1 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3 * Makefile.in (s-macro_list): Depend on cc1.
5 2014-03-23 Teresa Johnson <tejohnson@google.com>
7 * ipa-utils.c (ipa_print_order): Use specified dump file.
9 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
11 PR rtl-optimization/60601
12 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
14 * gcc.c (eval_spec_function): Initialize save_growing_value.
16 2014-03-22 Jakub Jelinek <jakub@redhat.com>
19 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
20 code == MINUS_EXPR, never swap op0 with op1.
22 * toplev.c (init_local_tick): Avoid signed integer multiplication
24 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
25 shift by first operand's bitsize.
27 2014-03-21 Jakub Jelinek <jakub@redhat.com>
30 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
32 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
33 TARGET_ISA_64BIT_P(x).
35 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
38 pattern for vector nor instead of subtract from splat(-1).
39 (altivec_expand_vec_perm_const_le): Likewise.
41 2014-03-21 Richard Henderson <rth@twiddle.net>
44 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
45 related insns after epilogue_completed.
47 2014-03-21 Martin Jambor <mjambor@suse.cz>
50 * cgraph.h (symtab_node): New flag body_removed.
51 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
53 * symtab.c (dump_symtab_base): Dump body_removed flag.
54 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
55 had their bodies removed.
57 2014-03-21 Martin Jambor <mjambor@suse.cz>
60 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
63 2014-03-21 Richard Biener <rguenther@suse.de>
65 PR tree-optimization/60577
66 * tree-core.h (struct tree_base): Document nothrow_flag use
67 in VAR_DECL_NONALIASED.
68 * tree.h (VAR_DECL_NONALIASED): New.
69 (may_be_aliased): Adjust.
70 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
72 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
74 * expr.c (expand_expr_real_1): Remove outdated comment.
76 2014-03-20 Jakub Jelinek <jakub@redhat.com>
79 * ira.c (adjust_cleared_regs): Call copy_rtx on
80 *reg_equiv[REGNO (loc)].src_p before passing it to
81 simplify_replace_fn_rtx.
84 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
85 into CONST, put pic register as first operand of PLUS. Use
86 gen_const_mem for both 32-bit and 64-bit PIC got loads.
88 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
90 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
92 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
94 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
95 around for store forwarding issue in the FPU on the UT699.
96 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
97 loads and operations if -mfix-ut699 is specified.
98 (divtf3_hq): Tweak attribute.
99 (sqrttf2_hq): Likewise.
101 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
103 * calls.c (store_one_arg): Remove incorrect const qualification on the
104 type of the temporary.
105 * cfgexpand.c (expand_return): Likewise.
106 * expr.c (expand_constructor): Likewise.
107 (expand_expr_real_1): Likewise.
109 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
111 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
114 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
117 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
119 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
121 * config/arm/aarch-common-protos.h
122 (alu_cost_table): Fix spelling of "extend".
123 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
125 2014-03-19 Richard Biener <rguenther@suse.de>
128 * tree-core.h (tree_type_common): Re-order pointer members
129 to reduce recursion depth during GC walks.
131 2014-03-19 Marek Polacek <polacek@redhat.com>
134 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
137 2014-03-19 Richard Biener <rguenther@suse.de>
140 * lto-streamer-in.c (input_function): In WPA stage do not drop
143 2014-03-19 Jakub Jelinek <jakub@redhat.com>
145 PR tree-optimization/60559
146 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
147 with build_zero_cst assignment.
149 2014-03-18 Kai Tietz <ktietz@redhat.com>
151 PR rtl-optimization/56356
152 * sdbout.c (sdbout_parms): Verify that parms'
153 incoming argument is valid.
154 (sdbout_reg_parms): Likewise.
156 2014-03-18 Richard Henderson <rth@redhat.com>
159 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
160 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
161 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
163 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
165 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
166 * doc/plugins.texi (Plugin callbacks): Mention
168 Italicize plugin event names in description. Explain that
169 PLUGIN_PRAGMAS has no sense for lto1. Explain
171 Remind that no GCC functions should be called after
173 Explain what pragmas with expansion are.
175 2014-03-18 Martin Liska <mliska@suse.cz>
177 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
178 gimple call statement is update.
179 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
180 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
182 2014-03-18 Jakub Jelinek <jakub@redhat.com>
185 * ubsan.c (ubsan_instrument_unreachable): Call
186 initialize_sanitizer_builtins.
187 (ubsan_pass): Likewise.
190 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
191 varpool_finalize_decl instead of rest_of_decl_compilation.
193 2014-03-18 Richard Biener <rguenther@suse.de>
195 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
196 by using bitmap_and_compl instead of bitmap_and_compl_into.
197 (df_rd_transfer_function): Likewise.
199 2014-03-18 Richard Biener <rguenther@suse.de>
201 * doc/lto.texi (fresolution): Fix typo.
203 2014-03-18 Richard Biener <rguenther@suse.de>
205 * doc/invoke.texi (flto): Update for changes in 4.9.
207 2014-03-18 Richard Biener <rguenther@suse.de>
209 * doc/loop.texi: Remove section on the removed lambda framework.
210 Update loop docs with recent changes in preserving loop structure.
212 2014-03-18 Richard Biener <rguenther@suse.de>
214 * doc/lto.texi (-fresolution): Document.
216 2014-03-18 Richard Biener <rguenther@suse.de>
218 * doc/contrib.texi: Adjust my name.
220 2014-03-18 Jakub Jelinek <jakub@redhat.com>
223 * internal-fn.c: Include diagnostic-core.h.
224 (expand_BUILTIN_EXPECT): New function.
225 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
226 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
227 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
228 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
230 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
231 Revert 3 argument __builtin_expect code.
232 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
233 * gimple-fold.c (gimple_fold_call): Likewise.
234 * tree.h (fold_builtin_expect): New prototype.
235 * builtins.c (build_builtin_expect_predicate): Add predictor
236 argument, if non-NULL, create 3 argument __builtin_expect.
237 (fold_builtin_expect): No longer static. Add ARG2 argument,
238 pass it through to build_builtin_expect_predicate.
239 (fold_builtin_2): Adjust caller.
240 (fold_builtin_3): Handle BUILT_IN_EXPECT.
241 * internal-fn.def (BUILTIN_EXPECT): New.
243 2014-03-18 Tobias Burnus <burnus@net-b.de>
246 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
247 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
248 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
250 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
253 * predict.c (combine_predictions_for_bb): Fix up formatting.
254 (expr_expected_value_1, expr_expected_value): Add predictor argument,
255 fill what it points to if non-NULL.
256 (tree_predict_by_opcode): Adjust caller, use the predictor.
257 * predict.def (PRED_COMPARE_AND_SWAP): Add.
259 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
261 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
262 proper constant for the store mode.
264 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
266 * symtab.c (change_decl_assembler_name): Fix transparent alias
269 2014-03-16 Renlin Li <Renlin.Li@arm.com>
271 * config/aarch64/aarch64.c: Correct the comments about the
272 aarch64 stack layout.
274 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
276 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
277 check for GF_OMP_FOR_KIND_FOR.
279 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
281 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
282 ymm and zmm register names.
284 2014-03-17 Jakub Jelinek <jakub@redhat.com>
287 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
288 note creation for the 2010-08-31 changes.
290 2014-03-17 Marek Polacek <polacek@redhat.com>
293 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
294 as -fno-tree-loop-vectorize.
295 (expand_omp_simd): Likewise.
297 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
299 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
300 (eligible_for_call_delay): New prototype.
301 * config/sparc/sparc.c (tls_call_delay): Rename into...
302 (eligible_for_call_delay): ...this. Return false if the instruction
303 cannot be put in the delay slot of a branch.
304 (eligible_for_restore_insn): Simplify.
305 (eligible_for_return_delay): Return false if the instruction cannot be
306 put in the delay slot of a branch and simplify.
307 (eligible_for_sibcall_delay): Return false if the instruction cannot be
308 put in the delay slot of a branch.
309 * config/sparc/sparc.md (fix_ut699): New attribute.
310 (tls_call_delay): Delete.
311 (in_call_delay): Reimplement.
312 (eligible_for_sibcall_delay): Rename into...
313 (in_sibcall_delay): ...this.
314 (eligible_for_return_delay): Rename into...
315 (in_return_delay): ...this.
316 (in_branch_delay): Reimplement.
317 (in_uncond_branch_delay): Delete.
318 (in_annul_branch_delay): Delete.
320 2014-03-14 Richard Henderson <rth@redhat.com>
323 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
324 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
325 (*floathi<X87MODEF>2_i387_with_temp): Remove.
326 (floathi splitters): Remove.
327 (float<SWI48x>xf2): New pattern.
328 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
329 code that tried to handle DImode for 32-bit, but which was excluded
330 by the pattern's condition. Drop allocation of stack temporary.
331 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
332 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
333 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
334 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
335 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
336 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
337 (*float<SWI48><MODEF>2_sse_interunit): Remove.
338 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
339 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
340 (*float<SWI48x><X87MODEF>2_i387): Remove.
341 (all float _with_temp splitters): Remove.
342 (*float<SWI48x><MODEF>2_i387): New pattern.
343 (*float<SWI48><MODEF>2_sse): New pattern.
344 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
345 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
347 2014-03-14 Jakub Jelinek <jakub@redhat.com>
348 Marek Polacek <polacek@redhat.com>
351 * common.opt (dump_base_name_prefixed): New Variable.
352 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
353 if x_dump_base_name_prefixed is already set, set it at the end.
355 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
357 PR rtl-optimization/60508
358 * lra-constraints.c (get_reload_reg): Add new parameter
360 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
361 Pass the new parameter values.
363 2014-03-14 Richard Biener <rguenther@suse.de>
365 * common.opt: Revert unintented changes from r205065.
368 2014-03-14 Richard Biener <rguenther@suse.de>
371 * cfghooks.c (split_block): Properly adjust all loops the
372 block was a latch of.
374 2014-03-14 Martin Jambor <mjambor@suse.cz>
377 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
380 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
383 * config/avr/avr.c (avr_set_current_function): Pass function name
384 through default_strip_name_encoding before sanity checking instead
385 of skipping the first char of the assembler name.
387 2014-03-13 Richard Henderson <rth@redhat.com>
390 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
391 (ix86_force_to_memory, ix86_free_from_memory): Remove.
392 * config/i386/i386-protos.h: Likewise.
393 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
394 in the expander instead of a splitter.
395 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
396 any possibility of requiring a memory.
397 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
398 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
399 (fp branch splitters): Update for ix86_split_fp_branch.
400 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
401 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
402 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
403 (*fop_<MODEF>_2_i387): Remove f/r alternative.
404 (*fop_<MODEF>_3_i387): Likewise.
405 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
406 (splitters for the fop_* register patterns): Remove.
407 (fscalexf4_i387): Rename from *fscalexf4_i387.
408 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
410 2014-03-13 Jakub Jelinek <jakub@redhat.com>
412 PR tree-optimization/59779
413 * tree-dfa.c (get_ref_base_and_extent): Use double_int
414 type for bitsize and maxsize instead of HOST_WIDE_INT.
416 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
418 PR rtl-optimization/57320
419 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
420 the CFG after thread_prologue_and_epilogue_insns.
422 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
424 PR rtl-optimization/57189
425 * lra-constraints.c (process_alt_operands): Disfavor spilling
428 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
430 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic
433 2014-03-13 Jakub Jelinek <jakub@redhat.com>
435 PR tree-optimization/59025
437 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
438 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
440 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
443 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
444 calls of avr_out_plus_1.
446 2014-03-13 Bin Cheng <bin.cheng@arm.com>
448 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
449 BB's single pred and update the father loop's latch info later.
451 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
453 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
457 (VEC_base): Likewise.
458 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
459 registers, we need to swap double words in little endian mode.
461 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
462 to be a container mode for 128-bit integer operations added in ISA
463 2.07. Unlike TImode and PTImode, the preferred register set is
464 the Altivec/VMX registers for the 128-bit operations.
466 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
468 (rs6000_split_128bit_ok_p): Likewise.
470 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
471 macros for creating ISA 2.07 normal and overloaded builtin
472 functions with 3 arguments.
473 (BU_P8V_OVERLOAD_3): Likewise.
474 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
475 for use as overloaded functions.
476 (VPERM_1TI_UNS): Likewise.
477 (VSEL_1TI): Likewise.
478 (VSEL_1TI_UNS): Likewise.
479 (ST_INTERNAL_1ti): Likewise.
480 (LD_INTERNAL_1ti): Likewise.
481 (XXSEL_1TI): Likewise.
482 (XXSEL_1TI_UNS): Likewise.
483 (VPERM_1TI): Likewise.
484 (VPERM_1TI_UNS): Likewise.
485 (XXPERMDI_1TI): Likewise.
487 (LXVD2X_V1TI): Likewise.
488 (STXVD2X_V1TI): Likewise.
489 (VEC_INIT_V1TI): Likewise.
490 (VEC_SET_V1TI): Likewise.
491 (VEC_EXT_V1TI): Likewise.
492 (EQV_V1TI): Likewise.
493 (NAND_V1TI): Likewise.
494 (ORC_V1TI): Likewise.
495 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
496 added in ISA 2.07. Add both normal 'altivec' builtins, and the
500 (VADDEUQM): Likewise.
501 (VADDECUQ): Likewise.
502 (VSUBEUQM): Likewise.
503 (VSUBECUQ): Likewise.
505 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
506 __int128_t and __uint128_t types.
507 (__uint128_type): Likewise.
508 (altivec_categorize_keyword): Add support for vector __int128_t,
509 vector __uint128_t, vector __int128, and vector unsigned __int128
510 as a container type for TImode operations that need to be done in
511 VSX/Altivec registers.
512 (rs6000_macro_to_expand): Likewise.
513 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
514 to support 128-bit integer instructions vaddcuq, vadduqm,
515 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
516 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
518 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
519 for V1TImode, and set up preferences to use VSX/Altivec registers.
520 Setup VSX reload handlers.
521 (rs6000_debug_reg_global): Likewise.
522 (rs6000_init_hard_regno_mode_ok): Likewise.
523 (rs6000_preferred_simd_mode): Likewise.
524 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
525 (easy_altivec_constant): Likewise.
526 (output_vec_const_move): Likewise.
527 (rs6000_expand_vector_set): Convert V1TImode set and extract to
529 (rs6000_expand_vector_extract): Likewise.
530 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
532 (rs6000_const_vec): Add support for V1TImode.
533 (rs6000_emit_le_vsx_load): Swap double words when loading or
534 storing TImode/V1TImode.
535 (rs6000_emit_le_vsx_store): Likewise.
536 (rs6000_emit_le_vsx_move): Likewise.
537 (rs6000_emit_move): Add support for V1TImode.
538 (altivec_expand_ld_builtin): Likewise.
539 (altivec_expand_st_builtin): Likewise.
540 (altivec_expand_vec_init_builtin): Likewise.
541 (altivec_expand_builtin): Likewise.
542 (rs6000_init_builtins): Add support for V1TImode type. Add
543 support for ISA 2.07 128-bit integer builtins. Define type names
544 for the VSX/Altivec vector types.
545 (altivec_init_builtins): Add support for overloaded vector
546 functions with V1TImode type.
547 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
548 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
550 (rs6000_split_128bit_ok_p): Likewise.
551 (rs6000_handle_altivec_attribute): Create V1TImode from vector
552 __int128_t and vector __uint128_t.
554 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
562 (VS_scalar): Likewise.
563 (VS_double): Likewise.
564 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
566 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
567 we support the ISA 2.07 128-bit integer arithmetic instructions.
568 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
569 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
570 and TImode types for use with the builtin functions.
571 (V1TI_type_node): Likewise.
572 (unsigned_V1TI_type_node): Likewise.
573 (intTI_type_internal_node): Likewise.
574 (uintTI_type_internal_node): Likewise.
576 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
577 128-bit builtin functions.
578 (UNSPEC_VADDEUQM): Likewise.
579 (UNSPEC_VADDECUQ): Likewise.
580 (UNSPEC_VSUBCUQ): Likewise.
581 (UNSPEC_VSUBEUQM): Likewise.
582 (UNSPEC_VSUBECUQ): Likewise.
583 (VM): Add V1TImode to vector mode iterators.
586 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
587 (altivec_vaddcuq): Likewise.
588 (altivec_vsubuqm): Likewise.
589 (altivec_vsubcuq): Likewise.
590 (altivec_vaddeuqm): Likewise.
591 (altivec_vaddecuq): Likewise.
592 (altivec_vsubeuqm): Likewise.
593 (altivec_vsubecuq): Likewise.
595 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
597 (BOOL_128): Likewise.
598 (BOOL_REGS_OUTPUT): Likewise.
599 (BOOL_REGS_OP1): Likewise.
600 (BOOL_REGS_OP2): Likewise.
601 (BOOL_REGS_UNARY): Likewise.
602 (BOOL_REGS_AND_CR0): Likewise.
604 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
605 128-bit integer builtin support.
606 (vec_vadduqm): Likewise.
607 (vec_vaddecuq): Likewise.
608 (vec_vaddeuqm): Likewise.
609 (vec_vsubecuq): Likewise.
610 (vec_vsubeuqm): Likewise.
611 (vec_vsubcuq): Likewise.
612 (vec_vsubuqm): Likewise.
614 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
615 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
616 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
617 128-bit integer add/subtract to ISA 2.07.
619 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
621 * config/arc/arc.c (arc_predicate_delay_insns):
622 Fix third argument passed to conditionalize_nonjump.
624 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
626 * config/aarch64/aarch64-builtins.c
627 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
628 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
629 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
630 instead of __builtin_lfloor.
631 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
633 2014-03-12 Jakub Jelinek <jakub@redhat.com>
635 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
636 (tree_ssa_ifcombine_bb_1): New function.
637 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
638 is an empty forwarder block to then_bb or vice versa and then_bb
639 and else_bb are effectively swapped.
