1 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
3 * config/i386/i386.c (examine_argument): Return bool. Return true if
4 parameter should be passed in memory.
5 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
6 (construct_container): Update calls to examine_argument.
7 (function_arg_advance_64): Ditto.
8 (return_in_memory_32): Merge with ix86_return_in_memory.
9 (return_in_memory_64): Ditto.
10 (return_in_memory_ms_64): Ditto.
12 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
14 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
15 * coverage.c (coverage_compute_profile_id): Handle externally visible
18 2014-04-14 Martin Jambor <mjambor@suse.cz>
20 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
21 DECL_DISREGARD_INLINE_LIMITS functions.
23 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
26 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
28 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
31 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
32 optimize_insn_for_speed_p instead of
33 optimize_function_for_speed_p.
35 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
37 * doc/invoke.texi (free): Document AArch64.
39 2014-04-14 Richard Biener <rguenther@suse.de>
41 PR tree-optimization/60042
42 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
43 (insert_into_preds_of_block): Do not prevent PHI insertion
44 for REFERENCE exprs here ...
45 (eliminate_dom_walker::before_dom_children): ... but prevent
46 their use here under similar conditions when applied to the
47 IL after PRE optimizations.
49 2014-04-14 Richard Biener <rguenther@suse.de>
51 * passes.def: Move early points-to after early SRA.
53 2014-04-14 Richard Biener <rguenther@suse.de>
55 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
56 check for which sign-changes we allow when forwarding
57 a converted value into a switch.
59 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
61 * stor-layout.c (place_field): Finalize non-constant offset for the
64 2014-04-14 Richard Biener <rguenther@suse.de>
66 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
68 (expand_switch_using_bit_tests_p): Likewise.
69 (process_switch): Compute and pass on speed_p based on the
71 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
72 optimize_bb_for_speed_p.
74 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
76 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
77 * function.h (struct function): Rename has_force_vect_loops into
78 has_force_vectorize_loops.
79 * lto-streamer-in.c (input_cfg): Adjust for renaming.
80 (input_struct_function_base): Likewise.
81 * lto-streamer-out.c (output_cfg): Likewise.
82 (output_struct_function_base): Likewise.
83 * omp-low.c (expand_omp_simd): Likewise.
84 * tree-cfg.c (move_sese_region_to_fn): Likewise.
85 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
86 (version_loop_for_if_conversion): Likewise.
87 (tree_if_conversion): Likewise.
88 (main_tree_if_conversion): Likewise.
89 (gate_tree_if_conversion): Likewise.
90 * tree-inline.c (copy_loops): Likewise.
91 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
92 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
93 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
94 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
95 * tree-vectorizer.c (vectorize_loops): Likewise.
96 * tree-vectorizer.h (unlimited_cost_model): Likewise.
98 2014-04-14 Richard Biener <rguenther@suse.de>
101 * lto-streamer-out.c (wrap_refs): New function.
102 (lto_output): Wrap symbol references in global initializes in
103 type-preserving MEM_REFs.
105 2014-04-14 Christian Bruel <christian.bruel@st.com>
107 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
109 2014-04-14 Christian Bruel <christian.bruel@st.com>
111 * config/sh/sh.md (setmemqi): New expand pattern.
112 * config/sh/sh.h (CLEAR_RATIO): Define.
113 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
114 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
116 2014-04-14 Richard Biener <rguenther@suse.de>
119 * fold-const.c (negate_expr_p): Don't negate directional rounding
121 (fold_negate_expr): Likewise.
123 2014-04-14 Richard Biener <rguenther@suse.de>
125 PR tree-optimization/59817
126 PR tree-optimization/60453
127 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
128 recursion to catch all CHRECs in the scalar evolution and restrict
129 the predicate for the remains appropriately.
131 2014-04-12 Catherine Moore <clm@codesourcery.com>
133 * config/mips/constraints.md: Add new register constraint "kb".
134 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
135 (*movhi_internal): Likewise.
136 (*movqi_internal): Likewise.
137 * config/mips/mips.h (M16_STORE_REGS): New register class.
138 (REG_CLASS_NAMES): Add M16_STORE_REGS.
139 (REG_CLASS_CONTENTS): Likewise.
140 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
142 2014-04-11 Tobias Burnus <burnus@net-b.de>
145 * doc/invoke.texi (-Wformat-signedness): Document it.
146 (Wformat=2): Mention that this enables -Wformat-signedness.
148 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
150 * common/config/epiphany/epiphany-common.c
151 (epiphany_option_optimization_table): Enable section anchors by
152 default at -O1 or higher.
153 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
154 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
155 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
156 carries no extra cost.
157 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
158 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
159 * config/epiphany/predicates.md (memclob_operand): New predicate.
160 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
161 Use memclob_operand predicate and X constraint for operand 3.
163 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
165 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
166 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
169 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
171 PR rtl-optimization/60651
172 * mode-switching.c (optimize_mode_switching): Make sure to emit
173 sets of a lower numbered entity before sets of a higher numbered
174 entity to a mode of the same or lower priority.
175 When creating a seginfo for a basic block that starts with a code
176 label, move the insertion point past the code label.
177 (new_seginfo): Document and enforce requirement that
178 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
179 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
180 * doc/tm.texi: Regenerate.
182 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
185 * config/arc/arc.c (arc_save_restore): Fix assert typo.
187 2013-04-11 Jakub Jelinek <jakub@redhat.com>
189 * BASE-VER: Set to 4.10.0.
191 2014-04-11 Tobias Burnus <burnus@net-b.de>
194 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
195 * doc/gcc.texi (Service): Update description in the @menu
196 * doc/invoke.texi (Option Summary): Remove misplaced and
199 2014-04-11 Steve Ellcey <sellcey@mips.com>
200 Jakub Jelinek <jakub@redhat.com>
203 * expr.c (convert_move): Use emit_store_flag_force instead of
204 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
207 2014-04-11 Richard Biener <rguenther@suse.de>
210 * varasm.c (assemble_alias): Avoid endless error reporting
211 recursion by setting TREE_ASM_WRITTEN.
213 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
215 * config/s390/s390.md: Add a splitter for NOT rtx.
217 2014-04-11 Jakub Jelinek <jakub@redhat.com>
219 PR rtl-optimization/60663
220 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
222 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
223 Jakub Jelinek <jakub@redhat.com>
226 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
227 flag from decl_node to node.
229 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
232 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
233 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
234 ameliorating the cases where it can be.
236 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
239 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
241 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
242 (loadsync_<mode>): Change mode.
243 (load_quadpti, store_quadpti): New.
244 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
245 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
246 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
248 2014-04-09 Cong Hou <congh@google.com>
251 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
254 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
256 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
257 instead of vnor to exploit possible fusion opportunity in the
259 (altivec_expand_vec_perm_const_le): Likewise.
261 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
263 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
264 (loadsync_<mode>): Change mode.
265 (load_quadpti, store_quadpti): New.
266 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
267 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
269 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
272 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
273 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
274 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
276 2014-04-08 Richard Biener <rguenther@suse.de>
279 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
280 a 64bit widest int print double-int similar to on HWI64 hosts.
282 2014-04-08 Richard Biener <rguenther@suse.de>
284 PR tree-optimization/60785
285 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
286 default defs properly.
288 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
290 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
293 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
295 * ipa-devirt.c (maybe_record_node): When node is not recorded,
296 set completep to false rather than true.
298 2014-04-07 Douglas B Rupp <rupp@adacore.com>
301 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
302 ARM_TARGET2_DWARF_FORMAT.
304 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
307 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
308 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
311 2014-04-07 Richard Biener <rguenther@suse.de>
313 PR tree-optimization/60766
314 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
315 (may_eliminate_iv): Convert cand_value_at result to desired type.
317 2014-04-07 Jason Merrill <jason@redhat.com>
320 * common.opt (-fno-gnu-unique): Add.
321 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
323 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
325 * haifa-sched.c: Fix outdated function reference and minor
326 grammar errors in introductory comment.
328 2014-04-07 Richard Biener <rguenther@suse.de>
331 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
333 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
335 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
338 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
339 size accounting for thunks.
340 (pa_asm_output_mi_thunk): Use final_start_function() and
341 final_end_function() to output function start and end directives.
343 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
345 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
346 device specific ISA/ feature information. Remove short_sp and
347 errata_skip ds. Add avr_device_specific_features enum to have device
349 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
350 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
351 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
352 updated device specific info.
353 * config/avr/avr-mcus.def: Merge device specific details to
355 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
357 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
358 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
359 assembler if RMW isa supported by current device.
360 * config/avr/genmultilib.awk: Update as device info structure changed.
361 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
363 2014-04-04 Cong Hou <congh@google.com>
365 PR tree-optimization/60656
366 * tree-vect-stmts.c (supportable_widening_operation):
367 Fix a bug that elements in a vector with vect_used_by_reduction
368 property are incorrectly reordered when the operation on it is not
369 consistant with the one in reduction operation.
371 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
373 PR rtl-optimization/60155
374 * gcse.c (record_set_data): New function.
375 (single_set_gcse): New function.
376 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
377 (hoist_code): Likewise.
378 (get_pressure_class_and_nregs): Likewise.
380 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
382 * explow.c (probe_stack_range): Emit a final optimization blockage.
384 2014-04-04 Anthony Green <green@moxielogic.com>
386 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
389 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
392 * lto-cgraph.c (input_overwrite_node): Check that partitioning
393 flags are set only during streaming.
394 * ipa.c (process_references, walk_polymorphic_call_targets,
395 symtab_remove_unreachable_nodes): Drop bodies of always inline
396 after early inlining.
397 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
399 2014-04-04 Jakub Jelinek <jakub@redhat.com>
400 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
403 * dwarf2out.c (const_ok_for_output_1): Reject expressions
406 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
409 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
411 (cortex_a53_fdivd): Likewise.
413 2014-04-04 Martin Jambor <mjambor@suse.cz>
416 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
418 * cgraph.c (clone_of_p): Also return true if thunks match.
419 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
420 cgraph_function_or_thunk_node and an obsolete comment.
421 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
423 (build_function_decl_skip_args): Likewise.
424 (set_new_clone_decl_and_node_flags): New function.
425 (duplicate_thunk_for_node): Likewise.
426 (redirect_edge_duplicating_thunks): Likewise.
427 (cgraph_clone_node): New parameter args_to_skip, pass it to
428 redirect_edge_duplicating_thunks which is called instead of
429 cgraph_redirect_edge_callee.
430 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
431 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
433 2014-04-04 Jeff Law <law@redhat.com>
436 * config/arm/predicates.md (const_int_I_operand): New predicate.
437 (const_int_M_operand): Similarly.
438 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
440 (insv_t2, extv_reg, extzv_t2): Likewise.
441 (load_multiple_with_writeback): Similarly for const_int_I_operand.
442 (pop_multiple_with_writeback_and_return): Likewise.
443 (vfp_pop_multiple_with_writeback): Likewise
445 2014-04-04 Richard Biener <rguenther@suse.de>
448 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
449 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
451 * gimplify.h (gimple_add_tmp_var_fn): Declare.
452 * gimplify.c (gimple_add_tmp_var_fn): New function.
453 * gimple-expr.h (create_tmp_reg_fn): Declare.
454 * gimple-expr.c (create_tmp_reg_fn): New function.
455 * gimple-low.c (record_vars_into): Don't change cfun.
456 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
457 code generation without cfun.
459 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
462 * Makefile.in (install-driver): Fix shell scripting.
464 2014-04-03 Cong Hou <congh@google.com>
466 PR tree-optimization/60505
467 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
468 threshold of number of iterations below which no vectorization
470 * tree-vect-loop.c (new_loop_vec_info):
471 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
472 * tree-vect-loop.c (vect_analyze_loop_operations):
473 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
474 * tree-vect-loop.c (vect_transform_loop):
475 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
476 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
477 of iterations of the loop and see if we should build the epilogue.
479 2014-04-03 Richard Biener <rguenther@suse.de>
481 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
482 (streamer_tree_cache_create): Adjust.
483 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
484 to allow optional nodes array.
485 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
486 (streamer_tree_cache_append): Likewise.
487 (streamer_tree_cache_create): Create nodes array optionally
488 as specified by parameter.
489 * lto-streamer-out.c (create_output_block): Avoid maintaining
490 the node array in the writer cache.
491 (DFS_write_tree): Remove assertion.
492 (produce_asm_for_decls): Free the out decl state hash table early.
493 * lto-streamer-in.c (lto_data_in_create): Adjust for
494 streamer_tree_cache_create prototype change.
496 2014-04-03 Richard Biener <rguenther@suse.de>
498 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
499 set TREE_CHAIN to NULL_TREE.
501 2014-04-03 Richard Biener <rguenther@suse.de>
503 PR tree-optimization/60740
504 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
505 over all GIMPLE_COND operands.
507 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
509 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
510 (Weffc++): Remove Scott's numbering, merge lists and reference
516 * class.c (accessible_nvdtor_p): New.
517 (check_bases): Don't check base destructor here ...
518 (check_bases_and_members): ... check them here. Trigger on
519 Wnon-virtual-dtor flag.
520 (finish_struct_1): Use accessible_nvdtor_p.
523 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
524 * g++.dg/warn/Wnvdtor-2.C: New.
525 * g++.dg/warn/Wnvdtor-3.C: New.
526 * g++.dg/warn/Wnvdtor-4.C: New.
527 * g++.dg/warn/Weff1.C: Delete.
528 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
529 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
531 2014-04-03 Nick Clifton <nickc@redhat.com>
533 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
536 2014-04-03 Martin Jambor <mjambor@suse.cz>
538 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
539 mention gcc_unreachable before failing.
540 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
543 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
546 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
547 inconsistent code and instead mark the context inconsistent.
548 (possible_polymorphic_call_targets): For inconsistent contexts
549 return empty complete list.
551 2014-04-02 Anthony Green <green@moxielogic.com>
553 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
554 (extendqisi2, extendhisi2): Define.
555 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
556 (WCHAR_TYPE): Change to unsigned int.
558 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
560 PR tree-optimization/60733
561 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
562 insertion point for PHI candidates to be the end of the feeding
563 block for the PHI argument.
565 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
567 PR rtl-optimization/60650
568 * lra-constraints.c (process_alt_operands): Decrease reject for
569 earlyclobber matching.
571 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
573 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
575 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
577 * config/spu/spu.c (pad_bb): Do not crash when the last
578 insn is CODE_FOR_blockage.
580 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
582 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
583 lies outside the target mode.
585 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
588 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
589 software floating point or no floating point registers, do not
590 allow any type in the FPRs. Eliminate a test for SPE SIMD types
591 in GPRs that occurs after we tested for GPRs that would never be
594 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
595 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
596 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
597 specifically allow DDmode, since that does not use the SPE SIMD
600 2014-04-02 Richard Biener <rguenther@suse.de>
603 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
604 MODE_INTs. Properly use negv_optab.
605 (expand_abs): Likewise.
607 2014-04-02 Richard Biener <rguenther@suse.de>
610 * Makefile.in (install-driver): Guard extra installs with special
613 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
615 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
618 2014-04-01 Richard Henderson <rth@redhat.com>
621 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
622 alternative enabled before register allocation.
624 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
626 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
627 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
629 (nios2_large_got_address): Remove unneeded 'sym' parameter.
630 (nios2_got_address): Update nios2_large_got_address call site.
631 (nios2_delegitimize_address): New function.
632 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
633 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
634 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
636 2014-04-01 Martin Husemann <martin@duskware.de>
638 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
641 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
643 PR rtl-optimization/60604
644 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
645 check from register_operand.
646 (register_operand): Redefine in terms of general_operand.
647 (nonmemory_operand): Use register_operand for the non-constant cases.
649 2014-04-01 Richard Biener <rguenther@suse.de>
651 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
653 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
655 * doc/invoke.texi (mapp-regs): Clarify.
657 2014-03-31 Ulrich Drepper <drepper@gmail.com>
659 * config/i386/avx512fintrin.h (__v32hi): Define type.
661 (_mm512_set1_epi8): Define.
662 (_mm512_set1_epi16): Define.
663 (_mm512_set4_epi32): Define.
664 (_mm512_set4_epi64): Define.
665 (_mm512_set4_pd): Define.
666 (_mm512_set4_ps): Define.
667 (_mm512_setr4_epi64): Define.
668 (_mm512_setr4_epi32): Define.
669 (_mm512_setr4_pd): Define.
670 (_mm512_setr4_ps): Define.
671 (_mm512_setzero_epi32): Define.
673 2014-03-31 Martin Jambor <mjambor@suse.cz>
676 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
677 callsite_arguments_match_p. Updated all callers. Also check types of
678 corresponding formal parameters and actual arguments.
679 (not_all_callers_have_enough_arguments_p) Renamed to
680 some_callers_have_mismatched_arguments_p.
682 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
684 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
686 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
689 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
692 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
694 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
695 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
697 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
698 Use FMAMODE_NOVF512 mode iterator.
699 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
700 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
701 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
702 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
704 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
705 Use VF_128_256 mode iterator.
706 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
709 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
711 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
712 static chain if needed.
714 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
717 * lra-constraints.c (index_part_to_reg): New.
718 (process_address): Use it.
720 2014-03-27 Jeff Law <law@redhat.com>
721 Jakub Jelinek <jakub@redhat.com>
724 * expr.c (do_tablejump): Use simplify_gen_binary rather than
725 gen_rtx_{PLUS,MULT} to build up the address expression.
727 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
728 creating non-canonical RTL.
730 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
733 * ipa-inline.c (want_inline_small_function_p): Short circuit large
734 functions; reorganize to make cheap checks first.
735 (inline_small_functions): Do not estimate growth when dumping;
737 * ipa-inline.h (inline_summary): Add min_size.
738 (growth_likely_positive): New function.
739 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
740 (set_cond_stmt_execution_predicate): Cleanup.
741 (estimate_edge_size_and_time): Compute min_size.
742 (estimate_calls_size_and_time): Likewise.
743 (estimate_node_size_and_time): Likewise.
744 (inline_update_overall_summary): Update min_size.
745 (do_estimate_edge_time): Likewise.
746 (do_estimate_edge_size): Update.
747 (do_estimate_edge_hints): Update.
748 (growth_likely_positive): New function.
750 2014-03-28 Jakub Jelinek <jakub@redhat.com>
753 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
754 also if addr has VOIDmode.
756 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
758 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
759 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
761 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
762 instructions as well as AdvancedSIMD loads.
764 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
766 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
767 Use crypto_aese type.
768 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
769 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
770 crypto_aese, crypto_aesmc. Move to types.md.
771 * config/arm/types.md (crypto_aes): Split into crypto_aese,
773 * config/arm/iterators.md (crypto_type): Likewise.
775 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
777 * cgraph.c: Include expr.h and tree-dfa.h.
778 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
781 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
784 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
785 regs from checking multi-reg pseudos.
787 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
789 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
791 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
793 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
794 if it would clobber the stack pointer, even temporarily.
796 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
798 * mode-switching.c: Make small adjustments to the top comment.
800 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
802 * config/rs6000/constraints.md (wD constraint): New constraint to
803 match the constant integer to get the top DImode/DFmode out of a
804 vector in a VSX register.
806 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
807 match the constant integer to get the top DImode/DFmode out of a
808 vector in a VSX register.
810 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
813 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
816 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
817 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
819 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
820 Optimize vec_extract of 64-bit values, where the value being
821 extracted is in the top word, where we can use scalar
822 instructions. Add direct move and store support. Combine the big
823 endian/little endian vector select load support into a single insn.
824 (vsx_extract_<mode>_internal1): Likewise.
825 (vsx_extract_<mode>_internal2): Likewise.
826 (vsx_extract_<mode>_load): Likewise.
827 (vsx_extract_<mode>_store): Likewise.
828 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
829 combined into vsx_extract_<mode>_load.
830 (vsx_extract_<mode>_one_le): Likewise.
832 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
833 define the top 64-bit vector element.
835 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
838 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
839 Document vec_vbpermq builtin.
842 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
843 enable use of xxsldwi and xxpermdi builtin functions.
844 (vec_xxpermdi): Likewise.
