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2379 Handle BUILT_IN_POSIX_MEMALIGN.
2380 (find_func_clobbers): Likewise.
2381 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2382 (call_may_clobber_ref_p_1): Likewise.
2384 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2387 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2390 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2393 * lto-cgraph.c (lto_output_node): Use
2394 symtab_get_symbol_partitioning_class.
2395 (lto_output_varpool_node): likewise.
2396 (symtab_get_symbol_partitioning_class): Move here from
2398 * cgraph.h (symbol_partitioning_class): Likewise.
2399 (symtab_get_symbol_partitioning_class): Declare.
2401 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2403 * ggc.h (ggc_internal_cleared_alloc): New macro.
2404 * vec.h (vec_safe_copy): Handle memory stats.
2405 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2406 * target-globals.c (save_target_globals): Likewise.
2408 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2411 * expr.c (emit_move_resolve_push): Export; be bit more selective
2412 on when to clear alias set.
2413 * expr.h (emit_move_resolve_push): Declare.
2414 * function.h (struct function): Add tail_call_marked.
2415 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2416 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2417 * config/i386/i386.md (TImode move expander): De not call
2419 (FP push expanders): Preserve memory attributes.
2420 * config/i386/sse.md (push<mode>1): Remove.
2421 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2422 (ix86_expand_push): Remove.
2423 * config/i386/mmx.md (push<mode>1): Remove.
2425 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2427 PR rtl-optimization/60030
2428 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2429 lopart with paradoxical subreg before shifting it up by hprec.
2431 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2433 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2434 Remove extra newline at end of file.
2435 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2436 (arm_issue_rate): Handle cortexa57.
2437 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2438 (cortex-a57.cortex-a53): Likewise.
2440 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2443 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2444 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2446 (arm_expand_prologue): Adjust all callers.
2447 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2448 info, registers also at the lowest numbered registers side. Use
2449 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2453 * var-tracking.c (adjust_mems): Before adding a SET to
2454 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2456 2014-02-06 Alan Modra <amodra@gmail.com>
2459 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2460 change SDmode to DDmode when lra_in_progress.
2462 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2465 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2466 free_data_ref on the dr first, and before goto again also set dr
2467 to the next dr. For simd_lane_access, free old datarefs[i] before
2468 overwriting it. For get_vectype_for_scalar_type failure, don't
2469 free_data_ref if simd_lane_access.
2471 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2474 * tree.h (opts_for_fn): New inline function.
2475 (opt_for_fn): Define.
2476 * config/i386/i386.c (ix86_function_regparm): Use
2477 opt_for_fn (decl, optimize) instead of optimize.
2479 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2481 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2482 for SYMBOL_REF in large memory model.
2484 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2486 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2488 (cortex-a57): Likewise.
2489 (cortex-a57.cortex-a53): Likewise.
2491 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2492 Kugan Vivekanandarajah <kuganv@linaro.org>
2494 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2496 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2498 2014-02-06 Richard Biener <rguenther@suse.de>
2500 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2501 set_loop_copy and initialize_original_copy_tables.
2503 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2505 * config/aarch64/aarch64-simd.md
2506 (aarch64_ashr_simddi): Change QI to SI.
2508 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2509 Jakub Jelinek <jakub@redhat.com>
2512 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2515 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2517 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2518 CODE_FOR_altivec_vpku[hw]um to
2519 CODE_FOR_altivec_vpku[hw]um_direct.
2520 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2521 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2522 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2523 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2525 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2527 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2528 generation for -maltivec=be.
2529 (altivec_vsumsws): Simplify redundant test.
2531 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2533 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2534 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2535 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2536 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2537 gen_altivec_vpkuwum.
2538 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2540 (altivec_vpks<VI_char>ss): Likewise.
2541 (altivec_vpks<VI_char>us): Likewise.
2542 (altivec_vpku<VI_char>us): Likewise.
2543 (altivec_vpku<VI_char>um): Likewise.
2544 (altivec_vpku<VI_char>um_direct): New (copy of
2545 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2547 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2548 target is little endian and -maltivec=be is not specified.
2549 (*altivec_vupkhs<VU_char>_direct): New (copy of
2550 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2551 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2552 target is little endian and -maltivec=be is not specified.
2553 (*altivec_vupkls<VU_char>_direct): New (copy of
2554 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2555 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2556 little endian and -maltivec=be is not specified.
2557 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2558 little endian and -maltivec=be is not specified.
2560 2014-02-05 Richard Henderson <rth@redhat.com>
2563 * combine-stack-adj.c: Revert r206943.
2564 * sched-int.h (struct deps_desc): Add last_args_size.
2565 * sched-deps.c (init_deps): Initialize it.
2566 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2567 contain REG_ARGS_SIZE notes.
2569 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2571 * lto-cgraph.c (asm_nodes_output): Make global.
2572 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2573 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2574 (driver_handle_option): Handle OPT_fwpa.
2576 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2579 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2580 a comment typo and formatting issue. If odr_hash hasn't been
2581 created, return vNULL and set *completep to false.
2584 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2585 bb with no successors.
2587 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2590 * doc/invoke.texi (-march): Clarify documentation for ARM.
2594 2014-02-05 Richard Biener <rguenther@suse.de>
2596 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2597 when not vectorizing because of too many alias checks.
2598 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2599 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2601 2014-02-05 Nick Clifton <nickc@redhat.com>
2603 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2604 accept extended registers in any mode when compiling for the MN10300.
2606 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2608 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2609 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2610 sanitization attributes.
2611 (can_inline_edge_p): Likewise.
2612 (sanitize_attrs_match_for_inline_p): New function.
2614 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2616 * ipa-prop.c (detect_type_change): Shor circuit testing of
2617 type changes on THIS pointer.
2619 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2622 * config/pa/pa.c (legitimize_tls_address): Return original address
2623 if not passed a SYMBOL_REF rtx.
2624 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2626 (pa_emit_move_sequence): Simplify TLS source operands.
2627 (pa_legitimate_constant_p): Reject all TLS constants.
2628 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2629 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2631 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2633 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2634 groups when we know they are controlled by LTO.
2635 * varasm.c (default_binds_local_p_1): If object is in other partition,
2636 it will be resolved locally.
2638 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2640 * config/host-linux.c (linux_gt_pch_use_address): Don't
2641 use SSIZE_MAX because it is not always defined.
2643 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2646 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2647 threshold for pseudo splitting.
2648 (update_ebb_live_info): Process call argument hard registers and
2649 hard registers from insn definition too.
2650 (max_small_class_regs_num): New constant.
2651 (inherit_in_ebb): Update live hard regs through EBBs. Update
2652 reloads_num only for small register classes. Don't split for
2655 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2658 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2661 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2663 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2666 2014-02-04 Marek Polacek <polacek@redhat.com>
2668 * gdbinit.in (pel): Define.
2670 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2672 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2675 2014-02-04 Richard Biener <rguenther@suse.de>
2678 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2679 in function context local.
2680 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2681 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2682 similar to LTO_imported_decl_ref.
2684 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2686 PR tree-optimization/60002
2687 * cgraphclones.c (build_function_decl_skip_args): Clear
2690 PR tree-optimization/60023
2691 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2692 false to gsi_replace.
2693 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2694 has been in some EH region and vec_stmt could throw, add
2695 vec_stmt into the same EH region.
2696 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2697 has no lhs, ignore it.
2698 * internal-fn.c (expand_MASK_LOAD): Likewise.
2701 * tree-inline.c (copy_forbidden): Fail for
2702 __attribute__((optimize (0))) functions.
2705 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2706 is set, copy one less argument.
2707 (expand_simd_clones): Don't subtract clone_info->inbranch
2708 from simd_clone_struct_alloc argument.
2710 PR rtl-optimization/57915
2711 * recog.c (simplify_while_replacing): If all unary/binary/relational
2712 operation arguments are constant, attempt to simplify those.
2715 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2716 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2718 2014-02-04 Richard Biener <rguenther@suse.de>
2720 PR tree-optimization/60012
2721 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2722 TBAA disambiguation to all DDRs.
2724 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2727 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2728 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2730 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2733 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2735 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2737 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2738 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2740 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2743 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2744 to figure out targets of polymorphic calls with known decl.
2745 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2746 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2747 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2748 (get_polymorphic_call_info): ... here.
2749 (get_polymorphic_call_info_from_invariant): New function.
2751 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2753 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2754 lookup via vtable pointer; check for type consistency
2755 and turn inconsitent facts into UNREACHABLE.
2756 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2757 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2758 type inconsistent querries; return UNREACHABLE instead.
2760 2014-02-03 Richard Henderson <rth@twiddle.net>
2763 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2764 already processed this node.
2765 (normalize_one_pred_1): Pass along mark_set.
2766 (normalize_one_pred): Create and destroy a pointer_set_t.
2767 (normalize_one_pred_chain): Likewise.
2769 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2771 PR gcov-profile/58602
2772 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2774 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2777 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2778 -fno-devirtualize; try to devirtualize by the knowledge of
2779 virtual table pointer given by aggregate propagation.
2780 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2781 (ipa_print_node_jump_functions): Dump also offset that
2782 is relevant for polymorphic calls.
2783 (determine_known_aggregate_parts): Add arg_type parameter; use it
2784 instead of determining the type from pointer type.
2785 (ipa_compute_jump_functions_for_edge): Update call of
2786 determine_known_aggregate_parts.
2787 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2788 (gimple_get_virt_method_for_binfo): ... here; simplify using
2789 vtable_pointer_value_to_vtable.
2790 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2791 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2792 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2793 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2795 (vtable_pointer_value_to_binfo): ... here.
2796 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2798 2014-02-03 Teresa Johnson <tejohnson@google.com>
2800 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2801 redef of outer loop index variable.
