1 2014-03-25 Richard Biener <rguenther@suse.de>
4 * gimplify-me.c (gimple_regimplify_operands): Update the
7 2014-03-25 Martin Jambor <mjambor@suse.cz>
10 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
11 (lto_output_varpool_node): Likewise.
12 (input_overwrite_node): Likewise.
13 (input_varpool_node): Likewise.
15 2014-03-25 Richard Biener <rguenther@suse.de>
17 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like
21 2014-03-25 Jakub Jelinek <jakub@redhat.com>
23 * combine.c (simplify_compare_const): Add MODE argument.
24 Handle mode_width 0 as very large mode_width.
25 (try_combine, simplify_comparison): Adjust callers.
27 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
28 type to avoid signed integer overflow.
29 * explow.c (plus_constant): Likewise.
31 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33 * doc/generic.texi: Correct typos.
35 2014-03-24 Tobias Burnus <burnus@net-b.de>
37 * doc/invoke.texi (-flto): Expand section about
38 using static libraries with LTO.
40 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
42 PR rtl-optimization/60501
43 * optabs.def (addptr3_optab): New optab.
44 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
45 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
46 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
48 * lra.c (emit_add3_insn): Use the addptr pattern if available.
50 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
52 2014-03-24 Ulrich Drepper <drepper@gmail.com>
54 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
57 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
58 (_mm256_undefined_ps): Define.
59 (_mm256_undefined_pd): Define.
60 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
61 (_mm_undefined_pd): Define.
62 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
63 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
64 (_mm512_undefined_ps): Define.
65 (_mm512_undefined_pd): Define.
67 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
69 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
71 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
72 (lshr_simd): DI mode added.
73 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
74 (aarch64_ushr_simddi): Likewise.
75 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
76 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
77 (vshrd_n_u64): Likewise.
79 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
81 * Makefile.in (s-macro_list): Depend on cc1.
83 2014-03-23 Teresa Johnson <tejohnson@google.com>
85 * ipa-utils.c (ipa_print_order): Use specified dump file.
87 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
89 PR rtl-optimization/60601
90 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
92 * gcc.c (eval_spec_function): Initialize save_growing_value.
94 2014-03-22 Jakub Jelinek <jakub@redhat.com>
97 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
98 code == MINUS_EXPR, never swap op0 with op1.
100 * toplev.c (init_local_tick): Avoid signed integer multiplication
102 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
103 shift by first operand's bitsize.
105 2014-03-21 Jakub Jelinek <jakub@redhat.com>
108 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
110 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
111 TARGET_ISA_64BIT_P(x).
113 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
115 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
116 pattern for vector nor instead of subtract from splat(-1).
117 (altivec_expand_vec_perm_const_le): Likewise.
119 2014-03-21 Richard Henderson <rth@twiddle.net>
122 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
123 related insns after epilogue_completed.
125 2014-03-21 Martin Jambor <mjambor@suse.cz>
128 * cgraph.h (symtab_node): New flag body_removed.
129 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
130 when removing bodies.
131 * symtab.c (dump_symtab_base): Dump body_removed flag.
132 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
133 had their bodies removed.
135 2014-03-21 Martin Jambor <mjambor@suse.cz>
138 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
141 2014-03-21 Richard Biener <rguenther@suse.de>
143 PR tree-optimization/60577
144 * tree-core.h (struct tree_base): Document nothrow_flag use
145 in VAR_DECL_NONALIASED.
146 * tree.h (VAR_DECL_NONALIASED): New.
147 (may_be_aliased): Adjust.
148 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
150 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
152 * expr.c (expand_expr_real_1): Remove outdated comment.
154 2014-03-20 Jakub Jelinek <jakub@redhat.com>
157 * ira.c (adjust_cleared_regs): Call copy_rtx on
158 *reg_equiv[REGNO (loc)].src_p before passing it to
159 simplify_replace_fn_rtx.
162 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
163 into CONST, put pic register as first operand of PLUS. Use
164 gen_const_mem for both 32-bit and 64-bit PIC got loads.
166 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
168 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
170 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
172 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
173 around for store forwarding issue in the FPU on the UT699.
174 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
175 loads and operations if -mfix-ut699 is specified.
176 (divtf3_hq): Tweak attribute.
177 (sqrttf2_hq): Likewise.
179 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
181 * calls.c (store_one_arg): Remove incorrect const qualification on the
182 type of the temporary.
183 * cfgexpand.c (expand_return): Likewise.
184 * expr.c (expand_constructor): Likewise.
185 (expand_expr_real_1): Likewise.
187 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
189 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
192 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
195 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
197 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
199 * config/arm/aarch-common-protos.h
200 (alu_cost_table): Fix spelling of "extend".
201 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
203 2014-03-19 Richard Biener <rguenther@suse.de>
206 * tree-core.h (tree_type_common): Re-order pointer members
207 to reduce recursion depth during GC walks.
209 2014-03-19 Marek Polacek <polacek@redhat.com>
212 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
215 2014-03-19 Richard Biener <rguenther@suse.de>
218 * lto-streamer-in.c (input_function): In WPA stage do not drop
221 2014-03-19 Jakub Jelinek <jakub@redhat.com>
223 PR tree-optimization/60559
224 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
225 with build_zero_cst assignment.
227 2014-03-18 Kai Tietz <ktietz@redhat.com>
229 PR rtl-optimization/56356
230 * sdbout.c (sdbout_parms): Verify that parms'
231 incoming argument is valid.
232 (sdbout_reg_parms): Likewise.
234 2014-03-18 Richard Henderson <rth@redhat.com>
237 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
238 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
239 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
241 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
243 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
244 * doc/plugins.texi (Plugin callbacks): Mention
246 Italicize plugin event names in description. Explain that
247 PLUGIN_PRAGMAS has no sense for lto1. Explain
249 Remind that no GCC functions should be called after
251 Explain what pragmas with expansion are.
253 2014-03-18 Martin Liska <mliska@suse.cz>
255 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
256 gimple call statement is update.
257 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
258 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
260 2014-03-18 Jakub Jelinek <jakub@redhat.com>
263 * ubsan.c (ubsan_instrument_unreachable): Call
264 initialize_sanitizer_builtins.
265 (ubsan_pass): Likewise.
268 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
269 varpool_finalize_decl instead of rest_of_decl_compilation.
271 2014-03-18 Richard Biener <rguenther@suse.de>
273 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
274 by using bitmap_and_compl instead of bitmap_and_compl_into.
275 (df_rd_transfer_function): Likewise.
277 2014-03-18 Richard Biener <rguenther@suse.de>
279 * doc/lto.texi (fresolution): Fix typo.
281 2014-03-18 Richard Biener <rguenther@suse.de>
283 * doc/invoke.texi (flto): Update for changes in 4.9.
285 2014-03-18 Richard Biener <rguenther@suse.de>
287 * doc/loop.texi: Remove section on the removed lambda framework.
288 Update loop docs with recent changes in preserving loop structure.
290 2014-03-18 Richard Biener <rguenther@suse.de>
292 * doc/lto.texi (-fresolution): Document.
294 2014-03-18 Richard Biener <rguenther@suse.de>
296 * doc/contrib.texi: Adjust my name.
298 2014-03-18 Jakub Jelinek <jakub@redhat.com>
301 * internal-fn.c: Include diagnostic-core.h.
302 (expand_BUILTIN_EXPECT): New function.
303 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
304 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
305 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
306 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
308 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
309 Revert 3 argument __builtin_expect code.
310 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
311 * gimple-fold.c (gimple_fold_call): Likewise.
312 * tree.h (fold_builtin_expect): New prototype.
313 * builtins.c (build_builtin_expect_predicate): Add predictor
314 argument, if non-NULL, create 3 argument __builtin_expect.
315 (fold_builtin_expect): No longer static. Add ARG2 argument,
316 pass it through to build_builtin_expect_predicate.
317 (fold_builtin_2): Adjust caller.
318 (fold_builtin_3): Handle BUILT_IN_EXPECT.
319 * internal-fn.def (BUILTIN_EXPECT): New.
321 2014-03-18 Tobias Burnus <burnus@net-b.de>
324 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
325 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
326 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
328 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
331 * predict.c (combine_predictions_for_bb): Fix up formatting.
332 (expr_expected_value_1, expr_expected_value): Add predictor argument,
333 fill what it points to if non-NULL.
334 (tree_predict_by_opcode): Adjust caller, use the predictor.
335 * predict.def (PRED_COMPARE_AND_SWAP): Add.
337 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
339 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
340 proper constant for the store mode.
342 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
344 * symtab.c (change_decl_assembler_name): Fix transparent alias
347 2014-03-16 Renlin Li <Renlin.Li@arm.com>
349 * config/aarch64/aarch64.c: Correct the comments about the
350 aarch64 stack layout.
352 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
354 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
355 check for GF_OMP_FOR_KIND_FOR.
357 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
359 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
360 ymm and zmm register names.
362 2014-03-17 Jakub Jelinek <jakub@redhat.com>
365 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
366 note creation for the 2010-08-31 changes.
368 2014-03-17 Marek Polacek <polacek@redhat.com>
371 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
372 as -fno-tree-loop-vectorize.
373 (expand_omp_simd): Likewise.
375 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
377 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
378 (eligible_for_call_delay): New prototype.
379 * config/sparc/sparc.c (tls_call_delay): Rename into...
380 (eligible_for_call_delay): ...this. Return false if the instruction
381 cannot be put in the delay slot of a branch.
382 (eligible_for_restore_insn): Simplify.
383 (eligible_for_return_delay): Return false if the instruction cannot be
384 put in the delay slot of a branch and simplify.
385 (eligible_for_sibcall_delay): Return false if the instruction cannot be
386 put in the delay slot of a branch.
387 * config/sparc/sparc.md (fix_ut699): New attribute.
388 (tls_call_delay): Delete.
389 (in_call_delay): Reimplement.
390 (eligible_for_sibcall_delay): Rename into...
391 (in_sibcall_delay): ...this.
392 (eligible_for_return_delay): Rename into...
393 (in_return_delay): ...this.
394 (in_branch_delay): Reimplement.
395 (in_uncond_branch_delay): Delete.
396 (in_annul_branch_delay): Delete.
398 2014-03-14 Richard Henderson <rth@redhat.com>
401 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
402 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
403 (*floathi<X87MODEF>2_i387_with_temp): Remove.
404 (floathi splitters): Remove.
405 (float<SWI48x>xf2): New pattern.
406 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
407 code that tried to handle DImode for 32-bit, but which was excluded
408 by the pattern's condition. Drop allocation of stack temporary.
409 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
410 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
411 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
412 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
413 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
414 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
415 (*float<SWI48><MODEF>2_sse_interunit): Remove.
416 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
417 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
418 (*float<SWI48x><X87MODEF>2_i387): Remove.
419 (all float _with_temp splitters): Remove.
420 (*float<SWI48x><MODEF>2_i387): New pattern.
421 (*float<SWI48><MODEF>2_sse): New pattern.
422 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
423 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
425 2014-03-14 Jakub Jelinek <jakub@redhat.com>
426 Marek Polacek <polacek@redhat.com>
429 * common.opt (dump_base_name_prefixed): New Variable.
430 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
431 if x_dump_base_name_prefixed is already set, set it at the end.
433 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
435 PR rtl-optimization/60508
436 * lra-constraints.c (get_reload_reg): Add new parameter
438 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
439 Pass the new parameter values.
441 2014-03-14 Richard Biener <rguenther@suse.de>
443 * common.opt: Revert unintented changes from r205065.
446 2014-03-14 Richard Biener <rguenther@suse.de>
449 * cfghooks.c (split_block): Properly adjust all loops the
450 block was a latch of.
452 2014-03-14 Martin Jambor <mjambor@suse.cz>
455 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
458 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
461 * config/avr/avr.c (avr_set_current_function): Pass function name
462 through default_strip_name_encoding before sanity checking instead
463 of skipping the first char of the assembler name.
465 2014-03-13 Richard Henderson <rth@redhat.com>
468 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
469 (ix86_force_to_memory, ix86_free_from_memory): Remove.
470 * config/i386/i386-protos.h: Likewise.
471 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
472 in the expander instead of a splitter.
473 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
474 any possibility of requiring a memory.
475 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
476 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
477 (fp branch splitters): Update for ix86_split_fp_branch.
478 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
479 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
480 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
481 (*fop_<MODEF>_2_i387): Remove f/r alternative.
482 (*fop_<MODEF>_3_i387): Likewise.
483 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
484 (splitters for the fop_* register patterns): Remove.
485 (fscalexf4_i387): Rename from *fscalexf4_i387.
486 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
488 2014-03-13 Jakub Jelinek <jakub@redhat.com>
490 PR tree-optimization/59779
491 * tree-dfa.c (get_ref_base_and_extent): Use double_int
492 type for bitsize and maxsize instead of HOST_WIDE_INT.
494 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
496 PR rtl-optimization/57320
497 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
498 the CFG after thread_prologue_and_epilogue_insns.
500 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
502 PR rtl-optimization/57189
503 * lra-constraints.c (process_alt_operands): Disfavor spilling
506 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
508 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
511 2014-03-13 Jakub Jelinek <jakub@redhat.com>
513 PR tree-optimization/59025
515 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
516 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
518 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
521 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
522 calls of avr_out_plus_1.
524 2014-03-13 Bin Cheng <bin.cheng@arm.com>
526 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
527 BB's single pred and update the father loop's latch info later.
529 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
531 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
535 (VEC_base): Likewise.
536 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
537 registers, we need to swap double words in little endian mode.
539 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
540 to be a container mode for 128-bit integer operations added in ISA
541 2.07. Unlike TImode and PTImode, the preferred register set is
542 the Altivec/VMX registers for the 128-bit operations.
544 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
546 (rs6000_split_128bit_ok_p): Likewise.
548 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
549 macros for creating ISA 2.07 normal and overloaded builtin
550 functions with 3 arguments.
551 (BU_P8V_OVERLOAD_3): Likewise.
552 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
553 for use as overloaded functions.
554 (VPERM_1TI_UNS): Likewise.
555 (VSEL_1TI): Likewise.
556 (VSEL_1TI_UNS): Likewise.
557 (ST_INTERNAL_1ti): Likewise.
558 (LD_INTERNAL_1ti): Likewise.
559 (XXSEL_1TI): Likewise.
560 (XXSEL_1TI_UNS): Likewise.
561 (VPERM_1TI): Likewise.
562 (VPERM_1TI_UNS): Likewise.
563 (XXPERMDI_1TI): Likewise.
565 (LXVD2X_V1TI): Likewise.
566 (STXVD2X_V1TI): Likewise.
567 (VEC_INIT_V1TI): Likewise.
568 (VEC_SET_V1TI): Likewise.
569 (VEC_EXT_V1TI): Likewise.
570 (EQV_V1TI): Likewise.
571 (NAND_V1TI): Likewise.
572 (ORC_V1TI): Likewise.
573 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
574 added in ISA 2.07. Add both normal 'altivec' builtins, and the
578 (VADDEUQM): Likewise.
579 (VADDECUQ): Likewise.
580 (VSUBEUQM): Likewise.
581 (VSUBECUQ): Likewise.
583 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
584 __int128_t and __uint128_t types.
585 (__uint128_type): Likewise.
586 (altivec_categorize_keyword): Add support for vector __int128_t,
587 vector __uint128_t, vector __int128, and vector unsigned __int128
588 as a container type for TImode operations that need to be done in
589 VSX/Altivec registers.
590 (rs6000_macro_to_expand): Likewise.
591 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
592 to support 128-bit integer instructions vaddcuq, vadduqm,
593 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
594 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
596 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
597 for V1TImode, and set up preferences to use VSX/Altivec registers.
598 Setup VSX reload handlers.
599 (rs6000_debug_reg_global): Likewise.
600 (rs6000_init_hard_regno_mode_ok): Likewise.
601 (rs6000_preferred_simd_mode): Likewise.
602 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
603 (easy_altivec_constant): Likewise.
604 (output_vec_const_move): Likewise.
605 (rs6000_expand_vector_set): Convert V1TImode set and extract to
607 (rs6000_expand_vector_extract): Likewise.
608 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
610 (rs6000_const_vec): Add support for V1TImode.
611 (rs6000_emit_le_vsx_load): Swap double words when loading or
612 storing TImode/V1TImode.
613 (rs6000_emit_le_vsx_store): Likewise.
614 (rs6000_emit_le_vsx_move): Likewise.
615 (rs6000_emit_move): Add support for V1TImode.
616 (altivec_expand_ld_builtin): Likewise.
617 (altivec_expand_st_builtin): Likewise.
618 (altivec_expand_vec_init_builtin): Likewise.
619 (altivec_expand_builtin): Likewise.
620 (rs6000_init_builtins): Add support for V1TImode type. Add
621 support for ISA 2.07 128-bit integer builtins. Define type names
622 for the VSX/Altivec vector types.
623 (altivec_init_builtins): Add support for overloaded vector
624 functions with V1TImode type.
625 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
626 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
628 (rs6000_split_128bit_ok_p): Likewise.
629 (rs6000_handle_altivec_attribute): Create V1TImode from vector
630 __int128_t and vector __uint128_t.
632 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
640 (VS_scalar): Likewise.
641 (VS_double): Likewise.