641 2014-03-12 Christian Bruel <christian.bruel@st.com>
644 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
645 REG_CFA_DEF_CFA note.
646 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
647 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
649 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
651 PR tree-optimization/60454
652 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
654 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
656 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
657 Do not define target_cpu_default2 to generic.
658 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
659 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
660 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
662 2014-03-12 Jakub Jelinek <jakub@redhat.com>
663 Marc Glisse <marc.glisse@inria.fr>
665 PR tree-optimization/60502
666 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
667 instead of build_low_bits_mask.
669 2014-03-12 Jakub Jelinek <jakub@redhat.com>
672 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
673 if there are multiple uses, but op doesn't live on E edge.
674 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
675 clobber stmts before __builtin_unreachable.
677 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
679 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
680 hard_frame_pointer_rtx.
681 * cse.c (cse_insn): Remove volatile check.
682 * cselib.c (cselib_process_insn): Likewise.
683 * dse.c (scan_insn): Likewise.
685 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
687 * config/arc/arc.c (conditionalize_nonjump): New function,
689 (arc_ifcvt): ... this.
690 (arc_predicate_delay_insns): Use it.
692 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
694 * config/arc/predicates.md (extend_operand): During/after reload,
695 allow const_int_operand.
696 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
697 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
698 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
700 (umulsi3_highpart_i): Likewise.
702 2014-03-11 Richard Biener <rguenther@suse.de>
704 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
705 Add asserts to guard possible wrong-code bugs.
707 2014-03-11 Richard Biener <rguenther@suse.de>
709 PR tree-optimization/60429
710 PR tree-optimization/60485
711 * tree-ssa-structalias.c (set_union_with_increment): Properly
712 take into account all fields that overlap the shifted vars.
713 (do_sd_constraint): Likewise.
714 (do_ds_constraint): Likewise.
715 (get_constraint_for_ptr_offset): Likewise.
717 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
719 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
720 (nios2_compute_frame_layout):
721 Add calculation of cfun->machine->fp_save_offset.
722 (nios2_expand_prologue): Correct setting of frame pointer register
724 (nios2_expand_epilogue): Update recovery of stack pointer from
725 frame pointer accordingly.
726 (nios2_initial_elimination_offset): Update calculation of offset
727 for eliminating to HARD_FRAME_POINTER_REGNUM.
729 2014-03-10 Jakub Jelinek <jakub@redhat.com>
732 * ipa.c (symtab_remove_unreachable_nodes): Don't call
733 cgraph_get_create_node on VAR_DECLs.
735 2014-03-10 Richard Biener <rguenther@suse.de>
738 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
740 2014-03-08 Douglas B Rupp <rupp@gnat.com>
742 * config/vms/vms.opt (vms_float_format): New variable.
744 2014-03-08 Tobias Burnus <burnus@net-b.de>
746 * doc/invoke.texi (-fcilkplus): Update implementation status.
748 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
749 Richard Biener <rguenther@suse.de>
751 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
752 consistently accross all TUs.
753 (run_gcc): Enable -fshort-double automatically at link at link-time
754 and disallow override.
756 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
759 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
760 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
761 if they can't be used.
763 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
765 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
766 for Solaris 11/x86 ld.
767 * configure: Regenerate.
769 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
771 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
772 (LIB_TLS_SPEC): Save as ld_tls_libs.
773 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
774 (HAVE_AS_IX86_TLSLDM): New test.
775 * configure, config.in: Regenerate.
776 * config/i386/i386.c (legitimize_tls_address): Fall back to
777 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
778 cannot support TLS_MODEL_LOCAL_DYNAMIC.
779 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
780 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
782 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
784 * common.opt (fira-loop-pressure): Mark as optimization.
786 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
788 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
789 an OpenMP mappable type.
791 2014-03-06 Matthias Klose <doko@ubuntu.com>
793 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
794 MULTILIB_OSDIRNAMES is not defined.
796 2014-03-06 Jakub Jelinek <jakub@redhat.com>
797 Meador Inge <meadori@codesourcery.com>
800 * config/arm/arm.c (arm_tls_symbol_p): Remove.
801 (arm_legitimize_address): Call legitimize_tls_address for any
802 arm_tls_referenced_p expression, handle constant addend. Call it
803 before testing for !TARGET_ARM.
804 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
806 2014-03-06 Richard Biener <rguenther@suse.de>
812 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
814 * tree-streamer.c (record_common_node): Assert we don't record
815 nodes with type double.
816 (preload_common_node): Skip type double, complex double and double
817 pointer since it is now frontend dependent due to fshort-double option.
819 2014-03-06 Richard Biener <rguenther@suse.de>
821 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
822 or -fno-lto is specified and the linker has full plugin support.
823 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
824 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
825 * lto-wrapper.c (merge_and_complain): Merge compile-time
827 (run_gcc): And pass it through to the link options.
829 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
833 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
835 * cselib.c (remove_useless_values): Skip to avoid quadratic
836 behavior if the condition moved from...
837 (cselib_process_insn): ... here holds.
839 2014-03-05 Jakub Jelinek <jakub@redhat.com>
842 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
843 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
846 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
847 (TM_H): Add x86-tune.def.
849 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
851 * config/aarch64/aarch64.c (generic_tunings):
852 Use cortexa57_extra_costs.
854 2014-03-05 Jakub Jelinek <jakub@redhat.com>
857 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
858 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
859 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
862 2014-03-04 Heiher <r@hev.cc>
864 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
865 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
867 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
869 * config/i386/predicates.md (const2356_operand): Change to ...
870 (const2367_operand): ... this.
871 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
873 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
874 (*avx512pf_scatterpf<mode>sf): Ditto.
875 (avx512pf_scatterpf<mode>df): Ditto.
876 (*avx512pf_scatterpf<mode>df_mask): Ditto.
877 (*avx512pf_scatterpf<mode>df): Ditto.
878 * config/i386/i386.c (ix86_expand_builtin): Update
879 incorrect hint operand error message.
881 2014-03-04 Richard Biener <rguenther@suse.de>
883 * lto-section-in.c (lto_get_section_data): Fix const cast.
885 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
887 * tree-streamer.c (record_common_node): Assert we don't record
888 nodes with type double.
889 (preload_common_node): Skip type double, complex double and double
890 pointer since it is now frontend dependent due to fshort-double option.
892 2014-03-04 Richard Biener <rguenther@suse.de>
895 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
896 (lto_input_toplevel_asms): Likewise.
897 * lto-section-in.c (lto_get_section_data): Instead do it here
900 2014-03-04 Richard Biener <rguenther@suse.de>
902 PR tree-optimization/60382
903 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
904 dead PHIs a reduction.
906 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
908 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
910 (_mm_prefetch): Move out of GCC target("sse") pragma.
911 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
912 GCC target("prfchw") pragma.
913 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
915 * config/i386/i386.c (ix86_option_override_internal): Enable
916 -mprfchw with -mprefetchwt1.
918 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
920 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
923 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
925 * opts.h (CL_PCH_IGNORE): Define.
926 * targhooks.c (option_affects_pch_p):
927 Return false for options that have CL_PCH_IGNORE set.
928 * opt-functions.awk: Process PchIgnore.
929 * doc/options.texi: Document PchIgnore.
931 * config/arc/arc.opt (misize): Add PchIgnore property.
933 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
935 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
936 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
937 constraint on constants to permit them being loaded into
938 GENERAL_REGS or BASE_REGS.
940 2014-03-03 Nick Clifton <nickc@redhat.com>
942 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
943 anti-cacnonical alternatives.
944 (negandhi3_real): New pattern.
945 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
947 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
949 * config/avr/avr-mcus.def: Remove atxmega16x1.
950 * config/avr/avr-tables.opt: Regenerate.
951 * config/avr/t-multilib: Regenerate.
952 * doc/avr-mmcu.texi: Regenerate.
954 2014-03-03 Tobias Grosser <tobias@grosser.es>
955 Mircea Namolaru <mircea.namolaru@inria.fr>
957 PR tree-optimization/58028
958 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
961 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
963 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
964 not handled by recognizers.
966 2014-03-03 Jakub Jelinek <jakub@redhat.com>
969 * function.c (expand_function_end): Don't emit
970 clobber_return_register sequence if clobber_after is a BARRIER.
971 * cfgexpand.c (construct_exit_block): Append instructions before
972 return_label to prev_bb.
974 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
976 * config/rs6000/constraints.md: Document reserved use of "wc".
978 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
981 * ipa.c (function_and_variable_visibility): When dissolving comdat
982 group, also set all symbols to local.
984 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
989 2013-12-14 Jan Hubicka <jh@suse.cz>
991 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
993 2014-03-02 Jon Beniston <jon@beniston.com>
999 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1000 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1001 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1002 (simple_return, *simple_return): New patterns
1003 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1004 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1006 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1008 * dwarf2out.c (gen_subprogram_die): Tidy.
1010 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1013 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1014 (*mov_t_msb_neg_negc): ... this new insn.
1016 2014-02-28 Jason Merrill <jason@redhat.com>
1019 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1022 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1025 * dwarf2out.c (decltype_auto_die): New static.
1026 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1027 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1028 (is_cxx_auto): Likewise.
1030 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1032 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1033 we are not using general regs only.
1035 2014-02-28 Richard Biener <rguenther@suse.de>
1038 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1039 previous fix and only allow to remove trivial pre-headers
1040 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1041 (remove_forwarder_block): Properly update the latch of a loop.
1043 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1046 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1047 (cselib_preserved_hash_table): New.
1048 (preserve_constants_and_equivs): Move preserved vals to it.
1049 (cselib_find_slot): Look it up first.
1050 (cselib_init): Initialize it.
1051 (cselib_finish): Release it.
1052 (dump_cselib_table): Dump it.
1054 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1057 * cselib.c (remove_useless_values): Skip to avoid quadratic
1058 behavior if the condition moved from...
1059 (cselib_process_insn): ... here holds.
1061 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1064 * var-tracking.c (vt_initialize): Apply the same condition to
1065 preserve the CFA base value.
1067 2014-02-28 Joey Ye <joey.ye@arm.com>
1070 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1071 if reload in progress or completed.
1073 2014-02-28 Tobias Burnus <burnus@net-b.de>
1076 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1079 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1081 * doc/tm.texi.in (Condition Code Status): Update documention for
1082 relative locations of cc0-setter and cc0-user.
1084 2014-02-27 Jeff Law <law@redhat.com>
1086 PR rtl-optimization/52714
1087 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1088 into two independent simple sets, if I3 is a jump, ensure the
1089 pattern we place into I3 is a (set (pc) ...).
1091 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1092 Jeff Law <law@redhat.com>
1094 PR rtl-optimization/49847
1095 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1096 are in different blocks.
1097 * doc/tm.texi (Condition Code Status): Update documention for
1098 relative locations of cc0-setter and cc0-user.
1100 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1103 * lra.c (lra_emit_add): Check SUBREG too.
1105 2014-02-27 Andreas Schwab <schwab@suse.de>
1107 * config/m68k/m68k.c (m68k_option_override): Disable
1108 -flive-range-shrinkage for classic m68k.
1109 (m68k_override_options_after_change): Likewise.
1111 2014-02-27 Marek Polacek <polacek@redhat.com>
1114 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1115 -Wmaybe-uninitialized.
1117 2014-02-27 Alan Modra <amodra@gmail.com>
1120 * reload1.c (emit_input_reload_insns): When reload_override_in,
1121 set old to rl->in_reg when rl->in_reg is a subreg.
1123 2014-02-26 Richard Biener <rguenther@suse.de>
1126 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1128 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1130 * common/config/i386/predicates.md (const1256_operand): Remove.
1131 (const2356_operand): New.
1132 (const_1_to_2_operand): Remove.
1133 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1134 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1135 (*avx512pf_gatherpf<mode>sf): Ditto.
1136 (avx512pf_gatherpf<mode>df): Ditto.
1137 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1138 (*avx512pf_gatherpf<mode>df): Ditto.
1139 (avx512pf_scatterpf<mode>sf): Ditto.
1140 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1141 (*avx512pf_scatterpf<mode>sf): Ditto.
1142 (avx512pf_scatterpf<mode>df): Ditto.
1143 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1144 (*avx512pf_scatterpf<mode>df): Ditto.
1145 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1147 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1149 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1150 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1151 (_mm512_mask_testn_epi64_mask): Move to ...
1152 * config/i386/avx512cdintrin.h: Here.
1153 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1154 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1155 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1156 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1157 TARGET_AVX512F from TARGET_AVX512CD.
1159 2014-02-26 Richard Biener <rguenther@suse.de>
1162 * ipa.c (walk_polymorphic_call_targets): Properly guard
1163 call to inline_update_overall_summary.
1165 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1168 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1169 and latches only if requested. Fix latch if it is removed.
1170 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1171 LOOPS_HAVE_PREHEADERS.
1173 2014-02-25 Andrew Pinski <apinski@cavium.com>
1175 * builtins.c (expand_builtin_thread_pointer): Create a new target
1176 when the target is NULL.
1178 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1180 PR rtl-optimization/60317
1181 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1182 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1183 * lra-assigns.c: Include params.h.
1184 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1185 other reload pseudos considerations.
1187 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1189 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1190 to use canonical form for nor<mode>3.
1192 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1195 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1198 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1200 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1201 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1202 (ix86_handle_option): Handle OPT_mprefetchwt1.
1203 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1204 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1206 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1207 OPTION_MASK_ISA_PREFETCHWT1.
1208 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1209 (PTA_PREFETCHWT1): New.
1210 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1211 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1212 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1213 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1214 (*prefetch_avx512pf_<mode>_: Change into ...
1215 (*prefetch_prefetchwt1_<mode>: This.
1216 * config/i386/i386.opt (mprefetchwt1): New.
1217 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1218 (_mm_prefetch): Handle intent to write.
1219 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1221 2014-02-25 Richard Biener <rguenther@suse.de>
1224 * emit-rtl.c (mem_attrs_htab): Remove.
1225 (mem_attrs_htab_hash): Likewise.
1226 (mem_attrs_htab_eq): Likewise.
1227 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1228 (init_emit_once): Do not allocate mem_attrs_htab.
1230 2014-02-25 Richard Biener <rguenther@suse.de>
1233 * lto-opts.c (lto_write_options): Output non-explicit conservative
1234 -fwrapv, -fno-trapv and -fno-strict-overflow.
1235 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1236 (run_gcc): And pass them through.
1238 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1240 * sel-sched.c (calculate_new_fences): New parameter ptime.
1241 Calculate it as a maximum over all fence cycles.
1242 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1243 Print the final schedule timing when sched_verbose.
1245 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1247 PR rtl-optimization/60292
1248 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1249 bit fot the fence instruction.
1251 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1253 * calls.h: Fix typo in comment.
1255 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1257 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1258 adjusting offsetable addresses.
1260 2014-02-24 Guozhi Wei <carrot@google.com>
1262 * sparseset.h (sparseset_pop): Fix the wrong index.
1264 2014-02-24 Walter Lee <walt@tilera.com>
1266 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1267 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1269 * common/config/tilegx/tilegx-common.c
1270 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1271 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1273 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1274 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1275 (tilegx_gimplify_va_arg_expr): Handle big endian.
1276 (tilegx_expand_unaligned_load): Ditto.
1277 (tilegx_expand_unaligned_store): Ditto.
1278 (TARGET_RETURN_IN_MSB): New.
1279 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1280 (TARGET_ENDIAN_DEFAULT): New.
1281 (TARGET_BIG_ENDIAN): Handle big endian.
1282 (BYTES_BIG_ENDIAN): Ditto.
1283 (WORDS_BIG_ENDIAN): Ditto.
1284 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
1287 * config/tilegx/tilegx.md (extv): Handle big endian.
1289 (insn_st<n>): Ditto.
1290 (insn_st<n>_add<bitsuffix>): Ditto.
1291 (insn_stnt<n>): Ditto.
1292 (insn_stnt<n>_add<bitsuffix>):Ditto.
1293 (vec_interleave_highv8qi): Handle big endian.
1294 (vec_interleave_highv8qi_be): New.
1295 (vec_interleave_highv8qi_le): New.
1296 (insn_v1int_h): Handle big endian.
1297 (vec_interleave_lowv8qi): Handle big endian.
1298 (vec_interleave_lowv8qi_be): New.
1299 (vec_interleave_lowv8qi_le): New.
1300 (insn_v1int_l): Handle big endian.
1301 (vec_interleave_highv4hi): Handle big endian.
1302 (vec_interleave_highv4hi_be): New.
1303 (vec_interleave_highv4hi_le): New.
1304 (insn_v2int_h): Handle big endian.
1305 (vec_interleave_lowv4hi): Handle big endian.
1306 (vec_interleave_lowv4hi_be): New.
1307 (vec_interleave_lowv4hi_le): New.
1308 (insn_v2int_l): Handle big endian.
1309 (vec_interleave_highv2si): Handle big endian.
1310 (vec_interleave_highv2si_be): New.
1311 (vec_interleave_highv2si_le): New.
1312 (insn_v4int_h): Handle big endian.
1313 (vec_interleave_lowv2si): Handle big endian.
1314 (vec_interleave_lowv2si_be): New.
1315 (vec_interleave_lowv2si_le): New.
1316 (insn_v4int_l): Handle big endian.
1317 * config/tilegx/tilegx.opt (mbig-endian): New option.
1318 (mlittle-endian): New option.
1319 * doc/install.texi: Document tilegxbe-linux.
1320 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
1322 2014-02-24 Martin Jambor <mjambor@suse.cz>
1325 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
1326 there are no parameter descriptors.
1328 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
1330 PR rtl-optimization/60268
1331 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
1332 initialization to ...
1333 (sched_rgn_init): ... here.
1334 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
1336 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1338 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
1341 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
1343 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
1346 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1348 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
1349 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
1351 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
1353 * config/microblaze/predicates.md: Add cmp_op predicate.