2803 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2807 * doc/extend.texi (Function Attributes): Typo.
2809 2014-02-03 Cong Hou <congh@google.com>
2811 PR tree-optimization/60000
2812 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2813 if the vectorized statement is a store. A store statement can only
2814 appear at the end of pattern statements.
2816 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2818 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2819 (ix86_option_override_internal): Default long double to 64-bit for
2820 32-bit Bionic and to 128-bit for 64-bit Bionic.
2822 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2823 TARGET_LONG_DOUBLE_128 is true.
2824 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2826 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2827 (mlong-double-64): Negate -mlong-double-128.
2828 (mlong-double-128): New option.
2830 * config/i386/i386-c.c (ix86_target_macros): Define
2831 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2833 * doc/invoke.texi: Document -mlong-double-128.
2835 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2837 PR rtl-optimization/60024
2838 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2840 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2842 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2844 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2846 PR rtl-optimization/57662
2847 * sel-sched.c (code_motion_path_driver): Do not mark already not
2848 existing blocks in the visiting bitmap.
2850 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2852 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2853 on the insn being emitted.
2855 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2856 Will Deacon <will.deacon@arm.com>
2858 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2860 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2862 * config/arm/arm-tables.opt: Regenerate.
2864 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2866 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2867 for vector types other than V16QImode.
2868 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2869 define_expand, and call altivec_expand_vec_perm_le when producing
2870 code with little endian element order.
2871 (*altivec_vperm_<mode>_internal): New insn having previous
2872 behavior of altivec_vperm_<mode>.
2873 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2874 altivec_expand_vec_perm_le when producing code with little endian
2876 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2877 behavior of altivec_vperm_<mode>_uns.
2879 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2881 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2882 (altivec_vsumsws): Add handling for -maltivec=be with a little
2884 (altivec_vsumsws_direct): New.
2885 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2886 gen_altivec_vsumsws.
2888 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2890 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2891 vtable_pointer_value_to_binfo): New functions.
2892 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2893 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2895 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2897 * config/nios2/nios2.md (load_got_register): Initialize GOT
2898 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2899 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2901 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2903 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2904 preserverd by passthrough, do not propagate the type.
2906 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2908 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2909 (mips_atomic_assign_expand_fenv): New function.
2910 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2912 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2914 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2915 (__builtin_mips_set_fcsr): Likewise.
2916 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2917 MIPS_USI_FTYPE_VOID.
2918 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2919 (mips16_expand_set_fcsr): Likewise.
2920 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2921 (mips16_set_fcsr_stub): Likewise.
2922 (mips16_get_fcsr_one_only_stub): New class.
2923 (mips16_set_fcsr_one_only_stub): Likewise.
2924 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2925 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2926 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2927 (hard_float): New availability predicate.
2928 (mips_builtins): Add get_fcsr and set_fcsr.
2929 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2930 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2931 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2932 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2933 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2936 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2938 * config/mips/mips.c (mips_one_only_stub): New class.
2939 (mips_need_mips16_rdhwr_p): Replace with...
2940 (mips16_rdhwr_stub): ...this new variable.
2941 (mips16_stub_call_address): New function.
2942 (mips16_rdhwr_one_only_stub): New class.
2943 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2944 (mips_output_mips16_rdhwr): Delete.
2945 (mips_finish_stub): New function.
2946 (mips_code_end): Use it to handle rdhwr stubs.
2948 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2951 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2952 when calculating size of integer atomic types.
2954 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2956 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2958 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2960 PR tree-optimization/60003
2961 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2962 * profile.c (branch_prob): Use gimple_call_builtin_p
2963 to check for BUILT_IN_SETJMP_RECEIVER.
2964 * tree-inline.c (copy_bb): Call notice_special_calls.
2966 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2969 * lra-constraints.c (process_alt_operands): Update reload_sum only
2972 2014-01-31 Richard Henderson <rth@redhat.com>
2975 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2976 until after else_eh is processed.
2978 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2980 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2981 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2982 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2983 in smmintrin.h, remove them.
2984 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2985 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2986 * config/i386/i386.md (ROUND_SAE): Fix value.
2987 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2988 (const48_operand): New.
2989 * config/i386/subst.md (round), (round_expand): Use
2990 const_4_or_8_to_11_operand.
2991 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2993 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2995 * config/i386/constraints.md (Yk): Swap meaning with k.
2996 * config/i386/i386.md (movhi_internal): Change Yk to k.
2997 (movqi_internal): Ditto.
2998 (*k<logic><mode>): Ditto.
3001 (kandn<mode>): Ditto.
3002 (*<code>hi_1): Ditto.
3003 (*<code>qi_1): Ditto.
3004 (kxnor<mode>): Ditto.
3005 (kortestzhi): Ditto.
3006 (kortestchi): Ditto.
3008 (*one_cmplhi2_1): Ditto.
3009 (*one_cmplqi2_1): Ditto.
3010 * config/i386/sse.md (): Change k to Yk.
3011 (avx512f_load<mode>_mask): Ditto.
3012 (avx512f_blendm<mode>): Ditto.
3013 (avx512f_store<mode>_mask): Ditto.
3014 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3015 (avx512f_storedqu<mode>_mask): Ditto.
3016 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3018 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3019 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3020 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3021 (avx512f_maskcmp<mode>3): Ditto.
3022 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3023 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3024 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3025 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3026 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3027 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3028 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3029 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3030 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3031 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3032 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3033 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3034 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3035 (vec_extract_lo_<mode>_maskm): Ditto.
3036 (vec_extract_hi_<mode>_maskm): Ditto.
3037 (avx512f_vternlog<mode>_mask): Ditto.
3038 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3039 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3040 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3041 (avx512f_<code>v8div16qi2_mask): Ditto.
3042 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3043 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3044 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3045 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3046 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3047 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3048 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3049 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3050 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3051 (avx512cd_maskb_vec_dupv8di): Ditto.
3052 (avx512cd_maskw_vec_dupv16si): Ditto.
3053 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3054 (avx512f_vpermi2var<mode>3_mask): Ditto.
3055 (avx512f_vpermi2var<mode>3_mask): Ditto.
3056 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3057 (*avx512f_gathersi<mode>): Ditto.
3058 (*avx512f_gathersi<mode>_2): Ditto.
3059 (*avx512f_gatherdi<mode>): Ditto.
3060 (*avx512f_gatherdi<mode>_2): Ditto.
3061 (*avx512f_scattersi<mode>): Ditto.
3062 (*avx512f_scatterdi<mode>): Ditto.
3063 (avx512f_compress<mode>_mask): Ditto.
3064 (avx512f_compressstore<mode>_mask): Ditto.
3065 (avx512f_expand<mode>_mask): Ditto.
3066 * config/i386/subst.md (mask): Change k to Yk.
3067 (mask_scalar_merge): Ditto.
3070 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3072 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3074 2014-01-31 Richard Biener <rguenther@suse.de>
3077 * builtins.c (fold_builtin_memory_op): Make sure to not
3078 use a floating-point mode or a boolean or enumeral type for
3081 2014-01-30 DJ Delorie <dj@redhat.com>
3083 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3084 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3085 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3086 whenever main() has an epilogue.
3088 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3090 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3091 unused variable "field".
3092 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3093 (vsx_mergeh_<mode>): Likewise.
3094 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3095 (altivec_vmrghh): Likewise.
3096 (altivec_vmrghw): Likewise.
3097 (altivec_vmrglb): Likewise.
3098 (altivec_vmrglh): Likewise.
3099 (altivec_vmrglw): Likewise.
3100 (altivec_vspltb): Add missing uses.
3101 (altivec_vsplth): Likewise.
3102 (altivec_vspltw): Likewise.
3103 (altivec_vspltsf): Likewise.
3105 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3108 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3109 frame related instructions.
3111 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3113 PR rtl-optimization/59959
3114 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3115 any reload of register whose subreg is invalid.
3117 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3119 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3120 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3121 Add missing return type - void.
3123 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3125 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3126 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3127 remove element index adjustment for endian (now handled in vsx.md
3129 (altivec_expand_vec_perm_const): Use
3130 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3131 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3132 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3133 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3134 define_expand and a new define_insn *altivec_vspltb_internal;
3135 adjust for -maltivec=be on a little endian target.
3136 (altivec_vspltb_direct): New.
3137 (altivec_vsplth): Divide into a define_expand and a new
3138 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3139 little endian target.
3140 (altivec_vsplth_direct): New.
3141 (altivec_vspltw): Divide into a define_expand and a new
3142 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3143 little endian target.
3144 (altivec_vspltw_direct): New.
3145 (altivec_vspltsf): Divide into a define_expand and a new
3146 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3147 a little endian target.
3149 2014-01-30 Richard Biener <rguenther@suse.de>
3151 PR tree-optimization/59993
3152 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3153 can propagate form the earlier stmt and avoid the transform
3154 when the intermediate result is needed.
3156 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3158 * README.Portability: Fix typo.
3160 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3162 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3163 comparison_operator with ordered_comparison_operator.
3165 2014-01-30 Nick Clifton <nickc@redhat.com>
3167 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3168 Rename to mn10300_store_multiple_regs.
3169 * config/mn10300/mn10300.c: Likewise.
3170 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3171 store_multiple_regs.
3172 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3173 Call mn10300_store_multiple_regs.
3175 2014-01-30 Nick Clifton <nickc@redhat.com>
3176 DJ Delorie <dj@redhat.com>
3178 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3179 %fp 2 to keep registers after it properly word-aligned.
3180 (rl78_alloc_physical_registers_umul): Handle the case where both
3181 input operands are the same.