642 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
644 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
645 we support the ISA 2.07 128-bit integer arithmetic instructions.
646 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
647 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
648 and TImode types for use with the builtin functions.
649 (V1TI_type_node): Likewise.
650 (unsigned_V1TI_type_node): Likewise.
651 (intTI_type_internal_node): Likewise.
652 (uintTI_type_internal_node): Likewise.
654 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
655 128-bit builtin functions.
656 (UNSPEC_VADDEUQM): Likewise.
657 (UNSPEC_VADDECUQ): Likewise.
658 (UNSPEC_VSUBCUQ): Likewise.
659 (UNSPEC_VSUBEUQM): Likewise.
660 (UNSPEC_VSUBECUQ): Likewise.
661 (VM): Add V1TImode to vector mode iterators.
664 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
665 (altivec_vaddcuq): Likewise.
666 (altivec_vsubuqm): Likewise.
667 (altivec_vsubcuq): Likewise.
668 (altivec_vaddeuqm): Likewise.
669 (altivec_vaddecuq): Likewise.
670 (altivec_vsubeuqm): Likewise.
671 (altivec_vsubecuq): Likewise.
673 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
675 (BOOL_128): Likewise.
676 (BOOL_REGS_OUTPUT): Likewise.
677 (BOOL_REGS_OP1): Likewise.
678 (BOOL_REGS_OP2): Likewise.
679 (BOOL_REGS_UNARY): Likewise.
680 (BOOL_REGS_AND_CR0): Likewise.
682 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
683 128-bit integer builtin support.
684 (vec_vadduqm): Likewise.
685 (vec_vaddecuq): Likewise.
686 (vec_vaddeuqm): Likewise.
687 (vec_vsubecuq): Likewise.
688 (vec_vsubeuqm): Likewise.
689 (vec_vsubcuq): Likewise.
690 (vec_vsubuqm): Likewise.
692 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
693 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
694 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
695 128-bit integer add/subtract to ISA 2.07.
697 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
699 * config/arc/arc.c (arc_predicate_delay_insns):
700 Fix third argument passed to conditionalize_nonjump.
702 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
704 * config/aarch64/aarch64-builtins.c
705 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
706 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
707 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
708 instead of __builtin_lfloor.
709 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
711 2014-03-12 Jakub Jelinek <jakub@redhat.com>
713 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
714 (tree_ssa_ifcombine_bb_1): New function.
715 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
716 is an empty forwarder block to then_bb or vice versa and then_bb
717 and else_bb are effectively swapped.
719 2014-03-12 Christian Bruel <christian.bruel@st.com>
722 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
723 REG_CFA_DEF_CFA note.
724 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
725 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
727 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
729 PR tree-optimization/60454
730 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
732 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
734 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
735 Do not define target_cpu_default2 to generic.
736 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
737 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
738 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
740 2014-03-12 Jakub Jelinek <jakub@redhat.com>
741 Marc Glisse <marc.glisse@inria.fr>
743 PR tree-optimization/60502
744 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
745 instead of build_low_bits_mask.
747 2014-03-12 Jakub Jelinek <jakub@redhat.com>
750 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
751 if there are multiple uses, but op doesn't live on E edge.
752 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
753 clobber stmts before __builtin_unreachable.
755 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
757 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
758 hard_frame_pointer_rtx.
759 * cse.c (cse_insn): Remove volatile check.
760 * cselib.c (cselib_process_insn): Likewise.
761 * dse.c (scan_insn): Likewise.
763 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
765 * config/arc/arc.c (conditionalize_nonjump): New function,
767 (arc_ifcvt): ... this.
768 (arc_predicate_delay_insns): Use it.
770 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
772 * config/arc/predicates.md (extend_operand): During/after reload,
773 allow const_int_operand.
774 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
775 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
776 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
778 (umulsi3_highpart_i): Likewise.
780 2014-03-11 Richard Biener <rguenther@suse.de>
782 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
783 Add asserts to guard possible wrong-code bugs.
785 2014-03-11 Richard Biener <rguenther@suse.de>
787 PR tree-optimization/60429
788 PR tree-optimization/60485
789 * tree-ssa-structalias.c (set_union_with_increment): Properly
790 take into account all fields that overlap the shifted vars.
791 (do_sd_constraint): Likewise.
792 (do_ds_constraint): Likewise.
793 (get_constraint_for_ptr_offset): Likewise.
795 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
797 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
798 (nios2_compute_frame_layout):
799 Add calculation of cfun->machine->fp_save_offset.
800 (nios2_expand_prologue): Correct setting of frame pointer register
802 (nios2_expand_epilogue): Update recovery of stack pointer from
803 frame pointer accordingly.
804 (nios2_initial_elimination_offset): Update calculation of offset
805 for eliminating to HARD_FRAME_POINTER_REGNUM.
807 2014-03-10 Jakub Jelinek <jakub@redhat.com>
810 * ipa.c (symtab_remove_unreachable_nodes): Don't call
811 cgraph_get_create_node on VAR_DECLs.
813 2014-03-10 Richard Biener <rguenther@suse.de>
816 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
818 2014-03-08 Douglas B Rupp <rupp@gnat.com>
820 * config/vms/vms.opt (vms_float_format): New variable.
822 2014-03-08 Tobias Burnus <burnus@net-b.de>
824 * doc/invoke.texi (-fcilkplus): Update implementation status.
826 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
827 Richard Biener <rguenther@suse.de>
829 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
830 consistently accross all TUs.
831 (run_gcc): Enable -fshort-double automatically at link at link-time
832 and disallow override.
834 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
837 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
838 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
839 if they can't be used.
841 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
843 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
844 for Solaris 11/x86 ld.
845 * configure: Regenerate.
847 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
849 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
850 (LIB_TLS_SPEC): Save as ld_tls_libs.
851 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
852 (HAVE_AS_IX86_TLSLDM): New test.
853 * configure, config.in: Regenerate.
854 * config/i386/i386.c (legitimize_tls_address): Fall back to
855 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
856 cannot support TLS_MODEL_LOCAL_DYNAMIC.
857 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
858 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
860 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
862 * common.opt (fira-loop-pressure): Mark as optimization.
864 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
866 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
867 an OpenMP mappable type.
869 2014-03-06 Matthias Klose <doko@ubuntu.com>
871 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
872 MULTILIB_OSDIRNAMES is not defined.
874 2014-03-06 Jakub Jelinek <jakub@redhat.com>
875 Meador Inge <meadori@codesourcery.com>
878 * config/arm/arm.c (arm_tls_symbol_p): Remove.
879 (arm_legitimize_address): Call legitimize_tls_address for any
880 arm_tls_referenced_p expression, handle constant addend. Call it
881 before testing for !TARGET_ARM.
882 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
884 2014-03-06 Richard Biener <rguenther@suse.de>
890 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
892 * tree-streamer.c (record_common_node): Assert we don't record
893 nodes with type double.
894 (preload_common_node): Skip type double, complex double and double
895 pointer since it is now frontend dependent due to fshort-double option.
897 2014-03-06 Richard Biener <rguenther@suse.de>
899 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
900 or -fno-lto is specified and the linker has full plugin support.
901 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
902 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
903 * lto-wrapper.c (merge_and_complain): Merge compile-time
905 (run_gcc): And pass it through to the link options.
907 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
911 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
913 * cselib.c (remove_useless_values): Skip to avoid quadratic
914 behavior if the condition moved from...
915 (cselib_process_insn): ... here holds.
917 2014-03-05 Jakub Jelinek <jakub@redhat.com>
920 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
921 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
924 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
925 (TM_H): Add x86-tune.def.
927 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
929 * config/aarch64/aarch64.c (generic_tunings):
930 Use cortexa57_extra_costs.
932 2014-03-05 Jakub Jelinek <jakub@redhat.com>
935 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
936 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
937 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
940 2014-03-04 Heiher <r@hev.cc>
942 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
943 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
945 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
947 * config/i386/predicates.md (const2356_operand): Change to ...
948 (const2367_operand): ... this.
949 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
951 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
952 (*avx512pf_scatterpf<mode>sf): Ditto.
953 (avx512pf_scatterpf<mode>df): Ditto.
954 (*avx512pf_scatterpf<mode>df_mask): Ditto.
955 (*avx512pf_scatterpf<mode>df): Ditto.
956 * config/i386/i386.c (ix86_expand_builtin): Update
957 incorrect hint operand error message.
959 2014-03-04 Richard Biener <rguenther@suse.de>
961 * lto-section-in.c (lto_get_section_data): Fix const cast.
963 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
965 * tree-streamer.c (record_common_node): Assert we don't record
966 nodes with type double.
967 (preload_common_node): Skip type double, complex double and double
968 pointer since it is now frontend dependent due to fshort-double option.
970 2014-03-04 Richard Biener <rguenther@suse.de>
973 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
974 (lto_input_toplevel_asms): Likewise.
975 * lto-section-in.c (lto_get_section_data): Instead do it here
978 2014-03-04 Richard Biener <rguenther@suse.de>
980 PR tree-optimization/60382
981 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
982 dead PHIs a reduction.
984 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
986 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
988 (_mm_prefetch): Move out of GCC target("sse") pragma.
989 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
990 GCC target("prfchw") pragma.
991 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
993 * config/i386/i386.c (ix86_option_override_internal): Enable
994 -mprfchw with -mprefetchwt1.
996 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
998 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1001 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1003 * opts.h (CL_PCH_IGNORE): Define.
1004 * targhooks.c (option_affects_pch_p):
1005 Return false for options that have CL_PCH_IGNORE set.
1006 * opt-functions.awk: Process PchIgnore.
1007 * doc/options.texi: Document PchIgnore.
1009 * config/arc/arc.opt (misize): Add PchIgnore property.
1011 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1013 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1014 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1015 constraint on constants to permit them being loaded into
1016 GENERAL_REGS or BASE_REGS.
1018 2014-03-03 Nick Clifton <nickc@redhat.com>
1020 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1021 anti-cacnonical alternatives.
1022 (negandhi3_real): New pattern.
1023 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1025 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1027 * config/avr/avr-mcus.def: Remove atxmega16x1.
1028 * config/avr/avr-tables.opt: Regenerate.
1029 * config/avr/t-multilib: Regenerate.
1030 * doc/avr-mmcu.texi: Regenerate.
1032 2014-03-03 Tobias Grosser <tobias@grosser.es>
1033 Mircea Namolaru <mircea.namolaru@inria.fr>
1035 PR tree-optimization/58028
1036 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1039 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1041 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1042 not handled by recognizers.
1044 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1047 * function.c (expand_function_end): Don't emit
1048 clobber_return_register sequence if clobber_after is a BARRIER.
1049 * cfgexpand.c (construct_exit_block): Append instructions before
1050 return_label to prev_bb.
1052 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1054 * config/rs6000/constraints.md: Document reserved use of "wc".
1056 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1059 * ipa.c (function_and_variable_visibility): When dissolving comdat
1060 group, also set all symbols to local.
1062 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1067 2013-12-14 Jan Hubicka <jh@suse.cz>
1069 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1071 2014-03-02 Jon Beniston <jon@beniston.com>
1077 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1078 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1079 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1080 (simple_return, *simple_return): New patterns
1081 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1082 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1084 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1086 * dwarf2out.c (gen_subprogram_die): Tidy.
1088 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1091 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1092 (*mov_t_msb_neg_negc): ... this new insn.
1094 2014-02-28 Jason Merrill <jason@redhat.com>
1097 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1100 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1103 * dwarf2out.c (decltype_auto_die): New static.
1104 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1105 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1106 (is_cxx_auto): Likewise.
1108 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1110 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1111 we are not using general regs only.
1113 2014-02-28 Richard Biener <rguenther@suse.de>
1116 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1117 previous fix and only allow to remove trivial pre-headers
1118 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1119 (remove_forwarder_block): Properly update the latch of a loop.
1121 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1124 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1125 (cselib_preserved_hash_table): New.
1126 (preserve_constants_and_equivs): Move preserved vals to it.
1127 (cselib_find_slot): Look it up first.
1128 (cselib_init): Initialize it.
1129 (cselib_finish): Release it.
1130 (dump_cselib_table): Dump it.
1132 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1135 * cselib.c (remove_useless_values): Skip to avoid quadratic
1136 behavior if the condition moved from...
1137 (cselib_process_insn): ... here holds.
1139 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1142 * var-tracking.c (vt_initialize): Apply the same condition to
1143 preserve the CFA base value.
1145 2014-02-28 Joey Ye <joey.ye@arm.com>
1148 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1149 if reload in progress or completed.
1151 2014-02-28 Tobias Burnus <burnus@net-b.de>
1154 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1157 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1159 * doc/tm.texi.in (Condition Code Status): Update documention for
1160 relative locations of cc0-setter and cc0-user.
1162 2014-02-27 Jeff Law <law@redhat.com>
1164 PR rtl-optimization/52714
1165 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1166 into two independent simple sets, if I3 is a jump, ensure the
1167 pattern we place into I3 is a (set (pc) ...).
1169 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1170 Jeff Law <law@redhat.com>
1172 PR rtl-optimization/49847
1173 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1174 are in different blocks.
1175 * doc/tm.texi (Condition Code Status): Update documention for
1176 relative locations of cc0-setter and cc0-user.
1178 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1181 * lra.c (lra_emit_add): Check SUBREG too.
1183 2014-02-27 Andreas Schwab <schwab@suse.de>
1185 * config/m68k/m68k.c (m68k_option_override): Disable
1186 -flive-range-shrinkage for classic m68k.
1187 (m68k_override_options_after_change): Likewise.
1189 2014-02-27 Marek Polacek <polacek@redhat.com>
1192 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1193 -Wmaybe-uninitialized.
1195 2014-02-27 Alan Modra <amodra@gmail.com>
1198 * reload1.c (emit_input_reload_insns): When reload_override_in,
1199 set old to rl->in_reg when rl->in_reg is a subreg.
1201 2014-02-26 Richard Biener <rguenther@suse.de>
1204 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1206 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1208 * common/config/i386/predicates.md (const1256_operand): Remove.
1209 (const2356_operand): New.
1210 (const_1_to_2_operand): Remove.
1211 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1212 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1213 (*avx512pf_gatherpf<mode>sf): Ditto.
1214 (avx512pf_gatherpf<mode>df): Ditto.
1215 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1216 (*avx512pf_gatherpf<mode>df): Ditto.
1217 (avx512pf_scatterpf<mode>sf): Ditto.
1218 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1219 (*avx512pf_scatterpf<mode>sf): Ditto.
1220 (avx512pf_scatterpf<mode>df): Ditto.
1221 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1222 (*avx512pf_scatterpf<mode>df): Ditto.
1223 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1225 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1227 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1228 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1229 (_mm512_mask_testn_epi64_mask): Move to ...
1230 * config/i386/avx512cdintrin.h: Here.
1231 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1232 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1233 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1234 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1235 TARGET_AVX512F from TARGET_AVX512CD.
1237 2014-02-26 Richard Biener <rguenther@suse.de>
1240 * ipa.c (walk_polymorphic_call_targets): Properly guard
1241 call to inline_update_overall_summary.
1243 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1246 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1247 and latches only if requested. Fix latch if it is removed.
1248 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1249 LOOPS_HAVE_PREHEADERS.
1251 2014-02-25 Andrew Pinski <apinski@cavium.com>
1253 * builtins.c (expand_builtin_thread_pointer): Create a new target
1254 when the target is NULL.
1256 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1258 PR rtl-optimization/60317
1259 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1260 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1261 * lra-assigns.c: Include params.h.
1262 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1263 other reload pseudos considerations.
1265 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1267 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1268 to use canonical form for nor<mode>3.
1270 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1273 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1276 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1278 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1279 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1280 (ix86_handle_option): Handle OPT_mprefetchwt1.
1281 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1282 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1284 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1285 OPTION_MASK_ISA_PREFETCHWT1.
1286 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1287 (PTA_PREFETCHWT1): New.
1288 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1289 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1290 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1291 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1292 (*prefetch_avx512pf_<mode>_: Change into ...
1293 (*prefetch_prefetchwt1_<mode>: This.
1294 * config/i386/i386.opt (mprefetchwt1): New.
1295 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1296 (_mm_prefetch): Handle intent to write.
1297 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1299 2014-02-25 Richard Biener <rguenther@suse.de>
1302 * emit-rtl.c (mem_attrs_htab): Remove.
1303 (mem_attrs_htab_hash): Likewise.
1304 (mem_attrs_htab_eq): Likewise.
1305 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1306 (init_emit_once): Do not allocate mem_attrs_htab.
1308 2014-02-25 Richard Biener <rguenther@suse.de>
1311 * lto-opts.c (lto_write_options): Output non-explicit conservative
1312 -fwrapv, -fno-trapv and -fno-strict-overflow.
1313 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1314 (run_gcc): And pass them through.
1316 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1318 * sel-sched.c (calculate_new_fences): New parameter ptime.
1319 Calculate it as a maximum over all fence cycles.
1320 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1321 Print the final schedule timing when sched_verbose.
1323 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1325 PR rtl-optimization/60292
1326 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1327 bit fot the fence instruction.
1329 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1331 * calls.h: Fix typo in comment.
1333 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1335 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1336 adjusting offsetable addresses.
