1 2014-04-15 Richard Biener <rguenther@suse.de>
3 * alias.c (record_component_aliases): Do not walk BINFOs.
5 2014-04-15 Richard Biener <rguenther@suse.de>
7 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
8 Add struct function argument and adjust.
9 (find_func_aliases_for_call): Likewise.
10 (find_func_aliases): Likewise.
11 (find_func_clobbers): Likewise.
12 (intra_create_variable_infos): Likewise.
13 (compute_points_to_sets): Likewise.
14 (ipa_pta_execute): Adjust. Do not push/pop cfun.
16 2014-04-15 Richard Biener <rguenther@suse.de>
18 * tree.c (iterative_hash_expr): Use enum tree_code_class
19 to store TREE_CODE_CLASS.
20 (tree_block): Likewise.
21 (tree_set_block): Likewise.
22 * tree.h (fold_build_pointer_plus_loc): Use
23 convert_to_ptrofftype_loc.
25 2014-04-15 Jakub Jelinek <jakub@redhat.com>
28 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
31 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
33 * cfgloop.h (struct loop): Move force_vectorize down.
34 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
35 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
36 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
37 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
38 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
39 * tree-core.h (enum annot_expr_kind): Add new kind values.
40 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
41 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
43 * tree.def (ANNOTATE_EXPR): Tweak comment.
45 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
47 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
50 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
52 * tree.h (TYPE_IDENTIFIER): Declare.
53 * tree.c (subrange_type_for_debug_p): Use it.
54 * godump.c (go_format_type): Likewise.
55 * dwarf2out.c (is_cxx_auto, modified_type_die,
56 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
57 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
59 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
62 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
64 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
66 * config/i386/i386.c (examine_argument): Return bool. Return true if
67 parameter should be passed in memory.
68 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
69 (construct_container): Update calls to examine_argument.
70 (function_arg_advance_64): Ditto.
71 (return_in_memory_32): Merge with ix86_return_in_memory.
72 (return_in_memory_64): Ditto.
73 (return_in_memory_ms_64): Ditto.
75 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
77 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
78 * coverage.c (coverage_compute_profile_id): Handle externally visible
81 2014-04-14 Martin Jambor <mjambor@suse.cz>
83 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
84 DECL_DISREGARD_INLINE_LIMITS functions.
86 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
89 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
91 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
94 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
95 optimize_insn_for_speed_p instead of
96 optimize_function_for_speed_p.
98 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
100 * doc/invoke.texi (free): Document AArch64.
102 2014-04-14 Richard Biener <rguenther@suse.de>
104 PR tree-optimization/60042
105 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
106 (insert_into_preds_of_block): Do not prevent PHI insertion
107 for REFERENCE exprs here ...
108 (eliminate_dom_walker::before_dom_children): ... but prevent
109 their use here under similar conditions when applied to the
110 IL after PRE optimizations.
112 2014-04-14 Richard Biener <rguenther@suse.de>
114 * passes.def: Move early points-to after early SRA.
116 2014-04-14 Richard Biener <rguenther@suse.de>
118 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
119 check for which sign-changes we allow when forwarding
120 a converted value into a switch.
122 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
124 * stor-layout.c (place_field): Finalize non-constant offset for the
127 2014-04-14 Richard Biener <rguenther@suse.de>
129 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
131 (expand_switch_using_bit_tests_p): Likewise.
132 (process_switch): Compute and pass on speed_p based on the
134 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
135 optimize_bb_for_speed_p.
137 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
139 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
140 * function.h (struct function): Rename has_force_vect_loops into
141 has_force_vectorize_loops.
142 * lto-streamer-in.c (input_cfg): Adjust for renaming.
143 (input_struct_function_base): Likewise.
144 * lto-streamer-out.c (output_cfg): Likewise.
145 (output_struct_function_base): Likewise.
146 * omp-low.c (expand_omp_simd): Likewise.
147 * tree-cfg.c (move_sese_region_to_fn): Likewise.
148 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
149 (version_loop_for_if_conversion): Likewise.
150 (tree_if_conversion): Likewise.
151 (main_tree_if_conversion): Likewise.
152 (gate_tree_if_conversion): Likewise.
153 * tree-inline.c (copy_loops): Likewise.
154 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
155 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
156 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
157 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
158 * tree-vectorizer.c (vectorize_loops): Likewise.
159 * tree-vectorizer.h (unlimited_cost_model): Likewise.
161 2014-04-14 Richard Biener <rguenther@suse.de>
164 * lto-streamer-out.c (wrap_refs): New function.
165 (lto_output): Wrap symbol references in global initializes in
166 type-preserving MEM_REFs.
168 2014-04-14 Christian Bruel <christian.bruel@st.com>
170 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
172 2014-04-14 Christian Bruel <christian.bruel@st.com>
174 * config/sh/sh.md (setmemqi): New expand pattern.
175 * config/sh/sh.h (CLEAR_RATIO): Define.
176 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
177 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
179 2014-04-14 Richard Biener <rguenther@suse.de>
182 * fold-const.c (negate_expr_p): Don't negate directional rounding
184 (fold_negate_expr): Likewise.
186 2014-04-14 Richard Biener <rguenther@suse.de>
188 PR tree-optimization/59817
189 PR tree-optimization/60453
190 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
191 recursion to catch all CHRECs in the scalar evolution and restrict
192 the predicate for the remains appropriately.
194 2014-04-12 Catherine Moore <clm@codesourcery.com>
196 * config/mips/constraints.md: Add new register constraint "kb".
197 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
198 (*movhi_internal): Likewise.
199 (*movqi_internal): Likewise.
200 * config/mips/mips.h (M16_STORE_REGS): New register class.
201 (REG_CLASS_NAMES): Add M16_STORE_REGS.
202 (REG_CLASS_CONTENTS): Likewise.
203 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
205 2014-04-11 Tobias Burnus <burnus@net-b.de>
208 * doc/invoke.texi (-Wformat-signedness): Document it.
209 (Wformat=2): Mention that this enables -Wformat-signedness.
211 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
213 * common/config/epiphany/epiphany-common.c
214 (epiphany_option_optimization_table): Enable section anchors by
215 default at -O1 or higher.
216 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
217 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
218 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
219 carries no extra cost.