846 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
847 Document use of vec_xxsldwi and vec_xxpermdi builtins.
849 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
851 PR rtl-optimization/60650
852 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
854 (find_spills_for): New.
855 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
856 Spill all pseudos on the second iteration.
858 2014-03-27 Marek Polacek <polacek@redhat.com>
861 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
864 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
866 * config/s390/s390.c (s390_can_use_return_insn): Check for
867 call-saved FPRs on 31 bit.
869 2014-03-27 Jakub Jelinek <jakub@redhat.com>
872 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
873 if they need regimplification, just drop them instead of
874 calling gimple_regimplify_operands on them.
876 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
879 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
880 (aarch64_frame_pointer_required): Adjust logic.
881 (aarch64_can_eliminate): Adjust logic.
882 (aarch64_override_options_after_change): Adjust logic.
884 2014-03-27 Dehao Chen <dehao@google.com>
886 * ipa-inline.c (early_inliner): Update node's inline info.
888 2014-03-26 Dehao Chen <dehao@google.com>
890 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
891 compiler inserted conditional jumps for NAN float check.
893 2014-03-26 Jakub Jelinek <jakub@redhat.com>
895 * ubsan.h (ubsan_create_data): Change second argument's type
896 to const location_t *.
897 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
899 (ubsan_create_data): Change second argument to const location_t *PLOC.
900 Create Loc field whenever PLOC is non-NULL.
901 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
902 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
906 * real.c (real_to_integer2): Change type of low to UHWI.
908 2014-03-26 Tobias Burnus <burnus@net-b.de>
910 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
911 (CILK_SELF_SPECS): New define.
912 (driver_self_specs): Use it.
914 2014-03-26 Richard Biener <rguenther@suse.de>
916 * tree-pretty-print.c (percent_K_format): Implement special
917 case for LTO and its stripped down BLOCK tree.
919 2014-03-26 Jakub Jelinek <jakub@redhat.com>
922 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
924 * tree-vrp.c (simplify_internal_call_using_ranges): If only
925 one range is range_int_cst_p, but not both, at least optimize
926 addition/subtraction of 0 and multiplication by 0 or 1.
927 * gimple-fold.c (gimple_fold_call): Fold
928 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
929 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
930 INTEGER_CSTs, try to fold at least x * 0 and y - y.
932 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
934 PR rtl-optimization/60452
935 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
936 <case REG>: Return 1 for invalid offsets from the frame pointer.
938 2014-03-26 Marek Polacek <polacek@redhat.com>
941 * doc/extend.texi (C Extensions): Mention variable-length arrays in
944 2014-03-26 Marek Polacek <polacek@redhat.com>
947 * doc/extend.texi (Designated Inits): Describe what happens to omitted
950 2014-03-26 Marek Polacek <polacek@redhat.com>
953 * ira-color.c (update_conflict_hard_regno_costs): Perform the
954 multiplication in unsigned type.
956 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
958 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
960 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
962 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
964 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
967 * cif-code.def (UNREACHABLE) New code.
968 * ipa-inline.c (inline_small_functions): Skip edges to
969 __builtlin_unreachable.
970 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
971 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
972 predicate to __bulitin_unreachable.
973 (set_cond_stmt_execution_predicate): Fix issue when
974 invert_tree_comparison returns ERROR_MARK.
975 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
976 propagate to inline clones.
977 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
979 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
980 * cgraphclones.c (cgraph_clone_node): If call destination is already
981 ureachable, do not redirect it back.
982 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
985 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
987 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
988 Do not modify inline clones.
990 2014-03-25 Jakub Jelinek <jakub@redhat.com>
992 * config/i386/i386.md (general_sext_operand): New mode attr.
993 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
994 don't generate (sign_extend (const_int)).
995 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
996 operands[2]. Use We constraint instead of <i> and
997 <general_sext_operand> predicate instead of <general_operand>.
998 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
999 * config/i386/constraints.md (We): New constraint.
1000 * config/i386/predicates.md (x86_64_sext_operand,
1001 sext_operand): New predicates.
1003 2014-03-25 Martin Jambor <mjambor@suse.cz>
1006 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1007 inconsistent devirtualizations to __builtin_unreachable.
1009 2014-03-25 Marek Polacek <polacek@redhat.com>
1012 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1014 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1016 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1017 order of elements for big-endian.
1019 2014-03-25 Richard Biener <rguenther@suse.de>
1022 * gimplify-me.c (gimple_regimplify_operands): Update the
1025 2014-03-25 Martin Jambor <mjambor@suse.cz>
1028 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1029 (lto_output_varpool_node): Likewise.
1030 (input_overwrite_node): Likewise.
1031 (input_varpool_node): Likewise.
1033 2014-03-25 Richard Biener <rguenther@suse.de>
1035 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1036 (run_gcc): Likewise.
1038 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1040 * combine.c (simplify_compare_const): Add MODE argument.
1041 Handle mode_width 0 as very large mode_width.
1042 (try_combine, simplify_comparison): Adjust callers.
1044 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1045 type to avoid signed integer overflow.
1046 * explow.c (plus_constant): Likewise.
1048 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1050 * doc/generic.texi: Correct typos.
1052 2014-03-24 Tobias Burnus <burnus@net-b.de>
1054 * doc/invoke.texi (-flto): Expand section about
1055 using static libraries with LTO.
1057 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1059 PR rtl-optimization/60501
1060 * optabs.def (addptr3_optab): New optab.
1061 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1062 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1063 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1065 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1067 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1069 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1071 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1074 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1075 (_mm256_undefined_ps): Define.
1076 (_mm256_undefined_pd): Define.
1077 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1078 (_mm_undefined_pd): Define.
1079 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1080 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1081 (_mm512_undefined_ps): Define.
1082 (_mm512_undefined_pd): Define.
1083 Use _mm*_undefined_*.
1084 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1086 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1088 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1089 (lshr_simd): DI mode added.
1090 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1091 (aarch64_ushr_simddi): Likewise.
1092 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1093 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1094 (vshrd_n_u64): Likewise.
1096 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1098 * Makefile.in (s-macro_list): Depend on cc1.
1100 2014-03-23 Teresa Johnson <tejohnson@google.com>
1102 * ipa-utils.c (ipa_print_order): Use specified dump file.
1104 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1106 PR rtl-optimization/60601
1107 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1109 * gcc.c (eval_spec_function): Initialize save_growing_value.
1111 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1114 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1115 code == MINUS_EXPR, never swap op0 with op1.
1117 * toplev.c (init_local_tick): Avoid signed integer multiplication
1119 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1120 shift by first operand's bitsize.
1122 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1125 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1127 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1128 TARGET_ISA_64BIT_P(x).
1130 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1132 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1133 pattern for vector nor instead of subtract from splat(-1).
1134 (altivec_expand_vec_perm_const_le): Likewise.
1136 2014-03-21 Richard Henderson <rth@twiddle.net>
1139 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1140 related insns after epilogue_completed.
1142 2014-03-21 Martin Jambor <mjambor@suse.cz>
1145 * cgraph.h (symtab_node): New flag body_removed.
1146 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1147 when removing bodies.
1148 * symtab.c (dump_symtab_base): Dump body_removed flag.
1149 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1150 had their bodies removed.
1152 2014-03-21 Martin Jambor <mjambor@suse.cz>
1155 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1158 2014-03-21 Richard Biener <rguenther@suse.de>
1160 PR tree-optimization/60577
1161 * tree-core.h (struct tree_base): Document nothrow_flag use
1162 in VAR_DECL_NONALIASED.
1163 * tree.h (VAR_DECL_NONALIASED): New.
1164 (may_be_aliased): Adjust.
1165 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1167 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1169 * expr.c (expand_expr_real_1): Remove outdated comment.
1171 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1174 * ira.c (adjust_cleared_regs): Call copy_rtx on
1175 *reg_equiv[REGNO (loc)].src_p before passing it to
1176 simplify_replace_fn_rtx.
1179 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1180 into CONST, put pic register as first operand of PLUS. Use
1181 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1183 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1185 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1187 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1189 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1190 around for store forwarding issue in the FPU on the UT699.
1191 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1192 loads and operations if -mfix-ut699 is specified.
1193 (divtf3_hq): Tweak attribute.
1194 (sqrttf2_hq): Likewise.
1196 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1198 * calls.c (store_one_arg): Remove incorrect const qualification on the
1199 type of the temporary.
1200 * cfgexpand.c (expand_return): Likewise.
1201 * expr.c (expand_constructor): Likewise.
1202 (expand_expr_real_1): Likewise.
1204 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1206 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1209 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1212 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1214 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1216 * config/arm/aarch-common-protos.h
1217 (alu_cost_table): Fix spelling of "extend".
1218 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1220 2014-03-19 Richard Biener <rguenther@suse.de>
1223 * tree-core.h (tree_type_common): Re-order pointer members
1224 to reduce recursion depth during GC walks.
1226 2014-03-19 Marek Polacek <polacek@redhat.com>
1229 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1230 before accessing it.
1232 2014-03-19 Richard Biener <rguenther@suse.de>
1235 * lto-streamer-in.c (input_function): In WPA stage do not drop
1238 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1240 PR tree-optimization/60559
1241 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1242 with build_zero_cst assignment.
1244 2014-03-18 Kai Tietz <ktietz@redhat.com>
1246 PR rtl-optimization/56356
1247 * sdbout.c (sdbout_parms): Verify that parms'
1248 incoming argument is valid.
1249 (sdbout_reg_parms): Likewise.
1251 2014-03-18 Richard Henderson <rth@redhat.com>
1254 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1255 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1256 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1258 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1260 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1261 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1262 Italicize plugin event names in description. Explain that
1263 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1264 Remind that no GCC functions should be called after PLUGIN_FINISH.
1265 Explain what pragmas with expansion are.
1267 2014-03-18 Martin Liska <mliska@suse.cz>
1269 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1270 gimple call statement is update.
1271 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1272 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1274 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1277 * ubsan.c (ubsan_instrument_unreachable): Call
1278 initialize_sanitizer_builtins.
1279 (ubsan_pass): Likewise.
1282 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1283 varpool_finalize_decl instead of rest_of_decl_compilation.
1285 2014-03-18 Richard Biener <rguenther@suse.de>
1287 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1288 by using bitmap_and_compl instead of bitmap_and_compl_into.
1289 (df_rd_transfer_function): Likewise.
1291 2014-03-18 Richard Biener <rguenther@suse.de>
1293 * doc/lto.texi (fresolution): Fix typo.
1295 2014-03-18 Richard Biener <rguenther@suse.de>
1297 * doc/invoke.texi (flto): Update for changes in 4.9.
1299 2014-03-18 Richard Biener <rguenther@suse.de>
1301 * doc/loop.texi: Remove section on the removed lambda framework.
1302 Update loop docs with recent changes in preserving loop structure.
1304 2014-03-18 Richard Biener <rguenther@suse.de>
1306 * doc/lto.texi (-fresolution): Document.
1308 2014-03-18 Richard Biener <rguenther@suse.de>
1310 * doc/contrib.texi: Adjust my name.
1312 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1315 * internal-fn.c: Include diagnostic-core.h.
1316 (expand_BUILTIN_EXPECT): New function.
1317 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1318 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1319 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1320 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1322 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1323 Revert 3 argument __builtin_expect code.
1324 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1325 * gimple-fold.c (gimple_fold_call): Likewise.
1326 * tree.h (fold_builtin_expect): New prototype.
1327 * builtins.c (build_builtin_expect_predicate): Add predictor
1328 argument, if non-NULL, create 3 argument __builtin_expect.
1329 (fold_builtin_expect): No longer static. Add ARG2 argument,
1330 pass it through to build_builtin_expect_predicate.
1331 (fold_builtin_2): Adjust caller.
1332 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1333 * internal-fn.def (BUILTIN_EXPECT): New.
1335 2014-03-18 Tobias Burnus <burnus@net-b.de>
1338 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1339 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1340 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1342 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1345 * predict.c (combine_predictions_for_bb): Fix up formatting.
1346 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1347 fill what it points to if non-NULL.
1348 (tree_predict_by_opcode): Adjust caller, use the predictor.
1349 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1351 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1353 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1354 proper constant for the store mode.
1356 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1358 * symtab.c (change_decl_assembler_name): Fix transparent alias
1361 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1363 * config/aarch64/aarch64.c: Correct the comments about the
1364 aarch64 stack layout.
1366 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1368 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1369 check for GF_OMP_FOR_KIND_FOR.
1371 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1373 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1374 ymm and zmm register names.
1376 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1379 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1380 note creation for the 2010-08-31 changes.
1382 2014-03-17 Marek Polacek <polacek@redhat.com>
1385 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1386 as -fno-tree-loop-vectorize.
1387 (expand_omp_simd): Likewise.
1389 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1391 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1392 (eligible_for_call_delay): New prototype.
1393 * config/sparc/sparc.c (tls_call_delay): Rename into...
1394 (eligible_for_call_delay): ...this. Return false if the instruction
1395 cannot be put in the delay slot of a branch.
1396 (eligible_for_restore_insn): Simplify.
1397 (eligible_for_return_delay): Return false if the instruction cannot be
1398 put in the delay slot of a branch and simplify.
1399 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1400 put in the delay slot of a branch.
1401 * config/sparc/sparc.md (fix_ut699): New attribute.
1402 (tls_call_delay): Delete.
1403 (in_call_delay): Reimplement.
1404 (eligible_for_sibcall_delay): Rename into...
1405 (in_sibcall_delay): ...this.
1406 (eligible_for_return_delay): Rename into...
1407 (in_return_delay): ...this.
1408 (in_branch_delay): Reimplement.
1409 (in_uncond_branch_delay): Delete.
1410 (in_annul_branch_delay): Delete.
1412 2014-03-14 Richard Henderson <rth@redhat.com>
1415 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1416 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1417 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1418 (floathi splitters): Remove.
1419 (float<SWI48x>xf2): New pattern.
1420 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1421 code that tried to handle DImode for 32-bit, but which was excluded
1422 by the pattern's condition. Drop allocation of stack temporary.
1423 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1424 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1425 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1426 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1427 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1428 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1429 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1430 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1431 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1432 (*float<SWI48x><X87MODEF>2_i387): Remove.
1433 (all float _with_temp splitters): Remove.
1434 (*float<SWI48x><MODEF>2_i387): New pattern.
1435 (*float<SWI48><MODEF>2_sse): New pattern.
1436 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1437 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1439 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1440 Marek Polacek <polacek@redhat.com>
1443 * common.opt (dump_base_name_prefixed): New Variable.
1444 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1445 if x_dump_base_name_prefixed is already set, set it at the end.
1447 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1449 PR rtl-optimization/60508
1450 * lra-constraints.c (get_reload_reg): Add new parameter
1452 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1453 Pass the new parameter values.
1455 2014-03-14 Richard Biener <rguenther@suse.de>
1457 * common.opt: Revert unintented changes from r205065.
1460 2014-03-14 Richard Biener <rguenther@suse.de>
1463 * cfghooks.c (split_block): Properly adjust all loops the
1464 block was a latch of.
1466 2014-03-14 Martin Jambor <mjambor@suse.cz>
1469 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1472 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1475 * config/avr/avr.c (avr_set_current_function): Pass function name
1476 through default_strip_name_encoding before sanity checking instead
1477 of skipping the first char of the assembler name.
1479 2014-03-13 Richard Henderson <rth@redhat.com>
1482 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1483 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1484 * config/i386/i386-protos.h: Likewise.
1485 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1486 in the expander instead of a splitter.
1487 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1488 any possibility of requiring a memory.
1489 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1490 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1491 (fp branch splitters): Update for ix86_split_fp_branch.
1492 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1493 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1494 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1495 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1496 (*fop_<MODEF>_3_i387): Likewise.
1497 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1498 (splitters for the fop_* register patterns): Remove.
1499 (fscalexf4_i387): Rename from *fscalexf4_i387.
1500 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1502 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1504 PR tree-optimization/59779
1505 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1506 type for bitsize and maxsize instead of HOST_WIDE_INT.
1508 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1510 PR rtl-optimization/57320
1511 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1512 the CFG after thread_prologue_and_epilogue_insns.
1514 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1516 PR rtl-optimization/57189
1517 * lra-constraints.c (process_alt_operands): Disfavor spilling
1520 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1522 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1524 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1526 PR tree-optimization/59025
1528 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1529 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1531 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1534 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1535 calls of avr_out_plus_1.
1537 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1539 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1540 BB's single pred and update the father loop's latch info later.
1542 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1544 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1548 (VEC_base): Likewise.
1549 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1550 registers, we need to swap double words in little endian mode.
1552 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1553 to be a container mode for 128-bit integer operations added in ISA
1554 2.07. Unlike TImode and PTImode, the preferred register set is
1555 the Altivec/VMX registers for the 128-bit operations.
1557 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1559 (rs6000_split_128bit_ok_p): Likewise.
1561 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1562 macros for creating ISA 2.07 normal and overloaded builtin
1563 functions with 3 arguments.
1564 (BU_P8V_OVERLOAD_3): Likewise.
1565 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1566 for use as overloaded functions.
1567 (VPERM_1TI_UNS): Likewise.
1568 (VSEL_1TI): Likewise.
1569 (VSEL_1TI_UNS): Likewise.
1570 (ST_INTERNAL_1ti): Likewise.
1571 (LD_INTERNAL_1ti): Likewise.
1572 (XXSEL_1TI): Likewise.
1573 (XXSEL_1TI_UNS): Likewise.
1574 (VPERM_1TI): Likewise.
1575 (VPERM_1TI_UNS): Likewise.
1576 (XXPERMDI_1TI): Likewise.
1577 (SET_1TI): Likewise.
1578 (LXVD2X_V1TI): Likewise.
1579 (STXVD2X_V1TI): Likewise.
1580 (VEC_INIT_V1TI): Likewise.
1581 (VEC_SET_V1TI): Likewise.
1582 (VEC_EXT_V1TI): Likewise.
1583 (EQV_V1TI): Likewise.
1584 (NAND_V1TI): Likewise.
1585 (ORC_V1TI): Likewise.
1586 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1587 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1589 (VADDUQM): Likewise.
1590 (VSUBCUQ): Likewise.
1591 (VADDEUQM): Likewise.
1592 (VADDECUQ): Likewise.
1593 (VSUBEUQM): Likewise.
1594 (VSUBECUQ): Likewise.
1596 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1597 __int128_t and __uint128_t types.
1598 (__uint128_type): Likewise.
1599 (altivec_categorize_keyword): Add support for vector __int128_t,
1600 vector __uint128_t, vector __int128, and vector unsigned __int128
1601 as a container type for TImode operations that need to be done in
1602 VSX/Altivec registers.
1603 (rs6000_macro_to_expand): Likewise.
1604 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1605 to support 128-bit integer instructions vaddcuq, vadduqm,
1606 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1607 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1609 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1610 for V1TImode, and set up preferences to use VSX/Altivec registers.
1611 Setup VSX reload handlers.
1612 (rs6000_debug_reg_global): Likewise.
1613 (rs6000_init_hard_regno_mode_ok): Likewise.
1614 (rs6000_preferred_simd_mode): Likewise.
1615 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1616 (easy_altivec_constant): Likewise.
1617 (output_vec_const_move): Likewise.
1618 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1620 (rs6000_expand_vector_extract): Likewise.
1621 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1623 (rs6000_const_vec): Add support for V1TImode.
1624 (rs6000_emit_le_vsx_load): Swap double words when loading or
1625 storing TImode/V1TImode.
1626 (rs6000_emit_le_vsx_store): Likewise.
1627 (rs6000_emit_le_vsx_move): Likewise.
1628 (rs6000_emit_move): Add support for V1TImode.
1629 (altivec_expand_ld_builtin): Likewise.
1630 (altivec_expand_st_builtin): Likewise.
1631 (altivec_expand_vec_init_builtin): Likewise.
1632 (altivec_expand_builtin): Likewise.
1633 (rs6000_init_builtins): Add support for V1TImode type. Add
1634 support for ISA 2.07 128-bit integer builtins. Define type names
1635 for the VSX/Altivec vector types.