3183 2014-01-30 Richard Biener <rguenther@suse.de>
3185 PR tree-optimization/59903
3186 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3189 2014-01-30 Jason Merrill <jason@redhat.com>
3192 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3195 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3197 2014-01-30 Richard Biener <rguenther@suse.de>
3199 PR tree-optimization/59951
3200 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3202 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3205 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3206 SFmode to DFmode case.
3208 2014-01-29 DJ Delorie <dj@redhat.com>
3210 * config/msp430/msp430.opt (-minrt): New.
3211 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3213 (ENDFILE_SPEC): Likewise.
3215 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3217 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3218 (estimate_function_body_sizes): Use it.
3220 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3223 * dwarf2out.c (is_cxx_auto): New.
3224 (is_base_type): Use it.
3225 (gen_type_die_with_usage): Likewise.
3227 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3229 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3230 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3231 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3232 -maltivec=be with LE targets.
3233 (vsx_mergeh_<mode>): Likewise.
3234 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3235 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3236 (altivec_vmrghb): Replace with define_expand and new
3237 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3238 (altivec_vmrghb_direct): New define_insn.
3239 (altivec_vmrghh): Replace with define_expand and new
3240 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3241 (altivec_vmrghh_direct): New define_insn.
3242 (altivec_vmrghw): Replace with define_expand and new
3243 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3244 (altivec_vmrghw_direct): New define_insn.
3245 (*altivec_vmrghsf): Adjust for endianness.
3246 (altivec_vmrglb): Replace with define_expand and new
3247 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3248 (altivec_vmrglb_direct): New define_insn.
3249 (altivec_vmrglh): Replace with define_expand and new
3250 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3251 (altivec_vmrglh_direct): New define_insn.
3252 (altivec_vmrglw): Replace with define_expand and new
3253 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3254 (altivec_vmrglw_direct): New define_insn.
3255 (*altivec_vmrglsf): Adjust for endianness.
3256 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3257 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3258 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3259 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3260 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3261 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3262 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3263 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3265 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3267 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3268 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3271 2014-01-29 Richard Biener <rguenther@suse.de>
3273 PR tree-optimization/58742
3274 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3275 associate_pointerplus_align.
3276 (associate_pointerplus_diff): New function.
3277 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3278 and associate_pointerplus_diff.
3280 2014-01-29 Richard Biener <rguenther@suse.de>
3282 * lto-streamer.h (LTO_major_version): Bump to 3.
3283 (LTO_minor_version): Reset to 0.
3285 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3287 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3288 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3289 (arm_file_start): Generate correct asm header for armv7ve.
3290 * config/arm/bpabi.h: Add multilib support for armv7ve.
3291 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3292 and cortex-a15 to armv7ve.
3293 * config/arm/t-aprofile: Add multilib support for armv7ve.
3294 * doc/invoke.texi: Document -march=armv7ve.
3296 2014-01-29 Richard Biener <rguenther@suse.de>
3298 PR tree-optimization/58742
3299 * tree-ssa-forwprop.c (associate_plusminus): Return true
3300 if we changed sth, defer EH cleanup to ...
3301 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3302 (simplify_mult): New function.
3304 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3307 PR tree-optimization/59920
3308 * tree.c (build_common_builtin_nodes): Remove
3309 __builtin_setjmp_dispatcher initialization.
3310 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3311 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3312 instead of gsi_after_labels + manually skipping debug stmts.
3313 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3314 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3315 * tree-inline.c (copy_edges_for_bb): Remove
3316 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3317 argument. Ignore computed_goto_p stmts. Don't call
3318 make_abnormal_goto_edges. If a call might need abnormal edges
3319 for non-local gotos, see if it already has an edge to
3320 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3321 with true argument, don't do anything then, otherwise add
3322 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3323 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3325 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3326 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3327 (lower_stmt): Don't set data->calls_builtin_setjmp.
3328 (lower_builtin_setjmp): Adjust comment.
3329 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3330 * tree-cfg.c (found_computed_goto): Remove.
3331 (factor_computed_gotos): Remove.
3332 (make_goto_expr_edges): Return bool, true for computed gotos.
3333 Don't call make_abnormal_goto_edges.
3334 (build_gimple_cfg): Don't set found_computed_goto, don't call
3335 factor_computed_gotos.
3336 (computed_goto_p): No longer static.
3337 (make_blocks): Don't set found_computed_goto.
3338 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3339 (make_edges): If make_goto_expr_edges returns true, push bb
3340 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3341 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3342 vector. Record mapping between bbs and OpenMP regions if there
3343 are any, adjust make_gimple_omp_edges caller. Call
3344 handle_abnormal_edges.
3345 (make_abnormal_goto_edges): Remove.
3346 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3347 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3348 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3349 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3350 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3351 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3352 filling *region also set *region_idx to (*region)->entry->index.
3355 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3356 For REGs set ORIGINAL_REGNO.
3358 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3360 * doc/md.texi: Mention that a target shouldn't implement
3361 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3364 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3366 PR tree-optimization/59594
3367 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3368 a copy of the datarefs vector rather than the vector itself.
3370 2014-01-28 Jason Merrill <jason@redhat.com>
3373 * dwarf2out.c (auto_die): New static.
3374 (gen_type_die_with_usage): Handle C++1y 'auto'.
3375 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3378 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3381 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3382 (SPEC_X32): Likewise.
3383 (SPEC_64): Likewise.
3384 * config/i386/i386.c (ix86_option_override_internal): Turn off
3385 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3387 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3388 * config/i386/i386.h (TARGET_16BIT): New macro.
3389 (TARGET_16BIT_P): Likewise.
3390 * config/i386/i386.opt: Add m16.
3391 * doc/invoke.texi: Document -m16.
3393 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3395 PR preprocessor/59935
3396 * input.c (location_get_source_line): Bail out on when line number
3397 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3399 2014-01-28 Richard Biener <rguenther@suse.de>
3401 PR tree-optimization/58742
3402 * tree-ssa-forwprop.c (associate_plusminus): Handle
3403 pointer subtraction of the form (T)(P + A) - (T)P.
3405 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3407 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3410 2014-01-28 Richard Biener <rguenther@suse.de>
3413 2014-01-28 Richard Biener <rguenther@suse.de>
3415 PR rtl-optimization/45364
3416 PR rtl-optimization/59890
3417 * var-tracking.c (local_get_addr_clear_given_value): Handle
3418 already cleared slot.
3419 (val_reset): Handle not allocated local_get_addr_cache.
3420 (vt_find_locations): Use post-order on the inverted CFG.
3422 2014-01-28 Richard Biener <rguenther@suse.de>
3424 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3426 2014-01-28 Richard Biener <rguenther@suse.de>
3428 PR rtl-optimization/45364
3429 PR rtl-optimization/59890
3430 * var-tracking.c (local_get_addr_clear_given_value): Handle
3431 already cleared slot.
3432 (val_reset): Handle not allocated local_get_addr_cache.
3433 (vt_find_locations): Use post-order on the inverted CFG.
3435 2014-01-28 Alan Modra <amodra@gmail.com>
3437 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3438 * configure.ac <recursive call for build != host>: Define
3439 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3440 and LD_FOR_BUILD too.
3441 * configure: Regenerate.
3443 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3445 * config/i386/i386.c (get_builtin_code_for_version): Separate
3446 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3447 Broadwell from Haswell.
3449 2014-01-27 Steve Ellcey <sellcey@mips.com>
3451 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3452 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3453 * config/mips/mips.c (mips_option_override): Change setting
3455 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3456 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3457 Change from Mask to Var.
3459 2014-01-27 Jeff Law <law@redhat.com>
3461 * ipa-inline.c (inline_small_functions): Fix typo.
3463 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3465 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3466 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3467 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3468 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3469 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3470 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3471 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3472 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3473 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3474 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3475 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3476 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3477 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3478 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3479 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3480 (_mm512_storeu_epi64): Ditto.
3481 (_mm512_cmpge_epi32_mask): Ditto.
3482 (_mm512_cmpge_epu32_mask): Ditto.
3483 (_mm512_cmpge_epi64_mask): Ditto.
3484 (_mm512_cmpge_epu64_mask): Ditto.
3485 (_mm512_cmple_epi32_mask): Ditto.
3486 (_mm512_cmple_epu32_mask): Ditto.
3487 (_mm512_cmple_epi64_mask): Ditto.
3488 (_mm512_cmple_epu64_mask): Ditto.
3489 (_mm512_cmplt_epi32_mask): Ditto.
3490 (_mm512_cmplt_epu32_mask): Ditto.
3491 (_mm512_cmplt_epi64_mask): Ditto.
3492 (_mm512_cmplt_epu64_mask): Ditto.
3493 (_mm512_cmpneq_epi32_mask): Ditto.
3494 (_mm512_cmpneq_epu32_mask): Ditto.
3495 (_mm512_cmpneq_epi64_mask): Ditto.
3496 (_mm512_cmpneq_epu64_mask): Ditto.
3497 (_mm512_expand_pd): Ditto.
3498 (_mm512_expand_ps): Ditto.
3499 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3500 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3501 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3502 * config/i386/i386.c (ix86_builtins): Add
3503 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3504 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3505 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3506 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3507 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3508 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3509 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3510 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3511 IX86_BUILTIN_PMOVUSQW512_MEM.
3512 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3513 __builtin_ia32_pmovsqd512mem_mask,
3514 __builtin_ia32_pmovqd512mem_mask,
3515 __builtin_ia32_pmovusqw512mem_mask,
3516 __builtin_ia32_pmovsqw512mem_mask,
3517 __builtin_ia32_pmovqw512mem_mask,
3518 __builtin_ia32_pmovusdw512mem_mask,
3519 __builtin_ia32_pmovsdw512mem_mask,
3520 __builtin_ia32_pmovdw512mem_mask,
3521 __builtin_ia32_pmovqb512mem_mask,
3522 __builtin_ia32_pmovusqb512mem_mask,
3523 __builtin_ia32_pmovsqb512mem_mask,
3524 __builtin_ia32_pmovusdb512mem_mask,
3525 __builtin_ia32_pmovsdb512mem_mask,
3526 __builtin_ia32_pmovdb512mem_mask.