220 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
221 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
222 * config/epiphany/predicates.md (memclob_operand): New predicate.
223 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
224 Use memclob_operand predicate and X constraint for operand 3.
226 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
228 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
229 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
232 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
234 PR rtl-optimization/60651
235 * mode-switching.c (optimize_mode_switching): Make sure to emit
236 sets of a lower numbered entity before sets of a higher numbered
237 entity to a mode of the same or lower priority.
238 When creating a seginfo for a basic block that starts with a code
239 label, move the insertion point past the code label.
240 (new_seginfo): Document and enforce requirement that
241 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
242 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
243 * doc/tm.texi: Regenerate.
245 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
248 * config/arc/arc.c (arc_save_restore): Fix assert typo.
250 2013-04-11 Jakub Jelinek <jakub@redhat.com>
252 * BASE-VER: Set to 4.10.0.
254 2014-04-11 Tobias Burnus <burnus@net-b.de>
257 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
258 * doc/gcc.texi (Service): Update description in the @menu
259 * doc/invoke.texi (Option Summary): Remove misplaced and
262 2014-04-11 Steve Ellcey <sellcey@mips.com>
263 Jakub Jelinek <jakub@redhat.com>
266 * expr.c (convert_move): Use emit_store_flag_force instead of
267 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
270 2014-04-11 Richard Biener <rguenther@suse.de>
273 * varasm.c (assemble_alias): Avoid endless error reporting
274 recursion by setting TREE_ASM_WRITTEN.
276 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
278 * config/s390/s390.md: Add a splitter for NOT rtx.
280 2014-04-11 Jakub Jelinek <jakub@redhat.com>
282 PR rtl-optimization/60663
283 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
285 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
286 Jakub Jelinek <jakub@redhat.com>
289 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
290 flag from decl_node to node.
292 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
295 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
296 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
297 ameliorating the cases where it can be.
299 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
302 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
304 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
305 (loadsync_<mode>): Change mode.
306 (load_quadpti, store_quadpti): New.
307 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
308 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
309 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
311 2014-04-09 Cong Hou <congh@google.com>
314 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
317 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
319 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
320 instead of vnor to exploit possible fusion opportunity in the
322 (altivec_expand_vec_perm_const_le): Likewise.
324 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
326 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
327 (loadsync_<mode>): Change mode.
328 (load_quadpti, store_quadpti): New.
329 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
330 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
332 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
335 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
336 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
337 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
339 2014-04-08 Richard Biener <rguenther@suse.de>
342 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
343 a 64bit widest int print double-int similar to on HWI64 hosts.
345 2014-04-08 Richard Biener <rguenther@suse.de>
347 PR tree-optimization/60785
348 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
349 default defs properly.
351 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
353 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
356 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
358 * ipa-devirt.c (maybe_record_node): When node is not recorded,
359 set completep to false rather than true.
361 2014-04-07 Douglas B Rupp <rupp@adacore.com>
364 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
365 ARM_TARGET2_DWARF_FORMAT.
367 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
370 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
371 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
374 2014-04-07 Richard Biener <rguenther@suse.de>
376 PR tree-optimization/60766
377 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
378 (may_eliminate_iv): Convert cand_value_at result to desired type.
380 2014-04-07 Jason Merrill <jason@redhat.com>
383 * common.opt (-fno-gnu-unique): Add.
384 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
386 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
388 * haifa-sched.c: Fix outdated function reference and minor
389 grammar errors in introductory comment.
391 2014-04-07 Richard Biener <rguenther@suse.de>
394 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
396 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
398 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
401 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
402 size accounting for thunks.
403 (pa_asm_output_mi_thunk): Use final_start_function() and
404 final_end_function() to output function start and end directives.
406 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
408 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
409 device specific ISA/ feature information. Remove short_sp and
410 errata_skip ds. Add avr_device_specific_features enum to have device
412 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
413 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
414 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
415 updated device specific info.
416 * config/avr/avr-mcus.def: Merge device specific details to
418 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
420 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
421 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
422 assembler if RMW isa supported by current device.
423 * config/avr/genmultilib.awk: Update as device info structure changed.
424 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
426 2014-04-04 Cong Hou <congh@google.com>
428 PR tree-optimization/60656
429 * tree-vect-stmts.c (supportable_widening_operation):
430 Fix a bug that elements in a vector with vect_used_by_reduction
431 property are incorrectly reordered when the operation on it is not
432 consistant with the one in reduction operation.
434 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
436 PR rtl-optimization/60155
437 * gcse.c (record_set_data): New function.
438 (single_set_gcse): New function.
439 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
440 (hoist_code): Likewise.
441 (get_pressure_class_and_nregs): Likewise.
443 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
445 * explow.c (probe_stack_range): Emit a final optimization blockage.
447 2014-04-04 Anthony Green <green@moxielogic.com>
449 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
452 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
455 * lto-cgraph.c (input_overwrite_node): Check that partitioning
456 flags are set only during streaming.
457 * ipa.c (process_references, walk_polymorphic_call_targets,
458 symtab_remove_unreachable_nodes): Drop bodies of always inline
459 after early inlining.
460 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
462 2014-04-04 Jakub Jelinek <jakub@redhat.com>
463 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
466 * dwarf2out.c (const_ok_for_output_1): Reject expressions
469 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
472 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
474 (cortex_a53_fdivd): Likewise.
476 2014-04-04 Martin Jambor <mjambor@suse.cz>
479 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
481 * cgraph.c (clone_of_p): Also return true if thunks match.
482 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
483 cgraph_function_or_thunk_node and an obsolete comment.
484 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
486 (build_function_decl_skip_args): Likewise.
487 (set_new_clone_decl_and_node_flags): New function.
488 (duplicate_thunk_for_node): Likewise.
489 (redirect_edge_duplicating_thunks): Likewise.
490 (cgraph_clone_node): New parameter args_to_skip, pass it to
491 redirect_edge_duplicating_thunks which is called instead of
492 cgraph_redirect_edge_callee.
493 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
494 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
496 2014-04-04 Jeff Law <law@redhat.com>
499 * config/arm/predicates.md (const_int_I_operand): New predicate.
500 (const_int_M_operand): Similarly.
501 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
503 (insv_t2, extv_reg, extzv_t2): Likewise.