1354 * config/microblaze/microblaze.md: Add branch_compare instruction
1355 which uses cmp_op predicate and emits cmp insn before branch.
1356 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
1357 to microblaze_expand_conditional_branch and consolidate logic.
1358 (microblaze_expand_conditional_branch): emit branch_compare
1359 insn instead of handling cmp op separate from branch insn.
1361 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1363 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
1366 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1368 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
1369 define_insn with define_expand and new define_insn
1370 *altivec_lve<VI_char>x_internal.
1371 (altivec_stve<VI_char>x): Replace define_insn with define_expand
1372 and new define_insn *altivec_stve<VI_char>x_internal.
1373 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
1375 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
1377 (altivec_expand_stvex_be): New function.
1379 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
1381 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
1382 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
1383 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
1384 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
1386 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
1389 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
1390 instead of emit_move_insn.
1392 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1394 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
1396 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
1397 gen_altivec_vsumsws.
1399 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1401 * config/rs6000/altivec.md (altivec_lvxl): Rename as
1402 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
1403 (altivec_lvxl_<mode>): New define_expand incorporating
1404 -maltivec=be semantics where needed.
1405 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
1406 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
1407 semantics where needed.
1408 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
1409 (altivec_stvx_<mode>): New define_expand incorporating
1410 -maltivec=be semantics where needed.
1411 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
1412 VM2 iterator instead of V4SI.
1413 (altivec_stvxl_<mode>): New define_expand incorporating
1414 -maltivec=be semantics where needed.
1415 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
1416 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
1417 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
1418 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
1419 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
1420 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
1421 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
1422 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
1423 ALTIVEC_BUILTIN_STVXL.
1424 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
1425 (altivec_expand_stvx_be): Likewise.
1426 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
1427 (altivec_expand_lvx_be): Likewise.
1428 (altivec_expand_stvx_be): Likewise.
1429 (altivec_expand_builtin): Add cases for
1430 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
1431 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
1432 (altivec_init_builtins): Add definitions for
1433 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
1434 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
1436 2014-02-21 Catherine Moore <clm@codesourcery.com>
1438 * doc/invoke.texi (mvirt, mno-virt): Document.
1439 * config/mips/mips.opt (mvirt): New option.
1440 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
1442 2014-02-21 Richard Biener <rguenther@suse.de>
1444 PR tree-optimization/60276
1445 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
1446 (STMT_VINFO_MIN_NEG_DIST): New macro.
1447 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
1448 STMT_VINFO_MIN_NEG_DIST.
1449 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
1450 made for negative dependence distances still hold.
1452 2014-02-21 Richard Biener <rguenther@suse.de>
1455 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
1456 DECL_INITIAL for globals not in the current function context.
1458 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1460 PR tree-optimization/56490
1461 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
1462 * tree-ssa-uninit.c: Include params.h.
1463 (compute_control_dep_chain): Add num_calls argument, return false
1464 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
1465 num_calls to recursive call.
1466 (find_predicates): Change dep_chain into normal array,
1467 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
1468 variable and adjust compute_control_dep_chain caller.
1469 (find_def_preds): Likewise.
1471 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
1473 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
1474 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
1476 2014-02-21 Nick Clifton <nickc@redhat.com>
1478 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
1479 (pushhi1): Likewise.
1480 (popqi1): Add mode to pre_dec.
1483 2014-02-21 Jakub Jelinek <jakub@redhat.com>
1485 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
1486 mode for mask of V8SFmode permutation.
1488 2014-02-20 Richard Henderson <rth@redhat.com>
1491 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
1492 a new pseudo for OLDVAL.
1494 2014-02-20 Jakub Jelinek <jakub@redhat.com>
1497 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
1498 gen_reg_rtx if d->testing_p.
1499 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
1500 if d->testing_p and we will certainly return true.
1501 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
1504 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
1506 * emit-rtl.c (gen_reg_rtx): Assert that
1507 crtl->emit.regno_pointer_align_length is non-zero.
1509 2014-02-20 Richard Henderson <rth@redhat.com>
1512 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
1513 on failure the store back into EXPECT.
1515 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
1516 Sandra Loosemore <sandra@codesourcery.com>
1518 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
1519 * config/nios2/nios2.c (nios2_function_profiler): Add
1520 -fPIC (flag_pic == 2) support.
1521 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
1522 (nios2_large_offset_p): New function.
1523 (nios2_unspec_reloc_p): Move up position, update to use
1524 nios2_large_offset_p.
1525 (nios2_unspec_address): Remove function.
1526 (nios2_unspec_offset): New function.
1527 (nios2_large_got_address): New function.
1528 (nios2_got_address): Add large offset support.
1529 (nios2_legitimize_tls_address): Update usage of removed and new
1531 (nios2_symbol_binds_local_p): New function.
1532 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
1533 (nios2_legitimize_address): Update to use nios2_large_offset_p.
1534 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
1535 (nios2_print_operand): Merge H/L processing, add hiadj/lo
1536 processing for (const (unspec ...)).
1537 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
1539 2014-02-20 Richard Biener <rguenther@suse.de>
1541 * tree-cfg.c (replace_uses_by): Mark altered BBs before
1542 doing the substitution.
1543 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
1545 2014-02-20 Martin Jambor <mjambor@suse.cz>
1548 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
1549 info when checking whether lattices are bottom.
1551 2014-02-20 Richard Biener <rguenther@suse.de>
1554 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
1557 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
1560 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
1561 parameter specifying the scaling.
1562 (inline_call): Update.
1563 (want_inline_recursively): Guard division by zero.
1564 (recursive_inlining): Update.
1565 * ipa-inline.h (clone_inlined_nodes): Update.
1567 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1570 * config/i386/i386.c (classify_argument): Pass structures of size
1571 64 bytes or less in register.
1573 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
1574 Kirill Yukhin <kirill.yukhin@intel.com>
1576 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
1577 (_mm_rcp28_round_ss): Ditto.
1578 (_mm_rsqrt28_round_sd): Ditto.
1579 (_mm_rsqrt28_round_ss): Ditto.
1580 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
1581 (_mm_rcp14_round_ss): Ditto.
1582 (_mm_rsqrt14_round_sd): Ditto.
1583 (_mm_rsqrt14_round_ss): Ditto.
1584 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
1585 the first input operand, get rid of match_dup.
1586 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
1588 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
1590 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
1591 operand as the first input operand, set type attribute.
1592 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
1594 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
1595 operand as the first input operand, set type attribute.
1597 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1599 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
1602 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
1605 * config/i386/i386.c (construct_container): Remove TFmode check
1606 for X86_64_INTEGER_CLASS.
1608 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
1611 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
1612 only when -Wpsabi is enabled.
1614 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
1617 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
1618 passing arrays in registers are the same as for structs, so remove the
1619 special case for them.
1621 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
1623 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
1624 destination type, extract only the valid bits if the source type is not
1625 integral and has a different mode.
1627 2014-02-19 Richard Biener <rguenther@suse.de>
1630 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
1633 2014-02-19 Richard Biener <rguenther@suse.de>
1636 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
1637 (ipa_modify_call_arguments): Emit an argument load explicitely and
1638 preserve virtual SSA form there and for the replacement call.
1639 Do not update SSA form nor free dominance info.
1641 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1643 * ipa.c (function_and_variable_visibility): Also clear WEAK
1644 flag when disolving COMDAT_GROUP.
1646 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1648 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
1649 * ipa-prop.c (ipa_set_jf_known_type): Return early when
1651 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
1652 do more sanity checks.
1653 (detect_type_change): Return true when giving up early.
1654 (compute_complex_assign_jump_func): Fix type parameter of
1655 ipa_set_ancestor_jf.
1656 (compute_complex_ancestor_jump_func): Likewise.
1657 (update_jump_functions_after_inlining): Fix updating of
1659 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
1661 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
1663 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
1664 inline clones when edge disappears.
1666 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1669 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
1670 Split 64-bit moves into 2 patterns. Do not allow the use of
1671 direct move for TDmode in little endian, since the decimal value
1672 has little endian bytes within a word, but the 64-bit pieces are
1673 ordered in a big endian fashion, and normal subreg's of TDmode are
1675 (mov<mode>_64bit_dm): Likewise.
1676 (movtd_64bit_nodm): Likewise.
1678 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1680 PR tree-optimization/60174
1681 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
1682 statement of an SSA_NAME that occurs in an abnormal PHI node.
1684 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1687 * final.c (SEEN_BB): Remove.
1688 (SEEN_NOTE, SEEN_EMITTED): Renumber.
1689 (final_scan_insn): Don't force_source_line on second
1690 NOTE_INSN_BASIC_BLOCK.
1692 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
1695 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
1696 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
1697 (type_natural_mode): Warn ABI change when %zmm register is not
1698 available for AVX512F vector value passing.
1700 2014-02-18 Kai Tietz <ktietz@redhat.com>
1703 * config/i386/i386.c (ix86_expand_prologue): Use value in
1704 rax register as displacement when restoring %r10 or %rax.
1705 Fix wrong offset when restoring both registers.
1707 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
1709 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
1710 assertion with conditional return.
1712 2014-02-18 Jakub Jelinek <jakub@redhat.com>
1713 Uros Bizjak <ubizjak@gmail.com>
1716 * config/i386/driver-i386.c (host_detect_local_cpu): If
1717 YMM state is not saved by the OS, also clear has_f16c. Move
1718 CPUID 0x80000001 handling before YMM state saving checking.
1720 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
1722 PR rtl-optimization/58960
1723 * haifa-sched.c (alloc_global_sched_pressure_data): New,
1724 factored out from ...
1725 (sched_init): ... here.
1726 (free_global_sched_pressure_data): New, factored out from ...
1727 (sched_finish): ... here.
1728 * sched-int.h (free_global_sched_pressure_data): Declare.
1729 * sched-rgn.c (nr_regions_initial): New static global.
1730 (haifa_find_rgns): Initialize it.
1731 (schedule_region): Disable sched-pressure for the newly
1734 2014-02-17 Richard Biener <rguenther@suse.de>
1736 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
1737 release SSA defs of pattern stmts.
1739 2014-02-17 Richard Biener <rguenther@suse.de>
1741 * tree-inline.c (expand_call_inline): Release the virtual
1742 operand defined by the call we are about to inline.
1744 2014-02-17 Richard Biener <rguenther@suse.de>
1746 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
1748 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
1749 Ilya Tocar <ilya.tocar@intel.com>
1751 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
1752 arguments order in builtin.
1753 (_mm512_permutexvar_epi64): Ditto.
1754 (_mm512_mask_permutexvar_epi64): Ditto
1755 (_mm512_maskz_permutexvar_epi32): Ditto
1756 (_mm512_permutexvar_epi32): Ditto
1757 (_mm512_mask_permutexvar_epi32): Ditto
1759 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1761 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
1762 (p8_vmrgow): Likewise.
1764 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1766 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
1769 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1772 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
1773 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
1774 into 64-bit and 32-bit moves. On 64-bit moves, add support for
1775 using direct move instructions on ISA 2.07. Also adjust
1776 instruction length for 64-bit.
1777 (mov<mode>_64bit, TFmode/TDmode): Likewise.
1778 (mov<mode>_32bit, TFmode/TDmode): Likewise.
1780 2014-02-15 Alan Modra <amodra@gmail.com>
1784 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
1785 find_replacement on parts of insn rtl that might be reloaded.
1787 2014-02-15 Richard Biener <rguenther@suse.de>
1789 PR tree-optimization/60183
1790 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
1791 (tree_ssa_phiprop): Calculate and free post-dominators.
1793 2014-02-14 Jeff Law <law@redhat.com>
1795 PR rtl-optimization/60131
1796 * ree.c (get_extended_src_reg): New function.
1797 (combine_reaching_defs): Use it rather than assuming location of REG.
1798 (find_and_remove_re): Verify first operand of extension is
1799 a REG before adding the insns to the copy list.
1801 2014-02-14 Roland McGrath <mcgrathr@google.com>
1803 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
1804 * configure: Regenerated.
1805 * config.in: Regenerated.
1806 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
1807 instead of ASM_SHORT.
1809 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
1810 Richard Earnshaw <rearnsha@arm.com>
1812 PR rtl-optimization/59535
1813 * lra-constraints.c (process_alt_operands): Encourage alternative
1814 when unassigned pseudo class is superset of the alternative class.
1815 (inherit_reload_reg): Don't inherit when optimizing for code size.
1816 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
1817 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
1818 modes not less than 4 for Thumb1.
1820 2014-02-14 Kyle McMartin <kyle@redhat.com>
1823 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
1825 2014-02-14 Richard Biener <rguenther@suse.de>
1827 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
1828 (get_frame_arg): Drop the assert with langhook types_compatible_p.
1829 Do not strip INDIRECT_REFs.
1831 2014-02-14 Richard Biener <rguenther@suse.de>
1834 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
1835 DECL_FUNCTION_SPECIFIC_TARGET.
1836 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
1837 * tree-streamer-out.c (pack_ts_target_option): Remove.
1838 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
1839 (write_ts_function_decl_tree_pointers): Do not stream
1840 DECL_FUNCTION_SPECIFIC_TARGET.
1841 * tree-streamer-in.c (unpack_ts_target_option): Remove.
1842 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
1843 (lto_input_ts_function_decl_tree_pointers): Do not stream
1844 DECL_FUNCTION_SPECIFIC_TARGET.
1846 2014-02-14 Jakub Jelinek <jakub@redhat.com>
1848 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
1849 (get_initial_def_for_induction, vectorizable_induction): Ignore
1850 debug stmts when looking for exit_phi.
1851 (vectorizable_live_operation): Fix up condition.
1853 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1855 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
1856 nreverse() because it changes the content of original tree list.
1858 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1860 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
1861 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
1863 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
1865 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
1866 GNU coding standards.
1868 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1871 * dwarf2out.c (gen_subprogram_die): Don't call
1872 add_calling_convention_attribute if subr_die is old_die.
1874 2014-02-13 Sharad Singhai <singhai@google.com>
1876 * doc/optinfo.texi: Fix order of nodes.
1878 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
1880 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
1881 operands[2], not operands[3].
1883 2014-02-13 Richard Biener <rguenther@suse.de>
1886 * doc/install.texi (ISL): Update recommended version to 0.12.2,
1887 mention the possibility of an in-tree build.
1888 (CLooG): Update recommended version to 0.18.1, mention the
1889 possibility of an in-tree build and clarify that the ISL
1890 bundled with CLooG does not work.
1892 2014-02-13 Jakub Jelinek <jakub@redhat.com>
1895 * expr.c (compress_float_constant): If x is a hard register,
1896 extend into a pseudo and then move to x.
1898 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
1900 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
1901 caused by bad second argument to warning_at() with -mhotpatch and
1902 nested functions (e.g. with gfortran).
1904 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
1906 * opts.c (option_name): Remove "enabled by default" rider.
1908 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
1910 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
1912 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
1913 Uros Bizjak <ubizjak@gmail.com>
1916 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
1917 * configure: Regenerated.
1919 2014-02-12 Richard Biener <rguenther@suse.de>
1921 * vec.c (vec_prefix::calculate_allocation): Move as
1922 inline variant to vec.h.
1923 (vec_prefix::calculate_allocation_1): New out-of-line version.
1924 * vec.h (vec_prefix::calculate_allocation_1): Declare.
1925 (vec_prefix::m_has_auto_buf): Rename to ...
1926 (vec_prefix::m_using_auto_storage): ... this.
1927 (vec_prefix::calculate_allocation): Inline the easy cases
1928 and dispatch to calculate_allocation_1 which doesn't need the
1930 (va_heap::reserve): Use gcc_checking_assert.
1931 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
1932 m_using_auto_storage.
1933 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
1935 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
1936 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
1937 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
1939 2014-02-12 Richard Biener <rguenther@suse.de>
1941 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
1942 when we found a dependence.
1944 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
1946 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
1948 (maybe_fold_stmt): ... into this new function.
1949 * omp-low.c (lower_omp): Update comment.
1951 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
1954 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
1957 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
1959 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
1960 identifiers in comments.
1961 (cortexa53_extra_costs): Likewise.
1962 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
1963 (cortexa7_extra_costs): Likewise.
1964 (cortexa12_extra_costs): Likewise.
1965 (cortexa15_extra_costs): Likewise.
1966 (v7m_extra_costs): Likewise.
1968 2014-02-12 Richard Biener <rguenther@suse.de>
1971 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
1972 of posix_memalign being successful.
1973 (lower_stmt): Restrict lowering of posix_memalign to when
1974 -ftree-bit-ccp is enabled.
1976 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1978 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
1980 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
1982 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
1984 PR rtl-optimization/60116
1985 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
1986 other_insn once the combination has been validated.
1988 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
1991 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
1993 * ipa-devirt.c: Include demangle.h
1994 (odr_violation_reported): New static variable.
1995 (add_type_duplicate): Update odr_violations.
1996 (maybe_record_node): Add completep parameter; update it.
1997 (record_target_from_binfo): Add COMPLETEP parameter;
1998 update it as needed.
1999 (possible_polymorphic_call_targets_1): Likewise.
2000 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2001 rename FINAL to COMPLETE.
2002 (record_targets_from_bases): Sanity check we found the binfo;
2003 fix COMPLETEP updating.
2004 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2005 parameter, fix computing of COMPLETEP.
2006 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2007 at LTO time do demangling.
2008 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2009 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2011 (gimple_get_virt_method_for_binfo): Likewise.
2012 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2013 gimple_get_virt_method_for_vtable): Update prototypes.
2015 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2018 * genautomata.c (add_presence_absence): Fix typo with
2019 {final_}presence_list.
2021 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2024 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2025 for VSX/Altivec vectors that land in GPR registers.
2027 2014-02-11 Richard Henderson <rth@redhat.com>
2028 Jakub Jelinek <jakub@redhat.com>
2031 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2032 around drhs if type conversion to lacc->type is not useless.