1636 (altivec_init_builtins): Add support for overloaded vector
1637 functions with V1TImode type.
1638 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1639 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1641 (rs6000_split_128bit_ok_p): Likewise.
1642 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1643 __int128_t and vector __uint128_t.
1645 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1646 and mode attributes.
1653 (VS_scalar): Likewise.
1654 (VS_double): Likewise.
1655 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1657 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1658 we support the ISA 2.07 128-bit integer arithmetic instructions.
1659 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1660 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1661 and TImode types for use with the builtin functions.
1662 (V1TI_type_node): Likewise.
1663 (unsigned_V1TI_type_node): Likewise.
1664 (intTI_type_internal_node): Likewise.
1665 (uintTI_type_internal_node): Likewise.
1667 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1668 128-bit builtin functions.
1669 (UNSPEC_VADDEUQM): Likewise.
1670 (UNSPEC_VADDECUQ): Likewise.
1671 (UNSPEC_VSUBCUQ): Likewise.
1672 (UNSPEC_VSUBEUQM): Likewise.
1673 (UNSPEC_VSUBECUQ): Likewise.
1674 (VM): Add V1TImode to vector mode iterators.
1676 (VI_unit): Likewise.
1677 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1678 (altivec_vaddcuq): Likewise.
1679 (altivec_vsubuqm): Likewise.
1680 (altivec_vsubcuq): Likewise.
1681 (altivec_vaddeuqm): Likewise.
1682 (altivec_vaddecuq): Likewise.
1683 (altivec_vsubeuqm): Likewise.
1684 (altivec_vsubecuq): Likewise.
1686 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1688 (BOOL_128): Likewise.
1689 (BOOL_REGS_OUTPUT): Likewise.
1690 (BOOL_REGS_OP1): Likewise.
1691 (BOOL_REGS_OP2): Likewise.
1692 (BOOL_REGS_UNARY): Likewise.
1693 (BOOL_REGS_AND_CR0): Likewise.
1695 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1696 128-bit integer builtin support.
1697 (vec_vadduqm): Likewise.
1698 (vec_vaddecuq): Likewise.
1699 (vec_vaddeuqm): Likewise.
1700 (vec_vsubecuq): Likewise.
1701 (vec_vsubeuqm): Likewise.
1702 (vec_vsubcuq): Likewise.
1703 (vec_vsubuqm): Likewise.
1705 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1706 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1707 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1708 128-bit integer add/subtract to ISA 2.07.
1710 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1712 * config/arc/arc.c (arc_predicate_delay_insns):
1713 Fix third argument passed to conditionalize_nonjump.
1715 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1717 * config/aarch64/aarch64-builtins.c
1718 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1719 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1720 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1721 instead of __builtin_lfloor.
1722 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1724 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1726 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1727 (tree_ssa_ifcombine_bb_1): New function.
1728 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1729 is an empty forwarder block to then_bb or vice versa and then_bb
1730 and else_bb are effectively swapped.
1732 2014-03-12 Christian Bruel <christian.bruel@st.com>
1735 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1736 REG_CFA_DEF_CFA note.
1737 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1738 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1740 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1742 PR tree-optimization/60454
1743 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1745 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1747 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1748 Do not define target_cpu_default2 to generic.
1749 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1750 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1751 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1753 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1754 Marc Glisse <marc.glisse@inria.fr>
1756 PR tree-optimization/60502
1757 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1758 instead of build_low_bits_mask.
1760 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1763 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1764 if there are multiple uses, but op doesn't live on E edge.
1765 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1766 clobber stmts before __builtin_unreachable.
1768 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1770 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1771 hard_frame_pointer_rtx.
1772 * cse.c (cse_insn): Remove volatile check.
1773 * cselib.c (cselib_process_insn): Likewise.
1774 * dse.c (scan_insn): Likewise.
1776 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1778 * config/arc/arc.c (conditionalize_nonjump): New function,
1780 (arc_ifcvt): ... this.
1781 (arc_predicate_delay_insns): Use it.
1783 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1785 * config/arc/predicates.md (extend_operand): During/after reload,
1786 allow const_int_operand.
1787 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1788 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1789 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1791 (umulsi3_highpart_i): Likewise.
1793 2014-03-11 Richard Biener <rguenther@suse.de>
1795 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1796 Add asserts to guard possible wrong-code bugs.
1798 2014-03-11 Richard Biener <rguenther@suse.de>
1800 PR tree-optimization/60429
1801 PR tree-optimization/60485
1802 * tree-ssa-structalias.c (set_union_with_increment): Properly
1803 take into account all fields that overlap the shifted vars.
1804 (do_sd_constraint): Likewise.
1805 (do_ds_constraint): Likewise.
1806 (get_constraint_for_ptr_offset): Likewise.
1808 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1810 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1811 (nios2_compute_frame_layout):
1812 Add calculation of cfun->machine->fp_save_offset.
1813 (nios2_expand_prologue): Correct setting of frame pointer register
1815 (nios2_expand_epilogue): Update recovery of stack pointer from
1816 frame pointer accordingly.
1817 (nios2_initial_elimination_offset): Update calculation of offset
1818 for eliminating to HARD_FRAME_POINTER_REGNUM.
1820 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1823 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1824 cgraph_get_create_node on VAR_DECLs.
1826 2014-03-10 Richard Biener <rguenther@suse.de>
1829 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1831 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1833 * config/vms/vms.opt (vms_float_format): New variable.
1835 2014-03-08 Tobias Burnus <burnus@net-b.de>
1837 * doc/invoke.texi (-fcilkplus): Update implementation status.
1839 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1840 Richard Biener <rguenther@suse.de>
1842 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1843 consistently accross all TUs.
1844 (run_gcc): Enable -fshort-double automatically at link at link-time
1845 and disallow override.
1847 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1850 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1851 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1852 if they can't be used.
1854 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1856 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1857 for Solaris 11/x86 ld.
1858 * configure: Regenerate.
1860 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1862 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1863 (LIB_TLS_SPEC): Save as ld_tls_libs.
1864 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1865 (HAVE_AS_IX86_TLSLDM): New test.
1866 * configure, config.in: Regenerate.
1867 * config/i386/i386.c (legitimize_tls_address): Fall back to
1868 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1869 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1870 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1871 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1873 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1875 * common.opt (fira-loop-pressure): Mark as optimization.
1877 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1879 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1880 an OpenMP mappable type.
1882 2014-03-06 Matthias Klose <doko@ubuntu.com>
1884 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1885 MULTILIB_OSDIRNAMES is not defined.
1887 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1888 Meador Inge <meadori@codesourcery.com>
1891 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1892 (arm_legitimize_address): Call legitimize_tls_address for any
1893 arm_tls_referenced_p expression, handle constant addend. Call it
1894 before testing for !TARGET_ARM.
1895 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1897 2014-03-06 Richard Biener <rguenther@suse.de>
1903 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1905 * tree-streamer.c (record_common_node): Assert we don't record
1906 nodes with type double.
1907 (preload_common_node): Skip type double, complex double and double
1908 pointer since it is now frontend dependent due to fshort-double option.
1910 2014-03-06 Richard Biener <rguenther@suse.de>
1912 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1913 or -fno-lto is specified and the linker has full plugin support.
1914 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1915 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1916 * lto-wrapper.c (merge_and_complain): Merge compile-time
1917 optimization levels.
1918 (run_gcc): And pass it through to the link options.
1920 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1924 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1926 * cselib.c (remove_useless_values): Skip to avoid quadratic
1927 behavior if the condition moved from...
1928 (cselib_process_insn): ... here holds.
1930 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1933 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1934 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1937 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1938 (TM_H): Add x86-tune.def.
1940 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1942 * config/aarch64/aarch64.c (generic_tunings):
1943 Use cortexa57_extra_costs.
1945 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1948 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1949 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1950 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1953 2014-03-04 Heiher <r@hev.cc>
1955 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1956 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1958 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1960 * config/i386/predicates.md (const2356_operand): Change to ...
1961 (const2367_operand): ... this.
1962 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1964 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1965 (*avx512pf_scatterpf<mode>sf): Ditto.
1966 (avx512pf_scatterpf<mode>df): Ditto.
1967 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1968 (*avx512pf_scatterpf<mode>df): Ditto.
1969 * config/i386/i386.c (ix86_expand_builtin): Update
1970 incorrect hint operand error message.
1972 2014-03-04 Richard Biener <rguenther@suse.de>
1974 * lto-section-in.c (lto_get_section_data): Fix const cast.
1976 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1978 * tree-streamer.c (record_common_node): Assert we don't record
1979 nodes with type double.
1980 (preload_common_node): Skip type double, complex double and double
1981 pointer since it is now frontend dependent due to fshort-double option.
1983 2014-03-04 Richard Biener <rguenther@suse.de>
1986 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1987 (lto_input_toplevel_asms): Likewise.
1988 * lto-section-in.c (lto_get_section_data): Instead do it here
1991 2014-03-04 Richard Biener <rguenther@suse.de>
1993 PR tree-optimization/60382
1994 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1995 dead PHIs a reduction.
1997 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1999 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2001 (_mm_prefetch): Move out of GCC target("sse") pragma.
2002 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2003 GCC target("prfchw") pragma.
2004 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2006 * config/i386/i386.c (ix86_option_override_internal): Enable
2007 -mprfchw with -mprefetchwt1.
2009 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2011 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2014 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2016 * opts.h (CL_PCH_IGNORE): Define.
2017 * targhooks.c (option_affects_pch_p):
2018 Return false for options that have CL_PCH_IGNORE set.
2019 * opt-functions.awk: Process PchIgnore.
2020 * doc/options.texi: Document PchIgnore.
2022 * config/arc/arc.opt (misize): Add PchIgnore property.
2024 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2026 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2027 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2028 constraint on constants to permit them being loaded into
2029 GENERAL_REGS or BASE_REGS.
2031 2014-03-03 Nick Clifton <nickc@redhat.com>
2033 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2034 anti-cacnonical alternatives.
2035 (negandhi3_real): New pattern.
2036 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2038 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2040 * config/avr/avr-mcus.def: Remove atxmega16x1.
2041 * config/avr/avr-tables.opt: Regenerate.
2042 * config/avr/t-multilib: Regenerate.
2043 * doc/avr-mmcu.texi: Regenerate.
2045 2014-03-03 Tobias Grosser <tobias@grosser.es>
2046 Mircea Namolaru <mircea.namolaru@inria.fr>
2048 PR tree-optimization/58028
2049 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2052 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2054 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2055 not handled by recognizers.
2057 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2060 * function.c (expand_function_end): Don't emit
2061 clobber_return_register sequence if clobber_after is a BARRIER.
2062 * cfgexpand.c (construct_exit_block): Append instructions before
2063 return_label to prev_bb.
2065 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2067 * config/rs6000/constraints.md: Document reserved use of "wc".
2069 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2072 * ipa.c (function_and_variable_visibility): When dissolving comdat
2073 group, also set all symbols to local.
2075 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2080 2013-12-14 Jan Hubicka <jh@suse.cz>
2082 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2084 2014-03-02 Jon Beniston <jon@beniston.com>
2090 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2091 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2092 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2093 (simple_return, *simple_return): New patterns
2094 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2095 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2097 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2099 * dwarf2out.c (gen_subprogram_die): Tidy.
2101 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2104 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2105 (*mov_t_msb_neg_negc): ... this new insn.
2107 2014-02-28 Jason Merrill <jason@redhat.com>
2110 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2113 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2116 * dwarf2out.c (decltype_auto_die): New static.
2117 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2118 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2119 (is_cxx_auto): Likewise.
2121 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2123 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2124 we are not using general regs only.
2126 2014-02-28 Richard Biener <rguenther@suse.de>
2129 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2130 previous fix and only allow to remove trivial pre-headers
2131 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2132 (remove_forwarder_block): Properly update the latch of a loop.
2134 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2137 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2138 (cselib_preserved_hash_table): New.
2139 (preserve_constants_and_equivs): Move preserved vals to it.
2140 (cselib_find_slot): Look it up first.
2141 (cselib_init): Initialize it.
2142 (cselib_finish): Release it.
2143 (dump_cselib_table): Dump it.
2145 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2148 * cselib.c (remove_useless_values): Skip to avoid quadratic
2149 behavior if the condition moved from...
2150 (cselib_process_insn): ... here holds.
2152 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2155 * var-tracking.c (vt_initialize): Apply the same condition to
2156 preserve the CFA base value.
2158 2014-02-28 Joey Ye <joey.ye@arm.com>
2161 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2162 if reload in progress or completed.
2164 2014-02-28 Tobias Burnus <burnus@net-b.de>
2167 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2170 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2172 * doc/tm.texi.in (Condition Code Status): Update documention for
2173 relative locations of cc0-setter and cc0-user.
2175 2014-02-27 Jeff Law <law@redhat.com>
2177 PR rtl-optimization/52714
2178 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2179 into two independent simple sets, if I3 is a jump, ensure the
2180 pattern we place into I3 is a (set (pc) ...).
2182 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2183 Jeff Law <law@redhat.com>
2185 PR rtl-optimization/49847
2186 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2187 are in different blocks.
2188 * doc/tm.texi (Condition Code Status): Update documention for
2189 relative locations of cc0-setter and cc0-user.
2191 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2194 * lra.c (lra_emit_add): Check SUBREG too.
2196 2014-02-27 Andreas Schwab <schwab@suse.de>
2198 * config/m68k/m68k.c (m68k_option_override): Disable
2199 -flive-range-shrinkage for classic m68k.
2200 (m68k_override_options_after_change): Likewise.
2202 2014-02-27 Marek Polacek <polacek@redhat.com>
2205 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2206 -Wmaybe-uninitialized.
2208 2014-02-27 Alan Modra <amodra@gmail.com>
2211 * reload1.c (emit_input_reload_insns): When reload_override_in,
2212 set old to rl->in_reg when rl->in_reg is a subreg.
2214 2014-02-26 Richard Biener <rguenther@suse.de>
2217 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2219 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2221 * common/config/i386/predicates.md (const1256_operand): Remove.
2222 (const2356_operand): New.
2223 (const_1_to_2_operand): Remove.
2224 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2225 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2226 (*avx512pf_gatherpf<mode>sf): Ditto.
2227 (avx512pf_gatherpf<mode>df): Ditto.
2228 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2229 (*avx512pf_gatherpf<mode>df): Ditto.
2230 (avx512pf_scatterpf<mode>sf): Ditto.
2231 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2232 (*avx512pf_scatterpf<mode>sf): Ditto.
2233 (avx512pf_scatterpf<mode>df): Ditto.
2234 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2235 (*avx512pf_scatterpf<mode>df): Ditto.
2236 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2238 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2240 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2241 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2242 (_mm512_mask_testn_epi64_mask): Move to ...
2243 * config/i386/avx512cdintrin.h: Here.
2244 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2245 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2246 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2247 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2248 TARGET_AVX512F from TARGET_AVX512CD.
2250 2014-02-26 Richard Biener <rguenther@suse.de>
2253 * ipa.c (walk_polymorphic_call_targets): Properly guard
2254 call to inline_update_overall_summary.
2256 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2259 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2260 and latches only if requested. Fix latch if it is removed.
2261 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2262 LOOPS_HAVE_PREHEADERS.
2264 2014-02-25 Andrew Pinski <apinski@cavium.com>
2266 * builtins.c (expand_builtin_thread_pointer): Create a new target
2267 when the target is NULL.
2269 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2271 PR rtl-optimization/60317
2272 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2273 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2274 * lra-assigns.c: Include params.h.
2275 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2276 other reload pseudos considerations.
2278 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2280 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2281 to use canonical form for nor<mode>3.
2283 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2286 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2289 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2291 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2292 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2293 (ix86_handle_option): Handle OPT_mprefetchwt1.
2294 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2295 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2297 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2298 OPTION_MASK_ISA_PREFETCHWT1.
2299 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2300 (PTA_PREFETCHWT1): New.
2301 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2302 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2303 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2304 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2305 (*prefetch_avx512pf_<mode>_: Change into ...
2306 (*prefetch_prefetchwt1_<mode>: This.
2307 * config/i386/i386.opt (mprefetchwt1): New.
2308 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2309 (_mm_prefetch): Handle intent to write.
2310 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2312 2014-02-25 Richard Biener <rguenther@suse.de>
2315 * emit-rtl.c (mem_attrs_htab): Remove.
2316 (mem_attrs_htab_hash): Likewise.
2317 (mem_attrs_htab_eq): Likewise.
2318 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2319 (init_emit_once): Do not allocate mem_attrs_htab.
2321 2014-02-25 Richard Biener <rguenther@suse.de>
2324 * lto-opts.c (lto_write_options): Output non-explicit conservative
2325 -fwrapv, -fno-trapv and -fno-strict-overflow.
2326 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2327 (run_gcc): And pass them through.
2329 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2331 * sel-sched.c (calculate_new_fences): New parameter ptime.
2332 Calculate it as a maximum over all fence cycles.
2333 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2334 Print the final schedule timing when sched_verbose.
2336 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2338 PR rtl-optimization/60292
2339 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2340 bit fot the fence instruction.
2342 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2344 * calls.h: Fix typo in comment.
2346 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2348 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2349 adjusting offsetable addresses.
2351 2014-02-24 Guozhi Wei <carrot@google.com>
2353 * sparseset.h (sparseset_pop): Fix the wrong index.
2355 2014-02-24 Walter Lee <walt@tilera.com>
2357 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2358 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2360 * common/config/tilegx/tilegx-common.c
2361 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2362 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2364 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2365 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2366 (tilegx_gimplify_va_arg_expr): Handle big endian.
2367 (tilegx_expand_unaligned_load): Ditto.
2368 (tilegx_expand_unaligned_store): Ditto.
2369 (TARGET_RETURN_IN_MSB): New.
2370 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2371 (TARGET_ENDIAN_DEFAULT): New.
2372 (TARGET_BIG_ENDIAN): Handle big endian.
2373 (BYTES_BIG_ENDIAN): Ditto.
2374 (WORDS_BIG_ENDIAN): Ditto.
2375 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2378 * config/tilegx/tilegx.md (extv): Handle big endian.
2380 (insn_st<n>): Ditto.
2381 (insn_st<n>_add<bitsuffix>): Ditto.
2382 (insn_stnt<n>): Ditto.
2383 (insn_stnt<n>_add<bitsuffix>):Ditto.
2384 (vec_interleave_highv8qi): Handle big endian.
2385 (vec_interleave_highv8qi_be): New.
2386 (vec_interleave_highv8qi_le): New.
2387 (insn_v1int_h): Handle big endian.
2388 (vec_interleave_lowv8qi): Handle big endian.
2389 (vec_interleave_lowv8qi_be): New.
2390 (vec_interleave_lowv8qi_le): New.
2391 (insn_v1int_l): Handle big endian.
2392 (vec_interleave_highv4hi): Handle big endian.
2393 (vec_interleave_highv4hi_be): New.
2394 (vec_interleave_highv4hi_le): New.
2395 (insn_v2int_h): Handle big endian.
2396 (vec_interleave_lowv4hi): Handle big endian.
2397 (vec_interleave_lowv4hi_be): New.
2398 (vec_interleave_lowv4hi_le): New.
2399 (insn_v2int_l): Handle big endian.
2400 (vec_interleave_highv2si): Handle big endian.
2401 (vec_interleave_highv2si_be): New.
2402 (vec_interleave_highv2si_le): New.
2403 (insn_v4int_h): Handle big endian.
2404 (vec_interleave_lowv2si): Handle big endian.
2405 (vec_interleave_lowv2si_be): New.
2406 (vec_interleave_lowv2si_le): New.
2407 (insn_v4int_l): Handle big endian.
2408 * config/tilegx/tilegx.opt (mbig-endian): New option.
2409 (mlittle-endian): New option.
2410 * doc/install.texi: Document tilegxbe-linux.
2411 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2413 2014-02-24 Martin Jambor <mjambor@suse.cz>
2416 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2417 there are no parameter descriptors.
2419 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2421 PR rtl-optimization/60268
2422 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2423 initialization to ...