504 (load_multiple_with_writeback): Similarly for const_int_I_operand.
505 (pop_multiple_with_writeback_and_return): Likewise.
506 (vfp_pop_multiple_with_writeback): Likewise
508 2014-04-04 Richard Biener <rguenther@suse.de>
511 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
512 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
514 * gimplify.h (gimple_add_tmp_var_fn): Declare.
515 * gimplify.c (gimple_add_tmp_var_fn): New function.
516 * gimple-expr.h (create_tmp_reg_fn): Declare.
517 * gimple-expr.c (create_tmp_reg_fn): New function.
518 * gimple-low.c (record_vars_into): Don't change cfun.
519 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
520 code generation without cfun.
522 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
525 * Makefile.in (install-driver): Fix shell scripting.
527 2014-04-03 Cong Hou <congh@google.com>
529 PR tree-optimization/60505
530 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
531 threshold of number of iterations below which no vectorization
533 * tree-vect-loop.c (new_loop_vec_info):
534 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
535 * tree-vect-loop.c (vect_analyze_loop_operations):
536 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
537 * tree-vect-loop.c (vect_transform_loop):
538 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
539 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
540 of iterations of the loop and see if we should build the epilogue.
542 2014-04-03 Richard Biener <rguenther@suse.de>
544 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
545 (streamer_tree_cache_create): Adjust.
546 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
547 to allow optional nodes array.
548 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
549 (streamer_tree_cache_append): Likewise.
550 (streamer_tree_cache_create): Create nodes array optionally
551 as specified by parameter.
552 * lto-streamer-out.c (create_output_block): Avoid maintaining
553 the node array in the writer cache.
554 (DFS_write_tree): Remove assertion.
555 (produce_asm_for_decls): Free the out decl state hash table early.
556 * lto-streamer-in.c (lto_data_in_create): Adjust for
557 streamer_tree_cache_create prototype change.
559 2014-04-03 Richard Biener <rguenther@suse.de>
561 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
562 set TREE_CHAIN to NULL_TREE.
564 2014-04-03 Richard Biener <rguenther@suse.de>
566 PR tree-optimization/60740
567 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
568 over all GIMPLE_COND operands.
570 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
572 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
573 (Weffc++): Remove Scott's numbering, merge lists and reference
579 * class.c (accessible_nvdtor_p): New.
580 (check_bases): Don't check base destructor here ...
581 (check_bases_and_members): ... check them here. Trigger on
582 Wnon-virtual-dtor flag.
583 (finish_struct_1): Use accessible_nvdtor_p.
586 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
587 * g++.dg/warn/Wnvdtor-2.C: New.
588 * g++.dg/warn/Wnvdtor-3.C: New.
589 * g++.dg/warn/Wnvdtor-4.C: New.
590 * g++.dg/warn/Weff1.C: Delete.
591 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
592 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
594 2014-04-03 Nick Clifton <nickc@redhat.com>
596 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
599 2014-04-03 Martin Jambor <mjambor@suse.cz>
601 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
602 mention gcc_unreachable before failing.
603 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
606 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
609 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
610 inconsistent code and instead mark the context inconsistent.
611 (possible_polymorphic_call_targets): For inconsistent contexts
612 return empty complete list.
614 2014-04-02 Anthony Green <green@moxielogic.com>
616 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
617 (extendqisi2, extendhisi2): Define.
618 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
619 (WCHAR_TYPE): Change to unsigned int.
621 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
623 PR tree-optimization/60733
624 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
625 insertion point for PHI candidates to be the end of the feeding
626 block for the PHI argument.
628 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
630 PR rtl-optimization/60650
631 * lra-constraints.c (process_alt_operands): Decrease reject for
632 earlyclobber matching.
634 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
636 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
638 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
640 * config/spu/spu.c (pad_bb): Do not crash when the last
641 insn is CODE_FOR_blockage.
643 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
645 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
646 lies outside the target mode.
648 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
651 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
652 software floating point or no floating point registers, do not
653 allow any type in the FPRs. Eliminate a test for SPE SIMD types
654 in GPRs that occurs after we tested for GPRs that would never be
657 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
658 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
659 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
660 specifically allow DDmode, since that does not use the SPE SIMD
663 2014-04-02 Richard Biener <rguenther@suse.de>
666 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
667 MODE_INTs. Properly use negv_optab.
668 (expand_abs): Likewise.
670 2014-04-02 Richard Biener <rguenther@suse.de>
673 * Makefile.in (install-driver): Guard extra installs with special
676 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
678 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
681 2014-04-01 Richard Henderson <rth@redhat.com>
684 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
685 alternative enabled before register allocation.
687 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
689 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
690 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
692 (nios2_large_got_address): Remove unneeded 'sym' parameter.
693 (nios2_got_address): Update nios2_large_got_address call site.
694 (nios2_delegitimize_address): New function.
695 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
696 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
697 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
699 2014-04-01 Martin Husemann <martin@duskware.de>
701 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
704 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
706 PR rtl-optimization/60604
707 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
708 check from register_operand.
709 (register_operand): Redefine in terms of general_operand.
710 (nonmemory_operand): Use register_operand for the non-constant cases.
712 2014-04-01 Richard Biener <rguenther@suse.de>
714 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
716 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
718 * doc/invoke.texi (mapp-regs): Clarify.
720 2014-03-31 Ulrich Drepper <drepper@gmail.com>
722 * config/i386/avx512fintrin.h (__v32hi): Define type.
724 (_mm512_set1_epi8): Define.
725 (_mm512_set1_epi16): Define.
726 (_mm512_set4_epi32): Define.
727 (_mm512_set4_epi64): Define.
728 (_mm512_set4_pd): Define.
729 (_mm512_set4_ps): Define.
730 (_mm512_setr4_epi64): Define.
731 (_mm512_setr4_epi32): Define.
732 (_mm512_setr4_pd): Define.
733 (_mm512_setr4_ps): Define.
734 (_mm512_setzero_epi32): Define.
736 2014-03-31 Martin Jambor <mjambor@suse.cz>
739 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
740 callsite_arguments_match_p. Updated all callers. Also check types of
741 corresponding formal parameters and actual arguments.
742 (not_all_callers_have_enough_arguments_p) Renamed to
743 some_callers_have_mismatched_arguments_p.