2034 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2036 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2038 (cortex-a57.cortex-a53): Likewise.
2039 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2041 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2043 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2046 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2048 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2049 add_options_for_arm_vfp3.
2051 2014-02-11 Jeff Law <law@redhat.com>
2054 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2055 object with an undesirable mode.
2057 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2060 * config/i386/sol2-9.h: New file.
2061 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2062 *-*-solaris2.9*): Use it.
2064 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2066 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2067 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2069 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2071 * config/microblaze/microblaze.c: Extend mcpu version format
2073 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2075 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2077 2014-02-10 Richard Henderson <rth@redhat.com>
2080 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2081 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2082 ms-abi vs -mno-accumulate-outgoing-args.
2083 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2084 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2087 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2090 * cfgexpand.c (expand_asm_operands): Attach source location to
2091 ASM_INPUT rtx objects.
2092 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2094 2014-02-10 Nick Clifton <nickc@redhat.com>
2096 * config/mn10300/mn10300.c (popcount): New function.
2097 (mn10300_expand_prologue): Include saved registers in stack usage
2100 2014-02-10 Jeff Law <law@redhat.com>
2103 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2104 when changing the SET_DEST of a prior insn to avoid an input reload.
2106 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2108 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2109 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2110 -mcall-openbsd, or -mcall-linux.
2111 (CC1_ENDIAN_BIG_SPEC): Remove.
2112 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2113 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2114 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2115 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2116 and %cc1_endian_default.
2117 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2119 2014-02-10 Richard Biener <rguenther@suse.de>
2121 PR tree-optimization/60115
2122 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2123 MEM_REF handling. Properly verify that the accesses are not
2124 out of the objects bound.
2126 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2128 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2131 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2133 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2134 proper constants and fix formatting.
2135 (possible_polymorphic_call_targets): Fix formatting.
2137 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2138 Ilya Tocar <ilya.tocar@intel.com>
2140 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2141 (_mm512_loadu_epi32): Renamed into...
2142 (_mm512_loadu_si512): This.
2143 (_mm512_storeu_epi32): Renamed into...
2144 (_mm512_storeu_si512): This.
2145 (_mm512_maskz_ceil_ps): Removed.
2146 (_mm512_maskz_ceil_pd): Ditto.
2147 (_mm512_maskz_floor_ps): Ditto.
2148 (_mm512_maskz_floor_pd): Ditto.
2149 (_mm512_floor_round_ps): Ditto.
2150 (_mm512_floor_round_pd): Ditto.
2151 (_mm512_ceil_round_ps): Ditto.
2152 (_mm512_ceil_round_pd): Ditto.
2153 (_mm512_mask_floor_round_ps): Ditto.
2154 (_mm512_mask_floor_round_pd): Ditto.
2155 (_mm512_mask_ceil_round_ps): Ditto.
2156 (_mm512_mask_ceil_round_pd): Ditto.
2157 (_mm512_maskz_floor_round_ps): Ditto.
2158 (_mm512_maskz_floor_round_pd): Ditto.
2159 (_mm512_maskz_ceil_round_ps): Ditto.
2160 (_mm512_maskz_ceil_round_pd): Ditto.
2161 (_mm512_expand_pd): Ditto.
2162 (_mm512_expand_ps): Ditto.
2163 * config/i386/i386.c (ix86_builtins): Remove
2164 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2165 (bdesc_args): Ditto.
2166 * config/i386/predicates.md (const1256_operand): New.
2167 (const_1_to_2_operand): Ditto.
2168 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2169 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2170 (*avx512pf_gatherpf<mode>sf): Ditto.
2171 (avx512pf_gatherpf<mode>df): Ditto.
2172 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2173 (*avx512pf_gatherpf<mode>df): Ditto.
2174 (avx512pf_scatterpf<mode>sf): Ditto.
2175 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2176 (*avx512pf_scatterpf<mode>sf): Ditto.
2177 (avx512pf_scatterpf<mode>df): Ditto.
2178 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2179 (*avx512pf_scatterpf<mode>df): Ditto.
2180 (avx512f_expand<mode>): Removed.
2181 (<shift_insn><mode>3<mask_name>): Change predicate type.
2183 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2185 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2186 not at the end of datarefs vector use ordered_remove to avoid
2187 reordering datarefs vector.
2190 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2191 mark local addressable non-static vars as GOVD_PRIVATE
2192 instead of GOVD_LOCAL.
2193 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2194 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2198 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2199 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2200 assume_aligned or alloc_align attributes.
2201 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2202 arguments. Handle also assume_aligned and alloc_align attributes.
2203 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2204 calls to functions with assume_aligned or alloc_align attributes.
2205 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2207 2014-02-08 Terry Guo <terry.guo@arm.com>
2209 * doc/invoke.texi: Document ARM -march=armv7e-m.
2211 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2213 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2214 flag on __cilkrts_rethrow builtin.
2217 * ipa-cp.c (determine_versionability): Fail at -O0
2218 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2219 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2222 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2225 * tree-inline.c (copy_forbidden): Fail for
2226 __attribute__((optimize (0))) functions.
2228 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2230 * varpool.c: Include pointer-set.h.
2231 (varpool_remove_unreferenced_decls): Variables in other partitions
2232 will not be output; be however careful to not lose information
2235 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2237 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2238 lookup in the vtable constructor.
2240 2014-02-07 Jeff Law <law@redhat.com>
2243 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2244 define_insn_and_split.
2246 * ipa-inline.c (inline_small_functions): Fix typos.
2248 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2250 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2251 (s390_can_use_return_insn): Declare.
2252 * config/s390/s390.h (EPILOGUE_USES): Define.
2253 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2255 (s390_chunkify_start): Handle return JUMP_LABELs.
2256 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2257 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2258 (s390_can_use_return_insn): New functions.
2259 (s390_fix_long_loop_prediction): Handle conditional returns.
2260 (TARGET_SET_UP_BY_PROLOGUE): Define.
2261 * config/s390/s390.md (ANY_RETURN): New code iterator.
2262 (*creturn, *csimple_return, return, simple_return): New patterns.
2264 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2266 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2267 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2268 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2269 REG_CFA_RESTORE list when deciding not to restore a register.
2271 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2273 * config/s390/s390.c: Include tree-pass.h and context.h.
2274 (s390_early_mach): New function, split out from...
2275 (s390_emit_prologue): ...here.
2276 (pass_data_s390_early_mach): New pass structure.
2277 (pass_s390_early_mach): New class.
2278 (s390_option_override): Create and register early_mach pass.
2279 Move to end of file.
2281 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2283 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2284 to match for the exit block.
2286 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2288 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
2289 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
2290 Reject misaligned operands.
2292 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2294 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
2296 2014-02-07 Richard Biener <rguenther@suse.de>
2299 * gimple-low.c (lower_builtin_posix_memalign): New function.
2300 (lower_stmt): Call it to lower posix_memalign in a way
2301 to make alignment info accessible.
2303 2014-02-07 Jakub Jelinek <jakub@redhat.com>
2306 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
2307 __builtin_setjmp_receiver.
2309 2014-02-07 Richard Biener <rguenther@suse.de>
2312 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
2313 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
2314 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
2315 Handle BUILT_IN_POSIX_MEMALIGN.
2316 (find_func_clobbers): Likewise.
2317 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2318 (call_may_clobber_ref_p_1): Likewise.
2320 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2323 * ipa-devirt.c (record_target_from_binfo): Remove overactive
2326 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2329 * lto-cgraph.c (lto_output_node): Use
2330 symtab_get_symbol_partitioning_class.
2331 (lto_output_varpool_node): likewise.
2332 (symtab_get_symbol_partitioning_class): Move here from
2334 * cgraph.h (symbol_partitioning_class): Likewise.
2335 (symtab_get_symbol_partitioning_class): Declare.
2337 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2339 * ggc.h (ggc_internal_cleared_alloc): New macro.
2340 * vec.h (vec_safe_copy): Handle memory stats.
2341 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
2342 * target-globals.c (save_target_globals): Likewise.
2344 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
2347 * expr.c (emit_move_resolve_push): Export; be bit more selective
2348 on when to clear alias set.
2349 * expr.h (emit_move_resolve_push): Declare.
2350 * function.h (struct function): Add tail_call_marked.
2351 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
2352 * config/i386/i386-protos.h (ix86_expand_push): Remove.
2353 * config/i386/i386.md (TImode move expander): De not call
2355 (FP push expanders): Preserve memory attributes.
2356 * config/i386/sse.md (push<mode>1): Remove.
2357 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
2358 (ix86_expand_push): Remove.
2359 * config/i386/mmx.md (push<mode>1): Remove.
2361 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2363 PR rtl-optimization/60030
2364 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
2365 lopart with paradoxical subreg before shifting it up by hprec.
2367 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2369 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
2370 Remove extra newline at end of file.
2371 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
2372 (arm_issue_rate): Handle cortexa57.
2373 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
2374 (cortex-a57.cortex-a53): Likewise.
2376 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2379 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
2380 don't record in REG_FRAME_RELATED_EXPR registers not set in that
2382 (arm_expand_prologue): Adjust all callers.
2383 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
2384 info, registers also at the lowest numbered registers side. Use
2385 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
2389 * var-tracking.c (adjust_mems): Before adding a SET to
2390 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
2392 2014-02-06 Alan Modra <amodra@gmail.com>
2395 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
2396 change SDmode to DDmode when lra_in_progress.
2398 2014-02-06 Jakub Jelinek <jakub@redhat.com>
2401 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
2402 free_data_ref on the dr first, and before goto again also set dr
2403 to the next dr. For simd_lane_access, free old datarefs[i] before
2404 overwriting it. For get_vectype_for_scalar_type failure, don't
2405 free_data_ref if simd_lane_access.
2407 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
2410 * tree.h (opts_for_fn): New inline function.
2411 (opt_for_fn): Define.
2412 * config/i386/i386.c (ix86_function_regparm): Use
2413 opt_for_fn (decl, optimize) instead of optimize.
2415 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
2417 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
2418 for SYMBOL_REF in large memory model.
2420 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2422 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
2424 (cortex-a57): Likewise.
2425 (cortex-a57.cortex-a53): Likewise.
2427 2014-02-06 Yury Gribov <y.gribov@samsung.com>
2428 Kugan Vivekanandarajah <kuganv@linaro.org>
2430 * config/arm/arm.c (arm_vector_alignment_reachable): Check
2432 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
2434 2014-02-06 Richard Biener <rguenther@suse.de>
2436 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
2437 set_loop_copy and initialize_original_copy_tables.
2439 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
2441 * config/aarch64/aarch64-simd.md
2442 (aarch64_ashr_simddi): Change QI to SI.
2444 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2445 Jakub Jelinek <jakub@redhat.com>
2448 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
2451 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2453 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
2454 CODE_FOR_altivec_vpku[hw]um to
2455 CODE_FOR_altivec_vpku[hw]um_direct.
2456 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
2457 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
2458 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
2459 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
2461 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2463 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
2464 generation for -maltivec=be.
2465 (altivec_vsumsws): Simplify redundant test.
2467 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2469 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
2470 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
2471 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
2472 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
2473 gen_altivec_vpkuwum.
2474 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
2476 (altivec_vpks<VI_char>ss): Likewise.
2477 (altivec_vpks<VI_char>us): Likewise.
2478 (altivec_vpku<VI_char>us): Likewise.
2479 (altivec_vpku<VI_char>um): Likewise.
2480 (altivec_vpku<VI_char>um_direct): New (copy of
2481 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
2483 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
2484 target is little endian and -maltivec=be is not specified.
2485 (*altivec_vupkhs<VU_char>_direct): New (copy of
2486 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
2487 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
2488 target is little endian and -maltivec=be is not specified.
2489 (*altivec_vupkls<VU_char>_direct): New (copy of
2490 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
2491 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
2492 little endian and -maltivec=be is not specified.
2493 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
2494 little endian and -maltivec=be is not specified.
2496 2014-02-05 Richard Henderson <rth@redhat.com>
2499 * combine-stack-adj.c: Revert r206943.
2500 * sched-int.h (struct deps_desc): Add last_args_size.
2501 * sched-deps.c (init_deps): Initialize it.
2502 (sched_analyze_insn): Add OUTPUT dependencies between insns that
2503 contain REG_ARGS_SIZE notes.
2505 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
2507 * lto-cgraph.c (asm_nodes_output): Make global.
2508 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
2509 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
2510 (driver_handle_option): Handle OPT_fwpa.
2512 2014-02-05 Jakub Jelinek <jakub@redhat.com>
2515 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
2516 a comment typo and formatting issue. If odr_hash hasn't been
2517 created, return vNULL and set *completep to false.
2520 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
2521 bb with no successors.
2523 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
2526 * doc/invoke.texi (-march): Clarify documentation for ARM.
2530 2014-02-05 Richard Biener <rguenther@suse.de>
2532 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
2533 when not vectorizing because of too many alias checks.
2534 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
2535 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
2537 2014-02-05 Nick Clifton <nickc@redhat.com>
2539 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
2540 accept extended registers in any mode when compiling for the MN10300.
2542 2014-02-05 Yury Gribov <y.gribov@samsung.com>
2544 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
2545 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
2546 sanitization attributes.
2547 (can_inline_edge_p): Likewise.
2548 (sanitize_attrs_match_for_inline_p): New function.
2550 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2552 * ipa-prop.c (detect_type_change): Shor circuit testing of
2553 type changes on THIS pointer.
2555 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
2558 * config/pa/pa.c (legitimize_tls_address): Return original address
2559 if not passed a SYMBOL_REF rtx.
2560 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
2562 (pa_emit_move_sequence): Simplify TLS source operands.
2563 (pa_legitimate_constant_p): Reject all TLS constants.
2564 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
2565 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
2567 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2569 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
2570 groups when we know they are controlled by LTO.
2571 * varasm.c (default_binds_local_p_1): If object is in other partition,
2572 it will be resolved locally.
2574 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2576 * config/host-linux.c (linux_gt_pch_use_address): Don't
2577 use SSIZE_MAX because it is not always defined.
2579 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
2582 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
2583 threshold for pseudo splitting.
2584 (update_ebb_live_info): Process call argument hard registers and
2585 hard registers from insn definition too.
2586 (max_small_class_regs_num): New constant.
2587 (inherit_in_ebb): Update live hard regs through EBBs. Update
2588 reloads_num only for small register classes. Don't split for
2591 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
2594 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
2597 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
2599 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
2602 2014-02-04 Marek Polacek <polacek@redhat.com>
2604 * gdbinit.in (pel): Define.
2606 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2608 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
2611 2014-02-04 Richard Biener <rguenther@suse.de>
2614 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
2615 in function context local.
2616 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
2617 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
2618 similar to LTO_imported_decl_ref.
2620 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2622 PR tree-optimization/60002
2623 * cgraphclones.c (build_function_decl_skip_args): Clear
2626 PR tree-optimization/60023
2627 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
2628 false to gsi_replace.
2629 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
2630 has been in some EH region and vec_stmt could throw, add
2631 vec_stmt into the same EH region.
2632 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
2633 has no lhs, ignore it.
2634 * internal-fn.c (expand_MASK_LOAD): Likewise.
2637 * tree-inline.c (copy_forbidden): Fail for
2638 __attribute__((optimize (0))) functions.
2641 * omp-low.c (simd_clone_struct_copy): If from->inbranch
2642 is set, copy one less argument.
2643 (expand_simd_clones): Don't subtract clone_info->inbranch
2644 from simd_clone_struct_alloc argument.
2646 PR rtl-optimization/57915
2647 * recog.c (simplify_while_replacing): If all unary/binary/relational
2648 operation arguments are constant, attempt to simplify those.
2651 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
2652 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
2654 2014-02-04 Richard Biener <rguenther@suse.de>
2656 PR tree-optimization/60012
2657 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
2658 TBAA disambiguation to all DDRs.
2660 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2663 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
2664 (LINK_SPEC): Use it for -shared, -shared-libgcc.
2666 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2669 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
2671 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2673 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
2674 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
2676 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2679 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
2680 to figure out targets of polymorphic calls with known decl.
2681 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2682 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
2683 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
2684 (get_polymorphic_call_info): ... here.
2685 (get_polymorphic_call_info_from_invariant): New function.
2687 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2689 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
2690 lookup via vtable pointer; check for type consistency
2691 and turn inconsitent facts into UNREACHABLE.
2692 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2693 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
2694 type inconsistent querries; return UNREACHABLE instead.
2696 2014-02-03 Richard Henderson <rth@twiddle.net>
2699 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
2700 already processed this node.
2701 (normalize_one_pred_1): Pass along mark_set.
2702 (normalize_one_pred): Create and destroy a pointer_set_t.
2703 (normalize_one_pred_chain): Likewise.
2705 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
2707 PR gcov-profile/58602
2708 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
2710 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
2713 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
2714 -fno-devirtualize; try to devirtualize by the knowledge of
2715 virtual table pointer given by aggregate propagation.
2716 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
2717 (ipa_print_node_jump_functions): Dump also offset that
2718 is relevant for polymorphic calls.
2719 (determine_known_aggregate_parts): Add arg_type parameter; use it
2720 instead of determining the type from pointer type.
2721 (ipa_compute_jump_functions_for_edge): Update call of
2722 determine_known_aggregate_parts.
2723 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
2724 (gimple_get_virt_method_for_binfo): ... here; simplify using
2725 vtable_pointer_value_to_vtable.
2726 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
2727 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
2728 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
2729 (vtable_pointer_value_to_vtable): Break out from ...; handle also
2731 (vtable_pointer_value_to_binfo): ... here.
2732 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
2734 2014-02-03 Teresa Johnson <tejohnson@google.com>
2736 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
2737 redef of outer loop index variable.
2739 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
2743 * doc/extend.texi (Function Attributes): Typo.