2424 (sched_rgn_init): ... here.
2425 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2427 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2429 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2432 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2434 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2437 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2439 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2440 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2442 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2444 * config/microblaze/predicates.md: Add cmp_op predicate.
2445 * config/microblaze/microblaze.md: Add branch_compare instruction
2446 which uses cmp_op predicate and emits cmp insn before branch.
2447 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2448 to microblaze_expand_conditional_branch and consolidate logic.
2449 (microblaze_expand_conditional_branch): emit branch_compare
2450 insn instead of handling cmp op separate from branch insn.
2452 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2454 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2457 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2459 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2460 define_insn with define_expand and new define_insn
2461 *altivec_lve<VI_char>x_internal.
2462 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2463 and new define_insn *altivec_stve<VI_char>x_internal.
2464 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2466 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2468 (altivec_expand_stvex_be): New function.
2470 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2472 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2473 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2474 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2475 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2477 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2480 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2481 instead of emit_move_insn.
2483 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2485 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2487 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2488 gen_altivec_vsumsws.
2490 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2492 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2493 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2494 (altivec_lvxl_<mode>): New define_expand incorporating
2495 -maltivec=be semantics where needed.
2496 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2497 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2498 semantics where needed.
2499 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2500 (altivec_stvx_<mode>): New define_expand incorporating
2501 -maltivec=be semantics where needed.
2502 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2503 VM2 iterator instead of V4SI.
2504 (altivec_stvxl_<mode>): New define_expand incorporating
2505 -maltivec=be semantics where needed.
2506 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2507 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2508 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2509 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2510 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2511 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2512 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2513 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2514 ALTIVEC_BUILTIN_STVXL.
2515 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2516 (altivec_expand_stvx_be): Likewise.
2517 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2518 (altivec_expand_lvx_be): Likewise.
2519 (altivec_expand_stvx_be): Likewise.
2520 (altivec_expand_builtin): Add cases for
2521 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2522 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2523 (altivec_init_builtins): Add definitions for
2524 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2525 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2527 2014-02-21 Catherine Moore <clm@codesourcery.com>
2529 * doc/invoke.texi (mvirt, mno-virt): Document.
2530 * config/mips/mips.opt (mvirt): New option.
2531 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2533 2014-02-21 Richard Biener <rguenther@suse.de>
2535 PR tree-optimization/60276
2536 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2537 (STMT_VINFO_MIN_NEG_DIST): New macro.
2538 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2539 STMT_VINFO_MIN_NEG_DIST.
2540 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2541 made for negative dependence distances still hold.
2543 2014-02-21 Richard Biener <rguenther@suse.de>
2546 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2547 DECL_INITIAL for globals not in the current function context.
2549 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2551 PR tree-optimization/56490
2552 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2553 * tree-ssa-uninit.c: Include params.h.
2554 (compute_control_dep_chain): Add num_calls argument, return false
2555 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2556 num_calls to recursive call.
2557 (find_predicates): Change dep_chain into normal array,
2558 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2559 variable and adjust compute_control_dep_chain caller.
2560 (find_def_preds): Likewise.
2562 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2564 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2565 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2567 2014-02-21 Nick Clifton <nickc@redhat.com>
2569 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2570 (pushhi1): Likewise.
2571 (popqi1): Add mode to pre_dec.
2574 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2576 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2577 mode for mask of V8SFmode permutation.
2579 2014-02-20 Richard Henderson <rth@redhat.com>
2582 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2583 a new pseudo for OLDVAL.
2585 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2588 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2589 gen_reg_rtx if d->testing_p.
2590 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2591 if d->testing_p and we will certainly return true.
2592 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2595 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2597 * emit-rtl.c (gen_reg_rtx): Assert that
2598 crtl->emit.regno_pointer_align_length is non-zero.
2600 2014-02-20 Richard Henderson <rth@redhat.com>
2603 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2604 on failure the store back into EXPECT.
2606 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2607 Sandra Loosemore <sandra@codesourcery.com>
2609 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2610 * config/nios2/nios2.c (nios2_function_profiler): Add
2611 -fPIC (flag_pic == 2) support.
2612 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2613 (nios2_large_offset_p): New function.
2614 (nios2_unspec_reloc_p): Move up position, update to use
2615 nios2_large_offset_p.
2616 (nios2_unspec_address): Remove function.
2617 (nios2_unspec_offset): New function.
2618 (nios2_large_got_address): New function.
2619 (nios2_got_address): Add large offset support.
2620 (nios2_legitimize_tls_address): Update usage of removed and new
2622 (nios2_symbol_binds_local_p): New function.
2623 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2624 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2625 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2626 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2627 processing for (const (unspec ...)).
2628 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2630 2014-02-20 Richard Biener <rguenther@suse.de>
2632 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2633 doing the substitution.
2634 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2636 2014-02-20 Martin Jambor <mjambor@suse.cz>
2639 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2640 info when checking whether lattices are bottom.
2642 2014-02-20 Richard Biener <rguenther@suse.de>
2645 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2648 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2651 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2652 parameter specifying the scaling.
2653 (inline_call): Update.
2654 (want_inline_recursively): Guard division by zero.
2655 (recursive_inlining): Update.
2656 * ipa-inline.h (clone_inlined_nodes): Update.
2658 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2661 * config/i386/i386.c (classify_argument): Pass structures of size
2662 64 bytes or less in register.
2664 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2665 Kirill Yukhin <kirill.yukhin@intel.com>
2667 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2668 (_mm_rcp28_round_ss): Ditto.
2669 (_mm_rsqrt28_round_sd): Ditto.
2670 (_mm_rsqrt28_round_ss): Ditto.
2671 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2672 (_mm_rcp14_round_ss): Ditto.
2673 (_mm_rsqrt14_round_sd): Ditto.
2674 (_mm_rsqrt14_round_ss): Ditto.
2675 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2676 the first input operand, get rid of match_dup.
2677 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2679 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2681 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2682 operand as the first input operand, set type attribute.
2683 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2685 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2686 operand as the first input operand, set type attribute.
2688 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2690 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2693 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2696 * config/i386/i386.c (construct_container): Remove TFmode check
2697 for X86_64_INTEGER_CLASS.
2699 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2702 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2703 only when -Wpsabi is enabled.
2705 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2708 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2709 passing arrays in registers are the same as for structs, so remove the
2710 special case for them.
2712 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2714 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2715 destination type, extract only the valid bits if the source type is not
2716 integral and has a different mode.
2718 2014-02-19 Richard Biener <rguenther@suse.de>
2721 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2724 2014-02-19 Richard Biener <rguenther@suse.de>
2727 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2728 (ipa_modify_call_arguments): Emit an argument load explicitely and
2729 preserve virtual SSA form there and for the replacement call.
2730 Do not update SSA form nor free dominance info.
2732 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2734 * ipa.c (function_and_variable_visibility): Also clear WEAK
2735 flag when disolving COMDAT_GROUP.
2737 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2739 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2740 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2742 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2743 do more sanity checks.
2744 (detect_type_change): Return true when giving up early.
2745 (compute_complex_assign_jump_func): Fix type parameter of
2746 ipa_set_ancestor_jf.
2747 (compute_complex_ancestor_jump_func): Likewise.
2748 (update_jump_functions_after_inlining): Fix updating of
2750 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2752 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2754 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2755 inline clones when edge disappears.
2757 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2760 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2761 Split 64-bit moves into 2 patterns. Do not allow the use of
2762 direct move for TDmode in little endian, since the decimal value
2763 has little endian bytes within a word, but the 64-bit pieces are
2764 ordered in a big endian fashion, and normal subreg's of TDmode are
2766 (mov<mode>_64bit_dm): Likewise.
2767 (movtd_64bit_nodm): Likewise.
2769 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2771 PR tree-optimization/60174
2772 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2773 statement of an SSA_NAME that occurs in an abnormal PHI node.
2775 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2778 * final.c (SEEN_BB): Remove.
2779 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2780 (final_scan_insn): Don't force_source_line on second
2781 NOTE_INSN_BASIC_BLOCK.
2783 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2786 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2787 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2788 (type_natural_mode): Warn ABI change when %zmm register is not
2789 available for AVX512F vector value passing.
2791 2014-02-18 Kai Tietz <ktietz@redhat.com>
2794 * config/i386/i386.c (ix86_expand_prologue): Use value in
2795 rax register as displacement when restoring %r10 or %rax.
2796 Fix wrong offset when restoring both registers.
2798 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2800 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2801 assertion with conditional return.
2803 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2804 Uros Bizjak <ubizjak@gmail.com>
2807 * config/i386/driver-i386.c (host_detect_local_cpu): If
2808 YMM state is not saved by the OS, also clear has_f16c. Move
2809 CPUID 0x80000001 handling before YMM state saving checking.
2811 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2813 PR rtl-optimization/58960
2814 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2815 factored out from ...
2816 (sched_init): ... here.
2817 (free_global_sched_pressure_data): New, factored out from ...
2818 (sched_finish): ... here.
2819 * sched-int.h (free_global_sched_pressure_data): Declare.
2820 * sched-rgn.c (nr_regions_initial): New static global.
2821 (haifa_find_rgns): Initialize it.
2822 (schedule_region): Disable sched-pressure for the newly
2825 2014-02-17 Richard Biener <rguenther@suse.de>
2827 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2828 release SSA defs of pattern stmts.
2830 2014-02-17 Richard Biener <rguenther@suse.de>
2832 * tree-inline.c (expand_call_inline): Release the virtual
2833 operand defined by the call we are about to inline.
2835 2014-02-17 Richard Biener <rguenther@suse.de>
2837 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2839 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2840 Ilya Tocar <ilya.tocar@intel.com>
2842 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2843 arguments order in builtin.
2844 (_mm512_permutexvar_epi64): Ditto.
2845 (_mm512_mask_permutexvar_epi64): Ditto
2846 (_mm512_maskz_permutexvar_epi32): Ditto
2847 (_mm512_permutexvar_epi32): Ditto
2848 (_mm512_mask_permutexvar_epi32): Ditto
2850 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2852 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2853 (p8_vmrgow): Likewise.
2855 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2857 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2860 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2863 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2864 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2865 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2866 using direct move instructions on ISA 2.07. Also adjust
2867 instruction length for 64-bit.
2868 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2869 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2871 2014-02-15 Alan Modra <amodra@gmail.com>
2875 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2876 find_replacement on parts of insn rtl that might be reloaded.
2878 2014-02-15 Richard Biener <rguenther@suse.de>
2880 PR tree-optimization/60183
2881 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2882 (tree_ssa_phiprop): Calculate and free post-dominators.
2884 2014-02-14 Jeff Law <law@redhat.com>
2886 PR rtl-optimization/60131
2887 * ree.c (get_extended_src_reg): New function.
2888 (combine_reaching_defs): Use it rather than assuming location of REG.
2889 (find_and_remove_re): Verify first operand of extension is
2890 a REG before adding the insns to the copy list.
2892 2014-02-14 Roland McGrath <mcgrathr@google.com>
2894 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2895 * configure: Regenerated.
2896 * config.in: Regenerated.
2897 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2898 instead of ASM_SHORT.
2900 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2901 Richard Earnshaw <rearnsha@arm.com>
2903 PR rtl-optimization/59535
2904 * lra-constraints.c (process_alt_operands): Encourage alternative
2905 when unassigned pseudo class is superset of the alternative class.
2906 (inherit_reload_reg): Don't inherit when optimizing for code size.
2907 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2908 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2909 modes not less than 4 for Thumb1.
2911 2014-02-14 Kyle McMartin <kyle@redhat.com>
2914 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2916 2014-02-14 Richard Biener <rguenther@suse.de>
2918 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2919 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2920 Do not strip INDIRECT_REFs.
2922 2014-02-14 Richard Biener <rguenther@suse.de>
2925 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2926 DECL_FUNCTION_SPECIFIC_TARGET.
2927 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2928 * tree-streamer-out.c (pack_ts_target_option): Remove.
2929 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2930 (write_ts_function_decl_tree_pointers): Do not stream
2931 DECL_FUNCTION_SPECIFIC_TARGET.
2932 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2933 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2934 (lto_input_ts_function_decl_tree_pointers): Do not stream
2935 DECL_FUNCTION_SPECIFIC_TARGET.
2937 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2939 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2940 (get_initial_def_for_induction, vectorizable_induction): Ignore
2941 debug stmts when looking for exit_phi.
2942 (vectorizable_live_operation): Fix up condition.
2944 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2946 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2947 nreverse() because it changes the content of original tree list.
2949 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2951 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2952 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2954 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2956 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2957 GNU coding standards.
2959 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2962 * dwarf2out.c (gen_subprogram_die): Don't call
2963 add_calling_convention_attribute if subr_die is old_die.
2965 2014-02-13 Sharad Singhai <singhai@google.com>
2967 * doc/optinfo.texi: Fix order of nodes.
2969 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2971 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2972 operands[2], not operands[3].
2974 2014-02-13 Richard Biener <rguenther@suse.de>
2977 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2978 mention the possibility of an in-tree build.
2979 (CLooG): Update recommended version to 0.18.1, mention the
2980 possibility of an in-tree build and clarify that the ISL
2981 bundled with CLooG does not work.
2983 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2986 * expr.c (compress_float_constant): If x is a hard register,
2987 extend into a pseudo and then move to x.
2989 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2991 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2992 caused by bad second argument to warning_at() with -mhotpatch and
2993 nested functions (e.g. with gfortran).
2995 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2997 * opts.c (option_name): Remove "enabled by default" rider.
2999 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3001 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3003 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3004 Uros Bizjak <ubizjak@gmail.com>
3007 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3008 * configure: Regenerated.
3010 2014-02-12 Richard Biener <rguenther@suse.de>
3012 * vec.c (vec_prefix::calculate_allocation): Move as
3013 inline variant to vec.h.
3014 (vec_prefix::calculate_allocation_1): New out-of-line version.
3015 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3016 (vec_prefix::m_has_auto_buf): Rename to ...
3017 (vec_prefix::m_using_auto_storage): ... this.
3018 (vec_prefix::calculate_allocation): Inline the easy cases
3019 and dispatch to calculate_allocation_1 which doesn't need the
3021 (va_heap::reserve): Use gcc_checking_assert.
3022 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3023 m_using_auto_storage.
3024 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3026 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3027 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3028 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3030 2014-02-12 Richard Biener <rguenther@suse.de>
3032 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3033 when we found a dependence.
3035 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3037 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3039 (maybe_fold_stmt): ... into this new function.
3040 * omp-low.c (lower_omp): Update comment.
3042 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3045 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3048 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3050 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3051 identifiers in comments.
3052 (cortexa53_extra_costs): Likewise.
3053 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3054 (cortexa7_extra_costs): Likewise.
3055 (cortexa12_extra_costs): Likewise.
3056 (cortexa15_extra_costs): Likewise.
3057 (v7m_extra_costs): Likewise.
3059 2014-02-12 Richard Biener <rguenther@suse.de>
3062 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3063 of posix_memalign being successful.
3064 (lower_stmt): Restrict lowering of posix_memalign to when
3065 -ftree-bit-ccp is enabled.
3067 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3069 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3071 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3073 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3075 PR rtl-optimization/60116
3076 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3077 other_insn once the combination has been validated.
3079 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3082 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3084 * ipa-devirt.c: Include demangle.h
3085 (odr_violation_reported): New static variable.
3086 (add_type_duplicate): Update odr_violations.
3087 (maybe_record_node): Add completep parameter; update it.
3088 (record_target_from_binfo): Add COMPLETEP parameter;
3089 update it as needed.
3090 (possible_polymorphic_call_targets_1): Likewise.
3091 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3092 rename FINAL to COMPLETE.
3093 (record_targets_from_bases): Sanity check we found the binfo;
3094 fix COMPLETEP updating.
3095 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3096 parameter, fix computing of COMPLETEP.
3097 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3098 at LTO time do demangling.
3099 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3100 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3102 (gimple_get_virt_method_for_binfo): Likewise.
3103 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3104 gimple_get_virt_method_for_vtable): Update prototypes.
3106 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3109 * genautomata.c (add_presence_absence): Fix typo with
3110 {final_}presence_list.
3112 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3115 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3116 for VSX/Altivec vectors that land in GPR registers.
3118 2014-02-11 Richard Henderson <rth@redhat.com>
3119 Jakub Jelinek <jakub@redhat.com>
3122 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3123 around drhs if type conversion to lacc->type is not useless.
3125 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3127 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3129 (cortex-a57.cortex-a53): Likewise.
3130 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3132 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3134 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3137 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3139 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3140 add_options_for_arm_vfp3.
3142 2014-02-11 Jeff Law <law@redhat.com>
3145 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3146 object with an undesirable mode.
3148 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3151 * config/i386/sol2-9.h: New file.
3152 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3153 *-*-solaris2.9*): Use it.
3155 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3157 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3158 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3160 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3162 * config/microblaze/microblaze.c: Extend mcpu version format
3164 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3166 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3168 2014-02-10 Richard Henderson <rth@redhat.com>
3171 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3172 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3173 ms-abi vs -mno-accumulate-outgoing-args.
3174 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3175 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3178 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3181 * cfgexpand.c (expand_asm_operands): Attach source location to
3182 ASM_INPUT rtx objects.
3183 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3185 2014-02-10 Nick Clifton <nickc@redhat.com>
3187 * config/mn10300/mn10300.c (popcount): New function.
3188 (mn10300_expand_prologue): Include saved registers in stack usage
3191 2014-02-10 Jeff Law <law@redhat.com>
3194 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3195 when changing the SET_DEST of a prior insn to avoid an input reload.
3197 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3199 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3200 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3201 -mcall-openbsd, or -mcall-linux.
3202 (CC1_ENDIAN_BIG_SPEC): Remove.
3203 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3204 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3205 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3206 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3207 and %cc1_endian_default.
3208 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3210 2014-02-10 Richard Biener <rguenther@suse.de>
3212 PR tree-optimization/60115
3213 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3214 MEM_REF handling. Properly verify that the accesses are not
3215 out of the objects bound.
3217 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3219 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3222 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3224 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3225 proper constants and fix formatting.
3226 (possible_polymorphic_call_targets): Fix formatting.
3228 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3229 Ilya Tocar <ilya.tocar@intel.com>
3231 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3232 (_mm512_loadu_epi32): Renamed into...
3233 (_mm512_loadu_si512): This.
3234 (_mm512_storeu_epi32): Renamed into...
3235 (_mm512_storeu_si512): This.
3236 (_mm512_maskz_ceil_ps): Removed.
3237 (_mm512_maskz_ceil_pd): Ditto.
3238 (_mm512_maskz_floor_ps): Ditto.
3239 (_mm512_maskz_floor_pd): Ditto.
3240 (_mm512_floor_round_ps): Ditto.
3241 (_mm512_floor_round_pd): Ditto.
3242 (_mm512_ceil_round_ps): Ditto.
3243 (_mm512_ceil_round_pd): Ditto.
3244 (_mm512_mask_floor_round_ps): Ditto.
3245 (_mm512_mask_floor_round_pd): Ditto.
3246 (_mm512_mask_ceil_round_ps): Ditto.
3247 (_mm512_mask_ceil_round_pd): Ditto.
3248 (_mm512_maskz_floor_round_ps): Ditto.
3249 (_mm512_maskz_floor_round_pd): Ditto.
3250 (_mm512_maskz_ceil_round_ps): Ditto.
3251 (_mm512_maskz_ceil_round_pd): Ditto.
3252 (_mm512_expand_pd): Ditto.
3253 (_mm512_expand_ps): Ditto.
3254 * config/i386/i386.c (ix86_builtins): Remove
3255 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3256 (bdesc_args): Ditto.
3257 * config/i386/predicates.md (const1256_operand): New.
3258 (const_1_to_2_operand): Ditto.
3259 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3260 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3261 (*avx512pf_gatherpf<mode>sf): Ditto.
3262 (avx512pf_gatherpf<mode>df): Ditto.
3263 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3264 (*avx512pf_gatherpf<mode>df): Ditto.
3265 (avx512pf_scatterpf<mode>sf): Ditto.