1338 2014-02-24 Guozhi Wei <carrot@google.com>
1340 * sparseset.h (sparseset_pop): Fix the wrong index.
1342 2014-02-24 Walter Lee <walt@tilera.com>
1344 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1345 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1347 * common/config/tilegx/tilegx-common.c
1348 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1349 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1351 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1352 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1353 (tilegx_gimplify_va_arg_expr): Handle big endian.
1354 (tilegx_expand_unaligned_load): Ditto.
1355 (tilegx_expand_unaligned_store): Ditto.
1356 (TARGET_RETURN_IN_MSB): New.
1357 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1358 (TARGET_ENDIAN_DEFAULT): New.
1359 (TARGET_BIG_ENDIAN): Handle big endian.
1360 (BYTES_BIG_ENDIAN): Ditto.
1361 (WORDS_BIG_ENDIAN): Ditto.
1362 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1365 * config/tilegx/tilegx.md (extv): Handle big endian.
1367 (insn_st<n>): Ditto.
1368 (insn_st<n>_add<bitsuffix>): Ditto.
1369 (insn_stnt<n>): Ditto.
1370 (insn_stnt<n>_add<bitsuffix>):Ditto.
1371 (vec_interleave_highv8qi): Handle big endian.
1372 (vec_interleave_highv8qi_be): New.
1373 (vec_interleave_highv8qi_le): New.
1374 (insn_v1int_h): Handle big endian.
1375 (vec_interleave_lowv8qi): Handle big endian.
1376 (vec_interleave_lowv8qi_be): New.
1377 (vec_interleave_lowv8qi_le): New.
1378 (insn_v1int_l): Handle big endian.
1379 (vec_interleave_highv4hi): Handle big endian.
1380 (vec_interleave_highv4hi_be): New.
1381 (vec_interleave_highv4hi_le): New.
1382 (insn_v2int_h): Handle big endian.
1383 (vec_interleave_lowv4hi): Handle big endian.
1384 (vec_interleave_lowv4hi_be): New.
1385 (vec_interleave_lowv4hi_le): New.
1386 (insn_v2int_l): Handle big endian.
1387 (vec_interleave_highv2si): Handle big endian.
1388 (vec_interleave_highv2si_be): New.
1389 (vec_interleave_highv2si_le): New.
1390 (insn_v4int_h): Handle big endian.
1391 (vec_interleave_lowv2si): Handle big endian.
1392 (vec_interleave_lowv2si_be): New.
1393 (vec_interleave_lowv2si_le): New.
1394 (insn_v4int_l): Handle big endian.
1395 * config/tilegx/tilegx.opt (mbig-endian): New option.
1396 (mlittle-endian): New option.
1397 * doc/install.texi: Document tilegxbe-linux.
1398 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1400 2014-02-24 Martin Jambor <mjambor@suse.cz>
1403 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1404 there are no parameter descriptors.
1406 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1408 PR rtl-optimization/60268
1409 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1410 initialization to ...
1411 (sched_rgn_init): ... here.
1412 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1414 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1416 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1419 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1421 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1424 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1426 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1427 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1429 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1431 * config/microblaze/predicates.md: Add cmp_op predicate.
1432 * config/microblaze/microblaze.md: Add branch_compare instruction
1433 which uses cmp_op predicate and emits cmp insn before branch.
1434 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1435 to microblaze_expand_conditional_branch and consolidate logic.
1436 (microblaze_expand_conditional_branch): emit branch_compare
1437 insn instead of handling cmp op separate from branch insn.
1439 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1441 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1444 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1446 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1447 define_insn with define_expand and new define_insn
1448 *altivec_lve<VI_char>x_internal.
1449 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1450 and new define_insn *altivec_stve<VI_char>x_internal.
1451 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1453 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1455 (altivec_expand_stvex_be): New function.
1457 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1459 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1460 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1461 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1462 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1464 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1467 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1468 instead of emit_move_insn.
1470 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1472 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1474 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1475 gen_altivec_vsumsws.
1477 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1479 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1480 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1481 (altivec_lvxl_<mode>): New define_expand incorporating
1482 -maltivec=be semantics where needed.
1483 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1484 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1485 semantics where needed.
1486 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1487 (altivec_stvx_<mode>): New define_expand incorporating
1488 -maltivec=be semantics where needed.
1489 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1490 VM2 iterator instead of V4SI.
1491 (altivec_stvxl_<mode>): New define_expand incorporating
1492 -maltivec=be semantics where needed.
1493 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1494 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1495 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1496 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1497 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1498 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1499 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1500 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1501 ALTIVEC_BUILTIN_STVXL.
1502 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1503 (altivec_expand_stvx_be): Likewise.
1504 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1505 (altivec_expand_lvx_be): Likewise.
1506 (altivec_expand_stvx_be): Likewise.
1507 (altivec_expand_builtin): Add cases for
1508 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1509 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1510 (altivec_init_builtins): Add definitions for
1511 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1512 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1514 2014-02-21 Catherine Moore <clm@codesourcery.com>
1516 * doc/invoke.texi (mvirt, mno-virt): Document.
1517 * config/mips/mips.opt (mvirt): New option.
1518 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1520 2014-02-21 Richard Biener <rguenther@suse.de>
1522 PR tree-optimization/60276
1523 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1524 (STMT_VINFO_MIN_NEG_DIST): New macro.
1525 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1526 STMT_VINFO_MIN_NEG_DIST.
1527 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1528 made for negative dependence distances still hold.
1530 2014-02-21 Richard Biener <rguenther@suse.de>
1533 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1534 DECL_INITIAL for globals not in the current function context.
1536 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1538 PR tree-optimization/56490
1539 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1540 * tree-ssa-uninit.c: Include params.h.
1541 (compute_control_dep_chain): Add num_calls argument, return false
1542 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1543 num_calls to recursive call.
1544 (find_predicates): Change dep_chain into normal array,
1545 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1546 variable and adjust compute_control_dep_chain caller.
1547 (find_def_preds): Likewise.
1549 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1551 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1552 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1554 2014-02-21 Nick Clifton <nickc@redhat.com>
1556 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1557 (pushhi1): Likewise.
1558 (popqi1): Add mode to pre_dec.
1561 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1563 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1564 mode for mask of V8SFmode permutation.
1566 2014-02-20 Richard Henderson <rth@redhat.com>
1569 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1570 a new pseudo for OLDVAL.
1572 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1575 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1576 gen_reg_rtx if d->testing_p.
1577 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1578 if d->testing_p and we will certainly return true.
1579 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1582 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1584 * emit-rtl.c (gen_reg_rtx): Assert that
1585 crtl->emit.regno_pointer_align_length is non-zero.
1587 2014-02-20 Richard Henderson <rth@redhat.com>
1590 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1591 on failure the store back into EXPECT.
1593 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1594 Sandra Loosemore <sandra@codesourcery.com>
1596 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1597 * config/nios2/nios2.c (nios2_function_profiler): Add
1598 -fPIC (flag_pic == 2) support.
1599 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1600 (nios2_large_offset_p): New function.
1601 (nios2_unspec_reloc_p): Move up position, update to use
1602 nios2_large_offset_p.
1603 (nios2_unspec_address): Remove function.
1604 (nios2_unspec_offset): New function.
1605 (nios2_large_got_address): New function.
1606 (nios2_got_address): Add large offset support.
1607 (nios2_legitimize_tls_address): Update usage of removed and new
1609 (nios2_symbol_binds_local_p): New function.
1610 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1611 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1612 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1613 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1614 processing for (const (unspec ...)).
1615 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1617 2014-02-20 Richard Biener <rguenther@suse.de>
1619 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1620 doing the substitution.
1621 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1623 2014-02-20 Martin Jambor <mjambor@suse.cz>
1626 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1627 info when checking whether lattices are bottom.
1629 2014-02-20 Richard Biener <rguenther@suse.de>
1632 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1635 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1638 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1639 parameter specifying the scaling.
1640 (inline_call): Update.
1641 (want_inline_recursively): Guard division by zero.
1642 (recursive_inlining): Update.
1643 * ipa-inline.h (clone_inlined_nodes): Update.
1645 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1648 * config/i386/i386.c (classify_argument): Pass structures of size
1649 64 bytes or less in register.
1651 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1652 Kirill Yukhin <kirill.yukhin@intel.com>
1654 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1655 (_mm_rcp28_round_ss): Ditto.
1656 (_mm_rsqrt28_round_sd): Ditto.
1657 (_mm_rsqrt28_round_ss): Ditto.
1658 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1659 (_mm_rcp14_round_ss): Ditto.
1660 (_mm_rsqrt14_round_sd): Ditto.
1661 (_mm_rsqrt14_round_ss): Ditto.
1662 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1663 the first input operand, get rid of match_dup.
1664 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1666 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1668 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1669 operand as the first input operand, set type attribute.
1670 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1672 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1673 operand as the first input operand, set type attribute.
1675 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1677 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1680 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1683 * config/i386/i386.c (construct_container): Remove TFmode check
1684 for X86_64_INTEGER_CLASS.
1686 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1689 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1690 only when -Wpsabi is enabled.
1692 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1695 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1696 passing arrays in registers are the same as for structs, so remove the
1697 special case for them.
1699 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1701 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1702 destination type, extract only the valid bits if the source type is not
1703 integral and has a different mode.
1705 2014-02-19 Richard Biener <rguenther@suse.de>
1708 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1711 2014-02-19 Richard Biener <rguenther@suse.de>
1714 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1715 (ipa_modify_call_arguments): Emit an argument load explicitely and
1716 preserve virtual SSA form there and for the replacement call.
1717 Do not update SSA form nor free dominance info.
1719 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1721 * ipa.c (function_and_variable_visibility): Also clear WEAK
1722 flag when disolving COMDAT_GROUP.
1724 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1726 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1727 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1729 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1730 do more sanity checks.
1731 (detect_type_change): Return true when giving up early.
1732 (compute_complex_assign_jump_func): Fix type parameter of
1733 ipa_set_ancestor_jf.
1734 (compute_complex_ancestor_jump_func): Likewise.
1735 (update_jump_functions_after_inlining): Fix updating of
1737 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1739 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1741 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1742 inline clones when edge disappears.
1744 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1747 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1748 Split 64-bit moves into 2 patterns. Do not allow the use of
1749 direct move for TDmode in little endian, since the decimal value
1750 has little endian bytes within a word, but the 64-bit pieces are
1751 ordered in a big endian fashion, and normal subreg's of TDmode are
1753 (mov<mode>_64bit_dm): Likewise.
1754 (movtd_64bit_nodm): Likewise.
1756 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1758 PR tree-optimization/60174
1759 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1760 statement of an SSA_NAME that occurs in an abnormal PHI node.
1762 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1765 * final.c (SEEN_BB): Remove.
1766 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1767 (final_scan_insn): Don't force_source_line on second
1768 NOTE_INSN_BASIC_BLOCK.
1770 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1773 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1774 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1775 (type_natural_mode): Warn ABI change when %zmm register is not
1776 available for AVX512F vector value passing.
1778 2014-02-18 Kai Tietz <ktietz@redhat.com>
1781 * config/i386/i386.c (ix86_expand_prologue): Use value in
1782 rax register as displacement when restoring %r10 or %rax.
1783 Fix wrong offset when restoring both registers.
1785 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1787 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1788 assertion with conditional return.
1790 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1791 Uros Bizjak <ubizjak@gmail.com>
1794 * config/i386/driver-i386.c (host_detect_local_cpu): If
1795 YMM state is not saved by the OS, also clear has_f16c. Move
1796 CPUID 0x80000001 handling before YMM state saving checking.
1798 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1800 PR rtl-optimization/58960
1801 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1802 factored out from ...
1803 (sched_init): ... here.
1804 (free_global_sched_pressure_data): New, factored out from ...
1805 (sched_finish): ... here.
1806 * sched-int.h (free_global_sched_pressure_data): Declare.
1807 * sched-rgn.c (nr_regions_initial): New static global.
1808 (haifa_find_rgns): Initialize it.
1809 (schedule_region): Disable sched-pressure for the newly
1812 2014-02-17 Richard Biener <rguenther@suse.de>
1814 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1815 release SSA defs of pattern stmts.
1817 2014-02-17 Richard Biener <rguenther@suse.de>
1819 * tree-inline.c (expand_call_inline): Release the virtual
1820 operand defined by the call we are about to inline.
1822 2014-02-17 Richard Biener <rguenther@suse.de>
1824 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1826 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1827 Ilya Tocar <ilya.tocar@intel.com>
1829 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1830 arguments order in builtin.
1831 (_mm512_permutexvar_epi64): Ditto.
1832 (_mm512_mask_permutexvar_epi64): Ditto
1833 (_mm512_maskz_permutexvar_epi32): Ditto
1834 (_mm512_permutexvar_epi32): Ditto
1835 (_mm512_mask_permutexvar_epi32): Ditto
1837 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1839 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1840 (p8_vmrgow): Likewise.
1842 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1844 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1847 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1850 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1851 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1852 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1853 using direct move instructions on ISA 2.07. Also adjust
1854 instruction length for 64-bit.
1855 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1856 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1858 2014-02-15 Alan Modra <amodra@gmail.com>
1862 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1863 find_replacement on parts of insn rtl that might be reloaded.
1865 2014-02-15 Richard Biener <rguenther@suse.de>
1867 PR tree-optimization/60183
1868 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1869 (tree_ssa_phiprop): Calculate and free post-dominators.
1871 2014-02-14 Jeff Law <law@redhat.com>
1873 PR rtl-optimization/60131
1874 * ree.c (get_extended_src_reg): New function.
1875 (combine_reaching_defs): Use it rather than assuming location of REG.
1876 (find_and_remove_re): Verify first operand of extension is
1877 a REG before adding the insns to the copy list.
1879 2014-02-14 Roland McGrath <mcgrathr@google.com>
1881 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1882 * configure: Regenerated.
1883 * config.in: Regenerated.
1884 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1885 instead of ASM_SHORT.
1887 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1888 Richard Earnshaw <rearnsha@arm.com>
1890 PR rtl-optimization/59535
1891 * lra-constraints.c (process_alt_operands): Encourage alternative
1892 when unassigned pseudo class is superset of the alternative class.
1893 (inherit_reload_reg): Don't inherit when optimizing for code size.
1894 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1895 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1896 modes not less than 4 for Thumb1.
1898 2014-02-14 Kyle McMartin <kyle@redhat.com>
1901 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1903 2014-02-14 Richard Biener <rguenther@suse.de>
1905 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1906 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1907 Do not strip INDIRECT_REFs.
1909 2014-02-14 Richard Biener <rguenther@suse.de>
1912 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1913 DECL_FUNCTION_SPECIFIC_TARGET.
1914 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1915 * tree-streamer-out.c (pack_ts_target_option): Remove.
1916 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1917 (write_ts_function_decl_tree_pointers): Do not stream
1918 DECL_FUNCTION_SPECIFIC_TARGET.
1919 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1920 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1921 (lto_input_ts_function_decl_tree_pointers): Do not stream
1922 DECL_FUNCTION_SPECIFIC_TARGET.
1924 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1926 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1927 (get_initial_def_for_induction, vectorizable_induction): Ignore
1928 debug stmts when looking for exit_phi.
1929 (vectorizable_live_operation): Fix up condition.
1931 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1933 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1934 nreverse() because it changes the content of original tree list.
1936 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1938 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1939 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1941 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1943 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1944 GNU coding standards.
1946 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1949 * dwarf2out.c (gen_subprogram_die): Don't call
1950 add_calling_convention_attribute if subr_die is old_die.
1952 2014-02-13 Sharad Singhai <singhai@google.com>
1954 * doc/optinfo.texi: Fix order of nodes.
1956 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1958 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1959 operands[2], not operands[3].
1961 2014-02-13 Richard Biener <rguenther@suse.de>
1964 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1965 mention the possibility of an in-tree build.
1966 (CLooG): Update recommended version to 0.18.1, mention the
1967 possibility of an in-tree build and clarify that the ISL
1968 bundled with CLooG does not work.
1970 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1973 * expr.c (compress_float_constant): If x is a hard register,
1974 extend into a pseudo and then move to x.
1976 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1978 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1979 caused by bad second argument to warning_at() with -mhotpatch and
1980 nested functions (e.g. with gfortran).
1982 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1984 * opts.c (option_name): Remove "enabled by default" rider.
1986 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1988 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1990 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1991 Uros Bizjak <ubizjak@gmail.com>
1994 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1995 * configure: Regenerated.
1997 2014-02-12 Richard Biener <rguenther@suse.de>
1999 * vec.c (vec_prefix::calculate_allocation): Move as
2000 inline variant to vec.h.
2001 (vec_prefix::calculate_allocation_1): New out-of-line version.
2002 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2003 (vec_prefix::m_has_auto_buf): Rename to ...
2004 (vec_prefix::m_using_auto_storage): ... this.
2005 (vec_prefix::calculate_allocation): Inline the easy cases
2006 and dispatch to calculate_allocation_1 which doesn't need the
2008 (va_heap::reserve): Use gcc_checking_assert.
2009 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2010 m_using_auto_storage.
2011 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2013 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2014 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2015 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2017 2014-02-12 Richard Biener <rguenther@suse.de>
2019 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2020 when we found a dependence.