2745 2014-02-03 Cong Hou <congh@google.com>
2747 PR tree-optimization/60000
2748 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
2749 if the vectorized statement is a store. A store statement can only
2750 appear at the end of pattern statements.
2752 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2754 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
2755 (ix86_option_override_internal): Default long double to 64-bit for
2756 32-bit Bionic and to 128-bit for 64-bit Bionic.
2758 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
2759 TARGET_LONG_DOUBLE_128 is true.
2760 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
2762 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
2763 (mlong-double-64): Negate -mlong-double-128.
2764 (mlong-double-128): New option.
2766 * config/i386/i386-c.c (ix86_target_macros): Define
2767 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
2769 * doc/invoke.texi: Document -mlong-double-128.
2771 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
2773 PR rtl-optimization/60024
2774 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
2776 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
2778 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
2780 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2782 PR rtl-optimization/57662
2783 * sel-sched.c (code_motion_path_driver): Do not mark already not
2784 existing blocks in the visiting bitmap.
2786 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
2788 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
2789 on the insn being emitted.
2791 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
2792 Will Deacon <will.deacon@arm.com>
2794 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
2796 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2798 * config/arm/arm-tables.opt: Regenerate.
2800 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2802 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
2803 for vector types other than V16QImode.
2804 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
2805 define_expand, and call altivec_expand_vec_perm_le when producing
2806 code with little endian element order.
2807 (*altivec_vperm_<mode>_internal): New insn having previous
2808 behavior of altivec_vperm_<mode>.
2809 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
2810 altivec_expand_vec_perm_le when producing code with little endian
2812 (*altivec_vperm_<mode>_uns_internal): New insn having previous
2813 behavior of altivec_vperm_<mode>_uns.
2815 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2817 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
2818 (altivec_vsumsws): Add handling for -maltivec=be with a little
2820 (altivec_vsumsws_direct): New.
2821 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
2822 gen_altivec_vsumsws.
2824 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2826 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
2827 vtable_pointer_value_to_binfo): New functions.
2828 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
2829 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
2831 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
2833 * config/nios2/nios2.md (load_got_register): Initialize GOT
2834 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
2835 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
2837 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
2839 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
2840 preserverd by passthrough, do not propagate the type.
2842 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2844 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
2845 (mips_atomic_assign_expand_fenv): New function.
2846 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
2848 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2850 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
2851 (__builtin_mips_set_fcsr): Likewise.
2852 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
2853 MIPS_USI_FTYPE_VOID.
2854 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
2855 (mips16_expand_set_fcsr): Likewise.
2856 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
2857 (mips16_set_fcsr_stub): Likewise.
2858 (mips16_get_fcsr_one_only_stub): New class.
2859 (mips16_set_fcsr_one_only_stub): Likewise.
2860 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
2861 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
2862 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
2863 (hard_float): New availability predicate.
2864 (mips_builtins): Add get_fcsr and set_fcsr.
2865 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
2866 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
2867 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
2868 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
2869 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
2872 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
2874 * config/mips/mips.c (mips_one_only_stub): New class.
2875 (mips_need_mips16_rdhwr_p): Replace with...
2876 (mips16_rdhwr_stub): ...this new variable.
2877 (mips16_stub_call_address): New function.
2878 (mips16_rdhwr_one_only_stub): New class.
2879 (mips_expand_thread_pointer): Use mips16_stub_call_address.
2880 (mips_output_mips16_rdhwr): Delete.
2881 (mips_finish_stub): New function.
2882 (mips_code_end): Use it to handle rdhwr stubs.
2884 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
2887 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
2888 when calculating size of integer atomic types.
2890 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
2892 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
2894 2014-02-01 Jakub Jelinek <jakub@redhat.com>
2896 PR tree-optimization/60003
2897 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
2898 * profile.c (branch_prob): Use gimple_call_builtin_p
2899 to check for BUILT_IN_SETJMP_RECEIVER.
2900 * tree-inline.c (copy_bb): Call notice_special_calls.
2902 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
2905 * lra-constraints.c (process_alt_operands): Update reload_sum only
2908 2014-01-31 Richard Henderson <rth@redhat.com>
2911 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
2912 until after else_eh is processed.
2914 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2916 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
2917 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
2918 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
2919 in smmintrin.h, remove them.
2920 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
2921 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
2922 * config/i386/i386.md (ROUND_SAE): Fix value.
2923 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
2924 (const48_operand): New.
2925 * config/i386/subst.md (round), (round_expand): Use
2926 const_4_or_8_to_11_operand.
2927 (round_saeonly), (round_saeonly_expand): Use const48_operand.
2929 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
2931 * config/i386/constraints.md (Yk): Swap meaning with k.
2932 * config/i386/i386.md (movhi_internal): Change Yk to k.
2933 (movqi_internal): Ditto.
2934 (*k<logic><mode>): Ditto.
2937 (kandn<mode>): Ditto.
2938 (*<code>hi_1): Ditto.
2939 (*<code>qi_1): Ditto.
2940 (kxnor<mode>): Ditto.
2941 (kortestzhi): Ditto.
2942 (kortestchi): Ditto.
2944 (*one_cmplhi2_1): Ditto.
2945 (*one_cmplqi2_1): Ditto.
2946 * config/i386/sse.md (): Change k to Yk.
2947 (avx512f_load<mode>_mask): Ditto.
2948 (avx512f_blendm<mode>): Ditto.
2949 (avx512f_store<mode>_mask): Ditto.
2950 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
2951 (avx512f_storedqu<mode>_mask): Ditto.
2952 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
2954 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
2955 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
2956 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
2957 (avx512f_maskcmp<mode>3): Ditto.
2958 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
2959 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
2960 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
2961 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
2962 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
2963 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
2964 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
2965 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
2966 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
2967 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
2968 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
2969 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
2970 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
2971 (vec_extract_lo_<mode>_maskm): Ditto.
2972 (vec_extract_hi_<mode>_maskm): Ditto.
2973 (avx512f_vternlog<mode>_mask): Ditto.
2974 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
2975 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
2976 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
2977 (avx512f_<code>v8div16qi2_mask): Ditto.
2978 (avx512f_<code>v8div16qi2_mask_store): Ditto.
2979 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
2980 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
2981 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
2982 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
2983 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2984 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2985 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2986 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2987 (avx512cd_maskb_vec_dupv8di): Ditto.
2988 (avx512cd_maskw_vec_dupv16si): Ditto.
2989 (avx512f_vpermi2var<mode>3_maskz): Ditto.
2990 (avx512f_vpermi2var<mode>3_mask): Ditto.
2991 (avx512f_vpermi2var<mode>3_mask): Ditto.
2992 (avx512f_vpermt2var<mode>3_maskz): Ditto.
2993 (*avx512f_gathersi<mode>): Ditto.
2994 (*avx512f_gathersi<mode>_2): Ditto.
2995 (*avx512f_gatherdi<mode>): Ditto.
2996 (*avx512f_gatherdi<mode>_2): Ditto.
2997 (*avx512f_scattersi<mode>): Ditto.
2998 (*avx512f_scatterdi<mode>): Ditto.
2999 (avx512f_compress<mode>_mask): Ditto.
3000 (avx512f_compressstore<mode>_mask): Ditto.
3001 (avx512f_expand<mode>_mask): Ditto.
3002 * config/i386/subst.md (mask): Change k to Yk.
3003 (mask_scalar_merge): Ditto.
3006 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3008 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3010 2014-01-31 Richard Biener <rguenther@suse.de>
3013 * builtins.c (fold_builtin_memory_op): Make sure to not
3014 use a floating-point mode or a boolean or enumeral type for
3017 2014-01-30 DJ Delorie <dj@redhat.com>
3019 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3020 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3021 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3022 whenever main() has an epilogue.
3024 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3026 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3027 unused variable "field".
3028 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3029 (vsx_mergeh_<mode>): Likewise.
3030 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3031 (altivec_vmrghh): Likewise.
3032 (altivec_vmrghw): Likewise.
3033 (altivec_vmrglb): Likewise.
3034 (altivec_vmrglh): Likewise.
3035 (altivec_vmrglw): Likewise.
3036 (altivec_vspltb): Add missing uses.
3037 (altivec_vsplth): Likewise.
3038 (altivec_vspltw): Likewise.
3039 (altivec_vspltsf): Likewise.
3041 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3044 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3045 frame related instructions.
3047 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3049 PR rtl-optimization/59959
3050 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3051 any reload of register whose subreg is invalid.
3053 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3055 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3056 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3057 Add missing return type - void.
3059 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3061 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3062 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3063 remove element index adjustment for endian (now handled in vsx.md
3065 (altivec_expand_vec_perm_const): Use
3066 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3067 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3068 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3069 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3070 define_expand and a new define_insn *altivec_vspltb_internal;
3071 adjust for -maltivec=be on a little endian target.
3072 (altivec_vspltb_direct): New.
3073 (altivec_vsplth): Divide into a define_expand and a new
3074 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3075 little endian target.
3076 (altivec_vsplth_direct): New.
3077 (altivec_vspltw): Divide into a define_expand and a new
3078 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3079 little endian target.
3080 (altivec_vspltw_direct): New.
3081 (altivec_vspltsf): Divide into a define_expand and a new
3082 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3083 a little endian target.
3085 2014-01-30 Richard Biener <rguenther@suse.de>
3087 PR tree-optimization/59993
3088 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3089 can propagate form the earlier stmt and avoid the transform
3090 when the intermediate result is needed.
3092 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3094 * README.Portability: Fix typo.
3096 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3098 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3099 comparison_operator with ordered_comparison_operator.
3101 2014-01-30 Nick Clifton <nickc@redhat.com>
3103 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3104 Rename to mn10300_store_multiple_regs.
3105 * config/mn10300/mn10300.c: Likewise.
3106 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3107 store_multiple_regs.
3108 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3109 Call mn10300_store_multiple_regs.
3111 2014-01-30 Nick Clifton <nickc@redhat.com>
3112 DJ Delorie <dj@redhat.com>
3114 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3115 %fp 2 to keep registers after it properly word-aligned.
3116 (rl78_alloc_physical_registers_umul): Handle the case where both
3117 input operands are the same.
3119 2014-01-30 Richard Biener <rguenther@suse.de>
3121 PR tree-optimization/59903
3122 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3125 2014-01-30 Jason Merrill <jason@redhat.com>
3128 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3131 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3133 2014-01-30 Richard Biener <rguenther@suse.de>
3135 PR tree-optimization/59951
3136 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3138 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3141 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3142 SFmode to DFmode case.
3144 2014-01-29 DJ Delorie <dj@redhat.com>
3146 * config/msp430/msp430.opt (-minrt): New.
3147 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3149 (ENDFILE_SPEC): Likewise.
3151 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3153 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3154 (estimate_function_body_sizes): Use it.
3156 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3159 * dwarf2out.c (is_cxx_auto): New.
3160 (is_base_type): Use it.
3161 (gen_type_die_with_usage): Likewise.
3163 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3165 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3166 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3167 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3168 -maltivec=be with LE targets.
3169 (vsx_mergeh_<mode>): Likewise.
3170 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3171 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3172 (altivec_vmrghb): Replace with define_expand and new
3173 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3174 (altivec_vmrghb_direct): New define_insn.
3175 (altivec_vmrghh): Replace with define_expand and new
3176 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3177 (altivec_vmrghh_direct): New define_insn.
3178 (altivec_vmrghw): Replace with define_expand and new
3179 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3180 (altivec_vmrghw_direct): New define_insn.
3181 (*altivec_vmrghsf): Adjust for endianness.
3182 (altivec_vmrglb): Replace with define_expand and new
3183 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3184 (altivec_vmrglb_direct): New define_insn.
3185 (altivec_vmrglh): Replace with define_expand and new
3186 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3187 (altivec_vmrglh_direct): New define_insn.
3188 (altivec_vmrglw): Replace with define_expand and new
3189 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3190 (altivec_vmrglw_direct): New define_insn.
3191 (*altivec_vmrglsf): Adjust for endianness.
3192 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3193 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3194 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3195 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3196 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3197 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3198 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3199 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3201 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3203 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3204 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3207 2014-01-29 Richard Biener <rguenther@suse.de>
3209 PR tree-optimization/58742
3210 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3211 associate_pointerplus_align.
3212 (associate_pointerplus_diff): New function.
3213 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3214 and associate_pointerplus_diff.
3216 2014-01-29 Richard Biener <rguenther@suse.de>
3218 * lto-streamer.h (LTO_major_version): Bump to 3.
3219 (LTO_minor_version): Reset to 0.
3221 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3223 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3224 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3225 (arm_file_start): Generate correct asm header for armv7ve.
3226 * config/arm/bpabi.h: Add multilib support for armv7ve.
3227 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3228 and cortex-a15 to armv7ve.
3229 * config/arm/t-aprofile: Add multilib support for armv7ve.
3230 * doc/invoke.texi: Document -march=armv7ve.
3232 2014-01-29 Richard Biener <rguenther@suse.de>
3234 PR tree-optimization/58742
3235 * tree-ssa-forwprop.c (associate_plusminus): Return true
3236 if we changed sth, defer EH cleanup to ...
3237 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3238 (simplify_mult): New function.
3240 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3243 PR tree-optimization/59920
3244 * tree.c (build_common_builtin_nodes): Remove
3245 __builtin_setjmp_dispatcher initialization.
3246 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3247 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3248 instead of gsi_after_labels + manually skipping debug stmts.
3249 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3250 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3251 * tree-inline.c (copy_edges_for_bb): Remove
3252 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3253 argument. Ignore computed_goto_p stmts. Don't call
3254 make_abnormal_goto_edges. If a call might need abnormal edges
3255 for non-local gotos, see if it already has an edge to
3256 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3257 with true argument, don't do anything then, otherwise add
3258 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3259 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3261 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3262 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3263 (lower_stmt): Don't set data->calls_builtin_setjmp.
3264 (lower_builtin_setjmp): Adjust comment.
3265 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3266 * tree-cfg.c (found_computed_goto): Remove.
3267 (factor_computed_gotos): Remove.
3268 (make_goto_expr_edges): Return bool, true for computed gotos.
3269 Don't call make_abnormal_goto_edges.
3270 (build_gimple_cfg): Don't set found_computed_goto, don't call
3271 factor_computed_gotos.
3272 (computed_goto_p): No longer static.
3273 (make_blocks): Don't set found_computed_goto.
3274 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3275 (make_edges): If make_goto_expr_edges returns true, push bb
3276 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3277 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3278 vector. Record mapping between bbs and OpenMP regions if there
3279 are any, adjust make_gimple_omp_edges caller. Call
3280 handle_abnormal_edges.
3281 (make_abnormal_goto_edges): Remove.
3282 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3283 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3284 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3285 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3286 * internal-fn.def (ABNORMAL_DISPATCHER): New.
3287 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
3288 filling *region also set *region_idx to (*region)->entry->index.
3291 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
3292 For REGs set ORIGINAL_REGNO.
3294 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
3296 * doc/md.texi: Mention that a target shouldn't implement
3297 vec_widen_(s|u)mul_even/odd pair if it is less efficient
3300 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3302 PR tree-optimization/59594
3303 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
3304 a copy of the datarefs vector rather than the vector itself.
3306 2014-01-28 Jason Merrill <jason@redhat.com>
3309 * dwarf2out.c (auto_die): New static.
3310 (gen_type_die_with_usage): Handle C++1y 'auto'.
3311 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
3314 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
3317 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
3318 (SPEC_X32): Likewise.
3319 (SPEC_64): Likewise.
3320 * config/i386/i386.c (ix86_option_override_internal): Turn off
3321 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
3323 (x86_file_start): Output .code16gcc for TARGET_16BIT.
3324 * config/i386/i386.h (TARGET_16BIT): New macro.
3325 (TARGET_16BIT_P): Likewise.
3326 * config/i386/i386.opt: Add m16.
3327 * doc/invoke.texi: Document -m16.
3329 2014-01-28 Jakub Jelinek <jakub@redhat.com>
3331 PR preprocessor/59935
3332 * input.c (location_get_source_line): Bail out on when line number
3333 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
3335 2014-01-28 Richard Biener <rguenther@suse.de>
3337 PR tree-optimization/58742
3338 * tree-ssa-forwprop.c (associate_plusminus): Handle
3339 pointer subtraction of the form (T)(P + A) - (T)P.
3341 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3343 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
3346 2014-01-28 Richard Biener <rguenther@suse.de>
3349 2014-01-28 Richard Biener <rguenther@suse.de>
3351 PR rtl-optimization/45364
3352 PR rtl-optimization/59890
3353 * var-tracking.c (local_get_addr_clear_given_value): Handle
3354 already cleared slot.
3355 (val_reset): Handle not allocated local_get_addr_cache.
3356 (vt_find_locations): Use post-order on the inverted CFG.
3358 2014-01-28 Richard Biener <rguenther@suse.de>
3360 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
3362 2014-01-28 Richard Biener <rguenther@suse.de>
3364 PR rtl-optimization/45364
3365 PR rtl-optimization/59890
3366 * var-tracking.c (local_get_addr_clear_given_value): Handle
3367 already cleared slot.
3368 (val_reset): Handle not allocated local_get_addr_cache.
3369 (vt_find_locations): Use post-order on the inverted CFG.
3371 2014-01-28 Alan Modra <amodra@gmail.com>
3373 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
3374 * configure.ac <recursive call for build != host>: Define
3375 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
3376 and LD_FOR_BUILD too.
3377 * configure: Regenerate.
3379 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
3381 * config/i386/i386.c (get_builtin_code_for_version): Separate
3382 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
3383 Broadwell from Haswell.
3385 2014-01-27 Steve Ellcey <sellcey@mips.com>
3387 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
3388 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
3389 * config/mips/mips.c (mips_option_override): Change setting
3391 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
3392 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
3393 Change from Mask to Var.
3395 2014-01-27 Jeff Law <law@redhat.com>
3397 * ipa-inline.c (inline_small_functions): Fix typo.