3527 (bdesc_args): Add __builtin_ia32_expanddf512,
3528 __builtin_ia32_expandsf512.
3529 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3530 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3531 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3532 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3533 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3534 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3535 (avx512f_<code>v8div16qi2_mask_store): This.
3536 (avx512f_expand<mode>): New.
3538 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3540 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3542 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3543 (_mm512_prefetch_i32scatter_pd): Ditto.
3544 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3545 (_mm512_prefetch_i64scatter_pd): Ditto.
3546 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3547 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3548 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3549 (_mm512_prefetch_i32scatter_ps): Ditto.
3550 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3551 (_mm512_prefetch_i64scatter_ps): Ditto.
3552 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3553 * config/i386/i386-builtin-types.def: Define
3554 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3555 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3556 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3557 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3558 IX86_BUILTIN_SCATTERPFQPD.
3559 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3560 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3561 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3562 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3563 __builtin_ia32_scatterpfqps.
3564 (ix86_expand_builtin): Expand new built-ins.
3565 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3566 fix memory access data type.
3567 (*avx512pf_gatherpf<mode>_mask): Ditto.
3568 (*avx512pf_gatherpf<mode>): Ditto.
3569 (avx512pf_scatterpf<mode>): Ditto.
3570 (*avx512pf_scatterpf<mode>_mask): Ditto.
3571 (*avx512pf_scatterpf<mode>): Ditto.
3572 (GATHER_SCATTER_SF_MEM_MODE): New.
3573 (avx512pf_gatherpf<mode>df): Ditto.
3574 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3575 (*avx512pf_scatterpf<mode>df): Ditto.
3577 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3580 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3581 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3584 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3586 * common/config/arm/arm-common.c
3587 (arm_rewrite_mcpu): Handle multiple names.
3589 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3591 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3593 * gimple-builder.h (create_gimple_tmp): Delete.
3595 2014-01-27 Christian Bruel <christian.bruel@st.com>
3597 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3600 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3602 * config/pa/pa.md (call): Generate indirect long calls to non-local
3603 functions when outputing 32-bit code.
3604 (call_value): Likewise except for special call to buggy powf function.
3606 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3607 portable runtime and PIC indirect calls.
3608 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3609 and PIC call sequences. Use ldo instead of blr to set return register
3610 in PIC call sequence.
3612 2014-01-25 Walter Lee <walt@tilera.com>
3614 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3615 avoid clobbering a live register.
3617 2014-01-25 Walter Lee <walt@tilera.com>
3619 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3620 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3621 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3622 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3624 2014-01-25 Walter Lee <walt@tilera.com>
3626 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3627 arguments on even registers.
3628 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3630 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3631 (BIGGEST_ALIGNMENT): Ditto.
3632 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3634 2014-01-25 Walter Lee <walt@tilera.com>
3636 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3637 insns before bundling.
3638 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3640 2014-01-25 Walter Lee <walt@tilera.com>
3642 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3643 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3644 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3646 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3648 * config/mips/constraints.md (kl): Delete.
3649 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3650 define expands, using...
3651 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3652 instructions for MIPS16.
3653 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3654 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3656 2014-01-25 Walter Lee <walt@tilera.com>
3658 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3662 2014-01-25 Walter Lee <walt@tilera.com>
3664 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3665 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3667 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3669 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3672 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3674 * print-rtl.c (in_call_function_usage): New var.
3675 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3676 EXPR_LIST mode as mode and not as reg note name.
3679 * cfgloopmanip.c (copy_loop_info): If
3680 loop->warned_aggressive_loop_optimizations, make sure
3681 the flag is set in target loop too.
3683 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3685 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3687 * builtins.def: Likewise.
3688 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3689 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3690 * ira.c (ira_setup_eliminable_regset): Likewise.
3691 * omp-low.c (gate_expand_omp): Likewise.
3692 (execute_lower_omp): Likewise.
3693 (diagnose_sb_0): Likewise.
3694 (gate_diagnose_omp_blocks): Likewise.
3695 (simd_clone_clauses_extract): Likewise.
3698 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3700 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3701 correction for little endian...
3702 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3705 2014-01-24 Jeff Law <law@redhat.com>
3707 PR tree-optimization/59919
3708 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3709 for non-returning calls.
3711 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3713 * common/config/aarch64/aarch64-common.c
3714 (aarch64_rewrite_mcpu): Handle multiple names.
3715 * config/aarch64/aarch64.h
3716 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3718 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3720 * input.c (add_file_to_cache_tab): Handle the case where fopen
3723 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3726 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3727 from push operand if code of push isn't PRE_DEC.
3729 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3732 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3733 -mquad-memory-atomic. Update -mquad-memory documentation to say
3734 it is only used for non-atomic loads/stores.
3736 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3737 -mquad-memory or -mquad-memory-atomic switches.
3739 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3740 -mquad-memory-atomic to ISA 2.07 support.
3742 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3743 to separate support of normal quad word memory operations (ldq, stq)
3744 from the atomic quad word memory operations.
3746 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3747 support to separate non-atomic quad word operations from atomic
3748 quad word operations. Disable non-atomic quad word operations in
3749 little endian mode so that we don't have to swap words after the
3750 load and before the store.
3751 (quad_load_store_p): Add comment about atomic quad word support.
3752 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3753 options printed with -mdebug=reg.
3755 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3756 -mquad-memory-atomic as the test for whether we have quad word
3757 atomic instructions.
3758 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3759 or -mp8-vector are used, allow byte/half-word atomic operations.
3761 * config/rs6000/sync.md (load_lockedti): Insure that the address
3762 is a proper indexed or indirect address for the lqarx instruction.
3763 On little endian systems, swap the hi/lo registers after the lqarx
3765 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3766 insure the address is valid for the lqarx instruction.
3767 (store_conditionalti): Insure that the address is a proper indexed
3768 or indirect address for the stqcrx. instruction. On little endian
3769 systems, swap the hi/lo registers before doing the stqcrx.
3771 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3772 insure the address is valid for the stqcrx. instruction.
3774 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3775 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3776 type of quad memory support is available.
3778 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3781 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3782 there is a danger of looping.
3784 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3786 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3787 force flag_ira_loop_pressure if set via command line.
3789 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3791 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3792 (ashr_simd): New builtin handling DI mode.
3793 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3794 (aarch64_sshr_simddi): New match pattern.
3795 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3796 (vshrd_n_s64): Likewise.
3797 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3799 2014-01-23 Nick Clifton <nickc@redhat.com>
3801 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3802 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3803 favour of mcu specific scripts.
3804 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3807 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3808 Alex Velenko <Alex.Velenko@arm.com>
3810 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3811 (vaddv_s16): Likewise.
3812 (vaddv_s32): Likewise.
3813 (vaddv_u8): Likewise.
3814 (vaddv_u16): Likewise.
3815 (vaddv_u32): Likewise.
3816 (vaddvq_s8): Likewise.
3817 (vaddvq_s16): Likewise.
3818 (vaddvq_s32): Likewise.
3819 (vaddvq_s64): Likewise.
3820 (vaddvq_u8): Likewise.
3821 (vaddvq_u16): Likewise.
3822 (vaddvq_u32): Likewise.
3823 (vaddvq_u64): Likewise.
3824 (vaddv_f32): Likewise.
3825 (vaddvq_f32): Likewise.
3826 (vaddvq_f64): Likewise.
3827 (vmaxv_f32): Likewise.
3828 (vmaxv_s8): Likewise.
3829 (vmaxv_s16): Likewise.
3830 (vmaxv_s32): Likewise.
3831 (vmaxv_u8): Likewise.
3832 (vmaxv_u16): Likewise.
3833 (vmaxv_u32): Likewise.
3834 (vmaxvq_f32): Likewise.
3835 (vmaxvq_f64): Likewise.
3836 (vmaxvq_s8): Likewise.
3837 (vmaxvq_s16): Likewise.
3838 (vmaxvq_s32): Likewise.
3839 (vmaxvq_u8): Likewise.
3840 (vmaxvq_u16): Likewise.
3841 (vmaxvq_u32): Likewise.
3842 (vmaxnmv_f32): Likewise.
3843 (vmaxnmvq_f32): Likewise.
3844 (vmaxnmvq_f64): Likewise.
3845 (vminv_f32): Likewise.
3846 (vminv_s8): Likewise.
3847 (vminv_s16): Likewise.
3848 (vminv_s32): Likewise.
3849 (vminv_u8): Likewise.
3850 (vminv_u16): Likewise.
3851 (vminv_u32): Likewise.
3852 (vminvq_f32): Likewise.
3853 (vminvq_f64): Likewise.
3854 (vminvq_s8): Likewise.
3855 (vminvq_s16): Likewise.
3856 (vminvq_s32): Likewise.
3857 (vminvq_u8): Likewise.
3858 (vminvq_u16): Likewise.
3859 (vminvq_u32): Likewise.
3860 (vminnmv_f32): Likewise.
3861 (vminnmvq_f32): Likewise.
3862 (vminnmvq_f64): Likewise.
3864 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3866 * config/aarch64/aarch64-simd.md
3867 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3868 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3869 (*aarch64_mul3_elt<mode>): Likewise.
3870 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3871 (*aarch64_mul3_elt_to_64v2df): Likewise.
3872 (*aarch64_mla_elt<mode>): Likewise.
3873 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3874 (*aarch64_mls_elt<mode>): Likewise.
3875 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3876 (*aarch64_fma4_elt<mode>): Likewise.
3877 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3878 (*aarch64_fma4_elt_to_64v2df): Likewise.