3266 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3267 (*avx512pf_scatterpf<mode>sf): Ditto.
3268 (avx512pf_scatterpf<mode>df): Ditto.
3269 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3270 (*avx512pf_scatterpf<mode>df): Ditto.
3271 (avx512f_expand<mode>): Removed.
3272 (<shift_insn><mode>3<mask_name>): Change predicate type.
3274 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3276 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3277 not at the end of datarefs vector use ordered_remove to avoid
3278 reordering datarefs vector.
3281 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3282 mark local addressable non-static vars as GOVD_PRIVATE
3283 instead of GOVD_LOCAL.
3284 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3285 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3289 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3290 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3291 assume_aligned or alloc_align attributes.
3292 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3293 arguments. Handle also assume_aligned and alloc_align attributes.
3294 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3295 calls to functions with assume_aligned or alloc_align attributes.
3296 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3298 2014-02-08 Terry Guo <terry.guo@arm.com>
3300 * doc/invoke.texi: Document ARM -march=armv7e-m.
3302 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3304 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3305 flag on __cilkrts_rethrow builtin.
3308 * ipa-cp.c (determine_versionability): Fail at -O0
3309 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3310 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3313 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3316 * tree-inline.c (copy_forbidden): Fail for
3317 __attribute__((optimize (0))) functions.
3319 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3321 * varpool.c: Include pointer-set.h.
3322 (varpool_remove_unreferenced_decls): Variables in other partitions
3323 will not be output; be however careful to not lose information
3326 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3328 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3329 lookup in the vtable constructor.
3331 2014-02-07 Jeff Law <law@redhat.com>
3334 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3335 define_insn_and_split.
3337 * ipa-inline.c (inline_small_functions): Fix typos.
3339 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3341 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3342 (s390_can_use_return_insn): Declare.
3343 * config/s390/s390.h (EPILOGUE_USES): Define.
3344 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3346 (s390_chunkify_start): Handle return JUMP_LABELs.
3347 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3348 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3349 (s390_can_use_return_insn): New functions.
3350 (s390_fix_long_loop_prediction): Handle conditional returns.
3351 (TARGET_SET_UP_BY_PROLOGUE): Define.
3352 * config/s390/s390.md (ANY_RETURN): New code iterator.
3353 (*creturn, *csimple_return, return, simple_return): New patterns.
3355 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3357 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3358 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3359 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3360 REG_CFA_RESTORE list when deciding not to restore a register.
3362 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3364 * config/s390/s390.c: Include tree-pass.h and context.h.
3365 (s390_early_mach): New function, split out from...
3366 (s390_emit_prologue): ...here.
3367 (pass_data_s390_early_mach): New pass structure.
3368 (pass_s390_early_mach): New class.
3369 (s390_option_override): Create and register early_mach pass.
3370 Move to end of file.
3372 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3374 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3375 to match for the exit block.
3377 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3379 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3380 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3381 Reject misaligned operands.
3383 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3385 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3387 2014-02-07 Richard Biener <rguenther@suse.de>
3390 * gimple-low.c (lower_builtin_posix_memalign): New function.
3391 (lower_stmt): Call it to lower posix_memalign in a way
3392 to make alignment info accessible.
3394 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3397 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3398 __builtin_setjmp_receiver.
3400 2014-02-07 Richard Biener <rguenther@suse.de>
3403 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3404 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3405 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3406 Handle BUILT_IN_POSIX_MEMALIGN.
3407 (find_func_clobbers): Likewise.
3408 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3409 (call_may_clobber_ref_p_1): Likewise.
3411 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3414 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3417 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3420 * lto-cgraph.c (lto_output_node): Use
3421 symtab_get_symbol_partitioning_class.
3422 (lto_output_varpool_node): likewise.
3423 (symtab_get_symbol_partitioning_class): Move here from
3425 * cgraph.h (symbol_partitioning_class): Likewise.
3426 (symtab_get_symbol_partitioning_class): Declare.
3428 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3430 * ggc.h (ggc_internal_cleared_alloc): New macro.
3431 * vec.h (vec_safe_copy): Handle memory stats.
3432 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3433 * target-globals.c (save_target_globals): Likewise.
3435 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3438 * expr.c (emit_move_resolve_push): Export; be bit more selective
3439 on when to clear alias set.
3440 * expr.h (emit_move_resolve_push): Declare.
3441 * function.h (struct function): Add tail_call_marked.
3442 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3443 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3444 * config/i386/i386.md (TImode move expander): De not call
3446 (FP push expanders): Preserve memory attributes.
3447 * config/i386/sse.md (push<mode>1): Remove.
3448 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3449 (ix86_expand_push): Remove.
3450 * config/i386/mmx.md (push<mode>1): Remove.
3452 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3454 PR rtl-optimization/60030
3455 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3456 lopart with paradoxical subreg before shifting it up by hprec.
3458 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3460 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3461 Remove extra newline at end of file.
3462 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3463 (arm_issue_rate): Handle cortexa57.
3464 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3465 (cortex-a57.cortex-a53): Likewise.
3467 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3470 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3471 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3473 (arm_expand_prologue): Adjust all callers.
3474 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3475 info, registers also at the lowest numbered registers side. Use
3476 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3480 * var-tracking.c (adjust_mems): Before adding a SET to
3481 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3483 2014-02-06 Alan Modra <amodra@gmail.com>
3486 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3487 change SDmode to DDmode when lra_in_progress.
3489 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3492 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3493 free_data_ref on the dr first, and before goto again also set dr
3494 to the next dr. For simd_lane_access, free old datarefs[i] before
3495 overwriting it. For get_vectype_for_scalar_type failure, don't
3496 free_data_ref if simd_lane_access.
3498 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3501 * tree.h (opts_for_fn): New inline function.
3502 (opt_for_fn): Define.
3503 * config/i386/i386.c (ix86_function_regparm): Use
3504 opt_for_fn (decl, optimize) instead of optimize.
3506 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3508 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3509 for SYMBOL_REF in large memory model.
3511 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3513 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3515 (cortex-a57): Likewise.
3516 (cortex-a57.cortex-a53): Likewise.
3518 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3519 Kugan Vivekanandarajah <kuganv@linaro.org>
3521 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3523 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3525 2014-02-06 Richard Biener <rguenther@suse.de>
3527 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3528 set_loop_copy and initialize_original_copy_tables.
3530 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3532 * config/aarch64/aarch64-simd.md
3533 (aarch64_ashr_simddi): Change QI to SI.
3535 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3536 Jakub Jelinek <jakub@redhat.com>
3539 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3542 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3544 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3545 CODE_FOR_altivec_vpku[hw]um to
3546 CODE_FOR_altivec_vpku[hw]um_direct.
3547 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3548 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3549 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3550 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3552 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3554 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3555 generation for -maltivec=be.
3556 (altivec_vsumsws): Simplify redundant test.
3558 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3560 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3561 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3562 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3563 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3564 gen_altivec_vpkuwum.
3565 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3567 (altivec_vpks<VI_char>ss): Likewise.
3568 (altivec_vpks<VI_char>us): Likewise.
3569 (altivec_vpku<VI_char>us): Likewise.
3570 (altivec_vpku<VI_char>um): Likewise.
3571 (altivec_vpku<VI_char>um_direct): New (copy of
3572 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3574 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3575 target is little endian and -maltivec=be is not specified.
3576 (*altivec_vupkhs<VU_char>_direct): New (copy of
3577 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3578 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3579 target is little endian and -maltivec=be is not specified.
3580 (*altivec_vupkls<VU_char>_direct): New (copy of
3581 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3582 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3583 little endian and -maltivec=be is not specified.
3584 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3585 little endian and -maltivec=be is not specified.
3587 2014-02-05 Richard Henderson <rth@redhat.com>
3590 * combine-stack-adj.c: Revert r206943.
3591 * sched-int.h (struct deps_desc): Add last_args_size.
3592 * sched-deps.c (init_deps): Initialize it.
3593 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3594 contain REG_ARGS_SIZE notes.
3596 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3598 * lto-cgraph.c (asm_nodes_output): Make global.
3599 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3600 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3601 (driver_handle_option): Handle OPT_fwpa.
3603 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3606 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3607 a comment typo and formatting issue. If odr_hash hasn't been
3608 created, return vNULL and set *completep to false.
3611 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3612 bb with no successors.
3614 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3617 * doc/invoke.texi (-march): Clarify documentation for ARM.
3621 2014-02-05 Richard Biener <rguenther@suse.de>
3623 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3624 when not vectorizing because of too many alias checks.
3625 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3626 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3628 2014-02-05 Nick Clifton <nickc@redhat.com>
3630 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3631 accept extended registers in any mode when compiling for the MN10300.
3633 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3635 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3636 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3637 sanitization attributes.
3638 (can_inline_edge_p): Likewise.
3639 (sanitize_attrs_match_for_inline_p): New function.
3641 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3643 * ipa-prop.c (detect_type_change): Shor circuit testing of
3644 type changes on THIS pointer.
3646 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3649 * config/pa/pa.c (legitimize_tls_address): Return original address
3650 if not passed a SYMBOL_REF rtx.
3651 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3653 (pa_emit_move_sequence): Simplify TLS source operands.
3654 (pa_legitimate_constant_p): Reject all TLS constants.
3655 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3656 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3658 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3660 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3661 groups when we know they are controlled by LTO.
3662 * varasm.c (default_binds_local_p_1): If object is in other partition,
3663 it will be resolved locally.
3665 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3667 * config/host-linux.c (linux_gt_pch_use_address): Don't
3668 use SSIZE_MAX because it is not always defined.
3670 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3673 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3674 threshold for pseudo splitting.
3675 (update_ebb_live_info): Process call argument hard registers and
3676 hard registers from insn definition too.
3677 (max_small_class_regs_num): New constant.
3678 (inherit_in_ebb): Update live hard regs through EBBs. Update
3679 reloads_num only for small register classes. Don't split for
3682 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3685 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3688 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3690 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3693 2014-02-04 Marek Polacek <polacek@redhat.com>
3695 * gdbinit.in (pel): Define.
3697 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3699 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3702 2014-02-04 Richard Biener <rguenther@suse.de>
3705 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3706 in function context local.
3707 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3708 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3709 similar to LTO_imported_decl_ref.
3711 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3713 PR tree-optimization/60002
3714 * cgraphclones.c (build_function_decl_skip_args): Clear
3717 PR tree-optimization/60023
3718 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3719 false to gsi_replace.
3720 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3721 has been in some EH region and vec_stmt could throw, add
3722 vec_stmt into the same EH region.
3723 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3724 has no lhs, ignore it.
3725 * internal-fn.c (expand_MASK_LOAD): Likewise.
3728 * tree-inline.c (copy_forbidden): Fail for
3729 __attribute__((optimize (0))) functions.
3732 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3733 is set, copy one less argument.
3734 (expand_simd_clones): Don't subtract clone_info->inbranch
3735 from simd_clone_struct_alloc argument.
3737 PR rtl-optimization/57915
3738 * recog.c (simplify_while_replacing): If all unary/binary/relational
3739 operation arguments are constant, attempt to simplify those.
3742 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3743 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3745 2014-02-04 Richard Biener <rguenther@suse.de>
3747 PR tree-optimization/60012
3748 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3749 TBAA disambiguation to all DDRs.
3751 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3754 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3755 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3757 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3760 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3762 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3764 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3765 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3767 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3770 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3771 to figure out targets of polymorphic calls with known decl.
3772 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3773 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3774 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3775 (get_polymorphic_call_info): ... here.
3776 (get_polymorphic_call_info_from_invariant): New function.
3778 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3780 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3781 lookup via vtable pointer; check for type consistency
3782 and turn inconsitent facts into UNREACHABLE.
3783 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3784 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3785 type inconsistent querries; return UNREACHABLE instead.
3787 2014-02-03 Richard Henderson <rth@twiddle.net>
3790 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3791 already processed this node.
3792 (normalize_one_pred_1): Pass along mark_set.
3793 (normalize_one_pred): Create and destroy a pointer_set_t.
3794 (normalize_one_pred_chain): Likewise.
3796 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3798 PR gcov-profile/58602
3799 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3801 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3804 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3805 -fno-devirtualize; try to devirtualize by the knowledge of
3806 virtual table pointer given by aggregate propagation.
3807 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3808 (ipa_print_node_jump_functions): Dump also offset that
3809 is relevant for polymorphic calls.
3810 (determine_known_aggregate_parts): Add arg_type parameter; use it
3811 instead of determining the type from pointer type.
3812 (ipa_compute_jump_functions_for_edge): Update call of
3813 determine_known_aggregate_parts.
3814 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3815 (gimple_get_virt_method_for_binfo): ... here; simplify using
3816 vtable_pointer_value_to_vtable.
3817 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3818 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3819 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3820 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3822 (vtable_pointer_value_to_binfo): ... here.
3823 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3825 2014-02-03 Teresa Johnson <tejohnson@google.com>
3827 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3828 redef of outer loop index variable.
3830 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3834 * doc/extend.texi (Function Attributes): Typo.
3836 2014-02-03 Cong Hou <congh@google.com>
3838 PR tree-optimization/60000
3839 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3840 if the vectorized statement is a store. A store statement can only
3841 appear at the end of pattern statements.
3843 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3845 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3846 (ix86_option_override_internal): Default long double to 64-bit for
3847 32-bit Bionic and to 128-bit for 64-bit Bionic.
3849 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3850 TARGET_LONG_DOUBLE_128 is true.
3851 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3853 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3854 (mlong-double-64): Negate -mlong-double-128.
3855 (mlong-double-128): New option.
3857 * config/i386/i386-c.c (ix86_target_macros): Define
3858 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3860 * doc/invoke.texi: Document -mlong-double-128.
3862 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3864 PR rtl-optimization/60024
3865 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3867 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3869 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3871 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3873 PR rtl-optimization/57662
3874 * sel-sched.c (code_motion_path_driver): Do not mark already not
3875 existing blocks in the visiting bitmap.
3877 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3879 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3880 on the insn being emitted.
3882 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3883 Will Deacon <will.deacon@arm.com>
3885 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3887 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3889 * config/arm/arm-tables.opt: Regenerate.
3891 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3893 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3894 for vector types other than V16QImode.
3895 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3896 define_expand, and call altivec_expand_vec_perm_le when producing
3897 code with little endian element order.
3898 (*altivec_vperm_<mode>_internal): New insn having previous
3899 behavior of altivec_vperm_<mode>.
3900 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3901 altivec_expand_vec_perm_le when producing code with little endian
3903 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3904 behavior of altivec_vperm_<mode>_uns.
3906 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3908 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3909 (altivec_vsumsws): Add handling for -maltivec=be with a little
3911 (altivec_vsumsws_direct): New.
3912 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3913 gen_altivec_vsumsws.
3915 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3917 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3918 vtable_pointer_value_to_binfo): New functions.
3919 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3920 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3922 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3924 * config/nios2/nios2.md (load_got_register): Initialize GOT
3925 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3926 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3928 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3930 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3931 preserverd by passthrough, do not propagate the type.
3933 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3935 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3936 (mips_atomic_assign_expand_fenv): New function.
3937 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3939 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3941 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3942 (__builtin_mips_set_fcsr): Likewise.
3943 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3944 MIPS_USI_FTYPE_VOID.
3945 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3946 (mips16_expand_set_fcsr): Likewise.
3947 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3948 (mips16_set_fcsr_stub): Likewise.
3949 (mips16_get_fcsr_one_only_stub): New class.
3950 (mips16_set_fcsr_one_only_stub): Likewise.
3951 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3952 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3953 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3954 (hard_float): New availability predicate.
3955 (mips_builtins): Add get_fcsr and set_fcsr.
3956 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3957 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3958 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3959 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3960 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3963 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3965 * config/mips/mips.c (mips_one_only_stub): New class.
3966 (mips_need_mips16_rdhwr_p): Replace with...
3967 (mips16_rdhwr_stub): ...this new variable.
3968 (mips16_stub_call_address): New function.
3969 (mips16_rdhwr_one_only_stub): New class.
3970 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3971 (mips_output_mips16_rdhwr): Delete.
3972 (mips_finish_stub): New function.
3973 (mips_code_end): Use it to handle rdhwr stubs.
3975 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3978 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3979 when calculating size of integer atomic types.
3981 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3983 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3985 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3987 PR tree-optimization/60003
3988 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3989 * profile.c (branch_prob): Use gimple_call_builtin_p
3990 to check for BUILT_IN_SETJMP_RECEIVER.
3991 * tree-inline.c (copy_bb): Call notice_special_calls.
3993 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3996 * lra-constraints.c (process_alt_operands): Update reload_sum only
3999 2014-01-31 Richard Henderson <rth@redhat.com>
4002 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4003 until after else_eh is processed.
4005 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4007 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4008 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4009 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4010 in smmintrin.h, remove them.
4011 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4012 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4013 * config/i386/i386.md (ROUND_SAE): Fix value.
4014 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4015 (const48_operand): New.
4016 * config/i386/subst.md (round), (round_expand): Use
4017 const_4_or_8_to_11_operand.
4018 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4020 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4022 * config/i386/constraints.md (Yk): Swap meaning with k.
4023 * config/i386/i386.md (movhi_internal): Change Yk to k.
4024 (movqi_internal): Ditto.
4025 (*k<logic><mode>): Ditto.
4028 (kandn<mode>): Ditto.
4029 (*<code>hi_1): Ditto.
4030 (*<code>qi_1): Ditto.
4031 (kxnor<mode>): Ditto.
4032 (kortestzhi): Ditto.
4033 (kortestchi): Ditto.
4035 (*one_cmplhi2_1): Ditto.
4036 (*one_cmplqi2_1): Ditto.
4037 * config/i386/sse.md (): Change k to Yk.
4038 (avx512f_load<mode>_mask): Ditto.
4039 (avx512f_blendm<mode>): Ditto.
4040 (avx512f_store<mode>_mask): Ditto.
4041 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4042 (avx512f_storedqu<mode>_mask): Ditto.
4043 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4045 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4046 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4047 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4048 (avx512f_maskcmp<mode>3): Ditto.
4049 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4050 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4051 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4052 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4053 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4054 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4055 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4056 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4057 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4058 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4059 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4060 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4061 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4062 (vec_extract_lo_<mode>_maskm): Ditto.
4063 (vec_extract_hi_<mode>_maskm): Ditto.
4064 (avx512f_vternlog<mode>_mask): Ditto.
4065 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4066 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4067 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4068 (avx512f_<code>v8div16qi2_mask): Ditto.
4069 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4070 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4071 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4072 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4073 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4074 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4075 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4076 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4077 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4078 (avx512cd_maskb_vec_dupv8di): Ditto.
4079 (avx512cd_maskw_vec_dupv16si): Ditto.
4080 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4081 (avx512f_vpermi2var<mode>3_mask): Ditto.
4082 (avx512f_vpermi2var<mode>3_mask): Ditto.
4083 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4084 (*avx512f_gathersi<mode>): Ditto.
4085 (*avx512f_gathersi<mode>_2): Ditto.
4086 (*avx512f_gatherdi<mode>): Ditto.
4087 (*avx512f_gatherdi<mode>_2): Ditto.
4088 (*avx512f_scattersi<mode>): Ditto.
4089 (*avx512f_scatterdi<mode>): Ditto.
4090 (avx512f_compress<mode>_mask): Ditto.
4091 (avx512f_compressstore<mode>_mask): Ditto.
4092 (avx512f_expand<mode>_mask): Ditto.
4093 * config/i386/subst.md (mask): Change k to Yk.
4094 (mask_scalar_merge): Ditto.
4097 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4099 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4101 2014-01-31 Richard Biener <rguenther@suse.de>
4104 * builtins.c (fold_builtin_memory_op): Make sure to not
4105 use a floating-point mode or a boolean or enumeral type for
4108 2014-01-30 DJ Delorie <dj@redhat.com>
4110 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4111 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4112 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4113 whenever main() has an epilogue.
4115 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4117 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4118 unused variable "field".
4119 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4120 (vsx_mergeh_<mode>): Likewise.
4121 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4122 (altivec_vmrghh): Likewise.