745 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
747 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
749 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
752 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
755 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
757 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
758 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
760 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
761 Use FMAMODE_NOVF512 mode iterator.
762 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
763 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
764 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
765 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
767 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
768 Use VF_128_256 mode iterator.
769 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
772 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
774 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
775 static chain if needed.
777 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
780 * lra-constraints.c (index_part_to_reg): New.
781 (process_address): Use it.
783 2014-03-27 Jeff Law <law@redhat.com>
784 Jakub Jelinek <jakub@redhat.com>
787 * expr.c (do_tablejump): Use simplify_gen_binary rather than
788 gen_rtx_{PLUS,MULT} to build up the address expression.
790 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
791 creating non-canonical RTL.
793 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
796 * ipa-inline.c (want_inline_small_function_p): Short circuit large
797 functions; reorganize to make cheap checks first.
798 (inline_small_functions): Do not estimate growth when dumping;
800 * ipa-inline.h (inline_summary): Add min_size.
801 (growth_likely_positive): New function.
802 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
803 (set_cond_stmt_execution_predicate): Cleanup.
804 (estimate_edge_size_and_time): Compute min_size.
805 (estimate_calls_size_and_time): Likewise.
806 (estimate_node_size_and_time): Likewise.
807 (inline_update_overall_summary): Update min_size.
808 (do_estimate_edge_time): Likewise.
809 (do_estimate_edge_size): Update.
810 (do_estimate_edge_hints): Update.
811 (growth_likely_positive): New function.
813 2014-03-28 Jakub Jelinek <jakub@redhat.com>
816 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
817 also if addr has VOIDmode.
819 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
821 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
822 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
824 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
825 instructions as well as AdvancedSIMD loads.
827 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
829 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
830 Use crypto_aese type.
831 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
832 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
833 crypto_aese, crypto_aesmc. Move to types.md.
834 * config/arm/types.md (crypto_aes): Split into crypto_aese,
836 * config/arm/iterators.md (crypto_type): Likewise.
838 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
840 * cgraph.c: Include expr.h and tree-dfa.h.
841 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
844 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
847 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
848 regs from checking multi-reg pseudos.
850 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
852 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
854 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
856 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
857 if it would clobber the stack pointer, even temporarily.
859 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
861 * mode-switching.c: Make small adjustments to the top comment.
863 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
865 * config/rs6000/constraints.md (wD constraint): New constraint to
866 match the constant integer to get the top DImode/DFmode out of a
867 vector in a VSX register.
869 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
870 match the constant integer to get the top DImode/DFmode out of a
871 vector in a VSX register.
873 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
876 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
879 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
880 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
882 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
883 Optimize vec_extract of 64-bit values, where the value being
884 extracted is in the top word, where we can use scalar
885 instructions. Add direct move and store support. Combine the big
886 endian/little endian vector select load support into a single insn.
887 (vsx_extract_<mode>_internal1): Likewise.
888 (vsx_extract_<mode>_internal2): Likewise.
889 (vsx_extract_<mode>_load): Likewise.
890 (vsx_extract_<mode>_store): Likewise.
891 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
892 combined into vsx_extract_<mode>_load.
893 (vsx_extract_<mode>_one_le): Likewise.
895 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
896 define the top 64-bit vector element.
898 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
901 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
902 Document vec_vbpermq builtin.
905 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
906 enable use of xxsldwi and xxpermdi builtin functions.
907 (vec_xxpermdi): Likewise.
909 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
910 Document use of vec_xxsldwi and vec_xxpermdi builtins.
912 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
914 PR rtl-optimization/60650
915 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
917 (find_spills_for): New.
918 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
919 Spill all pseudos on the second iteration.
921 2014-03-27 Marek Polacek <polacek@redhat.com>
924 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
927 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
929 * config/s390/s390.c (s390_can_use_return_insn): Check for
930 call-saved FPRs on 31 bit.
932 2014-03-27 Jakub Jelinek <jakub@redhat.com>
935 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
936 if they need regimplification, just drop them instead of
937 calling gimple_regimplify_operands on them.
939 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
942 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
943 (aarch64_frame_pointer_required): Adjust logic.
944 (aarch64_can_eliminate): Adjust logic.
945 (aarch64_override_options_after_change): Adjust logic.
947 2014-03-27 Dehao Chen <dehao@google.com>
949 * ipa-inline.c (early_inliner): Update node's inline info.
951 2014-03-26 Dehao Chen <dehao@google.com>
953 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
954 compiler inserted conditional jumps for NAN float check.
956 2014-03-26 Jakub Jelinek <jakub@redhat.com>
958 * ubsan.h (ubsan_create_data): Change second argument's type
959 to const location_t *.
960 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
962 (ubsan_create_data): Change second argument to const location_t *PLOC.
963 Create Loc field whenever PLOC is non-NULL.
964 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
965 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
969 * real.c (real_to_integer2): Change type of low to UHWI.
971 2014-03-26 Tobias Burnus <burnus@net-b.de>
973 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
974 (CILK_SELF_SPECS): New define.
975 (driver_self_specs): Use it.
977 2014-03-26 Richard Biener <rguenther@suse.de>
979 * tree-pretty-print.c (percent_K_format): Implement special
980 case for LTO and its stripped down BLOCK tree.
982 2014-03-26 Jakub Jelinek <jakub@redhat.com>
985 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
987 * tree-vrp.c (simplify_internal_call_using_ranges): If only
988 one range is range_int_cst_p, but not both, at least optimize
989 addition/subtraction of 0 and multiplication by 0 or 1.
990 * gimple-fold.c (gimple_fold_call): Fold
991 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
992 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
993 INTEGER_CSTs, try to fold at least x * 0 and y - y.
995 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
997 PR rtl-optimization/60452
998 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
999 <case REG>: Return 1 for invalid offsets from the frame pointer.
1001 2014-03-26 Marek Polacek <polacek@redhat.com>
1004 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1007 2014-03-26 Marek Polacek <polacek@redhat.com>
1010 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1013 2014-03-26 Marek Polacek <polacek@redhat.com>
1016 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1017 multiplication in unsigned type.
1019 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1021 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
1023 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1025 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
1027 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1030 * cif-code.def (UNREACHABLE) New code.