3879 (*aarch64_fnma4_elt<mode>): Likewise.
3880 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3881 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3882 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3883 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3884 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3885 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3886 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3887 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3888 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3890 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3892 * config/aarch64/aarch64-simd.md
3893 (aarch64_be_checked_get_lane<mode>): New define_expand.
3894 * config/aarch64/aarch64-simd-builtins.def
3895 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3896 New builtin definition.
3897 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3898 Use new safe be builtin.
3900 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3902 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3904 (aarch64_be_st1<mode>): Likewise.
3905 (aarch_ld1<VALL:mode>): Define_expand modified.
3906 (aarch_st1<VALL:mode>): Likewise.
3907 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3908 (UNSPEC_ST1): Likewise.
3910 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3912 * config/microblaze/microblaze.md: Add trap insn and attribute
3914 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3916 PR preprocessor/58580
3917 * input.h (location_get_source_line): Take an additional line_size
3919 (void diagnostics_file_cache_fini): Declare new function.
3920 * input.c (struct fcache): New type.
3921 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3922 New static constants.
3923 (diagnostic_file_cache_init, total_lines_num)
3924 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3925 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3926 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3927 (get_next_line, read_next_line, goto_next_line, read_line_num):
3928 New static function definitions.
3929 (diagnostic_file_cache_fini): New function.
3930 (location_get_source_line): Take an additional output line_len
3931 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3933 * diagnostic.c (diagnostic_finish): Call
3934 diagnostic_file_cache_fini.
3935 (adjust_line): Take an additional input parameter for the length
3936 of the line, rather than calculating it with strlen.
3937 (diagnostic_show_locus): Adjust the use of
3938 location_get_source_line and adjust_line with respect to their new
3939 signature. While displaying a line now, do not stop at the first
3940 null byte. Rather, display the zero byte as a space and keep
3941 going until we reach the size of the line.
3942 * Makefile.in: Add vec.o to OBJS-libcommon
3944 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3945 Ilya Tocar <ilya.tocar@intel.com>
3947 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3948 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3949 (__builtin_ia32_kmov16): Ditto.
3950 * config/i386/i386.md (UNSPEC_KMOV): New.
3953 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3955 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3956 (_mm512_storeu_si512): Ditto.
3958 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3961 * rtl.h (get_referenced_operands): Declare.
3962 * recog.c (get_referenced_operands): New function.
3963 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3964 operands have been referenced when recording LO_SUM references.
3966 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3968 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3970 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3972 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3973 Enable for generic and recent AMD targets.
3975 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3977 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3978 ARG_SIZE note when adjustment was eliminated.
3980 2014-01-22 Jeff Law <law@redhat.com>
3982 PR tree-optimization/59597
3983 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3984 in file. Accept new argument REGISTERING and use it to modify
3985 dump output appropriately.
3986 (register_jump_thread): Corresponding changes.
3987 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3988 thread paths involving joiner blocks. Add code to dump cancelled
3989 jump threading paths.
3991 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3993 PR rtl-optimization/59477
3994 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3995 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3997 2014-01-22 Tom Tromey <tromey@redhat.com>
3999 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4001 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4003 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4005 PR rtl-optimization/59896
4006 * lra-constraints.c (process_alt_operands): Check unused note for
4007 matched operands of insn with no output reloads.
4009 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4011 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4012 (mips_move_from_gpr_cost): Likewise.
4014 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4016 PR rtl-optimization/59858
4017 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4018 ira_class_hard_regs_num.
4019 (process_alt_operands): Increase reject for dying matched operand.
4021 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4024 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4025 smaller than size, perform several stores or loads and stores
4026 at dst + count - size to store or copy all of size bytes, rather
4027 than just last modesize bytes.
4029 2014-01-20 DJ Delorie <dj@redhat.com>
4031 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4032 that CLOBBERs are REGs before propogating their values.
4034 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4037 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4038 (cgraph_inline_failed_type): New function.
4039 * cgraph.h (DEFCIFCODE): Add type.
4040 (cgraph_inline_failed_type_t): New enum.
4041 (cgraph_inline_failed_type): New prototype.
4042 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4043 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4044 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4045 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4046 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4047 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4048 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4049 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4050 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4051 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4052 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4053 OPTIMIZATION_MISMATCH.
4054 * tree-inline.c (expand_call_inline): Emit errors during
4055 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4057 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4060 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4061 mode attribute in insn output.
4063 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4065 * output.h (output_constant): Delete.
4066 * varasm.c (output_constant): Make private.
4068 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4070 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4072 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4075 * tree.h (fold_builtin_strcat): New prototype.
4076 * builtins.c (fold_builtin_strcat): No longer static. Add len
4077 argument, if non-NULL, don't call c_strlen. Optimize
4078 directly into __builtin_memcpy instead of __builtin_strcpy.
4079 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4080 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4082 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4084 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4085 for SImode_address_operand operands, having only a REG argument.
4087 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4089 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4090 loader name using mbig-endian.
4091 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4093 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4095 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4099 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4101 * config/aarch64/aarch64-protos.h
4102 (aarch64_cannot_change_mode_class_ptr): Declare.
4103 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4104 aarch64_cannot_change_mode_class_ptr): New.
4105 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4106 backend hook aarch64_cannot_change_mode_class.
4108 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4110 * common/config/aarch64/aarch64-common.c
4111 (aarch64_handle_option): Don't handle any option order logic here.
4112 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4113 selected_cpu, warn on architecture version mismatch.
4114 (aarch64_override_options): Fix parsing order for option strings.
4116 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4117 Iain Sandoe <iain@codesourcery.com>
4120 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4121 warning. Amend comment to reflect current functionality.
4123 2014-01-20 Richard Biener <rguenther@suse.de>
4126 * builtins.c (fold_builtin_strcat): Remove case better handled
4127 by tree-ssa-strlen.c.
4129 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4131 * config/aarch64/aarch64.opt
4132 (mcpu, march, mtune): Make case-insensitive.
4134 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4137 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4138 if operands[1] is a REG or ZERO_EXTEND of a REG.
4140 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4142 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4144 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4146 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4147 long non-pic millicode calls.
4149 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4151 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4153 2014-01-19 Kito Cheng <kito@0xlab.org>
4155 * builtins.c (expand_movstr): Check movstr expand done or fail.
4157 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4158 H.J. Lu <hongjiu.lu@intel.com>
4161 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4162 to DImode for zero-extended addresses.
4164 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4166 PR rtl-optimization/57763
4167 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4168 on the new indirect jump_insn and increment LABEL_NUSES (label).
4170 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4174 * config.gcc (x86_archs): New variable.
4175 (x86_64_archs): Likewise.
4176 (x86_cpus): Likewise.
4177 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4178 --with-arch/--with-cpu= options.
4179 Support --with-arch=/--with-cpu={nehalem,westmere,
4180 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4182 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4184 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4185 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4187 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4189 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4191 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4194 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4195 clear cpp_get_options (parse_in)->warn_unused_macros for
4196 ix86_target_macros_internal with cpp_define.
4198 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4200 * jump.c (delete_related_insns): Keep (use (insn))s.
4201 * reorg.c (redundant_insn): Check for barriers too.
4203 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4205 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4207 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4209 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4210 call to $$dyncall when TARGET_LONG_CALLS is true.
4212 2014-01-17 Jeff Law <law@redhat.com>
4214 * ree.c (combine_set_extension): Temporarily disable test for
4215 changing number of hard registers.
4217 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4220 * ipa-inline-analysis.c (inline_free_summary):
4221 Do not free summary of aliases.
4223 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4226 * gimplify.c (gimplify_expr): Use create_tmp_var
4227 instead of create_tmp_var_raw. If cond doesn't have
4228 integral type, don't add the IFN_ANNOTATE builtin at all.
4230 2014-01-17 Martin Jambor <mjambor@suse.cz>
4233 * ipa-cp.c (prev_edge_clone): New variable.
4234 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4235 Also resize prev_edge_clone vector.
4236 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4237 (ipcp_edge_removal_hook): New function.
4238 (ipcp_driver): Register ipcp_edge_removal_hook.
4240 2014-01-17 Andrew Pinski <apinski@cavium.com>
4241 Steve Ellcey <sellcey@mips.com>
4244 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4247 2014-01-17 Jeff Law <law@redhat.com>
4250 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4251 so that pass_ccp runs first.
4253 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4255 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4256 (ix86_adjust_cost): Use !TARGET_XXX.
4257 (do_reorder_for_imul): Likewise.
4258 (swap_top_of_ready_list): Likewise.
4259 (ix86_sched_reorder): Likewise.
4261 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4263 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4264 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4265 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4266 (intel_memset): New. Duplicate slm_memset.
4267 (intel_cost): New. Duplicate slm_cost.
4268 (m_INTEL): New macro.
4269 (processor_target_table): Add "intel".
4270 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4271 with PROCESSOR_INTEL for "intel".
4272 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4273 PROCESSOR_SILVERMONT.
4274 (ix86_issue_rate): Likewise.
4275 (ix86_adjust_cost): Likewise.
4276 (ia32_multipass_dfa_lookahead): Likewise.
4277 (swap_top_of_ready_list): Likewise.
4278 (ix86_sched_reorder): Likewise.
4279 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4280 instead of TARGET_OPT_AGU.
4281 * config/i386/i386.h (TARGET_INTEL): New.
4282 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4283 (processor_type): Add PROCESSOR_INTEL.
4284 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4285 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4287 2014-01-17 Marek Polacek <polacek@redhat.com>
4290 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4293 2014-01-17 Richard Biener <rguenther@suse.de>
4295 PR tree-optimization/46590
4296 * opts.c (default_options_table): Add entries for
4297 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4298 all enabled at -O1 but not for -Og.