4123 (altivec_vmrghw): Likewise.
4124 (altivec_vmrglb): Likewise.
4125 (altivec_vmrglh): Likewise.
4126 (altivec_vmrglw): Likewise.
4127 (altivec_vspltb): Add missing uses.
4128 (altivec_vsplth): Likewise.
4129 (altivec_vspltw): Likewise.
4130 (altivec_vspltsf): Likewise.
4132 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4135 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4136 frame related instructions.
4138 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4140 PR rtl-optimization/59959
4141 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4142 any reload of register whose subreg is invalid.
4144 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4146 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4147 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4148 Add missing return type - void.
4150 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4152 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4153 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4154 remove element index adjustment for endian (now handled in vsx.md
4156 (altivec_expand_vec_perm_const): Use
4157 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4158 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4159 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4160 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4161 define_expand and a new define_insn *altivec_vspltb_internal;
4162 adjust for -maltivec=be on a little endian target.
4163 (altivec_vspltb_direct): New.
4164 (altivec_vsplth): Divide into a define_expand and a new
4165 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4166 little endian target.
4167 (altivec_vsplth_direct): New.
4168 (altivec_vspltw): Divide into a define_expand and a new
4169 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4170 little endian target.
4171 (altivec_vspltw_direct): New.
4172 (altivec_vspltsf): Divide into a define_expand and a new
4173 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4174 a little endian target.
4176 2014-01-30 Richard Biener <rguenther@suse.de>
4178 PR tree-optimization/59993
4179 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4180 can propagate form the earlier stmt and avoid the transform
4181 when the intermediate result is needed.
4183 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4185 * README.Portability: Fix typo.
4187 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4189 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4190 comparison_operator with ordered_comparison_operator.
4192 2014-01-30 Nick Clifton <nickc@redhat.com>
4194 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4195 Rename to mn10300_store_multiple_regs.
4196 * config/mn10300/mn10300.c: Likewise.
4197 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4198 store_multiple_regs.
4199 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4200 Call mn10300_store_multiple_regs.
4202 2014-01-30 Nick Clifton <nickc@redhat.com>
4203 DJ Delorie <dj@redhat.com>
4205 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4206 %fp 2 to keep registers after it properly word-aligned.
4207 (rl78_alloc_physical_registers_umul): Handle the case where both
4208 input operands are the same.
4210 2014-01-30 Richard Biener <rguenther@suse.de>
4212 PR tree-optimization/59903
4213 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4216 2014-01-30 Jason Merrill <jason@redhat.com>
4219 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4222 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4224 2014-01-30 Richard Biener <rguenther@suse.de>
4226 PR tree-optimization/59951
4227 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4229 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4232 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4233 SFmode to DFmode case.
4235 2014-01-29 DJ Delorie <dj@redhat.com>
4237 * config/msp430/msp430.opt (-minrt): New.
4238 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4240 (ENDFILE_SPEC): Likewise.
4242 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4244 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4245 (estimate_function_body_sizes): Use it.
4247 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4250 * dwarf2out.c (is_cxx_auto): New.
4251 (is_base_type): Use it.
4252 (gen_type_die_with_usage): Likewise.
4254 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4256 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4257 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4258 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4259 -maltivec=be with LE targets.
4260 (vsx_mergeh_<mode>): Likewise.
4261 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4262 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4263 (altivec_vmrghb): Replace with define_expand and new
4264 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4265 (altivec_vmrghb_direct): New define_insn.
4266 (altivec_vmrghh): Replace with define_expand and new
4267 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4268 (altivec_vmrghh_direct): New define_insn.
4269 (altivec_vmrghw): Replace with define_expand and new
4270 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4271 (altivec_vmrghw_direct): New define_insn.
4272 (*altivec_vmrghsf): Adjust for endianness.
4273 (altivec_vmrglb): Replace with define_expand and new
4274 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4275 (altivec_vmrglb_direct): New define_insn.
4276 (altivec_vmrglh): Replace with define_expand and new
4277 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4278 (altivec_vmrglh_direct): New define_insn.
4279 (altivec_vmrglw): Replace with define_expand and new
4280 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4281 (altivec_vmrglw_direct): New define_insn.
4282 (*altivec_vmrglsf): Adjust for endianness.
4283 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4284 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4285 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4286 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4287 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4288 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4289 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4290 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4292 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4294 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4295 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4298 2014-01-29 Richard Biener <rguenther@suse.de>
4300 PR tree-optimization/58742
4301 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4302 associate_pointerplus_align.
4303 (associate_pointerplus_diff): New function.
4304 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4305 and associate_pointerplus_diff.
4307 2014-01-29 Richard Biener <rguenther@suse.de>
4309 * lto-streamer.h (LTO_major_version): Bump to 3.
4310 (LTO_minor_version): Reset to 0.
4312 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4314 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4315 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4316 (arm_file_start): Generate correct asm header for armv7ve.
4317 * config/arm/bpabi.h: Add multilib support for armv7ve.
4318 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4319 and cortex-a15 to armv7ve.
4320 * config/arm/t-aprofile: Add multilib support for armv7ve.
4321 * doc/invoke.texi: Document -march=armv7ve.
4323 2014-01-29 Richard Biener <rguenther@suse.de>
4325 PR tree-optimization/58742
4326 * tree-ssa-forwprop.c (associate_plusminus): Return true
4327 if we changed sth, defer EH cleanup to ...
4328 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4329 (simplify_mult): New function.
4331 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4334 PR tree-optimization/59920
4335 * tree.c (build_common_builtin_nodes): Remove
4336 __builtin_setjmp_dispatcher initialization.
4337 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4338 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4339 instead of gsi_after_labels + manually skipping debug stmts.
4340 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4341 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4342 * tree-inline.c (copy_edges_for_bb): Remove
4343 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4344 argument. Ignore computed_goto_p stmts. Don't call
4345 make_abnormal_goto_edges. If a call might need abnormal edges
4346 for non-local gotos, see if it already has an edge to
4347 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4348 with true argument, don't do anything then, otherwise add
4349 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4350 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4352 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4353 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4354 (lower_stmt): Don't set data->calls_builtin_setjmp.
4355 (lower_builtin_setjmp): Adjust comment.
4356 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4357 * tree-cfg.c (found_computed_goto): Remove.
4358 (factor_computed_gotos): Remove.
4359 (make_goto_expr_edges): Return bool, true for computed gotos.
4360 Don't call make_abnormal_goto_edges.
4361 (build_gimple_cfg): Don't set found_computed_goto, don't call
4362 factor_computed_gotos.
4363 (computed_goto_p): No longer static.
4364 (make_blocks): Don't set found_computed_goto.
4365 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4366 (make_edges): If make_goto_expr_edges returns true, push bb
4367 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4368 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4369 vector. Record mapping between bbs and OpenMP regions if there
4370 are any, adjust make_gimple_omp_edges caller. Call
4371 handle_abnormal_edges.
4372 (make_abnormal_goto_edges): Remove.
4373 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4374 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4375 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4376 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4377 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4378 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4379 filling *region also set *region_idx to (*region)->entry->index.
4382 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4383 For REGs set ORIGINAL_REGNO.
4385 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4387 * doc/md.texi: Mention that a target shouldn't implement
4388 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4391 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4393 PR tree-optimization/59594
4394 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4395 a copy of the datarefs vector rather than the vector itself.
4397 2014-01-28 Jason Merrill <jason@redhat.com>
4400 * dwarf2out.c (auto_die): New static.
4401 (gen_type_die_with_usage): Handle C++1y 'auto'.
4402 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4405 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4408 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4409 (SPEC_X32): Likewise.
4410 (SPEC_64): Likewise.
4411 * config/i386/i386.c (ix86_option_override_internal): Turn off
4412 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4414 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4415 * config/i386/i386.h (TARGET_16BIT): New macro.
4416 (TARGET_16BIT_P): Likewise.
4417 * config/i386/i386.opt: Add m16.
4418 * doc/invoke.texi: Document -m16.
4420 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4422 PR preprocessor/59935
4423 * input.c (location_get_source_line): Bail out on when line number
4424 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4426 2014-01-28 Richard Biener <rguenther@suse.de>
4428 PR tree-optimization/58742
4429 * tree-ssa-forwprop.c (associate_plusminus): Handle
4430 pointer subtraction of the form (T)(P + A) - (T)P.
4432 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4434 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4437 2014-01-28 Richard Biener <rguenther@suse.de>
4440 2014-01-28 Richard Biener <rguenther@suse.de>
4442 PR rtl-optimization/45364
4443 PR rtl-optimization/59890
4444 * var-tracking.c (local_get_addr_clear_given_value): Handle
4445 already cleared slot.
4446 (val_reset): Handle not allocated local_get_addr_cache.
4447 (vt_find_locations): Use post-order on the inverted CFG.
4449 2014-01-28 Richard Biener <rguenther@suse.de>
4451 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4453 2014-01-28 Richard Biener <rguenther@suse.de>
4455 PR rtl-optimization/45364
4456 PR rtl-optimization/59890
4457 * var-tracking.c (local_get_addr_clear_given_value): Handle
4458 already cleared slot.
4459 (val_reset): Handle not allocated local_get_addr_cache.
4460 (vt_find_locations): Use post-order on the inverted CFG.
4462 2014-01-28 Alan Modra <amodra@gmail.com>
4464 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4465 * configure.ac <recursive call for build != host>: Define
4466 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4467 and LD_FOR_BUILD too.
4468 * configure: Regenerate.
4470 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4472 * config/i386/i386.c (get_builtin_code_for_version): Separate
4473 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4474 Broadwell from Haswell.
4476 2014-01-27 Steve Ellcey <sellcey@mips.com>
4478 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4479 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4480 * config/mips/mips.c (mips_option_override): Change setting
4482 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4483 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4484 Change from Mask to Var.
4486 2014-01-27 Jeff Law <law@redhat.com>
4488 * ipa-inline.c (inline_small_functions): Fix typo.
4490 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4492 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4493 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4494 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4495 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4496 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4497 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4498 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4499 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4500 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4501 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4502 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4503 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4504 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4505 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4506 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4507 (_mm512_storeu_epi64): Ditto.
4508 (_mm512_cmpge_epi32_mask): Ditto.
4509 (_mm512_cmpge_epu32_mask): Ditto.
4510 (_mm512_cmpge_epi64_mask): Ditto.
4511 (_mm512_cmpge_epu64_mask): Ditto.
4512 (_mm512_cmple_epi32_mask): Ditto.
4513 (_mm512_cmple_epu32_mask): Ditto.
4514 (_mm512_cmple_epi64_mask): Ditto.
4515 (_mm512_cmple_epu64_mask): Ditto.
4516 (_mm512_cmplt_epi32_mask): Ditto.
4517 (_mm512_cmplt_epu32_mask): Ditto.
4518 (_mm512_cmplt_epi64_mask): Ditto.
4519 (_mm512_cmplt_epu64_mask): Ditto.
4520 (_mm512_cmpneq_epi32_mask): Ditto.
4521 (_mm512_cmpneq_epu32_mask): Ditto.
4522 (_mm512_cmpneq_epi64_mask): Ditto.
4523 (_mm512_cmpneq_epu64_mask): Ditto.
4524 (_mm512_expand_pd): Ditto.
4525 (_mm512_expand_ps): Ditto.
4526 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4527 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4528 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4529 * config/i386/i386.c (ix86_builtins): Add
4530 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4531 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4532 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4533 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4534 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4535 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4536 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4537 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4538 IX86_BUILTIN_PMOVUSQW512_MEM.
4539 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4540 __builtin_ia32_pmovsqd512mem_mask,
4541 __builtin_ia32_pmovqd512mem_mask,
4542 __builtin_ia32_pmovusqw512mem_mask,
4543 __builtin_ia32_pmovsqw512mem_mask,
4544 __builtin_ia32_pmovqw512mem_mask,
4545 __builtin_ia32_pmovusdw512mem_mask,
4546 __builtin_ia32_pmovsdw512mem_mask,
4547 __builtin_ia32_pmovdw512mem_mask,
4548 __builtin_ia32_pmovqb512mem_mask,
4549 __builtin_ia32_pmovusqb512mem_mask,
4550 __builtin_ia32_pmovsqb512mem_mask,
4551 __builtin_ia32_pmovusdb512mem_mask,
4552 __builtin_ia32_pmovsdb512mem_mask,
4553 __builtin_ia32_pmovdb512mem_mask.
4554 (bdesc_args): Add __builtin_ia32_expanddf512,
4555 __builtin_ia32_expandsf512.
4556 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4557 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4558 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4559 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4560 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4561 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4562 (avx512f_<code>v8div16qi2_mask_store): This.
4563 (avx512f_expand<mode>): New.
4565 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4567 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4569 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4570 (_mm512_prefetch_i32scatter_pd): Ditto.
4571 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4572 (_mm512_prefetch_i64scatter_pd): Ditto.
4573 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4574 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4575 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4576 (_mm512_prefetch_i32scatter_ps): Ditto.
4577 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4578 (_mm512_prefetch_i64scatter_ps): Ditto.
4579 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4580 * config/i386/i386-builtin-types.def: Define
4581 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4582 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4583 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4584 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4585 IX86_BUILTIN_SCATTERPFQPD.
4586 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4587 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4588 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4589 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4590 __builtin_ia32_scatterpfqps.
4591 (ix86_expand_builtin): Expand new built-ins.
4592 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4593 fix memory access data type.
4594 (*avx512pf_gatherpf<mode>_mask): Ditto.
4595 (*avx512pf_gatherpf<mode>): Ditto.
4596 (avx512pf_scatterpf<mode>): Ditto.
4597 (*avx512pf_scatterpf<mode>_mask): Ditto.
4598 (*avx512pf_scatterpf<mode>): Ditto.
4599 (GATHER_SCATTER_SF_MEM_MODE): New.
4600 (avx512pf_gatherpf<mode>df): Ditto.
4601 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4602 (*avx512pf_scatterpf<mode>df): Ditto.
4604 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4607 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4608 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4611 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4613 * common/config/arm/arm-common.c
4614 (arm_rewrite_mcpu): Handle multiple names.
4616 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4618 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4620 * gimple-builder.h (create_gimple_tmp): Delete.
4622 2014-01-27 Christian Bruel <christian.bruel@st.com>
4624 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4627 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4629 * config/pa/pa.md (call): Generate indirect long calls to non-local
4630 functions when outputing 32-bit code.
4631 (call_value): Likewise except for special call to buggy powf function.
4633 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4634 portable runtime and PIC indirect calls.
4635 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4636 and PIC call sequences. Use ldo instead of blr to set return register
4637 in PIC call sequence.
4639 2014-01-25 Walter Lee <walt@tilera.com>
4641 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4642 avoid clobbering a live register.
4644 2014-01-25 Walter Lee <walt@tilera.com>
4646 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4647 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4648 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4649 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4651 2014-01-25 Walter Lee <walt@tilera.com>
4653 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4654 arguments on even registers.
4655 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4657 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4658 (BIGGEST_ALIGNMENT): Ditto.
4659 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4661 2014-01-25 Walter Lee <walt@tilera.com>
4663 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4664 insns before bundling.
4665 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4667 2014-01-25 Walter Lee <walt@tilera.com>
4669 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4670 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4671 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4673 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4675 * config/mips/constraints.md (kl): Delete.
4676 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4677 define expands, using...
4678 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4679 instructions for MIPS16.
4680 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4681 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4683 2014-01-25 Walter Lee <walt@tilera.com>
4685 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4689 2014-01-25 Walter Lee <walt@tilera.com>
4691 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4692 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4694 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4696 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4699 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4701 * print-rtl.c (in_call_function_usage): New var.
4702 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4703 EXPR_LIST mode as mode and not as reg note name.
4706 * cfgloopmanip.c (copy_loop_info): If
4707 loop->warned_aggressive_loop_optimizations, make sure
4708 the flag is set in target loop too.
4710 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4712 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4714 * builtins.def: Likewise.
4715 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4716 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4717 * ira.c (ira_setup_eliminable_regset): Likewise.
4718 * omp-low.c (gate_expand_omp): Likewise.
4719 (execute_lower_omp): Likewise.
4720 (diagnose_sb_0): Likewise.
4721 (gate_diagnose_omp_blocks): Likewise.
4722 (simd_clone_clauses_extract): Likewise.
4725 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4727 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4728 correction for little endian...
4729 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4732 2014-01-24 Jeff Law <law@redhat.com>
4734 PR tree-optimization/59919
4735 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4736 for non-returning calls.
4738 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4740 * common/config/aarch64/aarch64-common.c
4741 (aarch64_rewrite_mcpu): Handle multiple names.
4742 * config/aarch64/aarch64.h
4743 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4745 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4747 * input.c (add_file_to_cache_tab): Handle the case where fopen
4750 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4753 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4754 from push operand if code of push isn't PRE_DEC.
4756 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4759 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4760 -mquad-memory-atomic. Update -mquad-memory documentation to say
4761 it is only used for non-atomic loads/stores.
4763 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4764 -mquad-memory or -mquad-memory-atomic switches.
4766 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4767 -mquad-memory-atomic to ISA 2.07 support.
4769 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4770 to separate support of normal quad word memory operations (ldq, stq)
4771 from the atomic quad word memory operations.
4773 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4774 support to separate non-atomic quad word operations from atomic
4775 quad word operations. Disable non-atomic quad word operations in
4776 little endian mode so that we don't have to swap words after the
4777 load and before the store.
4778 (quad_load_store_p): Add comment about atomic quad word support.
4779 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4780 options printed with -mdebug=reg.
4782 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4783 -mquad-memory-atomic as the test for whether we have quad word
4784 atomic instructions.
4785 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4786 or -mp8-vector are used, allow byte/half-word atomic operations.
4788 * config/rs6000/sync.md (load_lockedti): Insure that the address
4789 is a proper indexed or indirect address for the lqarx instruction.
4790 On little endian systems, swap the hi/lo registers after the lqarx
4792 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4793 insure the address is valid for the lqarx instruction.
4794 (store_conditionalti): Insure that the address is a proper indexed
4795 or indirect address for the stqcrx. instruction. On little endian
4796 systems, swap the hi/lo registers before doing the stqcrx.
4798 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4799 insure the address is valid for the stqcrx. instruction.
4801 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4802 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4803 type of quad memory support is available.
4805 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4808 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4809 there is a danger of looping.
4811 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4813 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4814 force flag_ira_loop_pressure if set via command line.
4816 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4818 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4819 (ashr_simd): New builtin handling DI mode.
4820 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4821 (aarch64_sshr_simddi): New match pattern.
4822 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4823 (vshrd_n_s64): Likewise.
4824 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4826 2014-01-23 Nick Clifton <nickc@redhat.com>
4828 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4829 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4830 favour of mcu specific scripts.
4831 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4834 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4835 Alex Velenko <Alex.Velenko@arm.com>
4837 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4838 (vaddv_s16): Likewise.
4839 (vaddv_s32): Likewise.
4840 (vaddv_u8): Likewise.
4841 (vaddv_u16): Likewise.
4842 (vaddv_u32): Likewise.
4843 (vaddvq_s8): Likewise.
4844 (vaddvq_s16): Likewise.
4845 (vaddvq_s32): Likewise.
4846 (vaddvq_s64): Likewise.
4847 (vaddvq_u8): Likewise.
4848 (vaddvq_u16): Likewise.
4849 (vaddvq_u32): Likewise.
4850 (vaddvq_u64): Likewise.
4851 (vaddv_f32): Likewise.
4852 (vaddvq_f32): Likewise.
4853 (vaddvq_f64): Likewise.
4854 (vmaxv_f32): Likewise.
4855 (vmaxv_s8): Likewise.
4856 (vmaxv_s16): Likewise.
4857 (vmaxv_s32): Likewise.
4858 (vmaxv_u8): Likewise.
4859 (vmaxv_u16): Likewise.
4860 (vmaxv_u32): Likewise.
4861 (vmaxvq_f32): Likewise.
4862 (vmaxvq_f64): Likewise.
4863 (vmaxvq_s8): Likewise.
4864 (vmaxvq_s16): Likewise.