3399 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
3401 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
3402 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
3403 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
3404 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
3405 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
3406 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
3407 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
3408 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
3409 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
3410 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
3411 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
3412 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
3413 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
3414 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
3415 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
3416 (_mm512_storeu_epi64): Ditto.
3417 (_mm512_cmpge_epi32_mask): Ditto.
3418 (_mm512_cmpge_epu32_mask): Ditto.
3419 (_mm512_cmpge_epi64_mask): Ditto.
3420 (_mm512_cmpge_epu64_mask): Ditto.
3421 (_mm512_cmple_epi32_mask): Ditto.
3422 (_mm512_cmple_epu32_mask): Ditto.
3423 (_mm512_cmple_epi64_mask): Ditto.
3424 (_mm512_cmple_epu64_mask): Ditto.
3425 (_mm512_cmplt_epi32_mask): Ditto.
3426 (_mm512_cmplt_epu32_mask): Ditto.
3427 (_mm512_cmplt_epi64_mask): Ditto.
3428 (_mm512_cmplt_epu64_mask): Ditto.
3429 (_mm512_cmpneq_epi32_mask): Ditto.
3430 (_mm512_cmpneq_epu32_mask): Ditto.
3431 (_mm512_cmpneq_epi64_mask): Ditto.
3432 (_mm512_cmpneq_epu64_mask): Ditto.
3433 (_mm512_expand_pd): Ditto.
3434 (_mm512_expand_ps): Ditto.
3435 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
3436 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
3437 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
3438 * config/i386/i386.c (ix86_builtins): Add
3439 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
3440 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
3441 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
3442 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
3443 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
3444 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
3445 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
3446 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
3447 IX86_BUILTIN_PMOVUSQW512_MEM.
3448 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
3449 __builtin_ia32_pmovsqd512mem_mask,
3450 __builtin_ia32_pmovqd512mem_mask,
3451 __builtin_ia32_pmovusqw512mem_mask,
3452 __builtin_ia32_pmovsqw512mem_mask,
3453 __builtin_ia32_pmovqw512mem_mask,
3454 __builtin_ia32_pmovusdw512mem_mask,
3455 __builtin_ia32_pmovsdw512mem_mask,
3456 __builtin_ia32_pmovdw512mem_mask,
3457 __builtin_ia32_pmovqb512mem_mask,
3458 __builtin_ia32_pmovusqb512mem_mask,
3459 __builtin_ia32_pmovsqb512mem_mask,
3460 __builtin_ia32_pmovusdb512mem_mask,
3461 __builtin_ia32_pmovsdb512mem_mask,
3462 __builtin_ia32_pmovdb512mem_mask.
3463 (bdesc_args): Add __builtin_ia32_expanddf512,
3464 __builtin_ia32_expandsf512.
3465 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
3466 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
3467 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
3468 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
3469 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
3470 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
3471 (avx512f_<code>v8div16qi2_mask_store): This.
3472 (avx512f_expand<mode>): New.
3474 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
3476 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
3478 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3479 (_mm512_prefetch_i32scatter_pd): Ditto.
3480 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3481 (_mm512_prefetch_i64scatter_pd): Ditto.
3482 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3483 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
3484 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3485 (_mm512_prefetch_i32scatter_ps): Ditto.
3486 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3487 (_mm512_prefetch_i64scatter_ps): Ditto.
3488 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3489 * config/i386/i386-builtin-types.def: Define
3490 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
3491 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
3492 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
3493 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
3494 IX86_BUILTIN_SCATTERPFQPD.
3495 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
3496 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
3497 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
3498 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
3499 __builtin_ia32_scatterpfqps.
3500 (ix86_expand_builtin): Expand new built-ins.
3501 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
3502 fix memory access data type.
3503 (*avx512pf_gatherpf<mode>_mask): Ditto.
3504 (*avx512pf_gatherpf<mode>): Ditto.
3505 (avx512pf_scatterpf<mode>): Ditto.
3506 (*avx512pf_scatterpf<mode>_mask): Ditto.
3507 (*avx512pf_scatterpf<mode>): Ditto.
3508 (GATHER_SCATTER_SF_MEM_MODE): New.
3509 (avx512pf_gatherpf<mode>df): Ditto.
3510 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3511 (*avx512pf_scatterpf<mode>df): Ditto.
3513 2014-01-27 Jakub Jelinek <jakub@redhat.com>
3516 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
3517 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
3520 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3522 * common/config/arm/arm-common.c
3523 (arm_rewrite_mcpu): Handle multiple names.
3525 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3527 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
3529 * gimple-builder.h (create_gimple_tmp): Delete.
3531 2014-01-27 Christian Bruel <christian.bruel@st.com>
3533 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
3536 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
3538 * config/pa/pa.md (call): Generate indirect long calls to non-local
3539 functions when outputing 32-bit code.
3540 (call_value): Likewise except for special call to buggy powf function.
3542 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
3543 portable runtime and PIC indirect calls.
3544 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
3545 and PIC call sequences. Use ldo instead of blr to set return register
3546 in PIC call sequence.
3548 2014-01-25 Walter Lee <walt@tilera.com>
3550 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
3551 avoid clobbering a live register.
3553 2014-01-25 Walter Lee <walt@tilera.com>
3555 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
3556 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
3557 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
3558 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
3560 2014-01-25 Walter Lee <walt@tilera.com>
3562 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
3563 arguments on even registers.
3564 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
3566 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
3567 (BIGGEST_ALIGNMENT): Ditto.
3568 (BIGGEST_FIELD_ALIGNMENT): Ditto.
3570 2014-01-25 Walter Lee <walt@tilera.com>
3572 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
3573 insns before bundling.
3574 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
3576 2014-01-25 Walter Lee <walt@tilera.com>
3578 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
3579 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
3580 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
3582 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3584 * config/mips/constraints.md (kl): Delete.
3585 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
3586 define expands, using...
3587 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
3588 instructions for MIPS16.
3589 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
3590 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
3592 2014-01-25 Walter Lee <walt@tilera.com>
3594 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
3598 2014-01-25 Walter Lee <walt@tilera.com>
3600 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
3601 (TARGET_EXPAND_TO_RTL_HOOK): Define.
3603 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
3605 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
3608 2014-01-25 Jakub Jelinek <jakub@redhat.com>
3610 * print-rtl.c (in_call_function_usage): New var.
3611 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
3612 EXPR_LIST mode as mode and not as reg note name.
3615 * cfgloopmanip.c (copy_loop_info): If
3616 loop->warned_aggressive_loop_optimizations, make sure
3617 the flag is set in target loop too.
3619 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
3621 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
3623 * builtins.def: Likewise.
3624 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
3625 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
3626 * ira.c (ira_setup_eliminable_regset): Likewise.
3627 * omp-low.c (gate_expand_omp): Likewise.
3628 (execute_lower_omp): Likewise.
3629 (diagnose_sb_0): Likewise.
3630 (gate_diagnose_omp_blocks): Likewise.
3631 (simd_clone_clauses_extract): Likewise.
3634 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3636 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
3637 correction for little endian...
3638 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
3641 2014-01-24 Jeff Law <law@redhat.com>
3643 PR tree-optimization/59919
3644 * tree-vrp.c (find_assert_locations_1): Do not register asserts
3645 for non-returning calls.
3647 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
3649 * common/config/aarch64/aarch64-common.c
3650 (aarch64_rewrite_mcpu): Handle multiple names.
3651 * config/aarch64/aarch64.h
3652 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
3654 2014-01-24 Dodji Seketeli <dodji@redhat.com>
3656 * input.c (add_file_to_cache_tab): Handle the case where fopen
3659 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
3662 * config/i386/i386.md (pushsf splitter): Get stack adjustment
3663 from push operand if code of push isn't PRE_DEC.
3665 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
3668 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
3669 -mquad-memory-atomic. Update -mquad-memory documentation to say
3670 it is only used for non-atomic loads/stores.
3672 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
3673 -mquad-memory or -mquad-memory-atomic switches.
3675 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
3676 -mquad-memory-atomic to ISA 2.07 support.
3678 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
3679 to separate support of normal quad word memory operations (ldq, stq)
3680 from the atomic quad word memory operations.
3682 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3683 support to separate non-atomic quad word operations from atomic
3684 quad word operations. Disable non-atomic quad word operations in
3685 little endian mode so that we don't have to swap words after the
3686 load and before the store.
3687 (quad_load_store_p): Add comment about atomic quad word support.
3688 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
3689 options printed with -mdebug=reg.
3691 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
3692 -mquad-memory-atomic as the test for whether we have quad word
3693 atomic instructions.
3694 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
3695 or -mp8-vector are used, allow byte/half-word atomic operations.
3697 * config/rs6000/sync.md (load_lockedti): Insure that the address
3698 is a proper indexed or indirect address for the lqarx instruction.
3699 On little endian systems, swap the hi/lo registers after the lqarx
3701 (load_lockedpti): Use indexed_or_indirect_operand predicate to
3702 insure the address is valid for the lqarx instruction.
3703 (store_conditionalti): Insure that the address is a proper indexed
3704 or indirect address for the stqcrx. instruction. On little endian
3705 systems, swap the hi/lo registers before doing the stqcrx.
3707 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
3708 insure the address is valid for the stqcrx. instruction.
3710 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
3711 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
3712 type of quad memory support is available.
3714 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
3717 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
3718 there is a danger of looping.
3720 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
3722 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
3723 force flag_ira_loop_pressure if set via command line.
3725 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3727 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
3728 (ashr_simd): New builtin handling DI mode.
3729 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
3730 (aarch64_sshr_simddi): New match pattern.
3731 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
3732 (vshrd_n_s64): Likewise.
3733 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
3735 2014-01-23 Nick Clifton <nickc@redhat.com>
3737 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
3738 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
3739 favour of mcu specific scripts.
3740 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
3743 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3744 Alex Velenko <Alex.Velenko@arm.com>
3746 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
3747 (vaddv_s16): Likewise.
3748 (vaddv_s32): Likewise.
3749 (vaddv_u8): Likewise.
3750 (vaddv_u16): Likewise.
3751 (vaddv_u32): Likewise.
3752 (vaddvq_s8): Likewise.
3753 (vaddvq_s16): Likewise.
3754 (vaddvq_s32): Likewise.
3755 (vaddvq_s64): Likewise.
3756 (vaddvq_u8): Likewise.
3757 (vaddvq_u16): Likewise.
3758 (vaddvq_u32): Likewise.
3759 (vaddvq_u64): Likewise.
3760 (vaddv_f32): Likewise.
3761 (vaddvq_f32): Likewise.
3762 (vaddvq_f64): Likewise.
3763 (vmaxv_f32): Likewise.
3764 (vmaxv_s8): Likewise.
3765 (vmaxv_s16): Likewise.
3766 (vmaxv_s32): Likewise.
3767 (vmaxv_u8): Likewise.
3768 (vmaxv_u16): Likewise.
3769 (vmaxv_u32): Likewise.
3770 (vmaxvq_f32): Likewise.
3771 (vmaxvq_f64): Likewise.
3772 (vmaxvq_s8): Likewise.
3773 (vmaxvq_s16): Likewise.
3774 (vmaxvq_s32): Likewise.
3775 (vmaxvq_u8): Likewise.
3776 (vmaxvq_u16): Likewise.
3777 (vmaxvq_u32): Likewise.
3778 (vmaxnmv_f32): Likewise.
3779 (vmaxnmvq_f32): Likewise.
3780 (vmaxnmvq_f64): Likewise.
3781 (vminv_f32): Likewise.
3782 (vminv_s8): Likewise.
3783 (vminv_s16): Likewise.
3784 (vminv_s32): Likewise.
3785 (vminv_u8): Likewise.
3786 (vminv_u16): Likewise.
3787 (vminv_u32): Likewise.
3788 (vminvq_f32): Likewise.
3789 (vminvq_f64): Likewise.
3790 (vminvq_s8): Likewise.
3791 (vminvq_s16): Likewise.
3792 (vminvq_s32): Likewise.
3793 (vminvq_u8): Likewise.
3794 (vminvq_u16): Likewise.
3795 (vminvq_u32): Likewise.
3796 (vminnmv_f32): Likewise.
3797 (vminnmvq_f32): Likewise.
3798 (vminnmvq_f64): Likewise.
3800 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
3802 * config/aarch64/aarch64-simd.md
3803 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
3804 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
3805 (*aarch64_mul3_elt<mode>): Likewise.
3806 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
3807 (*aarch64_mul3_elt_to_64v2df): Likewise.
3808 (*aarch64_mla_elt<mode>): Likewise.
3809 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3810 (*aarch64_mls_elt<mode>): Likewise.
3811 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3812 (*aarch64_fma4_elt<mode>): Likewise.
3813 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3814 (*aarch64_fma4_elt_to_64v2df): Likewise.
3815 (*aarch64_fnma4_elt<mode>): Likewise.
3816 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
3817 (*aarch64_fnma4_elt_to_64v2df): Likewise.
3818 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3819 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3820 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3821 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3822 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3823 (aarch64_sqdmull_lane<mode>_internal): Likewise.
3824 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3826 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
3828 * config/aarch64/aarch64-simd.md
3829 (aarch64_be_checked_get_lane<mode>): New define_expand.
3830 * config/aarch64/aarch64-simd-builtins.def
3831 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
3832 New builtin definition.
3833 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
3834 Use new safe be builtin.
3836 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
3838 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
3840 (aarch64_be_st1<mode>): Likewise.
3841 (aarch_ld1<VALL:mode>): Define_expand modified.
3842 (aarch_st1<VALL:mode>): Likewise.
3843 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
3844 (UNSPEC_ST1): Likewise.
3846 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
3848 * config/microblaze/microblaze.md: Add trap insn and attribute
3850 2014-01-23 Dodji Seketeli <dodji@redhat.com>
3852 PR preprocessor/58580
3853 * input.h (location_get_source_line): Take an additional line_size
3855 (void diagnostics_file_cache_fini): Declare new function.
3856 * input.c (struct fcache): New type.
3857 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
3858 New static constants.
3859 (diagnostic_file_cache_init, total_lines_num)
3860 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
3861 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
3862 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
3863 (get_next_line, read_next_line, goto_next_line, read_line_num):
3864 New static function definitions.
3865 (diagnostic_file_cache_fini): New function.
3866 (location_get_source_line): Take an additional output line_len
3867 parameter. Re-write using lookup_or_add_file_to_cache_tab and
3869 * diagnostic.c (diagnostic_finish): Call
3870 diagnostic_file_cache_fini.
3871 (adjust_line): Take an additional input parameter for the length
3872 of the line, rather than calculating it with strlen.
3873 (diagnostic_show_locus): Adjust the use of
3874 location_get_source_line and adjust_line with respect to their new
3875 signature. While displaying a line now, do not stop at the first
3876 null byte. Rather, display the zero byte as a space and keep
3877 going until we reach the size of the line.
3878 * Makefile.in: Add vec.o to OBJS-libcommon
3880 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3881 Ilya Tocar <ilya.tocar@intel.com>
3883 * config/i386/avx512fintrin.h (_mm512_kmov): New.
3884 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
3885 (__builtin_ia32_kmov16): Ditto.
3886 * config/i386/i386.md (UNSPEC_KMOV): New.
3889 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
3891 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
3892 (_mm512_storeu_si512): Ditto.
3894 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
3897 * rtl.h (get_referenced_operands): Declare.
3898 * recog.c (get_referenced_operands): New function.
3899 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
3900 operands have been referenced when recording LO_SUM references.
3902 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
3904 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
3906 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3908 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
3909 Enable for generic and recent AMD targets.
3911 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
3913 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
3914 ARG_SIZE note when adjustment was eliminated.
3916 2014-01-22 Jeff Law <law@redhat.com>
3918 PR tree-optimization/59597
3919 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
3920 in file. Accept new argument REGISTERING and use it to modify
3921 dump output appropriately.
3922 (register_jump_thread): Corresponding changes.
3923 (mark_threaded_blocks): Reinstate code to cancel unprofitable
3924 thread paths involving joiner blocks. Add code to dump cancelled
3925 jump threading paths.
3927 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
3929 PR rtl-optimization/59477
3930 * lra-constraints.c (inherit_in_ebb): Process call for living hard
3931 regs. Update reloads_num and potential_reload_hard_regs for all insns.
3933 2014-01-22 Tom Tromey <tromey@redhat.com>
3935 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
3937 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
3939 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3941 PR rtl-optimization/59896
3942 * lra-constraints.c (process_alt_operands): Check unused note for
3943 matched operands of insn with no output reloads.
3945 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
3947 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
3948 (mips_move_from_gpr_cost): Likewise.
3950 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
3952 PR rtl-optimization/59858
3953 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
3954 ira_class_hard_regs_num.
3955 (process_alt_operands): Increase reject for dying matched operand.
3957 2014-01-21 Jakub Jelinek <jakub@redhat.com>
3960 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
3961 smaller than size, perform several stores or loads and stores
3962 at dst + count - size to store or copy all of size bytes, rather
3963 than just last modesize bytes.
3965 2014-01-20 DJ Delorie <dj@redhat.com>
3967 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
3968 that CLOBBERs are REGs before propogating their values.
3970 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
3973 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
3974 (cgraph_inline_failed_type): New function.
3975 * cgraph.h (DEFCIFCODE): Add type.
3976 (cgraph_inline_failed_type_t): New enum.
3977 (cgraph_inline_failed_type): New prototype.
3978 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
3979 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
3980 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
3981 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
3982 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
3983 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
3984 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
3985 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
3986 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
3987 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
3988 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
3989 OPTIMIZATION_MISMATCH.
3990 * tree-inline.c (expand_call_inline): Emit errors during
3991 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
3993 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
3996 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
3997 mode attribute in insn output.