2022 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2024 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2026 (maybe_fold_stmt): ... into this new function.
2027 * omp-low.c (lower_omp): Update comment.
2029 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2032 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2035 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2037 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2038 identifiers in comments.
2039 (cortexa53_extra_costs): Likewise.
2040 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2041 (cortexa7_extra_costs): Likewise.
2042 (cortexa12_extra_costs): Likewise.
2043 (cortexa15_extra_costs): Likewise.
2044 (v7m_extra_costs): Likewise.
2046 2014-02-12 Richard Biener <rguenther@suse.de>
2049 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2050 of posix_memalign being successful.
2051 (lower_stmt): Restrict lowering of posix_memalign to when
2052 -ftree-bit-ccp is enabled.
2054 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2056 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2058 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2060 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2062 PR rtl-optimization/60116
2063 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2064 other_insn once the combination has been validated.
2066 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2069 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2071 * ipa-devirt.c: Include demangle.h
2072 (odr_violation_reported): New static variable.
2073 (add_type_duplicate): Update odr_violations.
2074 (maybe_record_node): Add completep parameter; update it.
2075 (record_target_from_binfo): Add COMPLETEP parameter;
2076 update it as needed.
2077 (possible_polymorphic_call_targets_1): Likewise.
2078 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2079 rename FINAL to COMPLETE.
2080 (record_targets_from_bases): Sanity check we found the binfo;
2081 fix COMPLETEP updating.
2082 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2083 parameter, fix computing of COMPLETEP.
2084 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2085 at LTO time do demangling.
2086 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2087 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2089 (gimple_get_virt_method_for_binfo): Likewise.
2090 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2091 gimple_get_virt_method_for_vtable): Update prototypes.
2093 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2096 * genautomata.c (add_presence_absence): Fix typo with
2097 {final_}presence_list.
2099 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2102 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2103 for VSX/Altivec vectors that land in GPR registers.
2105 2014-02-11 Richard Henderson <rth@redhat.com>
2106 Jakub Jelinek <jakub@redhat.com>
2109 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2110 around drhs if type conversion to lacc->type is not useless.
2112 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2114 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2116 (cortex-a57.cortex-a53): Likewise.
2117 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2119 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2121 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2124 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2126 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2127 add_options_for_arm_vfp3.
2129 2014-02-11 Jeff Law <law@redhat.com>
2132 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2133 object with an undesirable mode.
2135 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2138 * config/i386/sol2-9.h: New file.
2139 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2140 *-*-solaris2.9*): Use it.
2142 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2144 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2145 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2147 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2149 * config/microblaze/microblaze.c: Extend mcpu version format
2151 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2153 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2155 2014-02-10 Richard Henderson <rth@redhat.com>
2158 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2159 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2160 ms-abi vs -mno-accumulate-outgoing-args.
2161 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2162 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2165 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2168 * cfgexpand.c (expand_asm_operands): Attach source location to
2169 ASM_INPUT rtx objects.
2170 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2172 2014-02-10 Nick Clifton <nickc@redhat.com>
2174 * config/mn10300/mn10300.c (popcount): New function.
2175 (mn10300_expand_prologue): Include saved registers in stack usage
2178 2014-02-10 Jeff Law <law@redhat.com>
2181 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2182 when changing the SET_DEST of a prior insn to avoid an input reload.
2184 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2186 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2187 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2188 -mcall-openbsd, or -mcall-linux.
2189 (CC1_ENDIAN_BIG_SPEC): Remove.
2190 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2191 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2192 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2193 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2194 and %cc1_endian_default.
2195 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2197 2014-02-10 Richard Biener <rguenther@suse.de>
2199 PR tree-optimization/60115
2200 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2201 MEM_REF handling. Properly verify that the accesses are not
2202 out of the objects bound.
2204 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2206 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2209 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2211 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2212 proper constants and fix formatting.
2213 (possible_polymorphic_call_targets): Fix formatting.
2215 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2216 Ilya Tocar <ilya.tocar@intel.com>
2218 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2219 (_mm512_loadu_epi32): Renamed into...
2220 (_mm512_loadu_si512): This.
2221 (_mm512_storeu_epi32): Renamed into...
2222 (_mm512_storeu_si512): This.
2223 (_mm512_maskz_ceil_ps): Removed.
2224 (_mm512_maskz_ceil_pd): Ditto.
2225 (_mm512_maskz_floor_ps): Ditto.
2226 (_mm512_maskz_floor_pd): Ditto.
2227 (_mm512_floor_round_ps): Ditto.
2228 (_mm512_floor_round_pd): Ditto.
2229 (_mm512_ceil_round_ps): Ditto.
2230 (_mm512_ceil_round_pd): Ditto.
2231 (_mm512_mask_floor_round_ps): Ditto.
2232 (_mm512_mask_floor_round_pd): Ditto.
2233 (_mm512_mask_ceil_round_ps): Ditto.
2234 (_mm512_mask_ceil_round_pd): Ditto.
2235 (_mm512_maskz_floor_round_ps): Ditto.
2236 (_mm512_maskz_floor_round_pd): Ditto.
2237 (_mm512_maskz_ceil_round_ps): Ditto.
2238 (_mm512_maskz_ceil_round_pd): Ditto.
2239 (_mm512_expand_pd): Ditto.
2240 (_mm512_expand_ps): Ditto.
2241 * config/i386/i386.c (ix86_builtins): Remove
2242 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2243 (bdesc_args): Ditto.
2244 * config/i386/predicates.md (const1256_operand): New.
2245 (const_1_to_2_operand): Ditto.
2246 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2247 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2248 (*avx512pf_gatherpf<mode>sf): Ditto.
2249 (avx512pf_gatherpf<mode>df): Ditto.
2250 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2251 (*avx512pf_gatherpf<mode>df): Ditto.
2252 (avx512pf_scatterpf<mode>sf): Ditto.
2253 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2254 (*avx512pf_scatterpf<mode>sf): Ditto.
2255 (avx512pf_scatterpf<mode>df): Ditto.
2256 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2257 (*avx512pf_scatterpf<mode>df): Ditto.
2258 (avx512f_expand<mode>): Removed.
2259 (<shift_insn><mode>3<mask_name>): Change predicate type.
2261 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2263 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2264 not at the end of datarefs vector use ordered_remove to avoid
2265 reordering datarefs vector.
2268 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2269 mark local addressable non-static vars as GOVD_PRIVATE
2270 instead of GOVD_LOCAL.
2271 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2272 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2276 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2277 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2278 assume_aligned or alloc_align attributes.
2279 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2280 arguments. Handle also assume_aligned and alloc_align attributes.
2281 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2282 calls to functions with assume_aligned or alloc_align attributes.
2283 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2285 2014-02-08 Terry Guo <terry.guo@arm.com>
2287 * doc/invoke.texi: Document ARM -march=armv7e-m.
2289 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2291 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2292 flag on __cilkrts_rethrow builtin.
2295 * ipa-cp.c (determine_versionability): Fail at -O0
2296 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2297 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2300 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2303 * tree-inline.c (copy_forbidden): Fail for
2304 __attribute__((optimize (0))) functions.
2306 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2308 * varpool.c: Include pointer-set.h.
2309 (varpool_remove_unreferenced_decls): Variables in other partitions
2310 will not be output; be however careful to not lose information
2313 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2315 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2316 lookup in the vtable constructor.
2318 2014-02-07 Jeff Law <law@redhat.com>
2321 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2322 define_insn_and_split.
2324 * ipa-inline.c (inline_small_functions): Fix typos.
2326 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2328 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2329 (s390_can_use_return_insn): Declare.
2330 * config/s390/s390.h (EPILOGUE_USES): Define.
2331 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2333 (s390_chunkify_start): Handle return JUMP_LABELs.
2334 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2335 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2336 (s390_can_use_return_insn): New functions.
2337 (s390_fix_long_loop_prediction): Handle conditional returns.
2338 (TARGET_SET_UP_BY_PROLOGUE): Define.
2339 * config/s390/s390.md (ANY_RETURN): New code iterator.
2340 (*creturn, *csimple_return, return, simple_return): New patterns.
2342 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2344 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2345 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2346 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2347 REG_CFA_RESTORE list when deciding not to restore a register.
2349 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2351 * config/s390/s390.c: Include tree-pass.h and context.h.
2352 (s390_early_mach): New function, split out from...
2353 (s390_emit_prologue): ...here.
2354 (pass_data_s390_early_mach): New pass structure.
2355 (pass_s390_early_mach): New class.
2356 (s390_option_override): Create and register early_mach pass.
2357 Move to end of file.
2359 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2361 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2362 to match for the exit block.
2364 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2366 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2367 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2368 Reject misaligned operands.
2370 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2372 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2374 2014-02-07 Richard Biener <rguenther@suse.de>
2377 * gimple-low.c (lower_builtin_posix_memalign): New function.
2378 (lower_stmt): Call it to lower posix_memalign in a way
2379 to make alignment info accessible.
2381 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2384 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2385 __builtin_setjmp_receiver.
2387 2014-02-07 Richard Biener <rguenther@suse.de>
2390 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2391 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2392 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2393 Handle BUILT_IN_POSIX_MEMALIGN.
2394 (find_func_clobbers): Likewise.
2395 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2396 (call_may_clobber_ref_p_1): Likewise.
2398 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2401 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2404 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2407 * lto-cgraph.c (lto_output_node): Use
2408 symtab_get_symbol_partitioning_class.
2409 (lto_output_varpool_node): likewise.
2410 (symtab_get_symbol_partitioning_class): Move here from
2412 * cgraph.h (symbol_partitioning_class): Likewise.
2413 (symtab_get_symbol_partitioning_class): Declare.
2415 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2417 * ggc.h (ggc_internal_cleared_alloc): New macro.
2418 * vec.h (vec_safe_copy): Handle memory stats.
2419 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2420 * target-globals.c (save_target_globals): Likewise.
2422 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2425 * expr.c (emit_move_resolve_push): Export; be bit more selective
2426 on when to clear alias set.
2427 * expr.h (emit_move_resolve_push): Declare.
2428 * function.h (struct function): Add tail_call_marked.
2429 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2430 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2431 * config/i386/i386.md (TImode move expander): De not call
2433 (FP push expanders): Preserve memory attributes.
2434 * config/i386/sse.md (push<mode>1): Remove.
2435 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2436 (ix86_expand_push): Remove.
2437 * config/i386/mmx.md (push<mode>1): Remove.
2439 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2441 PR rtl-optimization/60030
2442 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2443 lopart with paradoxical subreg before shifting it up by hprec.
2445 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2447 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2448 Remove extra newline at end of file.
2449 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2450 (arm_issue_rate): Handle cortexa57.
2451 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2452 (cortex-a57.cortex-a53): Likewise.
2454 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2457 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2458 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2460 (arm_expand_prologue): Adjust all callers.
2461 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2462 info, registers also at the lowest numbered registers side. Use
2463 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2467 * var-tracking.c (adjust_mems): Before adding a SET to
2468 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2470 2014-02-06 Alan Modra <amodra@gmail.com>
2473 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2474 change SDmode to DDmode when lra_in_progress.
2476 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2479 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2480 free_data_ref on the dr first, and before goto again also set dr
2481 to the next dr. For simd_lane_access, free old datarefs[i] before
2482 overwriting it. For get_vectype_for_scalar_type failure, don't
2483 free_data_ref if simd_lane_access.
2485 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2488 * tree.h (opts_for_fn): New inline function.
2489 (opt_for_fn): Define.
2490 * config/i386/i386.c (ix86_function_regparm): Use
2491 opt_for_fn (decl, optimize) instead of optimize.
2493 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2495 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2496 for SYMBOL_REF in large memory model.
2498 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2500 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2502 (cortex-a57): Likewise.
2503 (cortex-a57.cortex-a53): Likewise.
2505 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2506 Kugan Vivekanandarajah <kuganv@linaro.org>
2508 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2510 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2512 2014-02-06 Richard Biener <rguenther@suse.de>
2514 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2515 set_loop_copy and initialize_original_copy_tables.
2517 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2519 * config/aarch64/aarch64-simd.md
2520 (aarch64_ashr_simddi): Change QI to SI.
2522 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2523 Jakub Jelinek <jakub@redhat.com>
2526 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2529 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2531 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2532 CODE_FOR_altivec_vpku[hw]um to
2533 CODE_FOR_altivec_vpku[hw]um_direct.
2534 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2535 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2536 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2537 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2539 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2541 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2542 generation for -maltivec=be.
2543 (altivec_vsumsws): Simplify redundant test.
2545 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2547 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2548 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2549 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2550 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2551 gen_altivec_vpkuwum.
2552 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2554 (altivec_vpks<VI_char>ss): Likewise.
2555 (altivec_vpks<VI_char>us): Likewise.
2556 (altivec_vpku<VI_char>us): Likewise.
2557 (altivec_vpku<VI_char>um): Likewise.
2558 (altivec_vpku<VI_char>um_direct): New (copy of
2559 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2561 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2562 target is little endian and -maltivec=be is not specified.
2563 (*altivec_vupkhs<VU_char>_direct): New (copy of
2564 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2565 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2566 target is little endian and -maltivec=be is not specified.
2567 (*altivec_vupkls<VU_char>_direct): New (copy of
2568 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2569 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2570 little endian and -maltivec=be is not specified.
2571 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2572 little endian and -maltivec=be is not specified.
2574 2014-02-05 Richard Henderson <rth@redhat.com>
2577 * combine-stack-adj.c: Revert r206943.
2578 * sched-int.h (struct deps_desc): Add last_args_size.
2579 * sched-deps.c (init_deps): Initialize it.
2580 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2581 contain REG_ARGS_SIZE notes.
2583 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2585 * lto-cgraph.c (asm_nodes_output): Make global.
2586 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2587 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2588 (driver_handle_option): Handle OPT_fwpa.
2590 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2593 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2594 a comment typo and formatting issue. If odr_hash hasn't been
2595 created, return vNULL and set *completep to false.
2598 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2599 bb with no successors.
2601 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2604 * doc/invoke.texi (-march): Clarify documentation for ARM.
2608 2014-02-05 Richard Biener <rguenther@suse.de>
2610 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2611 when not vectorizing because of too many alias checks.
2612 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2613 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2615 2014-02-05 Nick Clifton <nickc@redhat.com>
2617 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2618 accept extended registers in any mode when compiling for the MN10300.
2620 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2622 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2623 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2624 sanitization attributes.
2625 (can_inline_edge_p): Likewise.
2626 (sanitize_attrs_match_for_inline_p): New function.
2628 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2630 * ipa-prop.c (detect_type_change): Shor circuit testing of
2631 type changes on THIS pointer.
2633 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2636 * config/pa/pa.c (legitimize_tls_address): Return original address
2637 if not passed a SYMBOL_REF rtx.
2638 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2640 (pa_emit_move_sequence): Simplify TLS source operands.
2641 (pa_legitimate_constant_p): Reject all TLS constants.
2642 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2643 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2645 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2647 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2648 groups when we know they are controlled by LTO.
2649 * varasm.c (default_binds_local_p_1): If object is in other partition,
2650 it will be resolved locally.
2652 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2654 * config/host-linux.c (linux_gt_pch_use_address): Don't
2655 use SSIZE_MAX because it is not always defined.
2657 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2660 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2661 threshold for pseudo splitting.
2662 (update_ebb_live_info): Process call argument hard registers and
2663 hard registers from insn definition too.
2664 (max_small_class_regs_num): New constant.
2665 (inherit_in_ebb): Update live hard regs through EBBs. Update
2666 reloads_num only for small register classes. Don't split for
2669 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2672 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2675 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2677 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2680 2014-02-04 Marek Polacek <polacek@redhat.com>
2682 * gdbinit.in (pel): Define.
2684 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2686 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2689 2014-02-04 Richard Biener <rguenther@suse.de>
2692 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2693 in function context local.
2694 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2695 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2696 similar to LTO_imported_decl_ref.
2698 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2700 PR tree-optimization/60002
2701 * cgraphclones.c (build_function_decl_skip_args): Clear
2704 PR tree-optimization/60023
2705 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2706 false to gsi_replace.
2707 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2708 has been in some EH region and vec_stmt could throw, add
2709 vec_stmt into the same EH region.
2710 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2711 has no lhs, ignore it.
2712 * internal-fn.c (expand_MASK_LOAD): Likewise.
2715 * tree-inline.c (copy_forbidden): Fail for
2716 __attribute__((optimize (0))) functions.
2719 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2720 is set, copy one less argument.
2721 (expand_simd_clones): Don't subtract clone_info->inbranch
2722 from simd_clone_struct_alloc argument.
2724 PR rtl-optimization/57915
2725 * recog.c (simplify_while_replacing): If all unary/binary/relational
2726 operation arguments are constant, attempt to simplify those.
2729 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2730 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2732 2014-02-04 Richard Biener <rguenther@suse.de>
2734 PR tree-optimization/60012
2735 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2736 TBAA disambiguation to all DDRs.
2738 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2741 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2742 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2744 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2747 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2749 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2751 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2752 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2754 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2757 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2758 to figure out targets of polymorphic calls with known decl.
2759 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2760 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2761 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2762 (get_polymorphic_call_info): ... here.
2763 (get_polymorphic_call_info_from_invariant): New function.