4865 (vmaxvq_s32): Likewise.
4866 (vmaxvq_u8): Likewise.
4867 (vmaxvq_u16): Likewise.
4868 (vmaxvq_u32): Likewise.
4869 (vmaxnmv_f32): Likewise.
4870 (vmaxnmvq_f32): Likewise.
4871 (vmaxnmvq_f64): Likewise.
4872 (vminv_f32): Likewise.
4873 (vminv_s8): Likewise.
4874 (vminv_s16): Likewise.
4875 (vminv_s32): Likewise.
4876 (vminv_u8): Likewise.
4877 (vminv_u16): Likewise.
4878 (vminv_u32): Likewise.
4879 (vminvq_f32): Likewise.
4880 (vminvq_f64): Likewise.
4881 (vminvq_s8): Likewise.
4882 (vminvq_s16): Likewise.
4883 (vminvq_s32): Likewise.
4884 (vminvq_u8): Likewise.
4885 (vminvq_u16): Likewise.
4886 (vminvq_u32): Likewise.
4887 (vminnmv_f32): Likewise.
4888 (vminnmvq_f32): Likewise.
4889 (vminnmvq_f64): Likewise.
4891 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4893 * config/aarch64/aarch64-simd.md
4894 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4895 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4896 (*aarch64_mul3_elt<mode>): Likewise.
4897 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4898 (*aarch64_mul3_elt_to_64v2df): Likewise.
4899 (*aarch64_mla_elt<mode>): Likewise.
4900 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4901 (*aarch64_mls_elt<mode>): Likewise.
4902 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4903 (*aarch64_fma4_elt<mode>): Likewise.
4904 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4905 (*aarch64_fma4_elt_to_64v2df): Likewise.
4906 (*aarch64_fnma4_elt<mode>): Likewise.
4907 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4908 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4909 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4910 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4911 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4912 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4913 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4914 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4915 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4917 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4919 * config/aarch64/aarch64-simd.md
4920 (aarch64_be_checked_get_lane<mode>): New define_expand.
4921 * config/aarch64/aarch64-simd-builtins.def
4922 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4923 New builtin definition.
4924 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4925 Use new safe be builtin.
4927 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4929 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4931 (aarch64_be_st1<mode>): Likewise.
4932 (aarch_ld1<VALL:mode>): Define_expand modified.
4933 (aarch_st1<VALL:mode>): Likewise.
4934 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4935 (UNSPEC_ST1): Likewise.
4937 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4939 * config/microblaze/microblaze.md: Add trap insn and attribute
4941 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4943 PR preprocessor/58580
4944 * input.h (location_get_source_line): Take an additional line_size
4946 (void diagnostics_file_cache_fini): Declare new function.
4947 * input.c (struct fcache): New type.
4948 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4949 New static constants.
4950 (diagnostic_file_cache_init, total_lines_num)
4951 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4952 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4953 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4954 (get_next_line, read_next_line, goto_next_line, read_line_num):
4955 New static function definitions.
4956 (diagnostic_file_cache_fini): New function.
4957 (location_get_source_line): Take an additional output line_len
4958 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4960 * diagnostic.c (diagnostic_finish): Call
4961 diagnostic_file_cache_fini.
4962 (adjust_line): Take an additional input parameter for the length
4963 of the line, rather than calculating it with strlen.
4964 (diagnostic_show_locus): Adjust the use of
4965 location_get_source_line and adjust_line with respect to their new
4966 signature. While displaying a line now, do not stop at the first
4967 null byte. Rather, display the zero byte as a space and keep
4968 going until we reach the size of the line.
4969 * Makefile.in: Add vec.o to OBJS-libcommon
4971 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4972 Ilya Tocar <ilya.tocar@intel.com>
4974 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4975 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4976 (__builtin_ia32_kmov16): Ditto.
4977 * config/i386/i386.md (UNSPEC_KMOV): New.
4980 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4982 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4983 (_mm512_storeu_si512): Ditto.
4985 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4988 * rtl.h (get_referenced_operands): Declare.
4989 * recog.c (get_referenced_operands): New function.
4990 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4991 operands have been referenced when recording LO_SUM references.
4993 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4995 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4997 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4999 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5000 Enable for generic and recent AMD targets.
5002 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5004 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5005 ARG_SIZE note when adjustment was eliminated.
5007 2014-01-22 Jeff Law <law@redhat.com>
5009 PR tree-optimization/59597
5010 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5011 in file. Accept new argument REGISTERING and use it to modify
5012 dump output appropriately.
5013 (register_jump_thread): Corresponding changes.
5014 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5015 thread paths involving joiner blocks. Add code to dump cancelled
5016 jump threading paths.
5018 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5020 PR rtl-optimization/59477
5021 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5022 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5024 2014-01-22 Tom Tromey <tromey@redhat.com>
5026 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5028 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5030 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5032 PR rtl-optimization/59896
5033 * lra-constraints.c (process_alt_operands): Check unused note for
5034 matched operands of insn with no output reloads.
5036 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5038 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5039 (mips_move_from_gpr_cost): Likewise.
5041 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5043 PR rtl-optimization/59858
5044 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5045 ira_class_hard_regs_num.
5046 (process_alt_operands): Increase reject for dying matched operand.
5048 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5051 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5052 smaller than size, perform several stores or loads and stores
5053 at dst + count - size to store or copy all of size bytes, rather
5054 than just last modesize bytes.
5056 2014-01-20 DJ Delorie <dj@redhat.com>
5058 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5059 that CLOBBERs are REGs before propogating their values.
5061 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5064 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5065 (cgraph_inline_failed_type): New function.
5066 * cgraph.h (DEFCIFCODE): Add type.
5067 (cgraph_inline_failed_type_t): New enum.
5068 (cgraph_inline_failed_type): New prototype.
5069 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5070 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5071 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5072 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5073 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5074 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5075 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5076 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5077 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5078 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5079 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5080 OPTIMIZATION_MISMATCH.
5081 * tree-inline.c (expand_call_inline): Emit errors during
5082 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5084 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5087 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5088 mode attribute in insn output.
5090 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5092 * output.h (output_constant): Delete.
5093 * varasm.c (output_constant): Make private.
5095 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5097 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5099 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5102 * tree.h (fold_builtin_strcat): New prototype.
5103 * builtins.c (fold_builtin_strcat): No longer static. Add len
5104 argument, if non-NULL, don't call c_strlen. Optimize
5105 directly into __builtin_memcpy instead of __builtin_strcpy.
5106 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5107 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5109 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5111 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5112 for SImode_address_operand operands, having only a REG argument.
5114 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5116 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5117 loader name using mbig-endian.
5118 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5120 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5122 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5126 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5128 * config/aarch64/aarch64-protos.h
5129 (aarch64_cannot_change_mode_class_ptr): Declare.
5130 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5131 aarch64_cannot_change_mode_class_ptr): New.
5132 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5133 backend hook aarch64_cannot_change_mode_class.
5135 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5137 * common/config/aarch64/aarch64-common.c
5138 (aarch64_handle_option): Don't handle any option order logic here.
5139 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5140 selected_cpu, warn on architecture version mismatch.
5141 (aarch64_override_options): Fix parsing order for option strings.
5143 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5144 Iain Sandoe <iain@codesourcery.com>
5147 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5148 warning. Amend comment to reflect current functionality.
5150 2014-01-20 Richard Biener <rguenther@suse.de>
5153 * builtins.c (fold_builtin_strcat): Remove case better handled
5154 by tree-ssa-strlen.c.
5156 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5158 * config/aarch64/aarch64.opt
5159 (mcpu, march, mtune): Make case-insensitive.
5161 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5164 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5165 if operands[1] is a REG or ZERO_EXTEND of a REG.
5167 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5169 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5171 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5173 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5174 long non-pic millicode calls.
5176 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5178 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5180 2014-01-19 Kito Cheng <kito@0xlab.org>
5182 * builtins.c (expand_movstr): Check movstr expand done or fail.
5184 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5185 H.J. Lu <hongjiu.lu@intel.com>
5188 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5189 to DImode for zero-extended addresses.
5191 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5193 PR rtl-optimization/57763
5194 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5195 on the new indirect jump_insn and increment LABEL_NUSES (label).
5197 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5201 * config.gcc (x86_archs): New variable.
5202 (x86_64_archs): Likewise.
5203 (x86_cpus): Likewise.
5204 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5205 --with-arch/--with-cpu= options.
5206 Support --with-arch=/--with-cpu={nehalem,westmere,
5207 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5209 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5211 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5212 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5214 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5216 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5218 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5221 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5222 clear cpp_get_options (parse_in)->warn_unused_macros for
5223 ix86_target_macros_internal with cpp_define.
5225 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5227 * jump.c (delete_related_insns): Keep (use (insn))s.
5228 * reorg.c (redundant_insn): Check for barriers too.
5230 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5232 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5234 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5236 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5237 call to $$dyncall when TARGET_LONG_CALLS is true.
5239 2014-01-17 Jeff Law <law@redhat.com>
5241 * ree.c (combine_set_extension): Temporarily disable test for
5242 changing number of hard registers.
5244 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5247 * ipa-inline-analysis.c (inline_free_summary):
5248 Do not free summary of aliases.
5250 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5253 * gimplify.c (gimplify_expr): Use create_tmp_var
5254 instead of create_tmp_var_raw. If cond doesn't have
5255 integral type, don't add the IFN_ANNOTATE builtin at all.
5257 2014-01-17 Martin Jambor <mjambor@suse.cz>
5260 * ipa-cp.c (prev_edge_clone): New variable.
5261 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5262 Also resize prev_edge_clone vector.
5263 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5264 (ipcp_edge_removal_hook): New function.
5265 (ipcp_driver): Register ipcp_edge_removal_hook.
5267 2014-01-17 Andrew Pinski <apinski@cavium.com>
5268 Steve Ellcey <sellcey@mips.com>
5271 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5274 2014-01-17 Jeff Law <law@redhat.com>
5277 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5278 so that pass_ccp runs first.
5280 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5282 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5283 (ix86_adjust_cost): Use !TARGET_XXX.
5284 (do_reorder_for_imul): Likewise.
5285 (swap_top_of_ready_list): Likewise.
5286 (ix86_sched_reorder): Likewise.
5288 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5290 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5291 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5292 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5293 (intel_memset): New. Duplicate slm_memset.
5294 (intel_cost): New. Duplicate slm_cost.
5295 (m_INTEL): New macro.
5296 (processor_target_table): Add "intel".
5297 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5298 with PROCESSOR_INTEL for "intel".
5299 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5300 PROCESSOR_SILVERMONT.
5301 (ix86_issue_rate): Likewise.
5302 (ix86_adjust_cost): Likewise.
5303 (ia32_multipass_dfa_lookahead): Likewise.
5304 (swap_top_of_ready_list): Likewise.
5305 (ix86_sched_reorder): Likewise.
5306 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5307 instead of TARGET_OPT_AGU.
5308 * config/i386/i386.h (TARGET_INTEL): New.
5309 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5310 (processor_type): Add PROCESSOR_INTEL.
5311 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5312 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5314 2014-01-17 Marek Polacek <polacek@redhat.com>
5317 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5320 2014-01-17 Richard Biener <rguenther@suse.de>
5322 PR tree-optimization/46590
5323 * opts.c (default_options_table): Add entries for
5324 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5325 all enabled at -O1 but not for -Og.
5326 * common.opt (fbranch-count-reg): Remove Init(1).
5327 (fmove-loop-invariants): Likewise.
5328 (ftree-pta): Likewise.
5330 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5332 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5333 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5334 decls to at least the GCC 4.8 used alignments.
5337 * tree-nested.c (convert_nonlocal_reference_stmt,
5338 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5339 of GIMPLE_BIND stmts, adjust associated decls.
5341 2014-01-17 Richard Biener <rguenther@suse.de>
5343 PR tree-optimization/46590
5344 * vec.h (vec<>::bseach): New member function implementing
5345 binary search according to C89 bsearch.
5346 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5347 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5348 bitmap pointer again. Make accesses_in_loop a flat array.
5349 (mem_ref_obstack): New global.
5350 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5351 (mark_ref_stored): Likewise.
5352 (ref_indep_loop_p_2): Likewise.
5353 (set_ref_stored_in_loop): New helper function.
5354 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5355 (memref_free): Adjust.
5356 (record_mem_ref_loc): Simplify.
5357 (gather_mem_refs_stmt): Adjust.
5358 (sort_locs_in_loop_postorder_cmp): New function.
5359 (analyze_memory_references): Sort accesses_in_loop after
5360 loop postorder number.
5361 (find_ref_loc_in_loop_cmp): New function.
5362 (for_all_locs_in_loop): Find relevant cluster of locs in
5363 accesses_in_loop and iterate without recursion.
5364 (execute_sm): Avoid uninit warning.
5365 (struct ref_always_accessed): Simplify.
5366 (ref_always_accessed::operator ()): Likewise.
5367 (ref_always_accessed_p): Likewise.
5368 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5369 loop postorder numbers here.
5370 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5373 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5376 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5377 on decls for which assemble_alias has been called.
5379 2014-01-17 Nick Clifton <nickc@redhat.com>
5381 * config/msp430/msp430.opt: (mcpu): New option.
5382 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5383 (msp430_option_override): Parse target_cpu. If the MCU name
5384 matches a generic string, clear target_mcu.
5385 (msp430_attr): Allow numeric interrupt values up to 63.
5386 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5387 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5389 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5391 * config/msp430/msp430.md (popm): Use %J rather than %I.
5392 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5393 (addhi_cy_i): Use immediate_operand for operand 2.
5394 * doc/invoke.texi: Document -mcpu option.
5396 2014-01-17 Richard Biener <rguenther@suse.de>
5398 PR rtl-optimization/38518
5399 * df.h (df_analyze_loop): Declare.
5400 * df-core.c: Include cfgloop.h.
5401 (df_analyze_1): Split out main part of df_analyze.
5402 (df_analyze): Adjust.
5403 (loop_inverted_post_order_compute): New function.
5404 (loop_post_order_compute): Likewise.
5405 (df_analyze_loop): New function avoiding whole-function
5407 * loop-invariant.c (find_defs): Use df_analyze_loop.
5408 (find_invariants): Adjust.
5409 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5411 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5413 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5414 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5416 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5418 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5419 traversal when removing references.
5421 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5424 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5426 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5429 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5430 pushing a reload for an autoinc when we had previously reloaded an
5431 inner part of the address.
5433 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5435 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5437 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5438 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5439 when not giving up or versioning for alias only because of
5441 (vect_analyze_data_ref_dependences): Set to true.
5442 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5444 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5445 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5449 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5452 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5453 operand 0 predicate for gathers, use a new pseudo as subtarget.
5455 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5458 * lra-constraints.c (process_alt_operands): Add printing debug info.
5459 Check absence of input/output reloads for matched operands too.
5461 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5463 PR rtl-optimization/59835
5464 * ira.c (ira_init_register_move_cost): Increase cost for
5467 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5469 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5471 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5474 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5475 non-register objects. Use gen_(high/low)part more consistently.
5478 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5481 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5482 endian support, remove tests for WORDS_BIG_ENDIAN.
5483 (p8_mfvsrd_3_<mode>): Likewise.
5484 (reload_gpr_from_vsx<mode>): Likewise.
5485 (reload_gpr_from_vsxsf): Likewise.
5486 (p8_mfvsrd_4_disf): Likewise.
5488 2014-01-16 Richard Biener <rguenther@suse.de>
5490 PR rtl-optimization/46590
5491 * lcm.c (compute_antinout_edge): Use postorder iteration.
5492 (compute_laterin): Use inverted postorder iteration.
5494 2014-01-16 Nick Clifton <nickc@redhat.com>
5497 * varasm.c (output_constant): Return the number of bytes actually
5499 (output_constructor_array_range): Update the field size with the
5500 number of bytes emitted by output_constant.
5501 (output_constructor_regular_field): Likewise. Also do not
5502 complain if the total number of bytes emitted is now greater
5503 than the expected fieldpos.
5504 * output.h (output_constant): Update prototype and descriptive comment.
5506 2014-01-16 Marek Polacek <polacek@redhat.com>
5509 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5510 it is error_mark_node.
5512 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5514 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5515 VALID_AVX256_REG_OR_OI_MODE.
5517 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5519 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5520 current procedure should be profiled.
5522 2014-01-15 Andrew Pinski <apinski@cavium.com>
5524 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5525 of moving from/to the STACK_REG register class.
5527 2014-01-15 Richard Henderson <rth@redhat.com>
5530 * reginfo.c (global_regs_decl): Globalize.
5531 * rtl.h (global_regs_decl): Declare.
5532 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5533 reserved via global_regs.
5535 2014-01-15 Teresa Johnson <tejohnson@google.com>
5537 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5539 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5541 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5542 and vmulosh rather than call gen_vec_widen_smult_*.
5543 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5544 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5545 (vec_widen_smult_even_v16qi): Likewise.
5546 (vec_widen_umult_even_v8hi): Likewise.
5547 (vec_widen_smult_even_v8hi): Likewise.
5548 (vec_widen_umult_odd_v16qi): Likewise.
5549 (vec_widen_smult_odd_v16qi): Likewise.
5550 (vec_widen_umult_odd_v8hi): Likewise.
5551 (vec_widen_smult_odd_v8hi): Likewise.
5552 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5553 vmuloub rather than call gen_vec_widen_umult_*.
5554 (vec_widen_umult_lo_v16qi): Likewise.
5555 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5556 vmulosb rather than call gen_vec_widen_smult_*.
5557 (vec_widen_smult_lo_v16qi): Likewise.
5558 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5559 rather than call gen_vec_widen_umult_*.
5560 (vec_widen_umult_lo_v8hi): Likewise.
5561 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5562 rather than call gen_vec_widen_smult_*.
5563 (vec_widen_smult_lo_v8hi): Likewise.
5565 2014-01-15 Jeff Law <law@redhat.com>
5567 PR tree-optimization/59747
5568 * ree.c (find_and_remove_re): Properly handle case where a second
5569 eliminated extension requires widening a copy created for elimination
5570 of a prior extension.
5571 (combine_set_extension): Ensure that the number of hard regs needed
5572 for a destination register does not change when we widen it.
5574 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5576 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5577 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5578 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5579 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5580 (avr-*-rtems*): Likewise.
5581 (bfin*-rtems*): Likewise.
5582 (moxie-*-rtems*): Likewise.
5583 (h8300-*-rtems*): Likewise.
5584 (i[34567]86-*-rtems*): Likewise.
5585 (lm32-*-rtems*): Likewise.
5586 (m32r-*-rtems*): Likewise.
5587 (m68k-*-rtems*): Likewise.
5588 (microblaze*-*-rtems*): Likewise.
5589 (mips*-*-rtems*): Likewise.
5590 (powerpc-*-rtems*): Likewise.
5591 (sh-*-rtems*): Likewise.
5592 (sparc-*-rtems*): Likewise.
5593 (sparc64-*-rtems*): Likewise.
5594 (v850-*-rtems*): Likewise.
5595 (m32c-*-rtems*): Likewise.
5597 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5599 PR rtl-optimization/59511
5600 * ira.c (ira_init_register_move_cost): Use memory costs for some
5601 cases of register move cost calculations.
5602 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5603 instead of BB frequency.
5604 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5605 * lra-assigns.c (find_hard_regno_for): Ditto.
5607 2014-01-15 Richard Biener <rguenther@suse.de>
5609 PR tree-optimization/59822
5610 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5611 (vectorizable_load): Use it to hoist defs of uses of invariant
5612 loads out of the loop.
5614 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5615 Kugan Vivekanandarajah <kuganv@linaro.org>
5618 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5621 2014-01-15 Richard Biener <rguenther@suse.de>
5623 PR rtl-optimization/59802
5624 * lcm.c (compute_available): Use inverted postorder to seed
5625 the initial worklist.
5627 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5630 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5631 ADDR_REGS for invalid symrefs in non-PIC code.
5633 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5636 * builtins.c (determine_block_size): Initialize *probable_max_size
5637 even if len_rtx is CONST_INT.