1031 * ipa-inline.c (inline_small_functions): Skip edges to
1032 __builtlin_unreachable.
1033 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
1034 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
1035 predicate to __bulitin_unreachable.
1036 (set_cond_stmt_execution_predicate): Fix issue when
1037 invert_tree_comparison returns ERROR_MARK.
1038 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
1039 propagate to inline clones.
1040 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
1042 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
1043 * cgraphclones.c (cgraph_clone_node): If call destination is already
1044 ureachable, do not redirect it back.
1045 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
1048 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1050 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
1051 Do not modify inline clones.
1053 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1055 * config/i386/i386.md (general_sext_operand): New mode attr.
1056 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
1057 don't generate (sign_extend (const_int)).
1058 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
1059 operands[2]. Use We constraint instead of <i> and
1060 <general_sext_operand> predicate instead of <general_operand>.
1061 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
1062 * config/i386/constraints.md (We): New constraint.
1063 * config/i386/predicates.md (x86_64_sext_operand,
1064 sext_operand): New predicates.
1066 2014-03-25 Martin Jambor <mjambor@suse.cz>
1069 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1070 inconsistent devirtualizations to __builtin_unreachable.
1072 2014-03-25 Marek Polacek <polacek@redhat.com>
1075 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1077 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1079 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1080 order of elements for big-endian.
1082 2014-03-25 Richard Biener <rguenther@suse.de>
1085 * gimplify-me.c (gimple_regimplify_operands): Update the
1088 2014-03-25 Martin Jambor <mjambor@suse.cz>
1091 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1092 (lto_output_varpool_node): Likewise.
1093 (input_overwrite_node): Likewise.
1094 (input_varpool_node): Likewise.
1096 2014-03-25 Richard Biener <rguenther@suse.de>
1098 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1099 (run_gcc): Likewise.
1101 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1103 * combine.c (simplify_compare_const): Add MODE argument.
1104 Handle mode_width 0 as very large mode_width.
1105 (try_combine, simplify_comparison): Adjust callers.
1107 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1108 type to avoid signed integer overflow.
1109 * explow.c (plus_constant): Likewise.
1111 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1113 * doc/generic.texi: Correct typos.
1115 2014-03-24 Tobias Burnus <burnus@net-b.de>
1117 * doc/invoke.texi (-flto): Expand section about
1118 using static libraries with LTO.
1120 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1122 PR rtl-optimization/60501
1123 * optabs.def (addptr3_optab): New optab.
1124 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1125 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1126 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1128 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1130 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1132 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1134 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1137 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1138 (_mm256_undefined_ps): Define.
1139 (_mm256_undefined_pd): Define.
1140 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1141 (_mm_undefined_pd): Define.
1142 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1143 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1144 (_mm512_undefined_ps): Define.
1145 (_mm512_undefined_pd): Define.
1146 Use _mm*_undefined_*.
1147 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1149 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1151 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1152 (lshr_simd): DI mode added.
1153 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1154 (aarch64_ushr_simddi): Likewise.
1155 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1156 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1157 (vshrd_n_u64): Likewise.
1159 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1161 * Makefile.in (s-macro_list): Depend on cc1.
1163 2014-03-23 Teresa Johnson <tejohnson@google.com>
1165 * ipa-utils.c (ipa_print_order): Use specified dump file.
1167 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1169 PR rtl-optimization/60601
1170 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1172 * gcc.c (eval_spec_function): Initialize save_growing_value.
1174 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1177 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1178 code == MINUS_EXPR, never swap op0 with op1.
1180 * toplev.c (init_local_tick): Avoid signed integer multiplication
1182 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1183 shift by first operand's bitsize.
1185 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1188 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1190 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1191 TARGET_ISA_64BIT_P(x).
1193 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1195 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1196 pattern for vector nor instead of subtract from splat(-1).
1197 (altivec_expand_vec_perm_const_le): Likewise.
1199 2014-03-21 Richard Henderson <rth@twiddle.net>
1202 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1203 related insns after epilogue_completed.
1205 2014-03-21 Martin Jambor <mjambor@suse.cz>
1208 * cgraph.h (symtab_node): New flag body_removed.
1209 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1210 when removing bodies.
1211 * symtab.c (dump_symtab_base): Dump body_removed flag.
1212 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1213 had their bodies removed.
1215 2014-03-21 Martin Jambor <mjambor@suse.cz>
1218 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1221 2014-03-21 Richard Biener <rguenther@suse.de>
1223 PR tree-optimization/60577
1224 * tree-core.h (struct tree_base): Document nothrow_flag use
1225 in VAR_DECL_NONALIASED.
1226 * tree.h (VAR_DECL_NONALIASED): New.
1227 (may_be_aliased): Adjust.
1228 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1230 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1232 * expr.c (expand_expr_real_1): Remove outdated comment.
1234 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1237 * ira.c (adjust_cleared_regs): Call copy_rtx on
1238 *reg_equiv[REGNO (loc)].src_p before passing it to
1239 simplify_replace_fn_rtx.
1242 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1243 into CONST, put pic register as first operand of PLUS. Use
1244 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1246 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1248 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1250 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1252 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1253 around for store forwarding issue in the FPU on the UT699.
1254 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1255 loads and operations if -mfix-ut699 is specified.
1256 (divtf3_hq): Tweak attribute.
1257 (sqrttf2_hq): Likewise.
1259 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1261 * calls.c (store_one_arg): Remove incorrect const qualification on the
1262 type of the temporary.
1263 * cfgexpand.c (expand_return): Likewise.
1264 * expr.c (expand_constructor): Likewise.
1265 (expand_expr_real_1): Likewise.
1267 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1269 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1272 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1275 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1277 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1279 * config/arm/aarch-common-protos.h
1280 (alu_cost_table): Fix spelling of "extend".
1281 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1283 2014-03-19 Richard Biener <rguenther@suse.de>
1286 * tree-core.h (tree_type_common): Re-order pointer members
1287 to reduce recursion depth during GC walks.
1289 2014-03-19 Marek Polacek <polacek@redhat.com>
1292 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1293 before accessing it.
1295 2014-03-19 Richard Biener <rguenther@suse.de>
1298 * lto-streamer-in.c (input_function): In WPA stage do not drop
1301 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1303 PR tree-optimization/60559
1304 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1305 with build_zero_cst assignment.