4299 * common.opt (fbranch-count-reg): Remove Init(1).
4300 (fmove-loop-invariants): Likewise.
4301 (ftree-pta): Likewise.
4303 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4305 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4306 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4307 decls to at least the GCC 4.8 used alignments.
4310 * tree-nested.c (convert_nonlocal_reference_stmt,
4311 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4312 of GIMPLE_BIND stmts, adjust associated decls.
4314 2014-01-17 Richard Biener <rguenther@suse.de>
4316 PR tree-optimization/46590
4317 * vec.h (vec<>::bseach): New member function implementing
4318 binary search according to C89 bsearch.
4319 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4320 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4321 bitmap pointer again. Make accesses_in_loop a flat array.
4322 (mem_ref_obstack): New global.
4323 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4324 (mark_ref_stored): Likewise.
4325 (ref_indep_loop_p_2): Likewise.
4326 (set_ref_stored_in_loop): New helper function.
4327 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4328 (memref_free): Adjust.
4329 (record_mem_ref_loc): Simplify.
4330 (gather_mem_refs_stmt): Adjust.
4331 (sort_locs_in_loop_postorder_cmp): New function.
4332 (analyze_memory_references): Sort accesses_in_loop after
4333 loop postorder number.
4334 (find_ref_loc_in_loop_cmp): New function.
4335 (for_all_locs_in_loop): Find relevant cluster of locs in
4336 accesses_in_loop and iterate without recursion.
4337 (execute_sm): Avoid uninit warning.
4338 (struct ref_always_accessed): Simplify.
4339 (ref_always_accessed::operator ()): Likewise.
4340 (ref_always_accessed_p): Likewise.
4341 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4342 loop postorder numbers here.
4343 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4346 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4349 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4350 on decls for which assemble_alias has been called.
4352 2014-01-17 Nick Clifton <nickc@redhat.com>
4354 * config/msp430/msp430.opt: (mcpu): New option.
4355 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4356 (msp430_option_override): Parse target_cpu. If the MCU name
4357 matches a generic string, clear target_mcu.
4358 (msp430_attr): Allow numeric interrupt values up to 63.
4359 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4360 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4362 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4364 * config/msp430/msp430.md (popm): Use %J rather than %I.
4365 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4366 (addhi_cy_i): Use immediate_operand for operand 2.
4367 * doc/invoke.texi: Document -mcpu option.
4369 2014-01-17 Richard Biener <rguenther@suse.de>
4371 PR rtl-optimization/38518
4372 * df.h (df_analyze_loop): Declare.
4373 * df-core.c: Include cfgloop.h.
4374 (df_analyze_1): Split out main part of df_analyze.
4375 (df_analyze): Adjust.
4376 (loop_inverted_post_order_compute): New function.
4377 (loop_post_order_compute): Likewise.
4378 (df_analyze_loop): New function avoiding whole-function
4380 * loop-invariant.c (find_defs): Use df_analyze_loop.
4381 (find_invariants): Adjust.
4382 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4384 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4386 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4387 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4389 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4391 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4392 traversal when removing references.
4394 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4397 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4399 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4402 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4403 pushing a reload for an autoinc when we had previously reloaded an
4404 inner part of the address.
4406 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4408 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4410 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4411 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4412 when not giving up or versioning for alias only because of
4414 (vect_analyze_data_ref_dependences): Set to true.
4415 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4417 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4418 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4422 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4425 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4426 operand 0 predicate for gathers, use a new pseudo as subtarget.
4428 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4431 * lra-constraints.c (process_alt_operands): Add printing debug info.
4432 Check absence of input/output reloads for matched operands too.
4434 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4436 PR rtl-optimization/59835
4437 * ira.c (ira_init_register_move_cost): Increase cost for
4440 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4442 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4444 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4447 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4448 non-register objects. Use gen_(high/low)part more consistently.
4451 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4454 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4455 endian support, remove tests for WORDS_BIG_ENDIAN.
4456 (p8_mfvsrd_3_<mode>): Likewise.
4457 (reload_gpr_from_vsx<mode>): Likewise.
4458 (reload_gpr_from_vsxsf): Likewise.
4459 (p8_mfvsrd_4_disf): Likewise.
4461 2014-01-16 Richard Biener <rguenther@suse.de>
4463 PR rtl-optimization/46590
4464 * lcm.c (compute_antinout_edge): Use postorder iteration.
4465 (compute_laterin): Use inverted postorder iteration.
4467 2014-01-16 Nick Clifton <nickc@redhat.com>
4470 * varasm.c (output_constant): Return the number of bytes actually
4472 (output_constructor_array_range): Update the field size with the
4473 number of bytes emitted by output_constant.
4474 (output_constructor_regular_field): Likewise. Also do not
4475 complain if the total number of bytes emitted is now greater
4476 than the expected fieldpos.
4477 * output.h (output_constant): Update prototype and descriptive comment.
4479 2014-01-16 Marek Polacek <polacek@redhat.com>
4482 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4483 it is error_mark_node.
4485 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4487 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4488 VALID_AVX256_REG_OR_OI_MODE.
4490 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4492 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4493 current procedure should be profiled.
4495 2014-01-15 Andrew Pinski <apinski@cavium.com>
4497 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4498 of moving from/to the STACK_REG register class.
4500 2014-01-15 Richard Henderson <rth@redhat.com>
4503 * reginfo.c (global_regs_decl): Globalize.
4504 * rtl.h (global_regs_decl): Declare.
4505 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4506 reserved via global_regs.
4508 2014-01-15 Teresa Johnson <tejohnson@google.com>
4510 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4512 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4514 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4515 and vmulosh rather than call gen_vec_widen_smult_*.
4516 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4517 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4518 (vec_widen_smult_even_v16qi): Likewise.
4519 (vec_widen_umult_even_v8hi): Likewise.
4520 (vec_widen_smult_even_v8hi): Likewise.
4521 (vec_widen_umult_odd_v16qi): Likewise.
4522 (vec_widen_smult_odd_v16qi): Likewise.
4523 (vec_widen_umult_odd_v8hi): Likewise.
4524 (vec_widen_smult_odd_v8hi): Likewise.
4525 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4526 vmuloub rather than call gen_vec_widen_umult_*.
4527 (vec_widen_umult_lo_v16qi): Likewise.
4528 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4529 vmulosb rather than call gen_vec_widen_smult_*.
4530 (vec_widen_smult_lo_v16qi): Likewise.
4531 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4532 rather than call gen_vec_widen_umult_*.
4533 (vec_widen_umult_lo_v8hi): Likewise.
4534 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4535 rather than call gen_vec_widen_smult_*.
4536 (vec_widen_smult_lo_v8hi): Likewise.
4538 2014-01-15 Jeff Law <law@redhat.com>
4540 PR tree-optimization/59747
4541 * ree.c (find_and_remove_re): Properly handle case where a second
4542 eliminated extension requires widening a copy created for elimination
4543 of a prior extension.
4544 (combine_set_extension): Ensure that the number of hard regs needed
4545 for a destination register does not change when we widen it.
4547 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4549 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4550 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4551 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4552 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4553 (avr-*-rtems*): Likewise.
4554 (bfin*-rtems*): Likewise.
4555 (moxie-*-rtems*): Likewise.
4556 (h8300-*-rtems*): Likewise.
4557 (i[34567]86-*-rtems*): Likewise.
4558 (lm32-*-rtems*): Likewise.
4559 (m32r-*-rtems*): Likewise.
4560 (m68k-*-rtems*): Likewise.
4561 (microblaze*-*-rtems*): Likewise.
4562 (mips*-*-rtems*): Likewise.
4563 (powerpc-*-rtems*): Likewise.
4564 (sh-*-rtems*): Likewise.
4565 (sparc-*-rtems*): Likewise.
4566 (sparc64-*-rtems*): Likewise.
4567 (v850-*-rtems*): Likewise.
4568 (m32c-*-rtems*): Likewise.
4570 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4572 PR rtl-optimization/59511
4573 * ira.c (ira_init_register_move_cost): Use memory costs for some
4574 cases of register move cost calculations.
4575 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4576 instead of BB frequency.
4577 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4578 * lra-assigns.c (find_hard_regno_for): Ditto.
4580 2014-01-15 Richard Biener <rguenther@suse.de>
4582 PR tree-optimization/59822
4583 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4584 (vectorizable_load): Use it to hoist defs of uses of invariant
4585 loads out of the loop.
4587 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4588 Kugan Vivekanandarajah <kuganv@linaro.org>
4591 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4594 2014-01-15 Richard Biener <rguenther@suse.de>
4596 PR rtl-optimization/59802
4597 * lcm.c (compute_available): Use inverted postorder to seed
4598 the initial worklist.
4600 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4603 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4604 ADDR_REGS for invalid symrefs in non-PIC code.
4606 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4609 * builtins.c (determine_block_size): Initialize *probable_max_size
4610 even if len_rtx is CONST_INT.
4612 2014-01-14 Andrew Pinski <apinski@cavium.com>
4614 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4615 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4616 (cortexa53_tunings): Likewise.
4617 (aarch64_sched_issue_rate): New function.
4618 (TARGET_SCHED_ISSUE_RATE): Define.
4620 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4622 * ira-costs.c (find_costs_and_classes): Add missed
4623 ira_init_register_move_cost_if_necessary.
4625 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4628 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4630 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4633 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4634 to indicate if type is used for function return value. Warn ABI
4635 change if the vector mode isn't available for function return value.
4636 (ix86_function_arg_advance): Pass false to type_natural_mode.
4637 (ix86_function_arg): Likewise.
4638 (ix86_gimplify_va_arg): Likewise.
4639 (function_arg_32): Don't warn ABI change.
4640 (ix86_function_value): Pass true to type_natural_mode.