5639 2014-01-14 Andrew Pinski <apinski@cavium.com>
5641 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5642 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5643 (cortexa53_tunings): Likewise.
5644 (aarch64_sched_issue_rate): New function.
5645 (TARGET_SCHED_ISSUE_RATE): Define.
5647 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5649 * ira-costs.c (find_costs_and_classes): Add missed
5650 ira_init_register_move_cost_if_necessary.
5652 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5655 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5657 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5660 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5661 to indicate if type is used for function return value. Warn ABI
5662 change if the vector mode isn't available for function return value.
5663 (ix86_function_arg_advance): Pass false to type_natural_mode.
5664 (ix86_function_arg): Likewise.
5665 (ix86_gimplify_va_arg): Likewise.
5666 (function_arg_32): Don't warn ABI change.
5667 (ix86_function_value): Pass true to type_natural_mode.
5668 (ix86_return_in_memory): Likewise.
5669 (ix86_struct_value_rtx): Removed.
5670 (TARGET_STRUCT_VALUE_RTX): Likewise.
5672 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5674 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5675 converting a conditional jump into a conditional return.
5677 2014-01-14 Richard Biener <rguenther@suse.de>
5679 PR tree-optimization/58921
5680 PR tree-optimization/59006
5681 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5682 hoisting invariant stmts.
5683 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5684 invariant loads on the preheader edge if possible.
5686 2014-01-14 Joey Ye <joey.ye@arm.com>
5688 * doc/plugin.texi (Building GCC plugins): Update to C++.
5690 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5692 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5693 (_mm_rcp28_round_ss): Ditto.
5694 (_mm_rsqrt28_round_sd): Ditto.
5695 (_mm_rsqrt28_round_ss): Ditto.
5696 (_mm_rcp28_sd): Ditto.
5697 (_mm_rcp28_ss): Ditto.
5698 (_mm_rsqrt28_sd): Ditto.
5699 (_mm_rsqrt28_ss): Ditto.
5700 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5701 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5702 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5703 (IX86_BUILTIN_RCP28SD): Ditto.
5704 (IX86_BUILTIN_RCP28SS): Ditto.
5705 (IX86_BUILTIN_RSQRT28SD): Ditto.
5706 (IX86_BUILTIN_RSQRT28SS): Ditto.
5707 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5708 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5709 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5710 (ix86_expand_special_args_builtin): Expand new FTYPE.
5711 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5712 (srcp14<mode>): Make insn unary.
5713 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5714 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5715 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5716 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5717 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5718 Fix rounding: make it SAE only.
5719 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5721 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5723 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5724 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5725 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5726 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5727 (round_saeonly_mask_scalar_operand4): Ditto.
5728 (round_saeonly_mask_scalar_op3): Ditto.
5729 (round_saeonly_mask_scalar_op4): Ditto.
5731 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5733 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5734 Implement -maltivec=be for vec_insert and vec_extract.
5736 2014-01-10 DJ Delorie <dj@redhat.com>
5738 * config/msp430/msp430.md (call_internal): Don't allow memory
5739 references with SP as the base register.
5740 (call_value_internal): Likewise.
5741 * config/msp430/constraints.md (Yc): New. For memory references
5742 that don't use SP as a base register.
5744 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5745 "an integer without a # prefix"
5746 * config/msp430/msp430.md (epilogue_helper): Use it.
5748 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5751 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5752 AVX512F gather builtins.
5753 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5754 on gather decls with INTEGER_TYPE masktype.
5755 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5756 directly into the builtin rather than hoisting it before loop.
5758 PR tree-optimization/59387
5759 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5760 (scev_const_prop): If folded_casts and type has undefined overflow,
5761 use force_gimple_operand instead of force_gimple_operand_gsi and
5762 for each added stmt if it is assign with
5763 arith_code_with_undefined_signed_overflow, call
5764 rewrite_to_defined_overflow.
5765 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5766 gimple-fold.h instead.
5767 (arith_code_with_undefined_signed_overflow,
5768 rewrite_to_defined_overflow): Moved to ...
5769 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5770 rewrite_to_defined_overflow): ... here. No longer static.
5771 Include gimplify-me.h.
5772 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5773 rewrite_to_defined_overflow): New prototypes.
5775 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5777 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5779 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5781 * builtins.c (get_object_alignment_2): Minor tweak.
5782 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5784 2014-01-13 Christian Bruel <christian.bruel@st.com>
5786 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5787 optimized non constant lengths.
5789 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5792 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5793 load as __atomic_load_N if possible.
5795 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5797 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5799 (rs6000_expand_builtin): Adjust call.
5801 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5804 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5805 * config/rs6000/rs6000.c: Include target-globals.h.
5806 (rs6000_set_current_function): Instead of doing target_reinit
5807 unconditionally, use save_target_globals_default_opts and
5808 restore_target_globals.
5810 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5812 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5813 (rs6000_expand_builtin): Handle mffs and mtfsf.
5814 (rs6000_init_builtins): Define mffs and mtfsf.
5815 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5816 (rs6000_mffs): New pattern.
5817 (rs6000_mtfsf): New pattern.
5819 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5821 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5822 Start narrowing with START. Apply candidate-use pair
5823 and check overall cost in narrowing.
5824 (iv_ca_prune): Pass new argument.
5826 2014-01-10 Jeff Law <law@redhat.com>
5829 * ree.c (combine_reaching_defs): Ensure the defining statement
5830 occurs before the extension when optimizing extensions with
5831 different source and destination hard registers.
5833 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5836 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5837 vtables into the type inheritance graph.
5839 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5841 PR rtl-optimization/59754
5842 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5843 modes in the REGNO != REGNO case.
5845 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5847 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5849 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5851 PR tree-optimization/59745
5852 * tree-predcom.c (tree_predictive_commoning_loop): Call
5853 free_affine_expand_cache if giving up because components is NULL.
5855 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5856 GC in payload of target_globals struct instead of allocating them on
5857 the heap and the larger structs separately using GC.
5858 * target-globals.h (struct target_globals): Make regs, hard_regs,
5859 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5860 of GTY((skip)) and change type to void *.
5861 (reset_target_globals): Cast loads from those fields to corresponding
5864 2014-01-10 Steve Ellcey <sellcey@mips.com>
5867 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5868 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5869 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5871 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5874 * aarch64-modes.def (CC_Zmode): New flags mode.
5875 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5876 represents an equality.
5877 (aarch64_get_condition_code): Handle CC_Zmode.
5878 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5880 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5882 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5883 extraction in good case.
5885 2014-01-10 Richard Biener <rguenther@suse.de>
5887 PR tree-optimization/59374
5888 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5889 checking after SLP discovery. Mark stmts not participating
5890 in any SLP instance properly.
5892 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5894 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5895 when handling a SET rtx.
5897 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5899 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5900 (cortex-a57): Likewise.
5901 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5903 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5905 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5906 non-iwmmxt builtins.
5908 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5912 * ipa-devirt.c record_target_from_binfo): Take as argument
5913 stack of binfos and lookup matching one for virtual inheritance.
5914 (possible_polymorphic_call_targets_1): Update.
5916 2014-01-10 Huacai Chen <chenhc@lemote.com>
5918 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5919 kernel strings for Loongson-2E/2F/3A.
5921 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5924 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5925 is_gimple_call before calling gimple_call_internal_p.
5927 2014-01-09 Steve Ellcey <sellcey@mips.com>
5929 * Makefile.in (TREE_FLOW_H): Remove.
5930 (TREE_SSA_H): Add file names from tree-flow.h.
5931 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5932 * tree.h: Remove tree-flow.h reference.
5933 * hash-table.h: Remove tree-flow.h reference.
5934 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5935 reference with tree-ssa-loop.h.
5937 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5939 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5940 default element-order behavior for -maltivec.
5941 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5942 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5943 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5944 when targeting big endian, at least for now.
5945 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5947 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5950 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5951 var satisfies use_register_for_decl, just take into account type
5952 alignment, rather than decl alignment.
5954 PR tree-optimization/59622
5955 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5956 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5957 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5958 Don't devirtualize for inplace at all. For targets.length () == 1,
5959 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5961 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5963 * config/i386/i386.md (cpu): Remove the unused btver1.
5965 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5967 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5969 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5972 * tree-core.h (struct target_globals): New forward declaration.
5973 (struct tree_target_option): Add globals field.
5974 * tree.h (TREE_TARGET_GLOBALS): Define.
5975 (prepare_target_option_nodes_for_pch): New prototype.
5976 * target-globals.h (struct target_globals): Define even if
5978 * tree.c (prepare_target_option_node_for_pch,
5979 prepare_target_option_nodes_for_pch): New functions.
5980 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5981 * config/i386/i386.c: Include target-globals.h.
5982 (ix86_set_current_function): Instead of doing target_reinit
5983 unconditionally, use save_target_globals_default_opts and
5984 restore_target_globals.
5986 2014-01-09 Richard Biener <rguenther@suse.de>
5988 PR tree-optimization/59715
5989 * tree-cfg.h (split_critical_edges): Declare.
5990 * tree-cfg.c (split_critical_edges): Export.
5991 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5993 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5995 * cfgexpand.c (expand_stack_vars): Optionally disable
5996 asan stack protection.
5997 (expand_used_vars): Likewise.
5998 (partition_stack_vars): Likewise.
5999 * asan.c (asan_emit_stack_protection): Optionally disable
6000 after return stack usage.
6001 (instrument_derefs): Optionally disable memory access instrumentation.
6002 (instrument_builtin_call): Likewise.
6003 (instrument_strlen_call): Likewise.
6004 (asan_protect_global): Optionally disable global variables protection.
6005 * doc/invoke.texi: Added doc for new options.
6006 * params.def: Added new options.
6007 * params.h: Likewise.
6009 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6011 PR rtl-optimization/59724
6012 * ifcvt.c (cond_exec_process_if_block): Don't call
6013 flow_find_head_matching_sequence with 0 longest_match.
6014 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6015 non-active insns if !stop_after.
6016 (try_head_merge_bb): Revert 2014-01-07 changes.
6018 2014-01-08 Jeff Law <law@redhat.com>
6020 * ree.c (get_sub_rtx): New function, extracted from...
6021 (merge_def_and_ext): Here.
6022 (combine_reaching_defs): Use get_sub_rtx.
6024 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6026 * cgraph.h (varpool_variable_node): Do not choke on null node.
6028 2014-01-08 Catherine Moore <clm@codesourcery.com>
6030 * config/mips/mips.md (simple_return): Attempt to use JRC
6032 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6034 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6036 PR rtl-optimization/59137
6037 * reorg.c (steal_delay_list_from_target): Call update_block for
6039 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6041 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6043 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6044 two duplicate entries.
6046 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6049 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6051 * config/mips/mips.c (mips_truncated_op_cost): New function.
6052 (mips_rtx_costs): Adjust test for BADDU.
6053 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6055 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6057 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6058 (*baddu_si): ...this new pattern.
6060 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6063 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6065 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6068 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6070 (expand_expr, expand_normal): Adjust.
6071 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6072 inner_reference_p. Use inner_reference_p to expand inner references.
6073 (store_expr): Adjust.
6074 * cfgexpand.c (expand_call_stmt): Adjust.
6076 2014-01-08 Rong Xu <xur@google.com>
6078 * gcov-io.c (gcov_var): Move from gcov-io.h.
6079 (gcov_position): Ditto.
6080 (gcov_is_error): Ditto.
6081 (gcov_rewrite): Ditto.
6082 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6083 only part to libgcc/libgcov.h.
6085 2014-01-08 Marek Polacek <polacek@redhat.com>
6088 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6090 2014-01-08 Marek Polacek <polacek@redhat.com>
6093 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6095 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6097 PR rtl-optimization/59649
6098 * stor-layout.c (get_mode_bounds): For BImode return
6099 0 and STORE_FLAG_VALUE.
6101 2014-01-08 Richard Biener <rguenther@suse.de>
6104 * gimple.h (is_gimple_builtin_call): Remove.
6105 (gimple_builtin_call_types_compatible_p): New.
6106 (gimple_call_builtin_p): New overload.
6107 * gimple.c (is_gimple_builtin_call): Remove.
6108 (validate_call): Rename to ...
6109 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6111 (validate_type): New static function.
6112 (gimple_call_builtin_p): New overload and adjust.
6113 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6114 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6115 (gimple_fold_stmt_to_constant_1): Likewise.
6116 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6118 2014-01-08 Richard Biener <rguenther@suse.de>
6121 * gimplify.c (gimplify_expr): Gimplify register-register type
6122 VIEW_CONVERT_EXPRs to separate stmts.
6124 2014-01-07 Jeff Law <law@redhat.com>
6127 * ree.c (combine_set_extension): Handle case where source
6128 and destination registers in an extension insn are different.
6129 (combine_reaching_defs): Allow source and destination registers
6130 in extension to be different under limited circumstances.
6131 (add_removable_extension): Remove restriction that the
6132 source and destination registers in the extension are the same.
6133 (find_and_remove_re): Emit a copy from the extension's
6134 destination to its source after the defining insn if
6135 the source and destination registers are different.
6138 * ifcvt.c (merge_if_block): If we are merging a block with more than
6139 one successor with a block with no successors, remove any BARRIER
6140 after the second block.
6142 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6144 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6146 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6149 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6150 for 14-bit register offsets when INT14_OK_STRICT is false.
6152 2014-01-07 Roland Stigge <stigge@antcom.de>
6153 Michael Meissner <meissner@linux.vnet.ibm.com>
6156 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6157 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6159 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6161 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6164 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6166 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6167 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6168 rtx is const0_rtx or not.
6170 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6173 * target-globals.c (save_target_globals): Remove this_fn_optab
6175 * toplev.c: Include optabs.h.
6176 (target_reinit): Temporarily restore the global options if another
6177 set of options are in force.
6179 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6181 PR rtl-optimization/58668
6182 * cfgcleanup.c (flow_find_cross_jump): Don't count
6183 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6184 to determine what is counted.
6185 (flow_find_head_matching_sequence): Use active_insn_p to determine
6187 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6189 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6190 determine what is counted.
6192 PR tree-optimization/59643
6193 * tree-predcom.c (split_data_refs_to_components): If one dr is
6194 read and one write, determine_offset fails and the write isn't
6195 in the bad component, just put the read into the bad component.
6197 2014-01-07 Mike Stump <mikestump@comcast.net>
6198 Jakub Jelinek <jakub@redhat.com>
6201 * tree-core.h (struct tree_optimization_option): Change optabs
6202 type from unsigned char * to void *.
6203 * optabs.c (init_tree_optimization_optabs): Adjust
6204 TREE_OPTIMIZATION_OPTABS initialization.
6206 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6209 * config/i386/i386.h (struct machine_function): Add
6210 no_drap_save_restore field.
6211 * config/i386/i386.c (ix86_save_reg): Use
6212 !cfun->machine->no_drap_save_restore instead of
6213 crtl->stack_realign_needed.
6214 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6215 this function clears frame_pointer_needed. Set
6216 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6217 and DRAP reg is needed.
6219 2014-01-06 Marek Polacek <polacek@redhat.com>
6222 * doc/implement-c.texi: Mention that other integer types are
6223 permitted as bit-field types in strictly conforming mode.
6225 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6227 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6230 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6232 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6234 2014-01-06 Martin Jambor <mjambor@suse.cz>
6237 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6239 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6241 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6245 * var-tracking.c (add_stores): Preserve the value of the source even if
6246 we don't record the store.
6248 2014-01-06 Terry Guo <terry.guo@arm.com>
6250 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6252 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6255 * config/darwin.c (darwin_function_section): Adjust return values to
6256 correspond to optimisation changes made in r206070.
6258 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6260 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6261 from prefetch_block tune setting.
6262 (nocona_cost): Correct size of prefetch block to 64.
6264 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6266 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6267 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6268 used to save the static chain register in the computation of the offset
6269 from which the FP registers need to be restored.
6271 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6273 PR tree-optimization/59519
6274 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6275 ICE if get_current_def (current_new_name) is already non-NULL, as long
6276 as it is a phi result of some other phi in *new_exit_bb that has
6279 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6280 or vmovdqu* for misaligned_operand.
6281 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6282 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6283 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6284 aligned_mem for AVX512F masked aligned load and store builtins and for
6287 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6289 PR tree-optimization/59651
6290 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6291 Address range for negative step should be added by TYPE_SIZE_UNIT.
6293 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6295 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6296 overlapping registers also for registers other than the stack pointer.
6298 2014-01-03 Marek Polacek <polacek@redhat.com>
6301 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6304 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6307 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6310 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6311 (push splitter): Use <P:MODE_SIZE> instead of
6312 GET_MODE_SIZE (<P:MODE>mode).
6313 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6314 (mov -1, reg peephole2): Likewise.
6315 * config/i386/sse.md (*mov<mode>_internal,
6316 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6317 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6318 *<code><mode>3, *andnot<mode>3<mask_name>,
6319 <mask_codefor><code><mode>3<mask_name>): Likewise.
6320 * config/i386/subst.md (mask_mode512bit_condition,
6321 sd_mask_mode512bit_condition): Likewise.
6323 2014-01-02 Xinliang David Li <davidxl@google.com>
6325 PR tree-optimization/59303
6326 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6327 (dump_predicates): Better output format.
6328 (pred_equal_p): New function.
6329 (is_neq_relop_p): Ditto.
6330 (is_neq_zero_form_p): Ditto.
6331 (pred_expr_equal_p): Ditto.
6332 (pred_neg_p): Ditto.
6333 (simplify_pred): Ditto.
6334 (simplify_preds_2): Ditto.
6335 (simplify_preds_3): Ditto.
6336 (simplify_preds_4): Ditto.
6337 (simplify_preds): Ditto.
6339 (push_to_worklist): Ditto.
6340 (get_pred_info_from_cmp): Ditto.
6341 (is_degenerated_phi): Ditto.
6342 (normalize_one_pred_1): Ditto.
6343 (normalize_one_pred): Ditto.
6344 (normalize_one_pred_chain): Ditto.
6345 (normalize_preds): Ditto.
6346 (normalize_cond_1): Remove function.
6347 (normalize_cond): Ditto.
6348 (is_gcond_subset_of): Ditto.
6349 (is_subset_of_any): Ditto.
6350 (is_or_set_subset_of): Ditto.
6351 (is_and_set_subset_of): Ditto.
6352 (is_norm_cond_subset_of): Ditto.
6353 (pred_chain_length_cmp): Ditto.
6354 (convert_control_dep_chain_into_preds): Type change.
6355 (find_predicates): Ditto.
6356 (find_def_preds): Ditto.
6357 (destroy_predicates_vecs): Ditto.
6358 (find_matching_predicates_in_rest_chains): Ditto.
6359 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6360 (is_pred_expr_subset): Ditto.
6361 (is_pred_chain_subset_of): Ditto.
6362 (is_included_in): Ditto.
6363 (is_superset_of): Ditto.
6365 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6367 Update copyright years.
6369 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6371 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6372 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6373 config/arc/arc.md, config/arc/arc.opt,
6374 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6375 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6376 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6377 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6378 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6379 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6380 vtable-verify.c, vtable-verify.h: Use the standard form for the
6383 2014-01-02 Tobias Burnus <burnus@net-b.de>
6385 * gcc.c (process_command): Update copyright notice dates.
6386 * gcov-dump.c: Ditto.
6388 * doc/cpp.texi: Bump @copying's copyright year.
6389 * doc/cppinternals.texi: Ditto.
6390 * doc/gcc.texi: Ditto.
6391 * doc/gccint.texi: Ditto.
6392 * doc/gcov.texi: Ditto.
6393 * doc/install.texi: Ditto.
6394 * doc/invoke.texi: Ditto.
6396 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6398 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6400 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6402 * config/i386/sse.md (*mov<mode>_internal): Guard
6403 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6405 PR rtl-optimization/59647
6406 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6407 new_rtx into UNSIGNED_FLOAT rtxes.
6409 Copyright (C) 2014 Free Software Foundation, Inc.
6411 Copying and distribution of this file, with or without modification,
6412 are permitted in any medium without royalty provided the copyright
6413 notice and this notice are preserved.