2765 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2767 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2768 lookup via vtable pointer; check for type consistency
2769 and turn inconsitent facts into UNREACHABLE.
2770 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2771 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2772 type inconsistent querries; return UNREACHABLE instead.
2774 2014-02-03 Richard Henderson <rth@twiddle.net>
2777 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2778 already processed this node.
2779 (normalize_one_pred_1): Pass along mark_set.
2780 (normalize_one_pred): Create and destroy a pointer_set_t.
2781 (normalize_one_pred_chain): Likewise.
2783 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2785 PR gcov-profile/58602
2786 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2788 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2791 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2792 -fno-devirtualize; try to devirtualize by the knowledge of
2793 virtual table pointer given by aggregate propagation.
2794 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2795 (ipa_print_node_jump_functions): Dump also offset that
2796 is relevant for polymorphic calls.
2797 (determine_known_aggregate_parts): Add arg_type parameter; use it
2798 instead of determining the type from pointer type.
2799 (ipa_compute_jump_functions_for_edge): Update call of
2800 determine_known_aggregate_parts.
2801 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2802 (gimple_get_virt_method_for_binfo): ... here; simplify using
2803 vtable_pointer_value_to_vtable.
2804 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2805 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2806 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2807 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2809 (vtable_pointer_value_to_binfo): ... here.
2810 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2812 2014-02-03 Teresa Johnson <tejohnson@google.com>
2814 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2815 redef of outer loop index variable.
2817 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2821 * doc/extend.texi (Function Attributes): Typo.
2823 2014-02-03 Cong Hou <congh@google.com>
2825 PR tree-optimization/60000
2826 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2827 if the vectorized statement is a store. A store statement can only
2828 appear at the end of pattern statements.
2830 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2832 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2833 (ix86_option_override_internal): Default long double to 64-bit for
2834 32-bit Bionic and to 128-bit for 64-bit Bionic.
2836 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2837 TARGET_LONG_DOUBLE_128 is true.
2838 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2840 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2841 (mlong-double-64): Negate -mlong-double-128.
2842 (mlong-double-128): New option.
2844 * config/i386/i386-c.c (ix86_target_macros): Define
2845 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2847 * doc/invoke.texi: Document -mlong-double-128.
2849 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2851 PR rtl-optimization/60024
2852 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2854 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2856 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2858 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2860 PR rtl-optimization/57662
2861 * sel-sched.c (code_motion_path_driver): Do not mark already not
2862 existing blocks in the visiting bitmap.
2864 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2866 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2867 on the insn being emitted.
2869 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2870 Will Deacon <will.deacon@arm.com>
2872 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2874 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2876 * config/arm/arm-tables.opt: Regenerate.
2878 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2880 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2881 for vector types other than V16QImode.
2882 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2883 define_expand, and call altivec_expand_vec_perm_le when producing
2884 code with little endian element order.
2885 (*altivec_vperm_<mode>_internal): New insn having previous
2886 behavior of altivec_vperm_<mode>.
2887 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2888 altivec_expand_vec_perm_le when producing code with little endian
2890 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2891 behavior of altivec_vperm_<mode>_uns.
2893 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2895 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2896 (altivec_vsumsws): Add handling for -maltivec=be with a little
2898 (altivec_vsumsws_direct): New.
2899 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2900 gen_altivec_vsumsws.
2902 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2904 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2905 vtable_pointer_value_to_binfo): New functions.
2906 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2907 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2909 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2911 * config/nios2/nios2.md (load_got_register): Initialize GOT
2912 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2913 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2915 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2917 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2918 preserverd by passthrough, do not propagate the type.
2920 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2922 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2923 (mips_atomic_assign_expand_fenv): New function.
2924 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2926 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2928 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2929 (__builtin_mips_set_fcsr): Likewise.
2930 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2931 MIPS_USI_FTYPE_VOID.
2932 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2933 (mips16_expand_set_fcsr): Likewise.
2934 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2935 (mips16_set_fcsr_stub): Likewise.
2936 (mips16_get_fcsr_one_only_stub): New class.
2937 (mips16_set_fcsr_one_only_stub): Likewise.
2938 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2939 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2940 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2941 (hard_float): New availability predicate.
2942 (mips_builtins): Add get_fcsr and set_fcsr.
2943 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2944 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2945 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2946 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2947 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2950 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2952 * config/mips/mips.c (mips_one_only_stub): New class.
2953 (mips_need_mips16_rdhwr_p): Replace with...
2954 (mips16_rdhwr_stub): ...this new variable.
2955 (mips16_stub_call_address): New function.
2956 (mips16_rdhwr_one_only_stub): New class.
2957 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2958 (mips_output_mips16_rdhwr): Delete.
2959 (mips_finish_stub): New function.
2960 (mips_code_end): Use it to handle rdhwr stubs.
2962 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2965 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2966 when calculating size of integer atomic types.
2968 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2970 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2972 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2974 PR tree-optimization/60003
2975 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2976 * profile.c (branch_prob): Use gimple_call_builtin_p
2977 to check for BUILT_IN_SETJMP_RECEIVER.
2978 * tree-inline.c (copy_bb): Call notice_special_calls.
2980 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2983 * lra-constraints.c (process_alt_operands): Update reload_sum only
2986 2014-01-31 Richard Henderson <rth@redhat.com>
2989 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2990 until after else_eh is processed.
2992 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2994 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2995 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2996 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2997 in smmintrin.h, remove them.
2998 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2999 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3000 * config/i386/i386.md (ROUND_SAE): Fix value.
3001 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3002 (const48_operand): New.
3003 * config/i386/subst.md (round), (round_expand): Use
3004 const_4_or_8_to_11_operand.
3005 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3007 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3009 * config/i386/constraints.md (Yk): Swap meaning with k.
3010 * config/i386/i386.md (movhi_internal): Change Yk to k.
3011 (movqi_internal): Ditto.
3012 (*k<logic><mode>): Ditto.
3015 (kandn<mode>): Ditto.
3016 (*<code>hi_1): Ditto.
3017 (*<code>qi_1): Ditto.
3018 (kxnor<mode>): Ditto.
3019 (kortestzhi): Ditto.
3020 (kortestchi): Ditto.
3022 (*one_cmplhi2_1): Ditto.
3023 (*one_cmplqi2_1): Ditto.
3024 * config/i386/sse.md (): Change k to Yk.
3025 (avx512f_load<mode>_mask): Ditto.
3026 (avx512f_blendm<mode>): Ditto.
3027 (avx512f_store<mode>_mask): Ditto.
3028 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3029 (avx512f_storedqu<mode>_mask): Ditto.
3030 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3032 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3033 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3034 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3035 (avx512f_maskcmp<mode>3): Ditto.
3036 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3037 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3038 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3039 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3040 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3041 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3042 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3043 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3044 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3045 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3046 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3047 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3048 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3049 (vec_extract_lo_<mode>_maskm): Ditto.
3050 (vec_extract_hi_<mode>_maskm): Ditto.
3051 (avx512f_vternlog<mode>_mask): Ditto.
3052 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3053 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3054 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3055 (avx512f_<code>v8div16qi2_mask): Ditto.
3056 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3057 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3058 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3059 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3060 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3061 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3062 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3063 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3064 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3065 (avx512cd_maskb_vec_dupv8di): Ditto.
3066 (avx512cd_maskw_vec_dupv16si): Ditto.
3067 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3068 (avx512f_vpermi2var<mode>3_mask): Ditto.
3069 (avx512f_vpermi2var<mode>3_mask): Ditto.
3070 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3071 (*avx512f_gathersi<mode>): Ditto.
3072 (*avx512f_gathersi<mode>_2): Ditto.
3073 (*avx512f_gatherdi<mode>): Ditto.
3074 (*avx512f_gatherdi<mode>_2): Ditto.
3075 (*avx512f_scattersi<mode>): Ditto.
3076 (*avx512f_scatterdi<mode>): Ditto.
3077 (avx512f_compress<mode>_mask): Ditto.
3078 (avx512f_compressstore<mode>_mask): Ditto.
3079 (avx512f_expand<mode>_mask): Ditto.
3080 * config/i386/subst.md (mask): Change k to Yk.
3081 (mask_scalar_merge): Ditto.
3084 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3086 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3088 2014-01-31 Richard Biener <rguenther@suse.de>
3091 * builtins.c (fold_builtin_memory_op): Make sure to not
3092 use a floating-point mode or a boolean or enumeral type for
3095 2014-01-30 DJ Delorie <dj@redhat.com>
3097 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3098 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3099 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3100 whenever main() has an epilogue.
3102 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3104 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3105 unused variable "field".
3106 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3107 (vsx_mergeh_<mode>): Likewise.
3108 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3109 (altivec_vmrghh): Likewise.
3110 (altivec_vmrghw): Likewise.
3111 (altivec_vmrglb): Likewise.
3112 (altivec_vmrglh): Likewise.
3113 (altivec_vmrglw): Likewise.
3114 (altivec_vspltb): Add missing uses.
3115 (altivec_vsplth): Likewise.
3116 (altivec_vspltw): Likewise.
3117 (altivec_vspltsf): Likewise.
3119 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3122 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3123 frame related instructions.
3125 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3127 PR rtl-optimization/59959
3128 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3129 any reload of register whose subreg is invalid.
3131 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3133 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3134 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3135 Add missing return type - void.
3137 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3139 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3140 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3141 remove element index adjustment for endian (now handled in vsx.md
3143 (altivec_expand_vec_perm_const): Use
3144 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3145 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3146 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3147 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3148 define_expand and a new define_insn *altivec_vspltb_internal;
3149 adjust for -maltivec=be on a little endian target.
3150 (altivec_vspltb_direct): New.
3151 (altivec_vsplth): Divide into a define_expand and a new
3152 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3153 little endian target.
3154 (altivec_vsplth_direct): New.
3155 (altivec_vspltw): Divide into a define_expand and a new
3156 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3157 little endian target.
3158 (altivec_vspltw_direct): New.
3159 (altivec_vspltsf): Divide into a define_expand and a new
3160 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3161 a little endian target.
3163 2014-01-30 Richard Biener <rguenther@suse.de>
3165 PR tree-optimization/59993
3166 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3167 can propagate form the earlier stmt and avoid the transform
3168 when the intermediate result is needed.
3170 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3172 * README.Portability: Fix typo.
3174 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3176 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3177 comparison_operator with ordered_comparison_operator.
3179 2014-01-30 Nick Clifton <nickc@redhat.com>
3181 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3182 Rename to mn10300_store_multiple_regs.
3183 * config/mn10300/mn10300.c: Likewise.
3184 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3185 store_multiple_regs.
3186 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3187 Call mn10300_store_multiple_regs.
3189 2014-01-30 Nick Clifton <nickc@redhat.com>
3190 DJ Delorie <dj@redhat.com>
3192 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3193 %fp 2 to keep registers after it properly word-aligned.
3194 (rl78_alloc_physical_registers_umul): Handle the case where both
3195 input operands are the same.
3197 2014-01-30 Richard Biener <rguenther@suse.de>
3199 PR tree-optimization/59903
3200 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3203 2014-01-30 Jason Merrill <jason@redhat.com>
3206 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3209 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3211 2014-01-30 Richard Biener <rguenther@suse.de>
3213 PR tree-optimization/59951
3214 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3216 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3219 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3220 SFmode to DFmode case.
3222 2014-01-29 DJ Delorie <dj@redhat.com>
3224 * config/msp430/msp430.opt (-minrt): New.
3225 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3227 (ENDFILE_SPEC): Likewise.
3229 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3231 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3232 (estimate_function_body_sizes): Use it.
3234 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3237 * dwarf2out.c (is_cxx_auto): New.
3238 (is_base_type): Use it.
3239 (gen_type_die_with_usage): Likewise.
3241 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3243 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3244 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3245 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3246 -maltivec=be with LE targets.
3247 (vsx_mergeh_<mode>): Likewise.
3248 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3249 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3250 (altivec_vmrghb): Replace with define_expand and new
3251 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3252 (altivec_vmrghb_direct): New define_insn.
3253 (altivec_vmrghh): Replace with define_expand and new
3254 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3255 (altivec_vmrghh_direct): New define_insn.
3256 (altivec_vmrghw): Replace with define_expand and new
3257 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3258 (altivec_vmrghw_direct): New define_insn.
3259 (*altivec_vmrghsf): Adjust for endianness.
3260 (altivec_vmrglb): Replace with define_expand and new
3261 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3262 (altivec_vmrglb_direct): New define_insn.
3263 (altivec_vmrglh): Replace with define_expand and new
3264 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3265 (altivec_vmrglh_direct): New define_insn.
3266 (altivec_vmrglw): Replace with define_expand and new
3267 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3268 (altivec_vmrglw_direct): New define_insn.
3269 (*altivec_vmrglsf): Adjust for endianness.
3270 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3271 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3272 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3273 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3274 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3275 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3276 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3277 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3279 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3281 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3282 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3285 2014-01-29 Richard Biener <rguenther@suse.de>
3287 PR tree-optimization/58742
3288 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3289 associate_pointerplus_align.
3290 (associate_pointerplus_diff): New function.
3291 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3292 and associate_pointerplus_diff.
3294 2014-01-29 Richard Biener <rguenther@suse.de>
3296 * lto-streamer.h (LTO_major_version): Bump to 3.
3297 (LTO_minor_version): Reset to 0.
3299 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3301 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3302 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3303 (arm_file_start): Generate correct asm header for armv7ve.
3304 * config/arm/bpabi.h: Add multilib support for armv7ve.
3305 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3306 and cortex-a15 to armv7ve.
3307 * config/arm/t-aprofile: Add multilib support for armv7ve.
3308 * doc/invoke.texi: Document -march=armv7ve.
3310 2014-01-29 Richard Biener <rguenther@suse.de>
3312 PR tree-optimization/58742
3313 * tree-ssa-forwprop.c (associate_plusminus): Return true
3314 if we changed sth, defer EH cleanup to ...
3315 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3316 (simplify_mult): New function.
3318 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3321 PR tree-optimization/59920
3322 * tree.c (build_common_builtin_nodes): Remove
3323 __builtin_setjmp_dispatcher initialization.
3324 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3325 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3326 instead of gsi_after_labels + manually skipping debug stmts.
3327 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3328 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3329 * tree-inline.c (copy_edges_for_bb): Remove
3330 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3331 argument. Ignore computed_goto_p stmts. Don't call
3332 make_abnormal_goto_edges. If a call might need abnormal edges
3333 for non-local gotos, see if it already has an edge to
3334 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3335 with true argument, don't do anything then, otherwise add
3336 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3337 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3339 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3340 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3341 (lower_stmt): Don't set data->calls_builtin_setjmp.
3342 (lower_builtin_setjmp): Adjust comment.
3343 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3344 * tree-cfg.c (found_computed_goto): Remove.
3345 (factor_computed_gotos): Remove.
3346 (make_goto_expr_edges): Return bool, true for computed gotos.
3347 Don't call make_abnormal_goto_edges.
3348 (build_gimple_cfg): Don't set found_computed_goto, don't call
3349 factor_computed_gotos.
3350 (computed_goto_p): No longer static.
3351 (make_blocks): Don't set found_computed_goto.
3352 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3353 (make_edges): If make_goto_expr_edges returns true, push bb
3354 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3355 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3356 vector. Record mapping between bbs and OpenMP regions if there
3357 are any, adjust make_gimple_omp_edges caller. Call
3358 handle_abnormal_edges.
3359 (make_abnormal_goto_edges): Remove.
3360 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3361 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3362 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3363 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3364 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3365 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3366 filling *region also set *region_idx to (*region)->entry->index.
3369 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3370 For REGs set ORIGINAL_REGNO.
3372 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3374 * doc/md.texi: Mention that a target shouldn't implement
3375 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3378 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3380 PR tree-optimization/59594
3381 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3382 a copy of the datarefs vector rather than the vector itself.
3384 2014-01-28 Jason Merrill <jason@redhat.com>
3387 * dwarf2out.c (auto_die): New static.
3388 (gen_type_die_with_usage): Handle C++1y 'auto'.
3389 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3392 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3395 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3396 (SPEC_X32): Likewise.
3397 (SPEC_64): Likewise.
3398 * config/i386/i386.c (ix86_option_override_internal): Turn off
3399 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3401 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3402 * config/i386/i386.h (TARGET_16BIT): New macro.
3403 (TARGET_16BIT_P): Likewise.
3404 * config/i386/i386.opt: Add m16.
3405 * doc/invoke.texi: Document -m16.
3407 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3409 PR preprocessor/59935
3410 * input.c (location_get_source_line): Bail out on when line number
3411 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3413 2014-01-28 Richard Biener <rguenther@suse.de>
3415 PR tree-optimization/58742
3416 * tree-ssa-forwprop.c (associate_plusminus): Handle
3417 pointer subtraction of the form (T)(P + A) - (T)P.
3419 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3421 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3424 2014-01-28 Richard Biener <rguenther@suse.de>
3427 2014-01-28 Richard Biener <rguenther@suse.de>
3429 PR rtl-optimization/45364
3430 PR rtl-optimization/59890
3431 * var-tracking.c (local_get_addr_clear_given_value): Handle
3432 already cleared slot.
3433 (val_reset): Handle not allocated local_get_addr_cache.