4641 (ix86_return_in_memory): Likewise.
4642 (ix86_struct_value_rtx): Removed.
4643 (TARGET_STRUCT_VALUE_RTX): Likewise.
4645 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4647 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4648 converting a conditional jump into a conditional return.
4650 2014-01-14 Richard Biener <rguenther@suse.de>
4652 PR tree-optimization/58921
4653 PR tree-optimization/59006
4654 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4655 hoisting invariant stmts.
4656 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4657 invariant loads on the preheader edge if possible.
4659 2014-01-14 Joey Ye <joey.ye@arm.com>
4661 * doc/plugin.texi (Building GCC plugins): Update to C++.
4663 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4665 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4666 (_mm_rcp28_round_ss): Ditto.
4667 (_mm_rsqrt28_round_sd): Ditto.
4668 (_mm_rsqrt28_round_ss): Ditto.
4669 (_mm_rcp28_sd): Ditto.
4670 (_mm_rcp28_ss): Ditto.
4671 (_mm_rsqrt28_sd): Ditto.
4672 (_mm_rsqrt28_ss): Ditto.
4673 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4674 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4675 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4676 (IX86_BUILTIN_RCP28SD): Ditto.
4677 (IX86_BUILTIN_RCP28SS): Ditto.
4678 (IX86_BUILTIN_RSQRT28SD): Ditto.
4679 (IX86_BUILTIN_RSQRT28SS): Ditto.
4680 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4681 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4682 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4683 (ix86_expand_special_args_builtin): Expand new FTYPE.
4684 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4685 (srcp14<mode>): Make insn unary.
4686 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4687 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4688 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4689 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4690 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4691 Fix rounding: make it SAE only.
4692 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4694 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4696 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4697 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4698 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4699 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4700 (round_saeonly_mask_scalar_operand4): Ditto.
4701 (round_saeonly_mask_scalar_op3): Ditto.
4702 (round_saeonly_mask_scalar_op4): Ditto.
4704 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4706 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4707 Implement -maltivec=be for vec_insert and vec_extract.
4709 2014-01-10 DJ Delorie <dj@redhat.com>
4711 * config/msp430/msp430.md (call_internal): Don't allow memory
4712 references with SP as the base register.
4713 (call_value_internal): Likewise.
4714 * config/msp430/constraints.md (Yc): New. For memory references
4715 that don't use SP as a base register.
4717 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4718 "an integer without a # prefix"
4719 * config/msp430/msp430.md (epilogue_helper): Use it.
4721 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4724 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4725 AVX512F gather builtins.
4726 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4727 on gather decls with INTEGER_TYPE masktype.
4728 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4729 directly into the builtin rather than hoisting it before loop.
4731 PR tree-optimization/59387
4732 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4733 (scev_const_prop): If folded_casts and type has undefined overflow,
4734 use force_gimple_operand instead of force_gimple_operand_gsi and
4735 for each added stmt if it is assign with
4736 arith_code_with_undefined_signed_overflow, call
4737 rewrite_to_defined_overflow.
4738 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4739 gimple-fold.h instead.
4740 (arith_code_with_undefined_signed_overflow,
4741 rewrite_to_defined_overflow): Moved to ...
4742 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4743 rewrite_to_defined_overflow): ... here. No longer static.
4744 Include gimplify-me.h.
4745 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4746 rewrite_to_defined_overflow): New prototypes.
4748 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4750 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4752 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4754 * builtins.c (get_object_alignment_2): Minor tweak.
4755 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4757 2014-01-13 Christian Bruel <christian.bruel@st.com>
4759 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4760 optimized non constant lengths.
4762 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4765 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4766 load as __atomic_load_N if possible.
4768 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4770 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4772 (rs6000_expand_builtin): Adjust call.
4774 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4777 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4778 * config/rs6000/rs6000.c: Include target-globals.h.
4779 (rs6000_set_current_function): Instead of doing target_reinit
4780 unconditionally, use save_target_globals_default_opts and
4781 restore_target_globals.
4783 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4785 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4786 (rs6000_expand_builtin): Handle mffs and mtfsf.
4787 (rs6000_init_builtins): Define mffs and mtfsf.
4788 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4789 (rs6000_mffs): New pattern.
4790 (rs6000_mtfsf): New pattern.
4792 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4794 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4795 Start narrowing with START. Apply candidate-use pair
4796 and check overall cost in narrowing.
4797 (iv_ca_prune): Pass new argument.
4799 2014-01-10 Jeff Law <law@redhat.com>
4802 * ree.c (combine_reaching_defs): Ensure the defining statement
4803 occurs before the extension when optimizing extensions with
4804 different source and destination hard registers.
4806 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4809 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4810 vtables into the type inheritance graph.
4812 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4814 PR rtl-optimization/59754
4815 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4816 modes in the REGNO != REGNO case.
4818 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4820 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4822 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4824 PR tree-optimization/59745
4825 * tree-predcom.c (tree_predictive_commoning_loop): Call
4826 free_affine_expand_cache if giving up because components is NULL.
4828 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4829 GC in payload of target_globals struct instead of allocating them on
4830 the heap and the larger structs separately using GC.
4831 * target-globals.h (struct target_globals): Make regs, hard_regs,
4832 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4833 of GTY((skip)) and change type to void *.
4834 (reset_target_globals): Cast loads from those fields to corresponding
4837 2014-01-10 Steve Ellcey <sellcey@mips.com>
4840 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4841 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4842 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4844 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4847 * aarch64-modes.def (CC_Zmode): New flags mode.
4848 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4849 represents an equality.
4850 (aarch64_get_condition_code): Handle CC_Zmode.
4851 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4853 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4855 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4856 extraction in good case.
4858 2014-01-10 Richard Biener <rguenther@suse.de>
4860 PR tree-optimization/59374
4861 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4862 checking after SLP discovery. Mark stmts not participating
4863 in any SLP instance properly.
4865 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4867 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4868 when handling a SET rtx.
4870 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4872 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4873 (cortex-a57): Likewise.
4874 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4876 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4878 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4879 non-iwmmxt builtins.
4881 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4885 * ipa-devirt.c record_target_from_binfo): Take as argument
4886 stack of binfos and lookup matching one for virtual inheritance.
4887 (possible_polymorphic_call_targets_1): Update.
4889 2014-01-10 Huacai Chen <chenhc@lemote.com>
4891 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4892 kernel strings for Loongson-2E/2F/3A.
4894 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4897 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4898 is_gimple_call before calling gimple_call_internal_p.
4900 2014-01-09 Steve Ellcey <sellcey@mips.com>
4902 * Makefile.in (TREE_FLOW_H): Remove.
4903 (TREE_SSA_H): Add file names from tree-flow.h.
4904 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4905 * tree.h: Remove tree-flow.h reference.
4906 * hash-table.h: Remove tree-flow.h reference.
4907 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4908 reference with tree-ssa-loop.h.
4910 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4912 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4913 default element-order behavior for -maltivec.
4914 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4915 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4916 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4917 when targeting big endian, at least for now.
4918 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4920 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4923 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4924 var satisfies use_register_for_decl, just take into account type
4925 alignment, rather than decl alignment.
4927 PR tree-optimization/59622
4928 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4929 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4930 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4931 Don't devirtualize for inplace at all. For targets.length () == 1,
4932 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4934 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4936 * config/i386/i386.md (cpu): Remove the unused btver1.
4938 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4940 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4942 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4945 * tree-core.h (struct target_globals): New forward declaration.
4946 (struct tree_target_option): Add globals field.
4947 * tree.h (TREE_TARGET_GLOBALS): Define.
4948 (prepare_target_option_nodes_for_pch): New prototype.
4949 * target-globals.h (struct target_globals): Define even if
4951 * tree.c (prepare_target_option_node_for_pch,
4952 prepare_target_option_nodes_for_pch): New functions.
4953 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4954 * config/i386/i386.c: Include target-globals.h.
4955 (ix86_set_current_function): Instead of doing target_reinit
4956 unconditionally, use save_target_globals_default_opts and
4957 restore_target_globals.
4959 2014-01-09 Richard Biener <rguenther@suse.de>
4961 PR tree-optimization/59715
4962 * tree-cfg.h (split_critical_edges): Declare.
4963 * tree-cfg.c (split_critical_edges): Export.
4964 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4966 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4968 * cfgexpand.c (expand_stack_vars): Optionally disable
4969 asan stack protection.
4970 (expand_used_vars): Likewise.
4971 (partition_stack_vars): Likewise.
4972 * asan.c (asan_emit_stack_protection): Optionally disable
4973 after return stack usage.
4974 (instrument_derefs): Optionally disable memory access instrumentation.
4975 (instrument_builtin_call): Likewise.
4976 (instrument_strlen_call): Likewise.
4977 (asan_protect_global): Optionally disable global variables protection.
4978 * doc/invoke.texi: Added doc for new options.
4979 * params.def: Added new options.
4980 * params.h: Likewise.
4982 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4984 PR rtl-optimization/59724
4985 * ifcvt.c (cond_exec_process_if_block): Don't call
4986 flow_find_head_matching_sequence with 0 longest_match.
4987 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4988 non-active insns if !stop_after.
4989 (try_head_merge_bb): Revert 2014-01-07 changes.
4991 2014-01-08 Jeff Law <law@redhat.com>
4993 * ree.c (get_sub_rtx): New function, extracted from...
4994 (merge_def_and_ext): Here.
4995 (combine_reaching_defs): Use get_sub_rtx.
4997 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4999 * cgraph.h (varpool_variable_node): Do not choke on null node.
5001 2014-01-08 Catherine Moore <clm@codesourcery.com>
5003 * config/mips/mips.md (simple_return): Attempt to use JRC
5005 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5007 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5009 PR rtl-optimization/59137
5010 * reorg.c (steal_delay_list_from_target): Call update_block for
5012 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5014 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5016 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5017 two duplicate entries.