3999 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4001 * output.h (output_constant): Delete.
4002 * varasm.c (output_constant): Make private.
4004 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4006 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4008 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4011 * tree.h (fold_builtin_strcat): New prototype.
4012 * builtins.c (fold_builtin_strcat): No longer static. Add len
4013 argument, if non-NULL, don't call c_strlen. Optimize
4014 directly into __builtin_memcpy instead of __builtin_strcpy.
4015 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4016 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4018 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4020 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4021 for SImode_address_operand operands, having only a REG argument.
4023 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4025 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4026 loader name using mbig-endian.
4027 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4029 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4031 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4035 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4037 * config/aarch64/aarch64-protos.h
4038 (aarch64_cannot_change_mode_class_ptr): Declare.
4039 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4040 aarch64_cannot_change_mode_class_ptr): New.
4041 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4042 backend hook aarch64_cannot_change_mode_class.
4044 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4046 * common/config/aarch64/aarch64-common.c
4047 (aarch64_handle_option): Don't handle any option order logic here.
4048 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4049 selected_cpu, warn on architecture version mismatch.
4050 (aarch64_override_options): Fix parsing order for option strings.
4052 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4053 Iain Sandoe <iain@codesourcery.com>
4056 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4057 warning. Amend comment to reflect current functionality.
4059 2014-01-20 Richard Biener <rguenther@suse.de>
4062 * builtins.c (fold_builtin_strcat): Remove case better handled
4063 by tree-ssa-strlen.c.
4065 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4067 * config/aarch64/aarch64.opt
4068 (mcpu, march, mtune): Make case-insensitive.
4070 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4073 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4074 if operands[1] is a REG or ZERO_EXTEND of a REG.
4076 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4078 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4080 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4082 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4083 long non-pic millicode calls.
4085 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4087 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4089 2014-01-19 Kito Cheng <kito@0xlab.org>
4091 * builtins.c (expand_movstr): Check movstr expand done or fail.
4093 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4094 H.J. Lu <hongjiu.lu@intel.com>
4097 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4098 to DImode for zero-extended addresses.
4100 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4102 PR rtl-optimization/57763
4103 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4104 on the new indirect jump_insn and increment LABEL_NUSES (label).
4106 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4110 * config.gcc (x86_archs): New variable.
4111 (x86_64_archs): Likewise.
4112 (x86_cpus): Likewise.
4113 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4114 --with-arch/--with-cpu= options.
4115 Support --with-arch=/--with-cpu={nehalem,westmere,
4116 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4118 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4120 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4121 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4123 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4125 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4127 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4130 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4131 clear cpp_get_options (parse_in)->warn_unused_macros for
4132 ix86_target_macros_internal with cpp_define.
4134 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4136 * jump.c (delete_related_insns): Keep (use (insn))s.
4137 * reorg.c (redundant_insn): Check for barriers too.
4139 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4141 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4143 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4145 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4146 call to $$dyncall when TARGET_LONG_CALLS is true.
4148 2014-01-17 Jeff Law <law@redhat.com>
4150 * ree.c (combine_set_extension): Temporarily disable test for
4151 changing number of hard registers.
4153 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4156 * ipa-inline-analysis.c (inline_free_summary):
4157 Do not free summary of aliases.
4159 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4162 * gimplify.c (gimplify_expr): Use create_tmp_var
4163 instead of create_tmp_var_raw. If cond doesn't have
4164 integral type, don't add the IFN_ANNOTATE builtin at all.
4166 2014-01-17 Martin Jambor <mjambor@suse.cz>
4169 * ipa-cp.c (prev_edge_clone): New variable.
4170 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4171 Also resize prev_edge_clone vector.
4172 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4173 (ipcp_edge_removal_hook): New function.
4174 (ipcp_driver): Register ipcp_edge_removal_hook.
4176 2014-01-17 Andrew Pinski <apinski@cavium.com>
4177 Steve Ellcey <sellcey@mips.com>
4180 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4183 2014-01-17 Jeff Law <law@redhat.com>
4186 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4187 so that pass_ccp runs first.
4189 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4191 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4192 (ix86_adjust_cost): Use !TARGET_XXX.
4193 (do_reorder_for_imul): Likewise.
4194 (swap_top_of_ready_list): Likewise.
4195 (ix86_sched_reorder): Likewise.
4197 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4199 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4200 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4201 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4202 (intel_memset): New. Duplicate slm_memset.
4203 (intel_cost): New. Duplicate slm_cost.
4204 (m_INTEL): New macro.
4205 (processor_target_table): Add "intel".
4206 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4207 with PROCESSOR_INTEL for "intel".
4208 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4209 PROCESSOR_SILVERMONT.
4210 (ix86_issue_rate): Likewise.
4211 (ix86_adjust_cost): Likewise.
4212 (ia32_multipass_dfa_lookahead): Likewise.
4213 (swap_top_of_ready_list): Likewise.
4214 (ix86_sched_reorder): Likewise.
4215 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4216 instead of TARGET_OPT_AGU.
4217 * config/i386/i386.h (TARGET_INTEL): New.
4218 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4219 (processor_type): Add PROCESSOR_INTEL.
4220 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4221 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4223 2014-01-17 Marek Polacek <polacek@redhat.com>
4226 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4229 2014-01-17 Richard Biener <rguenther@suse.de>
4231 PR tree-optimization/46590
4232 * opts.c (default_options_table): Add entries for
4233 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4234 all enabled at -O1 but not for -Og.
4235 * common.opt (fbranch-count-reg): Remove Init(1).
4236 (fmove-loop-invariants): Likewise.
4237 (ftree-pta): Likewise.
4239 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4241 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4242 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4243 decls to at least the GCC 4.8 used alignments.
4246 * tree-nested.c (convert_nonlocal_reference_stmt,
4247 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4248 of GIMPLE_BIND stmts, adjust associated decls.
4250 2014-01-17 Richard Biener <rguenther@suse.de>
4252 PR tree-optimization/46590
4253 * vec.h (vec<>::bseach): New member function implementing
4254 binary search according to C89 bsearch.
4255 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4256 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4257 bitmap pointer again. Make accesses_in_loop a flat array.
4258 (mem_ref_obstack): New global.
4259 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4260 (mark_ref_stored): Likewise.
4261 (ref_indep_loop_p_2): Likewise.
4262 (set_ref_stored_in_loop): New helper function.
4263 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4264 (memref_free): Adjust.
4265 (record_mem_ref_loc): Simplify.
4266 (gather_mem_refs_stmt): Adjust.
4267 (sort_locs_in_loop_postorder_cmp): New function.
4268 (analyze_memory_references): Sort accesses_in_loop after
4269 loop postorder number.
4270 (find_ref_loc_in_loop_cmp): New function.
4271 (for_all_locs_in_loop): Find relevant cluster of locs in
4272 accesses_in_loop and iterate without recursion.
4273 (execute_sm): Avoid uninit warning.
4274 (struct ref_always_accessed): Simplify.
4275 (ref_always_accessed::operator ()): Likewise.
4276 (ref_always_accessed_p): Likewise.
4277 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4278 loop postorder numbers here.
4279 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4282 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4285 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4286 on decls for which assemble_alias has been called.
4288 2014-01-17 Nick Clifton <nickc@redhat.com>
4290 * config/msp430/msp430.opt: (mcpu): New option.
4291 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
4292 (msp430_option_override): Parse target_cpu. If the MCU name
4293 matches a generic string, clear target_mcu.
4294 (msp430_attr): Allow numeric interrupt values up to 63.
4295 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
4296 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
4298 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
4300 * config/msp430/msp430.md (popm): Use %J rather than %I.
4301 (addsi3): Use msp430_nonimmediate_operand for operand 2.
4302 (addhi_cy_i): Use immediate_operand for operand 2.
4303 * doc/invoke.texi: Document -mcpu option.
4305 2014-01-17 Richard Biener <rguenther@suse.de>
4307 PR rtl-optimization/38518
4308 * df.h (df_analyze_loop): Declare.
4309 * df-core.c: Include cfgloop.h.
4310 (df_analyze_1): Split out main part of df_analyze.
4311 (df_analyze): Adjust.
4312 (loop_inverted_post_order_compute): New function.
4313 (loop_post_order_compute): Likewise.
4314 (df_analyze_loop): New function avoiding whole-function
4316 * loop-invariant.c (find_defs): Use df_analyze_loop.
4317 (find_invariants): Adjust.
4318 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
4320 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
4322 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
4323 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
4325 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
4327 * ipa-ref.c (ipa_remove_stmt_references): Fix references
4328 traversal when removing references.
4330 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
4333 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
4335 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
4338 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
4339 pushing a reload for an autoinc when we had previously reloaded an
4340 inner part of the address.
4342 2014-01-16 Jakub Jelinek <jakub@redhat.com>
4344 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
4346 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
4347 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
4348 when not giving up or versioning for alias only because of
4350 (vect_analyze_data_ref_dependences): Set to true.
4351 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
4353 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
4354 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
4358 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
4361 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
4362 operand 0 predicate for gathers, use a new pseudo as subtarget.
4364 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4367 * lra-constraints.c (process_alt_operands): Add printing debug info.
4368 Check absence of input/output reloads for matched operands too.
4370 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
4372 PR rtl-optimization/59835
4373 * ira.c (ira_init_register_move_cost): Increase cost for
4376 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
4378 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
4380 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
4383 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
4384 non-register objects. Use gen_(high/low)part more consistently.
4387 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
4390 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
4391 endian support, remove tests for WORDS_BIG_ENDIAN.
4392 (p8_mfvsrd_3_<mode>): Likewise.
4393 (reload_gpr_from_vsx<mode>): Likewise.
4394 (reload_gpr_from_vsxsf): Likewise.
4395 (p8_mfvsrd_4_disf): Likewise.
4397 2014-01-16 Richard Biener <rguenther@suse.de>
4399 PR rtl-optimization/46590
4400 * lcm.c (compute_antinout_edge): Use postorder iteration.
4401 (compute_laterin): Use inverted postorder iteration.
4403 2014-01-16 Nick Clifton <nickc@redhat.com>
4406 * varasm.c (output_constant): Return the number of bytes actually
4408 (output_constructor_array_range): Update the field size with the
4409 number of bytes emitted by output_constant.
4410 (output_constructor_regular_field): Likewise. Also do not
4411 complain if the total number of bytes emitted is now greater
4412 than the expected fieldpos.
4413 * output.h (output_constant): Update prototype and descriptive comment.
4415 2014-01-16 Marek Polacek <polacek@redhat.com>
4418 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
4419 it is error_mark_node.
4421 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
4423 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
4424 VALID_AVX256_REG_OR_OI_MODE.
4426 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
4428 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
4429 current procedure should be profiled.
4431 2014-01-15 Andrew Pinski <apinski@cavium.com>
4433 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
4434 of moving from/to the STACK_REG register class.
4436 2014-01-15 Richard Henderson <rth@redhat.com>
4439 * reginfo.c (global_regs_decl): Globalize.
4440 * rtl.h (global_regs_decl): Declare.
4441 * ira.c (do_reload): Diagnose frame_pointer_needed and it
4442 reserved via global_regs.
4444 2014-01-15 Teresa Johnson <tejohnson@google.com>
4446 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
4448 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
4450 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
4451 and vmulosh rather than call gen_vec_widen_smult_*.
4452 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
4453 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
4454 (vec_widen_smult_even_v16qi): Likewise.
4455 (vec_widen_umult_even_v8hi): Likewise.
4456 (vec_widen_smult_even_v8hi): Likewise.
4457 (vec_widen_umult_odd_v16qi): Likewise.
4458 (vec_widen_smult_odd_v16qi): Likewise.
4459 (vec_widen_umult_odd_v8hi): Likewise.
4460 (vec_widen_smult_odd_v8hi): Likewise.
4461 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
4462 vmuloub rather than call gen_vec_widen_umult_*.
4463 (vec_widen_umult_lo_v16qi): Likewise.
4464 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
4465 vmulosb rather than call gen_vec_widen_smult_*.
4466 (vec_widen_smult_lo_v16qi): Likewise.
4467 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
4468 rather than call gen_vec_widen_umult_*.
4469 (vec_widen_umult_lo_v8hi): Likewise.
4470 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
4471 rather than call gen_vec_widen_smult_*.
4472 (vec_widen_smult_lo_v8hi): Likewise.
4474 2014-01-15 Jeff Law <law@redhat.com>
4476 PR tree-optimization/59747
4477 * ree.c (find_and_remove_re): Properly handle case where a second
4478 eliminated extension requires widening a copy created for elimination
4479 of a prior extension.
4480 (combine_set_extension): Ensure that the number of hard regs needed
4481 for a destination register does not change when we widen it.
4483 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4485 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
4486 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
4487 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
4488 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
4489 (avr-*-rtems*): Likewise.
4490 (bfin*-rtems*): Likewise.
4491 (moxie-*-rtems*): Likewise.
4492 (h8300-*-rtems*): Likewise.
4493 (i[34567]86-*-rtems*): Likewise.
4494 (lm32-*-rtems*): Likewise.
4495 (m32r-*-rtems*): Likewise.
4496 (m68k-*-rtems*): Likewise.
4497 (microblaze*-*-rtems*): Likewise.
4498 (mips*-*-rtems*): Likewise.
4499 (powerpc-*-rtems*): Likewise.
4500 (sh-*-rtems*): Likewise.
4501 (sparc-*-rtems*): Likewise.
4502 (sparc64-*-rtems*): Likewise.
4503 (v850-*-rtems*): Likewise.
4504 (m32c-*-rtems*): Likewise.
4506 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
4508 PR rtl-optimization/59511
4509 * ira.c (ira_init_register_move_cost): Use memory costs for some
4510 cases of register move cost calculations.
4511 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
4512 instead of BB frequency.
4513 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
4514 * lra-assigns.c (find_hard_regno_for): Ditto.
4516 2014-01-15 Richard Biener <rguenther@suse.de>
4518 PR tree-optimization/59822
4519 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
4520 (vectorizable_load): Use it to hoist defs of uses of invariant
4521 loads out of the loop.
4523 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
4524 Kugan Vivekanandarajah <kuganv@linaro.org>
4527 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
4530 2014-01-15 Richard Biener <rguenther@suse.de>
4532 PR rtl-optimization/59802
4533 * lcm.c (compute_available): Use inverted postorder to seed
4534 the initial worklist.
4536 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4539 * config/s390/s390.c (s390_preferred_reload_class): Don't return
4540 ADDR_REGS for invalid symrefs in non-PIC code.
4542 2014-01-15 Jakub Jelinek <jakub@redhat.com>
4545 * builtins.c (determine_block_size): Initialize *probable_max_size
4546 even if len_rtx is CONST_INT.
4548 2014-01-14 Andrew Pinski <apinski@cavium.com>
4550 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
4551 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
4552 (cortexa53_tunings): Likewise.
4553 (aarch64_sched_issue_rate): New function.
4554 (TARGET_SCHED_ISSUE_RATE): Define.
4556 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4558 * ira-costs.c (find_costs_and_classes): Add missed
4559 ira_init_register_move_cost_if_necessary.
4561 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
4564 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
4566 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
4569 * config/i386/i386.c (type_natural_mode): Add a bool parameter
4570 to indicate if type is used for function return value. Warn ABI
4571 change if the vector mode isn't available for function return value.
4572 (ix86_function_arg_advance): Pass false to type_natural_mode.
4573 (ix86_function_arg): Likewise.
4574 (ix86_gimplify_va_arg): Likewise.
4575 (function_arg_32): Don't warn ABI change.
4576 (ix86_function_value): Pass true to type_natural_mode.
4577 (ix86_return_in_memory): Likewise.
4578 (ix86_struct_value_rtx): Removed.
4579 (TARGET_STRUCT_VALUE_RTX): Likewise.
4581 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
4583 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
4584 converting a conditional jump into a conditional return.
4586 2014-01-14 Richard Biener <rguenther@suse.de>
4588 PR tree-optimization/58921
4589 PR tree-optimization/59006
4590 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
4591 hoisting invariant stmts.
4592 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
4593 invariant loads on the preheader edge if possible.
4595 2014-01-14 Joey Ye <joey.ye@arm.com>
4597 * doc/plugin.texi (Building GCC plugins): Update to C++.
4599 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
4601 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
4602 (_mm_rcp28_round_ss): Ditto.
4603 (_mm_rsqrt28_round_sd): Ditto.
4604 (_mm_rsqrt28_round_ss): Ditto.
4605 (_mm_rcp28_sd): Ditto.
4606 (_mm_rcp28_ss): Ditto.
4607 (_mm_rsqrt28_sd): Ditto.
4608 (_mm_rsqrt28_ss): Ditto.
4609 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
4610 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
4611 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
4612 (IX86_BUILTIN_RCP28SD): Ditto.
4613 (IX86_BUILTIN_RCP28SS): Ditto.
4614 (IX86_BUILTIN_RSQRT28SD): Ditto.
4615 (IX86_BUILTIN_RSQRT28SS): Ditto.
4616 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
4617 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
4618 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
4619 (ix86_expand_special_args_builtin): Expand new FTYPE.
4620 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
4621 (srcp14<mode>): Make insn unary.
4622 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
4623 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
4624 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
4625 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
4626 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
4627 Fix rounding: make it SAE only.
4628 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
4630 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
4632 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
4633 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
4634 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
4635 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
4636 (round_saeonly_mask_scalar_operand4): Ditto.
4637 (round_saeonly_mask_scalar_op3): Ditto.
4638 (round_saeonly_mask_scalar_op4): Ditto.
4640 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4642 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4643 Implement -maltivec=be for vec_insert and vec_extract.
4645 2014-01-10 DJ Delorie <dj@redhat.com>
4647 * config/msp430/msp430.md (call_internal): Don't allow memory
4648 references with SP as the base register.
4649 (call_value_internal): Likewise.