1307 2014-03-18 Kai Tietz <ktietz@redhat.com>
1309 PR rtl-optimization/56356
1310 * sdbout.c (sdbout_parms): Verify that parms'
1311 incoming argument is valid.
1312 (sdbout_reg_parms): Likewise.
1314 2014-03-18 Richard Henderson <rth@redhat.com>
1317 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1318 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1319 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1321 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1323 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1324 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1325 Italicize plugin event names in description. Explain that
1326 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1327 Remind that no GCC functions should be called after PLUGIN_FINISH.
1328 Explain what pragmas with expansion are.
1330 2014-03-18 Martin Liska <mliska@suse.cz>
1332 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1333 gimple call statement is update.
1334 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1335 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1337 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1340 * ubsan.c (ubsan_instrument_unreachable): Call
1341 initialize_sanitizer_builtins.
1342 (ubsan_pass): Likewise.
1345 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1346 varpool_finalize_decl instead of rest_of_decl_compilation.
1348 2014-03-18 Richard Biener <rguenther@suse.de>
1350 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1351 by using bitmap_and_compl instead of bitmap_and_compl_into.
1352 (df_rd_transfer_function): Likewise.
1354 2014-03-18 Richard Biener <rguenther@suse.de>
1356 * doc/lto.texi (fresolution): Fix typo.
1358 2014-03-18 Richard Biener <rguenther@suse.de>
1360 * doc/invoke.texi (flto): Update for changes in 4.9.
1362 2014-03-18 Richard Biener <rguenther@suse.de>
1364 * doc/loop.texi: Remove section on the removed lambda framework.
1365 Update loop docs with recent changes in preserving loop structure.
1367 2014-03-18 Richard Biener <rguenther@suse.de>
1369 * doc/lto.texi (-fresolution): Document.
1371 2014-03-18 Richard Biener <rguenther@suse.de>
1373 * doc/contrib.texi: Adjust my name.
1375 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1378 * internal-fn.c: Include diagnostic-core.h.
1379 (expand_BUILTIN_EXPECT): New function.
1380 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1381 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1382 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1383 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1385 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1386 Revert 3 argument __builtin_expect code.
1387 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1388 * gimple-fold.c (gimple_fold_call): Likewise.
1389 * tree.h (fold_builtin_expect): New prototype.
1390 * builtins.c (build_builtin_expect_predicate): Add predictor
1391 argument, if non-NULL, create 3 argument __builtin_expect.
1392 (fold_builtin_expect): No longer static. Add ARG2 argument,
1393 pass it through to build_builtin_expect_predicate.
1394 (fold_builtin_2): Adjust caller.
1395 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1396 * internal-fn.def (BUILTIN_EXPECT): New.
1398 2014-03-18 Tobias Burnus <burnus@net-b.de>
1401 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1402 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1403 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1405 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1408 * predict.c (combine_predictions_for_bb): Fix up formatting.
1409 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1410 fill what it points to if non-NULL.
1411 (tree_predict_by_opcode): Adjust caller, use the predictor.
1412 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1414 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1416 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1417 proper constant for the store mode.
1419 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1421 * symtab.c (change_decl_assembler_name): Fix transparent alias
1424 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1426 * config/aarch64/aarch64.c: Correct the comments about the
1427 aarch64 stack layout.
1429 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1431 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1432 check for GF_OMP_FOR_KIND_FOR.
1434 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1436 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1437 ymm and zmm register names.
1439 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1442 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1443 note creation for the 2010-08-31 changes.
1445 2014-03-17 Marek Polacek <polacek@redhat.com>
1448 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1449 as -fno-tree-loop-vectorize.
1450 (expand_omp_simd): Likewise.
1452 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1454 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1455 (eligible_for_call_delay): New prototype.
1456 * config/sparc/sparc.c (tls_call_delay): Rename into...
1457 (eligible_for_call_delay): ...this. Return false if the instruction
1458 cannot be put in the delay slot of a branch.
1459 (eligible_for_restore_insn): Simplify.
1460 (eligible_for_return_delay): Return false if the instruction cannot be
1461 put in the delay slot of a branch and simplify.
1462 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1463 put in the delay slot of a branch.
1464 * config/sparc/sparc.md (fix_ut699): New attribute.
1465 (tls_call_delay): Delete.
1466 (in_call_delay): Reimplement.
1467 (eligible_for_sibcall_delay): Rename into...
1468 (in_sibcall_delay): ...this.
1469 (eligible_for_return_delay): Rename into...
1470 (in_return_delay): ...this.
1471 (in_branch_delay): Reimplement.
1472 (in_uncond_branch_delay): Delete.
1473 (in_annul_branch_delay): Delete.
1475 2014-03-14 Richard Henderson <rth@redhat.com>
1478 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1479 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1480 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1481 (floathi splitters): Remove.
1482 (float<SWI48x>xf2): New pattern.
1483 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1484 code that tried to handle DImode for 32-bit, but which was excluded
1485 by the pattern's condition. Drop allocation of stack temporary.
1486 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1487 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1488 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1489 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1490 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1491 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1492 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1493 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1494 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1495 (*float<SWI48x><X87MODEF>2_i387): Remove.
1496 (all float _with_temp splitters): Remove.
1497 (*float<SWI48x><MODEF>2_i387): New pattern.
1498 (*float<SWI48><MODEF>2_sse): New pattern.
1499 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1500 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1502 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1503 Marek Polacek <polacek@redhat.com>
1506 * common.opt (dump_base_name_prefixed): New Variable.
1507 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1508 if x_dump_base_name_prefixed is already set, set it at the end.
1510 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1512 PR rtl-optimization/60508
1513 * lra-constraints.c (get_reload_reg): Add new parameter
1515 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1516 Pass the new parameter values.
1518 2014-03-14 Richard Biener <rguenther@suse.de>
1520 * common.opt: Revert unintented changes from r205065.
1523 2014-03-14 Richard Biener <rguenther@suse.de>
1526 * cfghooks.c (split_block): Properly adjust all loops the
1527 block was a latch of.