5019 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5022 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5024 * config/mips/mips.c (mips_truncated_op_cost): New function.
5025 (mips_rtx_costs): Adjust test for BADDU.
5026 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5028 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5030 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5031 (*baddu_si): ...this new pattern.
5033 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5036 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5038 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5041 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5043 (expand_expr, expand_normal): Adjust.
5044 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5045 inner_reference_p. Use inner_reference_p to expand inner references.
5046 (store_expr): Adjust.
5047 * cfgexpand.c (expand_call_stmt): Adjust.
5049 2014-01-08 Rong Xu <xur@google.com>
5051 * gcov-io.c (gcov_var): Move from gcov-io.h.
5052 (gcov_position): Ditto.
5053 (gcov_is_error): Ditto.
5054 (gcov_rewrite): Ditto.
5055 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5056 only part to libgcc/libgcov.h.
5058 2014-01-08 Marek Polacek <polacek@redhat.com>
5061 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5063 2014-01-08 Marek Polacek <polacek@redhat.com>
5066 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5068 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5070 PR rtl-optimization/59649
5071 * stor-layout.c (get_mode_bounds): For BImode return
5072 0 and STORE_FLAG_VALUE.
5074 2014-01-08 Richard Biener <rguenther@suse.de>
5077 * gimple.h (is_gimple_builtin_call): Remove.
5078 (gimple_builtin_call_types_compatible_p): New.
5079 (gimple_call_builtin_p): New overload.
5080 * gimple.c (is_gimple_builtin_call): Remove.
5081 (validate_call): Rename to ...
5082 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5084 (validate_type): New static function.
5085 (gimple_call_builtin_p): New overload and adjust.
5086 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5087 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5088 (gimple_fold_stmt_to_constant_1): Likewise.
5089 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5091 2014-01-08 Richard Biener <rguenther@suse.de>
5094 * gimplify.c (gimplify_expr): Gimplify register-register type
5095 VIEW_CONVERT_EXPRs to separate stmts.
5097 2014-01-07 Jeff Law <law@redhat.com>
5100 * ree.c (combine_set_extension): Handle case where source
5101 and destination registers in an extension insn are different.
5102 (combine_reaching_defs): Allow source and destination registers
5103 in extension to be different under limited circumstances.
5104 (add_removable_extension): Remove restriction that the
5105 source and destination registers in the extension are the same.
5106 (find_and_remove_re): Emit a copy from the extension's
5107 destination to its source after the defining insn if
5108 the source and destination registers are different.
5111 * ifcvt.c (merge_if_block): If we are merging a block with more than
5112 one successor with a block with no successors, remove any BARRIER
5113 after the second block.
5115 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5117 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5119 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5122 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5123 for 14-bit register offsets when INT14_OK_STRICT is false.
5125 2014-01-07 Roland Stigge <stigge@antcom.de>
5126 Michael Meissner <meissner@linux.vnet.ibm.com>
5129 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5130 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5132 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5134 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5137 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5139 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5140 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5141 rtx is const0_rtx or not.
5143 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5146 * target-globals.c (save_target_globals): Remove this_fn_optab
5148 * toplev.c: Include optabs.h.
5149 (target_reinit): Temporarily restore the global options if another
5150 set of options are in force.
5152 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5154 PR rtl-optimization/58668
5155 * cfgcleanup.c (flow_find_cross_jump): Don't count
5156 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5157 to determine what is counted.
5158 (flow_find_head_matching_sequence): Use active_insn_p to determine
5160 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5162 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5163 determine what is counted.
5165 PR tree-optimization/59643
5166 * tree-predcom.c (split_data_refs_to_components): If one dr is
5167 read and one write, determine_offset fails and the write isn't
5168 in the bad component, just put the read into the bad component.
5170 2014-01-07 Mike Stump <mikestump@comcast.net>
5171 Jakub Jelinek <jakub@redhat.com>
5174 * tree-core.h (struct tree_optimization_option): Change optabs
5175 type from unsigned char * to void *.
5176 * optabs.c (init_tree_optimization_optabs): Adjust
5177 TREE_OPTIMIZATION_OPTABS initialization.
5179 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5182 * config/i386/i386.h (struct machine_function): Add
5183 no_drap_save_restore field.
5184 * config/i386/i386.c (ix86_save_reg): Use
5185 !cfun->machine->no_drap_save_restore instead of
5186 crtl->stack_realign_needed.
5187 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5188 this function clears frame_pointer_needed. Set
5189 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5190 and DRAP reg is needed.
5192 2014-01-06 Marek Polacek <polacek@redhat.com>
5195 * doc/implement-c.texi: Mention that other integer types are
5196 permitted as bit-field types in strictly conforming mode.
5198 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5200 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5203 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5205 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5207 2014-01-06 Martin Jambor <mjambor@suse.cz>
5210 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5212 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5214 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5218 * var-tracking.c (add_stores): Preserve the value of the source even if
5219 we don't record the store.
5221 2014-01-06 Terry Guo <terry.guo@arm.com>
5223 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5225 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5228 * config/darwin.c (darwin_function_section): Adjust return values to
5229 correspond to optimisation changes made in r206070.
5231 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5233 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5234 from prefetch_block tune setting.
5235 (nocona_cost): Correct size of prefetch block to 64.
5237 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5239 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5240 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5241 used to save the static chain register in the computation of the offset
5242 from which the FP registers need to be restored.
5244 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5246 PR tree-optimization/59519
5247 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5248 ICE if get_current_def (current_new_name) is already non-NULL, as long
5249 as it is a phi result of some other phi in *new_exit_bb that has
5252 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5253 or vmovdqu* for misaligned_operand.
5254 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5255 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5256 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5257 aligned_mem for AVX512F masked aligned load and store builtins and for
5260 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5262 PR tree-optimization/59651
5263 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5264 Address range for negative step should be added by TYPE_SIZE_UNIT.
5266 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5268 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5269 overlapping registers also for registers other than the stack pointer.
5271 2014-01-03 Marek Polacek <polacek@redhat.com>
5274 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5277 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5280 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5283 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5284 (push splitter): Use <P:MODE_SIZE> instead of
5285 GET_MODE_SIZE (<P:MODE>mode).
5286 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5287 (mov -1, reg peephole2): Likewise.
5288 * config/i386/sse.md (*mov<mode>_internal,
5289 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5290 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5291 *<code><mode>3, *andnot<mode>3<mask_name>,
5292 <mask_codefor><code><mode>3<mask_name>): Likewise.
5293 * config/i386/subst.md (mask_mode512bit_condition,
5294 sd_mask_mode512bit_condition): Likewise.
5296 2014-01-02 Xinliang David Li <davidxl@google.com>
5298 PR tree-optimization/59303
5299 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5300 (dump_predicates): Better output format.
5301 (pred_equal_p): New function.
5302 (is_neq_relop_p): Ditto.
5303 (is_neq_zero_form_p): Ditto.
5304 (pred_expr_equal_p): Ditto.
5305 (pred_neg_p): Ditto.
5306 (simplify_pred): Ditto.
5307 (simplify_preds_2): Ditto.
5308 (simplify_preds_3): Ditto.
5309 (simplify_preds_4): Ditto.
5310 (simplify_preds): Ditto.
5312 (push_to_worklist): Ditto.
5313 (get_pred_info_from_cmp): Ditto.
5314 (is_degenerated_phi): Ditto.
5315 (normalize_one_pred_1): Ditto.
5316 (normalize_one_pred): Ditto.
5317 (normalize_one_pred_chain): Ditto.
5318 (normalize_preds): Ditto.
5319 (normalize_cond_1): Remove function.
5320 (normalize_cond): Ditto.
5321 (is_gcond_subset_of): Ditto.
5322 (is_subset_of_any): Ditto.
5323 (is_or_set_subset_of): Ditto.
5324 (is_and_set_subset_of): Ditto.
5325 (is_norm_cond_subset_of): Ditto.
5326 (pred_chain_length_cmp): Ditto.
5327 (convert_control_dep_chain_into_preds): Type change.
5328 (find_predicates): Ditto.
5329 (find_def_preds): Ditto.
5330 (destroy_predicates_vecs): Ditto.
5331 (find_matching_predicates_in_rest_chains): Ditto.
5332 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5333 (is_pred_expr_subset): Ditto.
5334 (is_pred_chain_subset_of): Ditto.
5335 (is_included_in): Ditto.
5336 (is_superset_of): Ditto.
5338 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5340 Update copyright years.
5342 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5344 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5345 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5346 config/arc/arc.md, config/arc/arc.opt,
5347 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5348 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5349 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5350 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5351 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5352 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5353 vtable-verify.c, vtable-verify.h: Use the standard form for the
5356 2014-01-02 Tobias Burnus <burnus@net-b.de>
5358 * gcc.c (process_command): Update copyright notice dates.
5359 * gcov-dump.c: Ditto.
5361 * doc/cpp.texi: Bump @copying's copyright year.
5362 * doc/cppinternals.texi: Ditto.
5363 * doc/gcc.texi: Ditto.
5364 * doc/gccint.texi: Ditto.
5365 * doc/gcov.texi: Ditto.
5366 * doc/install.texi: Ditto.
5367 * doc/invoke.texi: Ditto.
5369 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5371 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5373 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5375 * config/i386/sse.md (*mov<mode>_internal): Guard
5376 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5378 PR rtl-optimization/59647
5379 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5380 new_rtx into UNSIGNED_FLOAT rtxes.
5382 Copyright (C) 2014 Free Software Foundation, Inc.
5384 Copying and distribution of this file, with or without modification,
5385 are permitted in any medium without royalty provided the copyright
5386 notice and this notice are preserved.