3434 (vt_find_locations): Use post-order on the inverted CFG.
3436 2014-01-28 Richard Biener <rguenther@suse.de>
3438 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3440 2014-01-28 Richard Biener <rguenther@suse.de>
3442 PR rtl-optimization/45364
3443 PR rtl-optimization/59890
3444 * var-tracking.c (local_get_addr_clear_given_value): Handle
3445 already cleared slot.
3446 (val_reset): Handle not allocated local_get_addr_cache.
3447 (vt_find_locations): Use post-order on the inverted CFG.
3449 2014-01-28 Alan Modra <amodra@gmail.com>
3451 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3452 * configure.ac <recursive call for build != host>: Define
3453 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3454 and LD_FOR_BUILD too.
3455 * configure: Regenerate.
3457 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3459 * config/i386/i386.c (get_builtin_code_for_version): Separate
3460 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3461 Broadwell from Haswell.
3463 2014-01-27 Steve Ellcey <sellcey@mips.com>
3465 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3466 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3467 * config/mips/mips.c (mips_option_override): Change setting
3469 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3470 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3471 Change from Mask to Var.
3473 2014-01-27 Jeff Law <law@redhat.com>
3475 * ipa-inline.c (inline_small_functions): Fix typo.
3477 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3479 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3480 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3481 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3482 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3483 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3484 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3485 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3486 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3487 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3488 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3489 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3490 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3491 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3492 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3493 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3494 (_mm512_storeu_epi64): Ditto.
3495 (_mm512_cmpge_epi32_mask): Ditto.
3496 (_mm512_cmpge_epu32_mask): Ditto.
3497 (_mm512_cmpge_epi64_mask): Ditto.
3498 (_mm512_cmpge_epu64_mask): Ditto.
3499 (_mm512_cmple_epi32_mask): Ditto.
3500 (_mm512_cmple_epu32_mask): Ditto.
3501 (_mm512_cmple_epi64_mask): Ditto.
3502 (_mm512_cmple_epu64_mask): Ditto.
3503 (_mm512_cmplt_epi32_mask): Ditto.
3504 (_mm512_cmplt_epu32_mask): Ditto.
3505 (_mm512_cmplt_epi64_mask): Ditto.
3506 (_mm512_cmplt_epu64_mask): Ditto.
3507 (_mm512_cmpneq_epi32_mask): Ditto.
3508 (_mm512_cmpneq_epu32_mask): Ditto.
3509 (_mm512_cmpneq_epi64_mask): Ditto.
3510 (_mm512_cmpneq_epu64_mask): Ditto.
3511 (_mm512_expand_pd): Ditto.
3512 (_mm512_expand_ps): Ditto.
3513 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3514 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3515 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3516 * config/i386/i386.c (ix86_builtins): Add
3517 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3518 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3519 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3520 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3521 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3522 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3523 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3524 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3525 IX86_BUILTIN_PMOVUSQW512_MEM.
3526 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3527 __builtin_ia32_pmovsqd512mem_mask,
3528 __builtin_ia32_pmovqd512mem_mask,
3529 __builtin_ia32_pmovusqw512mem_mask,
3530 __builtin_ia32_pmovsqw512mem_mask,
3531 __builtin_ia32_pmovqw512mem_mask,
3532 __builtin_ia32_pmovusdw512mem_mask,
3533 __builtin_ia32_pmovsdw512mem_mask,
3534 __builtin_ia32_pmovdw512mem_mask,
3535 __builtin_ia32_pmovqb512mem_mask,
3536 __builtin_ia32_pmovusqb512mem_mask,
3537 __builtin_ia32_pmovsqb512mem_mask,
3538 __builtin_ia32_pmovusdb512mem_mask,
3539 __builtin_ia32_pmovsdb512mem_mask,
3540 __builtin_ia32_pmovdb512mem_mask.
3541 (bdesc_args): Add __builtin_ia32_expanddf512,
3542 __builtin_ia32_expandsf512.
3543 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3544 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3545 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3546 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3547 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3548 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3549 (avx512f_<code>v8div16qi2_mask_store): This.
3550 (avx512f_expand<mode>): New.
3552 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3554 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3556 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3557 (_mm512_prefetch_i32scatter_pd): Ditto.
3558 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3559 (_mm512_prefetch_i64scatter_pd): Ditto.
3560 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3561 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3562 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3563 (_mm512_prefetch_i32scatter_ps): Ditto.
3564 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3565 (_mm512_prefetch_i64scatter_ps): Ditto.
3566 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3567 * config/i386/i386-builtin-types.def: Define
3568 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3569 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3570 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3571 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3572 IX86_BUILTIN_SCATTERPFQPD.
3573 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3574 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3575 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3576 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3577 __builtin_ia32_scatterpfqps.
3578 (ix86_expand_builtin): Expand new built-ins.
3579 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3580 fix memory access data type.
3581 (*avx512pf_gatherpf<mode>_mask): Ditto.
3582 (*avx512pf_gatherpf<mode>): Ditto.
3583 (avx512pf_scatterpf<mode>): Ditto.
3584 (*avx512pf_scatterpf<mode>_mask): Ditto.
3585 (*avx512pf_scatterpf<mode>): Ditto.
3586 (GATHER_SCATTER_SF_MEM_MODE): New.
3587 (avx512pf_gatherpf<mode>df): Ditto.
3588 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3589 (*avx512pf_scatterpf<mode>df): Ditto.
3591 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3594 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3595 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3598 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3600 * common/config/arm/arm-common.c
3601 (arm_rewrite_mcpu): Handle multiple names.
3603 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3605 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3607 * gimple-builder.h (create_gimple_tmp): Delete.
3609 2014-01-27 Christian Bruel <christian.bruel@st.com>
3611 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3614 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3616 * config/pa/pa.md (call): Generate indirect long calls to non-local
3617 functions when outputing 32-bit code.
3618 (call_value): Likewise except for special call to buggy powf function.
3620 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3621 portable runtime and PIC indirect calls.
3622 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3623 and PIC call sequences. Use ldo instead of blr to set return register
3624 in PIC call sequence.
3626 2014-01-25 Walter Lee <walt@tilera.com>
3628 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3629 avoid clobbering a live register.
3631 2014-01-25 Walter Lee <walt@tilera.com>
3633 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3634 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3635 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3636 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3638 2014-01-25 Walter Lee <walt@tilera.com>
3640 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3641 arguments on even registers.
3642 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3644 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3645 (BIGGEST_ALIGNMENT): Ditto.
3646 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3648 2014-01-25 Walter Lee <walt@tilera.com>
3650 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3651 insns before bundling.
3652 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3654 2014-01-25 Walter Lee <walt@tilera.com>
3656 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3657 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3658 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3660 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3662 * config/mips/constraints.md (kl): Delete.
3663 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3664 define expands, using...
3665 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3666 instructions for MIPS16.
3667 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3668 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3670 2014-01-25 Walter Lee <walt@tilera.com>
3672 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3676 2014-01-25 Walter Lee <walt@tilera.com>
3678 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3679 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3681 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3683 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3686 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3688 * print-rtl.c (in_call_function_usage): New var.
3689 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3690 EXPR_LIST mode as mode and not as reg note name.
3693 * cfgloopmanip.c (copy_loop_info): If
3694 loop->warned_aggressive_loop_optimizations, make sure
3695 the flag is set in target loop too.
3697 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3699 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3701 * builtins.def: Likewise.
3702 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3703 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3704 * ira.c (ira_setup_eliminable_regset): Likewise.
3705 * omp-low.c (gate_expand_omp): Likewise.
3706 (execute_lower_omp): Likewise.
3707 (diagnose_sb_0): Likewise.
3708 (gate_diagnose_omp_blocks): Likewise.
3709 (simd_clone_clauses_extract): Likewise.
3712 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3714 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3715 correction for little endian...
3716 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3719 2014-01-24 Jeff Law <law@redhat.com>
3721 PR tree-optimization/59919
3722 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3723 for non-returning calls.
3725 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3727 * common/config/aarch64/aarch64-common.c
3728 (aarch64_rewrite_mcpu): Handle multiple names.
3729 * config/aarch64/aarch64.h
3730 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3732 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3734 * input.c (add_file_to_cache_tab): Handle the case where fopen
3737 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3740 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3741 from push operand if code of push isn't PRE_DEC.
3743 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3746 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3747 -mquad-memory-atomic. Update -mquad-memory documentation to say
3748 it is only used for non-atomic loads/stores.
3750 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3751 -mquad-memory or -mquad-memory-atomic switches.
3753 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3754 -mquad-memory-atomic to ISA 2.07 support.
3756 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3757 to separate support of normal quad word memory operations (ldq, stq)
3758 from the atomic quad word memory operations.
3760 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3761 support to separate non-atomic quad word operations from atomic
3762 quad word operations. Disable non-atomic quad word operations in
3763 little endian mode so that we don't have to swap words after the
3764 load and before the store.
3765 (quad_load_store_p): Add comment about atomic quad word support.
3766 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3767 options printed with -mdebug=reg.
3769 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3770 -mquad-memory-atomic as the test for whether we have quad word
3771 atomic instructions.
3772 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3773 or -mp8-vector are used, allow byte/half-word atomic operations.
3775 * config/rs6000/sync.md (load_lockedti): Insure that the address
3776 is a proper indexed or indirect address for the lqarx instruction.
3777 On little endian systems, swap the hi/lo registers after the lqarx
3779 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3780 insure the address is valid for the lqarx instruction.
3781 (store_conditionalti): Insure that the address is a proper indexed
3782 or indirect address for the stqcrx. instruction. On little endian
3783 systems, swap the hi/lo registers before doing the stqcrx.
3785 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3786 insure the address is valid for the stqcrx. instruction.
3788 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3789 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3790 type of quad memory support is available.
3792 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3795 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3796 there is a danger of looping.
3798 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3800 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3801 force flag_ira_loop_pressure if set via command line.
3803 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3805 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3806 (ashr_simd): New builtin handling DI mode.
3807 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3808 (aarch64_sshr_simddi): New match pattern.
3809 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3810 (vshrd_n_s64): Likewise.
3811 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3813 2014-01-23 Nick Clifton <nickc@redhat.com>
3815 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3816 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3817 favour of mcu specific scripts.
3818 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3821 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3822 Alex Velenko <Alex.Velenko@arm.com>
3824 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3825 (vaddv_s16): Likewise.
3826 (vaddv_s32): Likewise.
3827 (vaddv_u8): Likewise.
3828 (vaddv_u16): Likewise.
3829 (vaddv_u32): Likewise.
3830 (vaddvq_s8): Likewise.
3831 (vaddvq_s16): Likewise.
3832 (vaddvq_s32): Likewise.
3833 (vaddvq_s64): Likewise.
3834 (vaddvq_u8): Likewise.
3835 (vaddvq_u16): Likewise.
3836 (vaddvq_u32): Likewise.
3837 (vaddvq_u64): Likewise.
3838 (vaddv_f32): Likewise.
3839 (vaddvq_f32): Likewise.
3840 (vaddvq_f64): Likewise.
3841 (vmaxv_f32): Likewise.
3842 (vmaxv_s8): Likewise.
3843 (vmaxv_s16): Likewise.
3844 (vmaxv_s32): Likewise.
3845 (vmaxv_u8): Likewise.
3846 (vmaxv_u16): Likewise.
3847 (vmaxv_u32): Likewise.
3848 (vmaxvq_f32): Likewise.
3849 (vmaxvq_f64): Likewise.
3850 (vmaxvq_s8): Likewise.
3851 (vmaxvq_s16): Likewise.
3852 (vmaxvq_s32): Likewise.
3853 (vmaxvq_u8): Likewise.
3854 (vmaxvq_u16): Likewise.
3855 (vmaxvq_u32): Likewise.
3856 (vmaxnmv_f32): Likewise.
3857 (vmaxnmvq_f32): Likewise.
3858 (vmaxnmvq_f64): Likewise.
3859 (vminv_f32): Likewise.
3860 (vminv_s8): Likewise.
3861 (vminv_s16): Likewise.
3862 (vminv_s32): Likewise.
3863 (vminv_u8): Likewise.
3864 (vminv_u16): Likewise.
3865 (vminv_u32): Likewise.
3866 (vminvq_f32): Likewise.
3867 (vminvq_f64): Likewise.
3868 (vminvq_s8): Likewise.
3869 (vminvq_s16): Likewise.
3870 (vminvq_s32): Likewise.
3871 (vminvq_u8): Likewise.
3872 (vminvq_u16): Likewise.
3873 (vminvq_u32): Likewise.
3874 (vminnmv_f32): Likewise.
3875 (vminnmvq_f32): Likewise.
3876 (vminnmvq_f64): Likewise.
3878 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3880 * config/aarch64/aarch64-simd.md
3881 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3882 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3883 (*aarch64_mul3_elt<mode>): Likewise.
3884 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3885 (*aarch64_mul3_elt_to_64v2df): Likewise.
3886 (*aarch64_mla_elt<mode>): Likewise.
3887 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3888 (*aarch64_mls_elt<mode>): Likewise.
3889 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3890 (*aarch64_fma4_elt<mode>): Likewise.
3891 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3892 (*aarch64_fma4_elt_to_64v2df): Likewise.
3893 (*aarch64_fnma4_elt<mode>): Likewise.
3894 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3895 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3896 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3897 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3898 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3899 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3900 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3901 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3902 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3904 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3906 * config/aarch64/aarch64-simd.md
3907 (aarch64_be_checked_get_lane<mode>): New define_expand.
3908 * config/aarch64/aarch64-simd-builtins.def
3909 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3910 New builtin definition.
3911 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3912 Use new safe be builtin.
3914 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3916 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3918 (aarch64_be_st1<mode>): Likewise.
3919 (aarch_ld1<VALL:mode>): Define_expand modified.
3920 (aarch_st1<VALL:mode>): Likewise.
3921 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3922 (UNSPEC_ST1): Likewise.
3924 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3926 * config/microblaze/microblaze.md: Add trap insn and attribute
3928 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3930 PR preprocessor/58580
3931 * input.h (location_get_source_line): Take an additional line_size
3933 (void diagnostics_file_cache_fini): Declare new function.
3934 * input.c (struct fcache): New type.
3935 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3936 New static constants.
3937 (diagnostic_file_cache_init, total_lines_num)
3938 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3939 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3940 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3941 (get_next_line, read_next_line, goto_next_line, read_line_num):
3942 New static function definitions.
3943 (diagnostic_file_cache_fini): New function.
3944 (location_get_source_line): Take an additional output line_len
3945 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3947 * diagnostic.c (diagnostic_finish): Call
3948 diagnostic_file_cache_fini.
3949 (adjust_line): Take an additional input parameter for the length
3950 of the line, rather than calculating it with strlen.
3951 (diagnostic_show_locus): Adjust the use of
3952 location_get_source_line and adjust_line with respect to their new
3953 signature. While displaying a line now, do not stop at the first
3954 null byte. Rather, display the zero byte as a space and keep
3955 going until we reach the size of the line.
3956 * Makefile.in: Add vec.o to OBJS-libcommon
3958 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3959 Ilya Tocar <ilya.tocar@intel.com>
3961 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3962 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3963 (__builtin_ia32_kmov16): Ditto.
3964 * config/i386/i386.md (UNSPEC_KMOV): New.
3967 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3969 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3970 (_mm512_storeu_si512): Ditto.
3972 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3975 * rtl.h (get_referenced_operands): Declare.
3976 * recog.c (get_referenced_operands): New function.
3977 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3978 operands have been referenced when recording LO_SUM references.
3980 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3982 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3984 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3986 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3987 Enable for generic and recent AMD targets.
3989 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3991 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3992 ARG_SIZE note when adjustment was eliminated.
3994 2014-01-22 Jeff Law <law@redhat.com>
3996 PR tree-optimization/59597
3997 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3998 in file. Accept new argument REGISTERING and use it to modify
3999 dump output appropriately.
4000 (register_jump_thread): Corresponding changes.
4001 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4002 thread paths involving joiner blocks. Add code to dump cancelled
4003 jump threading paths.
4005 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4007 PR rtl-optimization/59477
4008 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4009 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4011 2014-01-22 Tom Tromey <tromey@redhat.com>
4013 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4015 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4017 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4019 PR rtl-optimization/59896
4020 * lra-constraints.c (process_alt_operands): Check unused note for
4021 matched operands of insn with no output reloads.
4023 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4025 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4026 (mips_move_from_gpr_cost): Likewise.
4028 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4030 PR rtl-optimization/59858
4031 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4032 ira_class_hard_regs_num.
4033 (process_alt_operands): Increase reject for dying matched operand.
4035 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4038 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4039 smaller than size, perform several stores or loads and stores
4040 at dst + count - size to store or copy all of size bytes, rather
4041 than just last modesize bytes.
4043 2014-01-20 DJ Delorie <dj@redhat.com>
4045 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4046 that CLOBBERs are REGs before propogating their values.
4048 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4051 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4052 (cgraph_inline_failed_type): New function.
4053 * cgraph.h (DEFCIFCODE): Add type.
4054 (cgraph_inline_failed_type_t): New enum.
4055 (cgraph_inline_failed_type): New prototype.
4056 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4057 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4058 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4059 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4060 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4061 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4062 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4063 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4064 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4065 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4066 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4067 OPTIMIZATION_MISMATCH.