1529 2014-03-14 Martin Jambor <mjambor@suse.cz>
1532 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1535 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1538 * config/avr/avr.c (avr_set_current_function): Pass function name
1539 through default_strip_name_encoding before sanity checking instead
1540 of skipping the first char of the assembler name.
1542 2014-03-13 Richard Henderson <rth@redhat.com>
1545 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1546 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1547 * config/i386/i386-protos.h: Likewise.
1548 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1549 in the expander instead of a splitter.
1550 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1551 any possibility of requiring a memory.
1552 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1553 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1554 (fp branch splitters): Update for ix86_split_fp_branch.
1555 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1556 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1557 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1558 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1559 (*fop_<MODEF>_3_i387): Likewise.
1560 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1561 (splitters for the fop_* register patterns): Remove.
1562 (fscalexf4_i387): Rename from *fscalexf4_i387.
1563 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1565 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1567 PR tree-optimization/59779
1568 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1569 type for bitsize and maxsize instead of HOST_WIDE_INT.
1571 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1573 PR rtl-optimization/57320
1574 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1575 the CFG after thread_prologue_and_epilogue_insns.
1577 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1579 PR rtl-optimization/57189
1580 * lra-constraints.c (process_alt_operands): Disfavor spilling
1583 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1585 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1587 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1589 PR tree-optimization/59025
1591 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1592 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1594 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1597 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1598 calls of avr_out_plus_1.
1600 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1602 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1603 BB's single pred and update the father loop's latch info later.
1605 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1607 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1611 (VEC_base): Likewise.
1612 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1613 registers, we need to swap double words in little endian mode.
1615 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1616 to be a container mode for 128-bit integer operations added in ISA
1617 2.07. Unlike TImode and PTImode, the preferred register set is
1618 the Altivec/VMX registers for the 128-bit operations.
1620 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1622 (rs6000_split_128bit_ok_p): Likewise.
1624 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1625 macros for creating ISA 2.07 normal and overloaded builtin
1626 functions with 3 arguments.
1627 (BU_P8V_OVERLOAD_3): Likewise.
1628 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1629 for use as overloaded functions.
1630 (VPERM_1TI_UNS): Likewise.
1631 (VSEL_1TI): Likewise.
1632 (VSEL_1TI_UNS): Likewise.
1633 (ST_INTERNAL_1ti): Likewise.
1634 (LD_INTERNAL_1ti): Likewise.
1635 (XXSEL_1TI): Likewise.
1636 (XXSEL_1TI_UNS): Likewise.
1637 (VPERM_1TI): Likewise.
1638 (VPERM_1TI_UNS): Likewise.
1639 (XXPERMDI_1TI): Likewise.
1640 (SET_1TI): Likewise.
1641 (LXVD2X_V1TI): Likewise.
1642 (STXVD2X_V1TI): Likewise.
1643 (VEC_INIT_V1TI): Likewise.
1644 (VEC_SET_V1TI): Likewise.
1645 (VEC_EXT_V1TI): Likewise.
1646 (EQV_V1TI): Likewise.
1647 (NAND_V1TI): Likewise.
1648 (ORC_V1TI): Likewise.
1649 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1650 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1652 (VADDUQM): Likewise.
1653 (VSUBCUQ): Likewise.
1654 (VADDEUQM): Likewise.
1655 (VADDECUQ): Likewise.
1656 (VSUBEUQM): Likewise.
1657 (VSUBECUQ): Likewise.
1659 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1660 __int128_t and __uint128_t types.
1661 (__uint128_type): Likewise.
1662 (altivec_categorize_keyword): Add support for vector __int128_t,
1663 vector __uint128_t, vector __int128, and vector unsigned __int128
1664 as a container type for TImode operations that need to be done in
1665 VSX/Altivec registers.
1666 (rs6000_macro_to_expand): Likewise.
1667 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1668 to support 128-bit integer instructions vaddcuq, vadduqm,
1669 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1670 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1672 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1673 for V1TImode, and set up preferences to use VSX/Altivec registers.
1674 Setup VSX reload handlers.
1675 (rs6000_debug_reg_global): Likewise.
1676 (rs6000_init_hard_regno_mode_ok): Likewise.
1677 (rs6000_preferred_simd_mode): Likewise.
1678 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1679 (easy_altivec_constant): Likewise.
1680 (output_vec_const_move): Likewise.
1681 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1683 (rs6000_expand_vector_extract): Likewise.
1684 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1686 (rs6000_const_vec): Add support for V1TImode.
1687 (rs6000_emit_le_vsx_load): Swap double words when loading or
1688 storing TImode/V1TImode.
1689 (rs6000_emit_le_vsx_store): Likewise.
1690 (rs6000_emit_le_vsx_move): Likewise.
1691 (rs6000_emit_move): Add support for V1TImode.
1692 (altivec_expand_ld_builtin): Likewise.
1693 (altivec_expand_st_builtin): Likewise.
1694 (altivec_expand_vec_init_builtin): Likewise.
1695 (altivec_expand_builtin): Likewise.
1696 (rs6000_init_builtins): Add support for V1TImode type. Add
1697 support for ISA 2.07 128-bit integer builtins. Define type names
1698 for the VSX/Altivec vector types.
1699 (altivec_init_builtins): Add support for overloaded vector
1700 functions with V1TImode type.
1701 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1702 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1704 (rs6000_split_128bit_ok_p): Likewise.
1705 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1706 __int128_t and vector __uint128_t.
1708 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1709 and mode attributes.
1716 (VS_scalar): Likewise.
1717 (VS_double): Likewise.
1718 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1720 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1721 we support the ISA 2.07 128-bit integer arithmetic instructions.
1722 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1723 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1724 and TImode types for use with the builtin functions.
1725 (V1TI_type_node): Likewise.
1726 (unsigned_V1TI_type_node): Likewise.
1727 (intTI_type_internal_node): Likewise.
1728 (uintTI_type_internal_node): Likewise.
1730 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1731 128-bit builtin functions.
1732 (UNSPEC_VADDEUQM): Likewise.
1733 (UNSPEC_VADDECUQ): Likewise.
1734 (UNSPEC_VSUBCUQ): Likewise.
1735 (UNSPEC_VSUBEUQM): Likewise.
1736 (UNSPEC_VSUBECUQ): Likewise.
1737 (VM): Add V1TImode to vector mode iterators.