4650 * config/msp430/constraints.md (Yc): New. For memory references
4651 that don't use SP as a base register.
4653 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
4654 "an integer without a # prefix"
4655 * config/msp430/msp430.md (epilogue_helper): Use it.
4657 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4660 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
4661 AVX512F gather builtins.
4662 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
4663 on gather decls with INTEGER_TYPE masktype.
4664 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
4665 directly into the builtin rather than hoisting it before loop.
4667 PR tree-optimization/59387
4668 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
4669 (scev_const_prop): If folded_casts and type has undefined overflow,
4670 use force_gimple_operand instead of force_gimple_operand_gsi and
4671 for each added stmt if it is assign with
4672 arith_code_with_undefined_signed_overflow, call
4673 rewrite_to_defined_overflow.
4674 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
4675 gimple-fold.h instead.
4676 (arith_code_with_undefined_signed_overflow,
4677 rewrite_to_defined_overflow): Moved to ...
4678 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
4679 rewrite_to_defined_overflow): ... here. No longer static.
4680 Include gimplify-me.h.
4681 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
4682 rewrite_to_defined_overflow): New prototypes.
4684 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4686 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
4688 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
4690 * builtins.c (get_object_alignment_2): Minor tweak.
4691 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
4693 2014-01-13 Christian Bruel <christian.bruel@st.com>
4695 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
4696 optimized non constant lengths.
4698 2014-01-13 Jakub Jelinek <jakub@redhat.com>
4701 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
4702 load as __atomic_load_N if possible.
4704 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4706 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
4708 (rs6000_expand_builtin): Adjust call.
4710 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
4713 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
4714 * config/rs6000/rs6000.c: Include target-globals.h.
4715 (rs6000_set_current_function): Instead of doing target_reinit
4716 unconditionally, use save_target_globals_default_opts and
4717 restore_target_globals.
4719 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
4721 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
4722 (rs6000_expand_builtin): Handle mffs and mtfsf.
4723 (rs6000_init_builtins): Define mffs and mtfsf.
4724 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
4725 (rs6000_mffs): New pattern.
4726 (rs6000_mtfsf): New pattern.
4728 2014-01-11 Bin Cheng <bin.cheng@arm.com>
4730 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
4731 Start narrowing with START. Apply candidate-use pair
4732 and check overall cost in narrowing.
4733 (iv_ca_prune): Pass new argument.
4735 2014-01-10 Jeff Law <law@redhat.com>
4738 * ree.c (combine_reaching_defs): Ensure the defining statement
4739 occurs before the extension when optimizing extensions with
4740 different source and destination hard registers.
4742 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4745 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
4746 vtables into the type inheritance graph.
4748 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4750 PR rtl-optimization/59754
4751 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
4752 modes in the REGNO != REGNO case.
4754 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4756 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
4758 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4760 PR tree-optimization/59745
4761 * tree-predcom.c (tree_predictive_commoning_loop): Call
4762 free_affine_expand_cache if giving up because components is NULL.
4764 * target-globals.c (save_target_globals): Allocate < 4KB structs using
4765 GC in payload of target_globals struct instead of allocating them on
4766 the heap and the larger structs separately using GC.
4767 * target-globals.h (struct target_globals): Make regs, hard_regs,
4768 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
4769 of GTY((skip)) and change type to void *.
4770 (reset_target_globals): Cast loads from those fields to corresponding
4773 2014-01-10 Steve Ellcey <sellcey@mips.com>
4776 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
4777 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
4778 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
4780 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
4783 * aarch64-modes.def (CC_Zmode): New flags mode.
4784 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
4785 represents an equality.
4786 (aarch64_get_condition_code): Handle CC_Zmode.
4787 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
4789 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
4791 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
4792 extraction in good case.
4794 2014-01-10 Richard Biener <rguenther@suse.de>
4796 PR tree-optimization/59374
4797 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
4798 checking after SLP discovery. Mark stmts not participating
4799 in any SLP instance properly.
4801 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4803 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
4804 when handling a SET rtx.
4806 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4808 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
4809 (cortex-a57): Likewise.
4810 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
4812 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4814 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
4815 non-iwmmxt builtins.
4817 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
4821 * ipa-devirt.c record_target_from_binfo): Take as argument
4822 stack of binfos and lookup matching one for virtual inheritance.
4823 (possible_polymorphic_call_targets_1): Update.
4825 2014-01-10 Huacai Chen <chenhc@lemote.com>
4827 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
4828 kernel strings for Loongson-2E/2F/3A.
4830 2014-01-10 Jakub Jelinek <jakub@redhat.com>
4833 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
4834 is_gimple_call before calling gimple_call_internal_p.
4836 2014-01-09 Steve Ellcey <sellcey@mips.com>
4838 * Makefile.in (TREE_FLOW_H): Remove.
4839 (TREE_SSA_H): Add file names from tree-flow.h.
4840 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
4841 * tree.h: Remove tree-flow.h reference.
4842 * hash-table.h: Remove tree-flow.h reference.
4843 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
4844 reference with tree-ssa-loop.h.
4846 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4848 * doc/invoke.texi: Add -maltivec={be,le} options, and document
4849 default element-order behavior for -maltivec.
4850 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
4851 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
4852 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
4853 when targeting big endian, at least for now.
4854 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
4856 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4859 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
4860 var satisfies use_register_for_decl, just take into account type
4861 alignment, rather than decl alignment.
4863 PR tree-optimization/59622
4864 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
4865 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
4866 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
4867 Don't devirtualize for inplace at all. For targets.length () == 1,
4868 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
4870 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4872 * config/i386/i386.md (cpu): Remove the unused btver1.
4874 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
4876 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
4878 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4881 * tree-core.h (struct target_globals): New forward declaration.
4882 (struct tree_target_option): Add globals field.
4883 * tree.h (TREE_TARGET_GLOBALS): Define.
4884 (prepare_target_option_nodes_for_pch): New prototype.
4885 * target-globals.h (struct target_globals): Define even if
4887 * tree.c (prepare_target_option_node_for_pch,
4888 prepare_target_option_nodes_for_pch): New functions.
4889 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
4890 * config/i386/i386.c: Include target-globals.h.
4891 (ix86_set_current_function): Instead of doing target_reinit
4892 unconditionally, use save_target_globals_default_opts and
4893 restore_target_globals.
4895 2014-01-09 Richard Biener <rguenther@suse.de>
4897 PR tree-optimization/59715
4898 * tree-cfg.h (split_critical_edges): Declare.
4899 * tree-cfg.c (split_critical_edges): Export.
4900 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
4902 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
4904 * cfgexpand.c (expand_stack_vars): Optionally disable
4905 asan stack protection.
4906 (expand_used_vars): Likewise.
4907 (partition_stack_vars): Likewise.
4908 * asan.c (asan_emit_stack_protection): Optionally disable
4909 after return stack usage.
4910 (instrument_derefs): Optionally disable memory access instrumentation.
4911 (instrument_builtin_call): Likewise.
4912 (instrument_strlen_call): Likewise.
4913 (asan_protect_global): Optionally disable global variables protection.
4914 * doc/invoke.texi: Added doc for new options.
4915 * params.def: Added new options.
4916 * params.h: Likewise.
4918 2014-01-09 Jakub Jelinek <jakub@redhat.com>
4920 PR rtl-optimization/59724
4921 * ifcvt.c (cond_exec_process_if_block): Don't call
4922 flow_find_head_matching_sequence with 0 longest_match.
4923 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
4924 non-active insns if !stop_after.
4925 (try_head_merge_bb): Revert 2014-01-07 changes.
4927 2014-01-08 Jeff Law <law@redhat.com>
4929 * ree.c (get_sub_rtx): New function, extracted from...
4930 (merge_def_and_ext): Here.
4931 (combine_reaching_defs): Use get_sub_rtx.
4933 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
4935 * cgraph.h (varpool_variable_node): Do not choke on null node.
4937 2014-01-08 Catherine Moore <clm@codesourcery.com>
4939 * config/mips/mips.md (simple_return): Attempt to use JRC
4941 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
4943 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4945 PR rtl-optimization/59137
4946 * reorg.c (steal_delay_list_from_target): Call update_block for
4948 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
4950 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4952 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4953 two duplicate entries.
4955 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
4958 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
4960 * config/mips/mips.c (mips_truncated_op_cost): New function.
4961 (mips_rtx_costs): Adjust test for BADDU.
4962 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
4964 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
4966 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
4967 (*baddu_si): ...this new pattern.
4969 2014-01-08 Jakub Jelinek <jakub@redhat.com>
4972 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
4974 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
4977 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
4979 (expand_expr, expand_normal): Adjust.
4980 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
4981 inner_reference_p. Use inner_reference_p to expand inner references.
4982 (store_expr): Adjust.
4983 * cfgexpand.c (expand_call_stmt): Adjust.
4985 2014-01-08 Rong Xu <xur@google.com>
4987 * gcov-io.c (gcov_var): Move from gcov-io.h.
4988 (gcov_position): Ditto.
4989 (gcov_is_error): Ditto.
4990 (gcov_rewrite): Ditto.
4991 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
4992 only part to libgcc/libgcov.h.
4994 2014-01-08 Marek Polacek <polacek@redhat.com>
4997 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
4999 2014-01-08 Marek Polacek <polacek@redhat.com>
5002 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5004 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5006 PR rtl-optimization/59649
5007 * stor-layout.c (get_mode_bounds): For BImode return
5008 0 and STORE_FLAG_VALUE.
5010 2014-01-08 Richard Biener <rguenther@suse.de>
5013 * gimple.h (is_gimple_builtin_call): Remove.
5014 (gimple_builtin_call_types_compatible_p): New.
5015 (gimple_call_builtin_p): New overload.
5016 * gimple.c (is_gimple_builtin_call): Remove.
5017 (validate_call): Rename to ...
5018 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5020 (validate_type): New static function.
5021 (gimple_call_builtin_p): New overload and adjust.
5022 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5023 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5024 (gimple_fold_stmt_to_constant_1): Likewise.
5025 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5027 2014-01-08 Richard Biener <rguenther@suse.de>
5030 * gimplify.c (gimplify_expr): Gimplify register-register type
5031 VIEW_CONVERT_EXPRs to separate stmts.
5033 2014-01-07 Jeff Law <law@redhat.com>
5036 * ree.c (combine_set_extension): Handle case where source
5037 and destination registers in an extension insn are different.
5038 (combine_reaching_defs): Allow source and destination registers
5039 in extension to be different under limited circumstances.
5040 (add_removable_extension): Remove restriction that the
5041 source and destination registers in the extension are the same.
5042 (find_and_remove_re): Emit a copy from the extension's
5043 destination to its source after the defining insn if
5044 the source and destination registers are different.
5047 * ifcvt.c (merge_if_block): If we are merging a block with more than
5048 one successor with a block with no successors, remove any BARRIER
5049 after the second block.
5051 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5053 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5055 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5058 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5059 for 14-bit register offsets when INT14_OK_STRICT is false.
5061 2014-01-07 Roland Stigge <stigge@antcom.de>
5062 Michael Meissner <meissner@linux.vnet.ibm.com>
5065 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5066 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5068 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5070 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5073 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5075 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5076 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5077 rtx is const0_rtx or not.
5079 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5082 * target-globals.c (save_target_globals): Remove this_fn_optab
5084 * toplev.c: Include optabs.h.
5085 (target_reinit): Temporarily restore the global options if another
5086 set of options are in force.
5088 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5090 PR rtl-optimization/58668
5091 * cfgcleanup.c (flow_find_cross_jump): Don't count
5092 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5093 to determine what is counted.
5094 (flow_find_head_matching_sequence): Use active_insn_p to determine
5096 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5098 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5099 determine what is counted.
5101 PR tree-optimization/59643
5102 * tree-predcom.c (split_data_refs_to_components): If one dr is
5103 read and one write, determine_offset fails and the write isn't
5104 in the bad component, just put the read into the bad component.
5106 2014-01-07 Mike Stump <mikestump@comcast.net>
5107 Jakub Jelinek <jakub@redhat.com>
5110 * tree-core.h (struct tree_optimization_option): Change optabs
5111 type from unsigned char * to void *.
5112 * optabs.c (init_tree_optimization_optabs): Adjust
5113 TREE_OPTIMIZATION_OPTABS initialization.
5115 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5118 * config/i386/i386.h (struct machine_function): Add
5119 no_drap_save_restore field.
5120 * config/i386/i386.c (ix86_save_reg): Use
5121 !cfun->machine->no_drap_save_restore instead of
5122 crtl->stack_realign_needed.
5123 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5124 this function clears frame_pointer_needed. Set
5125 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5126 and DRAP reg is needed.
5128 2014-01-06 Marek Polacek <polacek@redhat.com>
5131 * doc/implement-c.texi: Mention that other integer types are
5132 permitted as bit-field types in strictly conforming mode.
5134 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5136 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5139 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5141 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5143 2014-01-06 Martin Jambor <mjambor@suse.cz>
5146 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5148 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5150 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5154 * var-tracking.c (add_stores): Preserve the value of the source even if
5155 we don't record the store.
5157 2014-01-06 Terry Guo <terry.guo@arm.com>
5159 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5161 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5164 * config/darwin.c (darwin_function_section): Adjust return values to
5165 correspond to optimisation changes made in r206070.
5167 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5169 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5170 from prefetch_block tune setting.
5171 (nocona_cost): Correct size of prefetch block to 64.
5173 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5175 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5176 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5177 used to save the static chain register in the computation of the offset
5178 from which the FP registers need to be restored.
5180 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5182 PR tree-optimization/59519
5183 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5184 ICE if get_current_def (current_new_name) is already non-NULL, as long
5185 as it is a phi result of some other phi in *new_exit_bb that has
5188 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5189 or vmovdqu* for misaligned_operand.
5190 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5191 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5192 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5193 aligned_mem for AVX512F masked aligned load and store builtins and for
5196 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5198 PR tree-optimization/59651
5199 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5200 Address range for negative step should be added by TYPE_SIZE_UNIT.
5202 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5204 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5205 overlapping registers also for registers other than the stack pointer.
5207 2014-01-03 Marek Polacek <polacek@redhat.com>
5210 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5213 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5216 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5219 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5220 (push splitter): Use <P:MODE_SIZE> instead of
5221 GET_MODE_SIZE (<P:MODE>mode).
5222 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5223 (mov -1, reg peephole2): Likewise.
5224 * config/i386/sse.md (*mov<mode>_internal,
5225 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5226 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5227 *<code><mode>3, *andnot<mode>3<mask_name>,
5228 <mask_codefor><code><mode>3<mask_name>): Likewise.
5229 * config/i386/subst.md (mask_mode512bit_condition,
5230 sd_mask_mode512bit_condition): Likewise.
5232 2014-01-02 Xinliang David Li <davidxl@google.com>
5234 PR tree-optimization/59303
5235 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5236 (dump_predicates): Better output format.
5237 (pred_equal_p): New function.
5238 (is_neq_relop_p): Ditto.
5239 (is_neq_zero_form_p): Ditto.
5240 (pred_expr_equal_p): Ditto.
5241 (pred_neg_p): Ditto.
5242 (simplify_pred): Ditto.
5243 (simplify_preds_2): Ditto.
5244 (simplify_preds_3): Ditto.
5245 (simplify_preds_4): Ditto.
5246 (simplify_preds): Ditto.
5248 (push_to_worklist): Ditto.
5249 (get_pred_info_from_cmp): Ditto.
5250 (is_degenerated_phi): Ditto.
5251 (normalize_one_pred_1): Ditto.
5252 (normalize_one_pred): Ditto.
5253 (normalize_one_pred_chain): Ditto.
5254 (normalize_preds): Ditto.
5255 (normalize_cond_1): Remove function.
5256 (normalize_cond): Ditto.
5257 (is_gcond_subset_of): Ditto.
5258 (is_subset_of_any): Ditto.
5259 (is_or_set_subset_of): Ditto.
5260 (is_and_set_subset_of): Ditto.
5261 (is_norm_cond_subset_of): Ditto.
5262 (pred_chain_length_cmp): Ditto.
5263 (convert_control_dep_chain_into_preds): Type change.
5264 (find_predicates): Ditto.
5265 (find_def_preds): Ditto.
5266 (destroy_predicates_vecs): Ditto.
5267 (find_matching_predicates_in_rest_chains): Ditto.
5268 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5269 (is_pred_expr_subset): Ditto.
5270 (is_pred_chain_subset_of): Ditto.
5271 (is_included_in): Ditto.
5272 (is_superset_of): Ditto.
5274 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5276 Update copyright years.
5278 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5280 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5281 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5282 config/arc/arc.md, config/arc/arc.opt,
5283 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5284 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5285 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5286 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
5287 config/linux-protos.h, config/linux.c, config/winnt-c.c,
5288 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
5289 vtable-verify.c, vtable-verify.h: Use the standard form for the
5292 2014-01-02 Tobias Burnus <burnus@net-b.de>
5294 * gcc.c (process_command): Update copyright notice dates.
5295 * gcov-dump.c: Ditto.
5297 * doc/cpp.texi: Bump @copying's copyright year.
5298 * doc/cppinternals.texi: Ditto.
5299 * doc/gcc.texi: Ditto.
5300 * doc/gccint.texi: Ditto.
5301 * doc/gcov.texi: Ditto.
5302 * doc/install.texi: Ditto.
5303 * doc/invoke.texi: Ditto.
5305 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5307 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
5309 2014-01-01 Jakub Jelinek <jakub@redhat.com>
5311 * config/i386/sse.md (*mov<mode>_internal): Guard
5312 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
5314 PR rtl-optimization/59647
5315 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
5316 new_rtx into UNSIGNED_FLOAT rtxes.
5318 Copyright (C) 2014 Free Software Foundation, Inc.
5320 Copying and distribution of this file, with or without modification,
5321 are permitted in any medium without royalty provided the copyright
5322 notice and this notice are preserved.