4068 * tree-inline.c (expand_call_inline): Emit errors during
4069 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4071 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4074 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4075 mode attribute in insn output.
4077 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4079 * output.h (output_constant): Delete.
4080 * varasm.c (output_constant): Make private.
4082 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4084 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4086 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4089 * tree.h (fold_builtin_strcat): New prototype.
4090 * builtins.c (fold_builtin_strcat): No longer static. Add len
4091 argument, if non-NULL, don't call c_strlen. Optimize
4092 directly into __builtin_memcpy instead of __builtin_strcpy.
4093 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4094 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4096 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4098 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4099 for SImode_address_operand operands, having only a REG argument.
4101 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4103 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4104 loader name using mbig-endian.
4105 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4107 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4109 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4113 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4115 * config/aarch64/aarch64-protos.h
4116 (aarch64_cannot_change_mode_class_ptr): Declare.
4117 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4118 aarch64_cannot_change_mode_class_ptr): New.
4119 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4120 backend hook aarch64_cannot_change_mode_class.
4122 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4124 * common/config/aarch64/aarch64-common.c
4125 (aarch64_handle_option): Don't handle any option order logic here.
4126 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4127 selected_cpu, warn on architecture version mismatch.
4128 (aarch64_override_options): Fix parsing order for option strings.
4130 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4131 Iain Sandoe <iain@codesourcery.com>
4134 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4135 warning. Amend comment to reflect current functionality.
4137 2014-01-20 Richard Biener <rguenther@suse.de>
4140 * builtins.c (fold_builtin_strcat): Remove case better handled
4141 by tree-ssa-strlen.c.
4143 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4145 * config/aarch64/aarch64.opt
4146 (mcpu, march, mtune): Make case-insensitive.
4148 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4151 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4152 if operands[1] is a REG or ZERO_EXTEND of a REG.
4154 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4156 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4158 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4160 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4161 long non-pic millicode calls.
4163 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4165 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4167 2014-01-19 Kito Cheng <kito@0xlab.org>
4169 * builtins.c (expand_movstr): Check movstr expand done or fail.
4171 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4172 H.J. Lu <hongjiu.lu@intel.com>
4175 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4176 to DImode for zero-extended addresses.
4178 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4180 PR rtl-optimization/57763
4181 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4182 on the new indirect jump_insn and increment LABEL_NUSES (label).
4184 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4188 * config.gcc (x86_archs): New variable.
4189 (x86_64_archs): Likewise.
4190 (x86_cpus): Likewise.
4191 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4192 --with-arch/--with-cpu= options.
4193 Support --with-arch=/--with-cpu={nehalem,westmere,
4194 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4196 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4198 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4199 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4201 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4203 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4205 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4208 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4209 clear cpp_get_options (parse_in)->warn_unused_macros for
4210 ix86_target_macros_internal with cpp_define.
4212 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4214 * jump.c (delete_related_insns): Keep (use (insn))s.
4215 * reorg.c (redundant_insn): Check for barriers too.
4217 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4219 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4221 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4223 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4224 call to $$dyncall when TARGET_LONG_CALLS is true.
4226 2014-01-17 Jeff Law <law@redhat.com>
4228 * ree.c (combine_set_extension): Temporarily disable test for
4229 changing number of hard registers.
4231 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4234 * ipa-inline-analysis.c (inline_free_summary):
4235 Do not free summary of aliases.
4237 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4240 * gimplify.c (gimplify_expr): Use create_tmp_var
4241 instead of create_tmp_var_raw. If cond doesn't have
4242 integral type, don't add the IFN_ANNOTATE builtin at all.
4244 2014-01-17 Martin Jambor <mjambor@suse.cz>
4247 * ipa-cp.c (prev_edge_clone): New variable.
4248 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4249 Also resize prev_edge_clone vector.
4250 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4251 (ipcp_edge_removal_hook): New function.
4252 (ipcp_driver): Register ipcp_edge_removal_hook.
4254 2014-01-17 Andrew Pinski <apinski@cavium.com>
4255 Steve Ellcey <sellcey@mips.com>
4258 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4261 2014-01-17 Jeff Law <law@redhat.com>
4264 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4265 so that pass_ccp runs first.
4267 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4269 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4270 (ix86_adjust_cost): Use !TARGET_XXX.
4271 (do_reorder_for_imul): Likewise.
4272 (swap_top_of_ready_list): Likewise.
4273 (ix86_sched_reorder): Likewise.
4275 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4277 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4278 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4279 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4280 (intel_memset): New. Duplicate slm_memset.
4281 (intel_cost): New. Duplicate slm_cost.
4282 (m_INTEL): New macro.
4283 (processor_target_table): Add "intel".
4284 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4285 with PROCESSOR_INTEL for "intel".
4286 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4287 PROCESSOR_SILVERMONT.
4288 (ix86_issue_rate): Likewise.
4289 (ix86_adjust_cost): Likewise.
4290 (ia32_multipass_dfa_lookahead): Likewise.
4291 (swap_top_of_ready_list): Likewise.
4292 (ix86_sched_reorder): Likewise.
4293 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4294 instead of TARGET_OPT_AGU.
4295 * config/i386/i386.h (TARGET_INTEL): New.
4296 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4297 (processor_type): Add PROCESSOR_INTEL.
4298 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4299 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4301 2014-01-17 Marek Polacek <polacek@redhat.com>
4304 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4307 2014-01-17 Richard Biener <rguenther@suse.de>
4309 PR tree-optimization/46590
4310 * opts.c (default_options_table): Add entries for
4311 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4312 all enabled at -O1 but not for -Og.
4313 * common.opt (fbranch-count-reg): Remove Init(1).
4314 (fmove-loop-invariants): Likewise.
4315 (ftree-pta): Likewise.
4317 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4319 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4320 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4321 decls to at least the GCC 4.8 used alignments.
4324 * tree-nested.c (convert_nonlocal_reference_stmt,
4325 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4326 of GIMPLE_BIND stmts, adjust associated decls.
4328 2014-01-17 Richard Biener <rguenther@suse.de>
4330 PR tree-optimization/46590
4331 * vec.h (vec<>::bseach): New member function implementing
4332 binary search according to C89 bsearch.
4333 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4334 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4335 bitmap pointer again. Make accesses_in_loop a flat array.
4336 (mem_ref_obstack): New global.
4337 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4338 (mark_ref_stored): Likewise.
4339 (ref_indep_loop_p_2): Likewise.
4340 (set_ref_stored_in_loop): New helper function.
4341 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4342 (memref_free): Adjust.
4343 (record_mem_ref_loc): Simplify.
4344 (gather_mem_refs_stmt): Adjust.
4345 (sort_locs_in_loop_postorder_cmp): New function.
4346 (analyze_memory_references): Sort accesses_in_loop after
4347 loop postorder number.
4348 (find_ref_loc_in_loop_cmp): New function.
4349 (for_all_locs_in_loop): Find relevant cluster of locs in
4350 accesses_in_loop and iterate without recursion.
4351 (execute_sm): Avoid uninit warning.
4352 (struct ref_always_accessed): Simplify.
4353 (ref_always_accessed::operator ()): Likewise.
4354 (ref_always_accessed_p): Likewise.
4355 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4356 loop postorder numbers here.
4357 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4360 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4363 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4364 on decls for which assemble_alias has been called.
4366 2014-01-17 Nick Clifton <nickc@redhat.com>
4368 * config/msp430/msp430.opt: (mcpu): New option.
4369 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4370 (msp430_option_override): Parse target_cpu. If the MCU name
4371 matches a generic string, clear target_mcu.
4372 (msp430_attr): Allow numeric interrupt values up to 63.
4373 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4374 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4376 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4378 * config/msp430/msp430.md (popm): Use %J rather than %I.
4379 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4380 (addhi_cy_i): Use immediate_operand for operand 2.
4381 * doc/invoke.texi: Document -mcpu option.
4383 2014-01-17 Richard Biener <rguenther@suse.de>
4385 PR rtl-optimization/38518
4386 * df.h (df_analyze_loop): Declare.
4387 * df-core.c: Include cfgloop.h.
4388 (df_analyze_1): Split out main part of df_analyze.
4389 (df_analyze): Adjust.
4390 (loop_inverted_post_order_compute): New function.
4391 (loop_post_order_compute): Likewise.
4392 (df_analyze_loop): New function avoiding whole-function
4394 * loop-invariant.c (find_defs): Use df_analyze_loop.
4395 (find_invariants): Adjust.
4396 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4398 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4400 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4401 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4403 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4405 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4406 traversal when removing references.
4408 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4411 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4413 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4416 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4417 pushing a reload for an autoinc when we had previously reloaded an
4418 inner part of the address.
4420 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4422 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4424 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4425 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4426 when not giving up or versioning for alias only because of
4428 (vect_analyze_data_ref_dependences): Set to true.
4429 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4431 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4432 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4436 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4439 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4440 operand 0 predicate for gathers, use a new pseudo as subtarget.
4442 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4445 * lra-constraints.c (process_alt_operands): Add printing debug info.
4446 Check absence of input/output reloads for matched operands too.
4448 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4450 PR rtl-optimization/59835
4451 * ira.c (ira_init_register_move_cost): Increase cost for
4454 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4456 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4458 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4461 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4462 non-register objects. Use gen_(high/low)part more consistently.
4465 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4468 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4469 endian support, remove tests for WORDS_BIG_ENDIAN.
4470 (p8_mfvsrd_3_<mode>): Likewise.
4471 (reload_gpr_from_vsx<mode>): Likewise.
4472 (reload_gpr_from_vsxsf): Likewise.
4473 (p8_mfvsrd_4_disf): Likewise.
4475 2014-01-16 Richard Biener <rguenther@suse.de>
4477 PR rtl-optimization/46590
4478 * lcm.c (compute_antinout_edge): Use postorder iteration.
4479 (compute_laterin): Use inverted postorder iteration.
4481 2014-01-16 Nick Clifton <nickc@redhat.com>
4484 * varasm.c (output_constant): Return the number of bytes actually
4486 (output_constructor_array_range): Update the field size with the
4487 number of bytes emitted by output_constant.
4488 (output_constructor_regular_field): Likewise. Also do not
4489 complain if the total number of bytes emitted is now greater
4490 than the expected fieldpos.
4491 * output.h (output_constant): Update prototype and descriptive comment.
4493 2014-01-16 Marek Polacek <polacek@redhat.com>
4496 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4497 it is error_mark_node.
4499 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4501 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4502 VALID_AVX256_REG_OR_OI_MODE.
4504 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4506 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4507 current procedure should be profiled.
4509 2014-01-15 Andrew Pinski <apinski@cavium.com>
4511 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4512 of moving from/to the STACK_REG register class.
4514 2014-01-15 Richard Henderson <rth@redhat.com>
4517 * reginfo.c (global_regs_decl): Globalize.
4518 * rtl.h (global_regs_decl): Declare.
4519 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4520 reserved via global_regs.
4522 2014-01-15 Teresa Johnson <tejohnson@google.com>
4524 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4526 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4528 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4529 and vmulosh rather than call gen_vec_widen_smult_*.
4530 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4531 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4532 (vec_widen_smult_even_v16qi): Likewise.
4533 (vec_widen_umult_even_v8hi): Likewise.
4534 (vec_widen_smult_even_v8hi): Likewise.
4535 (vec_widen_umult_odd_v16qi): Likewise.
4536 (vec_widen_smult_odd_v16qi): Likewise.
4537 (vec_widen_umult_odd_v8hi): Likewise.
4538 (vec_widen_smult_odd_v8hi): Likewise.
4539 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4540 vmuloub rather than call gen_vec_widen_umult_*.
4541 (vec_widen_umult_lo_v16qi): Likewise.
4542 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4543 vmulosb rather than call gen_vec_widen_smult_*.
4544 (vec_widen_smult_lo_v16qi): Likewise.
4545 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4546 rather than call gen_vec_widen_umult_*.
4547 (vec_widen_umult_lo_v8hi): Likewise.
4548 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4549 rather than call gen_vec_widen_smult_*.
4550 (vec_widen_smult_lo_v8hi): Likewise.
4552 2014-01-15 Jeff Law <law@redhat.com>
4554 PR tree-optimization/59747
4555 * ree.c (find_and_remove_re): Properly handle case where a second
4556 eliminated extension requires widening a copy created for elimination
4557 of a prior extension.
4558 (combine_set_extension): Ensure that the number of hard regs needed
4559 for a destination register does not change when we widen it.
4561 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4563 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4564 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4565 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4566 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4567 (avr-*-rtems*): Likewise.
4568 (bfin*-rtems*): Likewise.
4569 (moxie-*-rtems*): Likewise.
4570 (h8300-*-rtems*): Likewise.
4571 (i[34567]86-*-rtems*): Likewise.
4572 (lm32-*-rtems*): Likewise.
4573 (m32r-*-rtems*): Likewise.
4574 (m68k-*-rtems*): Likewise.
4575 (microblaze*-*-rtems*): Likewise.
4576 (mips*-*-rtems*): Likewise.
4577 (powerpc-*-rtems*): Likewise.
4578 (sh-*-rtems*): Likewise.
4579 (sparc-*-rtems*): Likewise.
4580 (sparc64-*-rtems*): Likewise.
4581 (v850-*-rtems*): Likewise.
4582 (m32c-*-rtems*): Likewise.
4584 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4586 PR rtl-optimization/59511
4587 * ira.c (ira_init_register_move_cost): Use memory costs for some
4588 cases of register move cost calculations.
4589 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4590 instead of BB frequency.
4591 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4592 * lra-assigns.c (find_hard_regno_for): Ditto.
4594 2014-01-15 Richard Biener <rguenther@suse.de>
4596 PR tree-optimization/59822
4597 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4598 (vectorizable_load): Use it to hoist defs of uses of invariant
4599 loads out of the loop.
4601 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4602 Kugan Vivekanandarajah <kuganv@linaro.org>
4605 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4608 2014-01-15 Richard Biener <rguenther@suse.de>
4610 PR rtl-optimization/59802
4611 * lcm.c (compute_available): Use inverted postorder to seed
4612 the initial worklist.
4614 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4617 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4618 ADDR_REGS for invalid symrefs in non-PIC code.
4620 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4623 * builtins.c (determine_block_size): Initialize *probable_max_size
4624 even if len_rtx is CONST_INT.
4626 2014-01-14 Andrew Pinski <apinski@cavium.com>
4628 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4629 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4630 (cortexa53_tunings): Likewise.
4631 (aarch64_sched_issue_rate): New function.
4632 (TARGET_SCHED_ISSUE_RATE): Define.
4634 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4636 * ira-costs.c (find_costs_and_classes): Add missed
4637 ira_init_register_move_cost_if_necessary.
4639 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4642 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4644 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4647 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4648 to indicate if type is used for function return value. Warn ABI
4649 change if the vector mode isn't available for function return value.
4650 (ix86_function_arg_advance): Pass false to type_natural_mode.
4651 (ix86_function_arg): Likewise.
4652 (ix86_gimplify_va_arg): Likewise.
4653 (function_arg_32): Don't warn ABI change.
4654 (ix86_function_value): Pass true to type_natural_mode.
4655 (ix86_return_in_memory): Likewise.
4656 (ix86_struct_value_rtx): Removed.
4657 (TARGET_STRUCT_VALUE_RTX): Likewise.
4659 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4661 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4662 converting a conditional jump into a conditional return.
4664 2014-01-14 Richard Biener <rguenther@suse.de>
4666 PR tree-optimization/58921
4667 PR tree-optimization/59006
4668 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4669 hoisting invariant stmts.
4670 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4671 invariant loads on the preheader edge if possible.
4673 2014-01-14 Joey Ye <joey.ye@arm.com>
4675 * doc/plugin.texi (Building GCC plugins): Update to C++.
4677 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4679 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4680 (_mm_rcp28_round_ss): Ditto.
4681 (_mm_rsqrt28_round_sd): Ditto.
4682 (_mm_rsqrt28_round_ss): Ditto.
4683 (_mm_rcp28_sd): Ditto.
4684 (_mm_rcp28_ss): Ditto.
4685 (_mm_rsqrt28_sd): Ditto.
4686 (_mm_rsqrt28_ss): Ditto.
4687 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4688 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4689 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4690 (IX86_BUILTIN_RCP28SD): Ditto.
4691 (IX86_BUILTIN_RCP28SS): Ditto.
4692 (IX86_BUILTIN_RSQRT28SD): Ditto.
4693 (IX86_BUILTIN_RSQRT28SS): Ditto.
4694 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4695 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4696 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4697 (ix86_expand_special_args_builtin): Expand new FTYPE.
4698 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4699 (srcp14<mode>): Make insn unary.
4700 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4701 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4702 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4703 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4704 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4705 Fix rounding: make it SAE only.
4706 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4708 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4710 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4711 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4712 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4713 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4714 (round_saeonly_mask_scalar_operand4): Ditto.
4715 (round_saeonly_mask_scalar_op3): Ditto.
4716 (round_saeonly_mask_scalar_op4): Ditto.
4718 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4720 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4721 Implement -maltivec=be for vec_insert and vec_extract.
4723 2014-01-10 DJ Delorie <dj@redhat.com>
4725 * config/msp430/msp430.md (call_internal): Don't allow memory
4726 references with SP as the base register.
4727 (call_value_internal): Likewise.
4728 * config/msp430/constraints.md (Yc): New. For memory references
4729 that don't use SP as a base register.
4731 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4732 "an integer without a # prefix"
4733 * config/msp430/msp430.md (epilogue_helper): Use it.
4735 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4738 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4739 AVX512F gather builtins.
4740 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4741 on gather decls with INTEGER_TYPE masktype.
4742 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4743 directly into the builtin rather than hoisting it before loop.
4745 PR tree-optimization/59387
4746 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4747 (scev_const_prop): If folded_casts and type has undefined overflow,
4748 use force_gimple_operand instead of force_gimple_operand_gsi and
4749 for each added stmt if it is assign with
4750 arith_code_with_undefined_signed_overflow, call
4751 rewrite_to_defined_overflow.
4752 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4753 gimple-fold.h instead.
4754 (arith_code_with_undefined_signed_overflow,
4755 rewrite_to_defined_overflow): Moved to ...
4756 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4757 rewrite_to_defined_overflow): ... here. No longer static.
4758 Include gimplify-me.h.
4759 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4760 rewrite_to_defined_overflow): New prototypes.
4762 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4764 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4766 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4768 * builtins.c (get_object_alignment_2): Minor tweak.
4769 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4771 2014-01-13 Christian Bruel <christian.bruel@st.com>
4773 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4774 optimized non constant lengths.
4776 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4779 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4780 load as __atomic_load_N if possible.
4782 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4784 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4786 (rs6000_expand_builtin): Adjust call.
4788 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4791 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4792 * config/rs6000/rs6000.c: Include target-globals.h.
4793 (rs6000_set_current_function): Instead of doing target_reinit
4794 unconditionally, use save_target_globals_default_opts and
4795 restore_target_globals.
4797 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4799 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4800 (rs6000_expand_builtin): Handle mffs and mtfsf.
4801 (rs6000_init_builtins): Define mffs and mtfsf.
4802 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4803 (rs6000_mffs): New pattern.
4804 (rs6000_mtfsf): New pattern.
4806 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4808 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4809 Start narrowing with START. Apply candidate-use pair
4810 and check overall cost in narrowing.
4811 (iv_ca_prune): Pass new argument.
4813 2014-01-10 Jeff Law <law@redhat.com>
4816 * ree.c (combine_reaching_defs): Ensure the defining statement
4817 occurs before the extension when optimizing extensions with
4818 different source and destination hard registers.
4820 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4823 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4824 vtables into the type inheritance graph.
4826 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4828 PR rtl-optimization/59754
4829 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4830 modes in the REGNO != REGNO case.
4832 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4834 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4836 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4838 PR tree-optimization/59745
4839 * tree-predcom.c (tree_predictive_commoning_loop): Call
4840 free_affine_expand_cache if giving up because components is NULL.
4842 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4843 GC in payload of target_globals struct instead of allocating them on
4844 the heap and the larger structs separately using GC.
4845 * target-globals.h (struct target_globals): Make regs, hard_regs,
4846 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4847 of GTY((skip)) and change type to void *.
4848 (reset_target_globals): Cast loads from those fields to corresponding
4851 2014-01-10 Steve Ellcey <sellcey@mips.com>
4854 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4855 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4856 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4858 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4861 * aarch64-modes.def (CC_Zmode): New flags mode.
4862 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4863 represents an equality.
4864 (aarch64_get_condition_code): Handle CC_Zmode.
4865 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4867 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4869 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4870 extraction in good case.
4872 2014-01-10 Richard Biener <rguenther@suse.de>
4874 PR tree-optimization/59374
4875 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4876 checking after SLP discovery. Mark stmts not participating
4877 in any SLP instance properly.
4879 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4881 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4882 when handling a SET rtx.
4884 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4886 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4887 (cortex-a57): Likewise.
4888 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4890 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4892 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4893 non-iwmmxt builtins.
4895 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4899 * ipa-devirt.c record_target_from_binfo): Take as argument
4900 stack of binfos and lookup matching one for virtual inheritance.
4901 (possible_polymorphic_call_targets_1): Update.
4903 2014-01-10 Huacai Chen <chenhc@lemote.com>
4905 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4906 kernel strings for Loongson-2E/2F/3A.
4908 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4911 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4912 is_gimple_call before calling gimple_call_internal_p.
4914 2014-01-09 Steve Ellcey <sellcey@mips.com>
4916 * Makefile.in (TREE_FLOW_H): Remove.
4917 (TREE_SSA_H): Add file names from tree-flow.h.
4918 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4919 * tree.h: Remove tree-flow.h reference.
4920 * hash-table.h: Remove tree-flow.h reference.
4921 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4922 reference with tree-ssa-loop.h.
4924 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4926 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4927 default element-order behavior for -maltivec.
4928 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4929 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4930 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4931 when targeting big endian, at least for now.
4932 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4934 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4937 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4938 var satisfies use_register_for_decl, just take into account type
4939 alignment, rather than decl alignment.
4941 PR tree-optimization/59622
4942 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4943 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4944 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4945 Don't devirtualize for inplace at all. For targets.length () == 1,
4946 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4948 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4950 * config/i386/i386.md (cpu): Remove the unused btver1.
4952 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4954 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4956 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4959 * tree-core.h (struct target_globals): New forward declaration.
4960 (struct tree_target_option): Add globals field.
4961 * tree.h (TREE_TARGET_GLOBALS): Define.
4962 (prepare_target_option_nodes_for_pch): New prototype.
4963 * target-globals.h (struct target_globals): Define even if
4965 * tree.c (prepare_target_option_node_for_pch,
4966 prepare_target_option_nodes_for_pch): New functions.
4967 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4968 * config/i386/i386.c: Include target-globals.h.
4969 (ix86_set_current_function): Instead of doing target_reinit
4970 unconditionally, use save_target_globals_default_opts and
4971 restore_target_globals.
4973 2014-01-09 Richard Biener <rguenther@suse.de>
4975 PR tree-optimization/59715
4976 * tree-cfg.h (split_critical_edges): Declare.
4977 * tree-cfg.c (split_critical_edges): Export.
4978 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4980 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4982 * cfgexpand.c (expand_stack_vars): Optionally disable
4983 asan stack protection.
4984 (expand_used_vars): Likewise.
4985 (partition_stack_vars): Likewise.
4986 * asan.c (asan_emit_stack_protection): Optionally disable
4987 after return stack usage.
4988 (instrument_derefs): Optionally disable memory access instrumentation.
4989 (instrument_builtin_call): Likewise.
4990 (instrument_strlen_call): Likewise.
4991 (asan_protect_global): Optionally disable global variables protection.
4992 * doc/invoke.texi: Added doc for new options.
4993 * params.def: Added new options.
4994 * params.h: Likewise.
4996 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4998 PR rtl-optimization/59724
4999 * ifcvt.c (cond_exec_process_if_block): Don't call
5000 flow_find_head_matching_sequence with 0 longest_match.
5001 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5002 non-active insns if !stop_after.
5003 (try_head_merge_bb): Revert 2014-01-07 changes.
5005 2014-01-08 Jeff Law <law@redhat.com>
5007 * ree.c (get_sub_rtx): New function, extracted from...
5008 (merge_def_and_ext): Here.
5009 (combine_reaching_defs): Use get_sub_rtx.
5011 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5013 * cgraph.h (varpool_variable_node): Do not choke on null node.
5015 2014-01-08 Catherine Moore <clm@codesourcery.com>
5017 * config/mips/mips.md (simple_return): Attempt to use JRC
5019 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5021 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5023 PR rtl-optimization/59137
5024 * reorg.c (steal_delay_list_from_target): Call update_block for
5026 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5028 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5030 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5031 two duplicate entries.
5033 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5036 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5038 * config/mips/mips.c (mips_truncated_op_cost): New function.
5039 (mips_rtx_costs): Adjust test for BADDU.
5040 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5042 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5044 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5045 (*baddu_si): ...this new pattern.
5047 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5050 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5052 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5055 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5057 (expand_expr, expand_normal): Adjust.
5058 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5059 inner_reference_p. Use inner_reference_p to expand inner references.
5060 (store_expr): Adjust.
5061 * cfgexpand.c (expand_call_stmt): Adjust.
5063 2014-01-08 Rong Xu <xur@google.com>
5065 * gcov-io.c (gcov_var): Move from gcov-io.h.
5066 (gcov_position): Ditto.
5067 (gcov_is_error): Ditto.
5068 (gcov_rewrite): Ditto.
5069 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5070 only part to libgcc/libgcov.h.
5072 2014-01-08 Marek Polacek <polacek@redhat.com>
5075 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5077 2014-01-08 Marek Polacek <polacek@redhat.com>
5080 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5082 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5084 PR rtl-optimization/59649
5085 * stor-layout.c (get_mode_bounds): For BImode return
5086 0 and STORE_FLAG_VALUE.
5088 2014-01-08 Richard Biener <rguenther@suse.de>
5091 * gimple.h (is_gimple_builtin_call): Remove.
5092 (gimple_builtin_call_types_compatible_p): New.
5093 (gimple_call_builtin_p): New overload.
5094 * gimple.c (is_gimple_builtin_call): Remove.
5095 (validate_call): Rename to ...
5096 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5098 (validate_type): New static function.
5099 (gimple_call_builtin_p): New overload and adjust.
5100 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5101 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5102 (gimple_fold_stmt_to_constant_1): Likewise.
5103 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5105 2014-01-08 Richard Biener <rguenther@suse.de>
5108 * gimplify.c (gimplify_expr): Gimplify register-register type
5109 VIEW_CONVERT_EXPRs to separate stmts.
5111 2014-01-07 Jeff Law <law@redhat.com>
5114 * ree.c (combine_set_extension): Handle case where source
5115 and destination registers in an extension insn are different.
5116 (combine_reaching_defs): Allow source and destination registers
5117 in extension to be different under limited circumstances.
5118 (add_removable_extension): Remove restriction that the
5119 source and destination registers in the extension are the same.
5120 (find_and_remove_re): Emit a copy from the extension's
5121 destination to its source after the defining insn if
5122 the source and destination registers are different.
5125 * ifcvt.c (merge_if_block): If we are merging a block with more than
5126 one successor with a block with no successors, remove any BARRIER
5127 after the second block.
5129 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5131 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5133 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5136 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5137 for 14-bit register offsets when INT14_OK_STRICT is false.
5139 2014-01-07 Roland Stigge <stigge@antcom.de>
5140 Michael Meissner <meissner@linux.vnet.ibm.com>
5143 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5144 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5146 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5148 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5151 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5153 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5154 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5155 rtx is const0_rtx or not.
5157 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5160 * target-globals.c (save_target_globals): Remove this_fn_optab
5162 * toplev.c: Include optabs.h.
5163 (target_reinit): Temporarily restore the global options if another
5164 set of options are in force.
5166 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5168 PR rtl-optimization/58668
5169 * cfgcleanup.c (flow_find_cross_jump): Don't count
5170 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5171 to determine what is counted.
5172 (flow_find_head_matching_sequence): Use active_insn_p to determine
5174 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5176 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5177 determine what is counted.
5179 PR tree-optimization/59643
5180 * tree-predcom.c (split_data_refs_to_components): If one dr is
5181 read and one write, determine_offset fails and the write isn't
5182 in the bad component, just put the read into the bad component.
5184 2014-01-07 Mike Stump <mikestump@comcast.net>
5185 Jakub Jelinek <jakub@redhat.com>
5188 * tree-core.h (struct tree_optimization_option): Change optabs
5189 type from unsigned char * to void *.
5190 * optabs.c (init_tree_optimization_optabs): Adjust
5191 TREE_OPTIMIZATION_OPTABS initialization.
5193 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5196 * config/i386/i386.h (struct machine_function): Add
5197 no_drap_save_restore field.
5198 * config/i386/i386.c (ix86_save_reg): Use
5199 !cfun->machine->no_drap_save_restore instead of
5200 crtl->stack_realign_needed.
5201 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5202 this function clears frame_pointer_needed. Set
5203 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5204 and DRAP reg is needed.
5206 2014-01-06 Marek Polacek <polacek@redhat.com>
5209 * doc/implement-c.texi: Mention that other integer types are
5210 permitted as bit-field types in strictly conforming mode.
5212 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5214 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5217 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5219 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5221 2014-01-06 Martin Jambor <mjambor@suse.cz>
5224 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5226 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5228 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5232 * var-tracking.c (add_stores): Preserve the value of the source even if
5233 we don't record the store.
5235 2014-01-06 Terry Guo <terry.guo@arm.com>
5237 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5239 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5242 * config/darwin.c (darwin_function_section): Adjust return values to
5243 correspond to optimisation changes made in r206070.
5245 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5247 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5248 from prefetch_block tune setting.
5249 (nocona_cost): Correct size of prefetch block to 64.
5251 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5253 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5254 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5255 used to save the static chain register in the computation of the offset
5256 from which the FP registers need to be restored.
5258 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5260 PR tree-optimization/59519
5261 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5262 ICE if get_current_def (current_new_name) is already non-NULL, as long
5263 as it is a phi result of some other phi in *new_exit_bb that has
5266 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5267 or vmovdqu* for misaligned_operand.
5268 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5269 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5270 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5271 aligned_mem for AVX512F masked aligned load and store builtins and for
5274 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5276 PR tree-optimization/59651
5277 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5278 Address range for negative step should be added by TYPE_SIZE_UNIT.
5280 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5282 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5283 overlapping registers also for registers other than the stack pointer.
5285 2014-01-03 Marek Polacek <polacek@redhat.com>
5288 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5291 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5294 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5297 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5298 (push splitter): Use <P:MODE_SIZE> instead of
5299 GET_MODE_SIZE (<P:MODE>mode).
5300 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5301 (mov -1, reg peephole2): Likewise.
5302 * config/i386/sse.md (*mov<mode>_internal,
5303 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5304 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5305 *<code><mode>3, *andnot<mode>3<mask_name>,
5306 <mask_codefor><code><mode>3<mask_name>): Likewise.
5307 * config/i386/subst.md (mask_mode512bit_condition,
5308 sd_mask_mode512bit_condition): Likewise.
5310 2014-01-02 Xinliang David Li <davidxl@google.com>
5312 PR tree-optimization/59303
5313 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5314 (dump_predicates): Better output format.
5315 (pred_equal_p): New function.
5316 (is_neq_relop_p): Ditto.
5317 (is_neq_zero_form_p): Ditto.
5318 (pred_expr_equal_p): Ditto.
5319 (pred_neg_p): Ditto.
5320 (simplify_pred): Ditto.
5321 (simplify_preds_2): Ditto.
5322 (simplify_preds_3): Ditto.
5323 (simplify_preds_4): Ditto.
5324 (simplify_preds): Ditto.
5326 (push_to_worklist): Ditto.
5327 (get_pred_info_from_cmp): Ditto.
5328 (is_degenerated_phi): Ditto.
5329 (normalize_one_pred_1): Ditto.
5330 (normalize_one_pred): Ditto.
5331 (normalize_one_pred_chain): Ditto.
5332 (normalize_preds): Ditto.
5333 (normalize_cond_1): Remove function.
5334 (normalize_cond): Ditto.
5335 (is_gcond_subset_of): Ditto.
5336 (is_subset_of_any): Ditto.
5337 (is_or_set_subset_of): Ditto.
5338 (is_and_set_subset_of): Ditto.
5339 (is_norm_cond_subset_of): Ditto.
5340 (pred_chain_length_cmp): Ditto.
5341 (convert_control_dep_chain_into_preds): Type change.
5342 (find_predicates): Ditto.
5343 (find_def_preds): Ditto.
5344 (destroy_predicates_vecs): Ditto.
5345 (find_matching_predicates_in_rest_chains): Ditto.
5346 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5347 (is_pred_expr_subset): Ditto.
5348 (is_pred_chain_subset_of): Ditto.
5349 (is_included_in): Ditto.
5350 (is_superset_of): Ditto.
5352 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5354 Update copyright years.
5356 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5358 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5359 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5360 config/arc/arc.md, config/arc/arc.opt,
5361 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5362 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5363 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5364 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5365 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5366 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5367 vtable-verify.c, vtable-verify.h: Use the standard form for the
5370 2014-01-02 Tobias Burnus <burnus@net-b.de>
5372 * gcc.c (process_command): Update copyright notice dates.
5373 * gcov-dump.c: Ditto.
5375 * doc/cpp.texi: Bump @copying's copyright year.
5376 * doc/cppinternals.texi: Ditto.
5377 * doc/gcc.texi: Ditto.
5378 * doc/gccint.texi: Ditto.
5379 * doc/gcov.texi: Ditto.
5380 * doc/install.texi: Ditto.
5381 * doc/invoke.texi: Ditto.
5383 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5385 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5387 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5389 * config/i386/sse.md (*mov<mode>_internal): Guard
5390 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5392 PR rtl-optimization/59647
5393 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5394 new_rtx into UNSIGNED_FLOAT rtxes.
5396 Copyright (C) 2014 Free Software Foundation, Inc.
5398 Copying and distribution of this file, with or without modification,
5399 are permitted in any medium without royalty provided the copyright
5400 notice and this notice are preserved.