1739 (VI_unit): Likewise.
1740 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1741 (altivec_vaddcuq): Likewise.
1742 (altivec_vsubuqm): Likewise.
1743 (altivec_vsubcuq): Likewise.
1744 (altivec_vaddeuqm): Likewise.
1745 (altivec_vaddecuq): Likewise.
1746 (altivec_vsubeuqm): Likewise.
1747 (altivec_vsubecuq): Likewise.
1749 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1751 (BOOL_128): Likewise.
1752 (BOOL_REGS_OUTPUT): Likewise.
1753 (BOOL_REGS_OP1): Likewise.
1754 (BOOL_REGS_OP2): Likewise.
1755 (BOOL_REGS_UNARY): Likewise.
1756 (BOOL_REGS_AND_CR0): Likewise.
1758 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1759 128-bit integer builtin support.
1760 (vec_vadduqm): Likewise.
1761 (vec_vaddecuq): Likewise.
1762 (vec_vaddeuqm): Likewise.
1763 (vec_vsubecuq): Likewise.
1764 (vec_vsubeuqm): Likewise.
1765 (vec_vsubcuq): Likewise.
1766 (vec_vsubuqm): Likewise.
1768 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1769 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1770 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1771 128-bit integer add/subtract to ISA 2.07.
1773 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1775 * config/arc/arc.c (arc_predicate_delay_insns):
1776 Fix third argument passed to conditionalize_nonjump.
1778 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1780 * config/aarch64/aarch64-builtins.c
1781 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1782 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1783 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1784 instead of __builtin_lfloor.
1785 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1787 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1789 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1790 (tree_ssa_ifcombine_bb_1): New function.
1791 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1792 is an empty forwarder block to then_bb or vice versa and then_bb
1793 and else_bb are effectively swapped.
1795 2014-03-12 Christian Bruel <christian.bruel@st.com>
1798 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1799 REG_CFA_DEF_CFA note.
1800 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1801 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1803 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1805 PR tree-optimization/60454
1806 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1808 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1810 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1811 Do not define target_cpu_default2 to generic.
1812 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1813 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1814 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1816 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1817 Marc Glisse <marc.glisse@inria.fr>
1819 PR tree-optimization/60502
1820 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1821 instead of build_low_bits_mask.
1823 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1826 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1827 if there are multiple uses, but op doesn't live on E edge.
1828 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1829 clobber stmts before __builtin_unreachable.
1831 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1833 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1834 hard_frame_pointer_rtx.
1835 * cse.c (cse_insn): Remove volatile check.
1836 * cselib.c (cselib_process_insn): Likewise.
1837 * dse.c (scan_insn): Likewise.
1839 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1841 * config/arc/arc.c (conditionalize_nonjump): New function,
1843 (arc_ifcvt): ... this.
1844 (arc_predicate_delay_insns): Use it.
1846 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1848 * config/arc/predicates.md (extend_operand): During/after reload,
1849 allow const_int_operand.
1850 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1851 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1852 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1854 (umulsi3_highpart_i): Likewise.
1856 2014-03-11 Richard Biener <rguenther@suse.de>
1858 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1859 Add asserts to guard possible wrong-code bugs.
1861 2014-03-11 Richard Biener <rguenther@suse.de>
1863 PR tree-optimization/60429
1864 PR tree-optimization/60485
1865 * tree-ssa-structalias.c (set_union_with_increment): Properly
1866 take into account all fields that overlap the shifted vars.
1867 (do_sd_constraint): Likewise.
1868 (do_ds_constraint): Likewise.
1869 (get_constraint_for_ptr_offset): Likewise.
1871 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1873 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1874 (nios2_compute_frame_layout):
1875 Add calculation of cfun->machine->fp_save_offset.
1876 (nios2_expand_prologue): Correct setting of frame pointer register
1878 (nios2_expand_epilogue): Update recovery of stack pointer from
1879 frame pointer accordingly.
1880 (nios2_initial_elimination_offset): Update calculation of offset
1881 for eliminating to HARD_FRAME_POINTER_REGNUM.
1883 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1886 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1887 cgraph_get_create_node on VAR_DECLs.
1889 2014-03-10 Richard Biener <rguenther@suse.de>
1892 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1894 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1896 * config/vms/vms.opt (vms_float_format): New variable.
1898 2014-03-08 Tobias Burnus <burnus@net-b.de>
1900 * doc/invoke.texi (-fcilkplus): Update implementation status.
1902 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1903 Richard Biener <rguenther@suse.de>
1905 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1906 consistently accross all TUs.
1907 (run_gcc): Enable -fshort-double automatically at link at link-time
1908 and disallow override.
1910 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1913 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1914 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1915 if they can't be used.
1917 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1919 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1920 for Solaris 11/x86 ld.
1921 * configure: Regenerate.
1923 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1925 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1926 (LIB_TLS_SPEC): Save as ld_tls_libs.
1927 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1928 (HAVE_AS_IX86_TLSLDM): New test.
1929 * configure, config.in: Regenerate.
1930 * config/i386/i386.c (legitimize_tls_address): Fall back to
1931 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1932 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1933 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1934 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1936 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1938 * common.opt (fira-loop-pressure): Mark as optimization.
1940 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1942 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1943 an OpenMP mappable type.
1945 2014-03-06 Matthias Klose <doko@ubuntu.com>
1947 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1948 MULTILIB_OSDIRNAMES is not defined.
1950 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1951 Meador Inge <meadori@codesourcery.com>
1954 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1955 (arm_legitimize_address): Call legitimize_tls_address for any
1956 arm_tls_referenced_p expression, handle constant addend. Call it
1957 before testing for !TARGET_ARM.
1958 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1960 2014-03-06 Richard Biener <rguenther@suse.de>
1966 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1968 * tree-streamer.c (record_common_node): Assert we don't record
1969 nodes with type double.
1970 (preload_common_node): Skip type double, complex double and double
1971 pointer since it is now frontend dependent due to fshort-double option.
1973 2014-03-06 Richard Biener <rguenther@suse.de>
1975 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1976 or -fno-lto is specified and the linker has full plugin support.
1977 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1978 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1979 * lto-wrapper.c (merge_and_complain): Merge compile-time
1980 optimization levels.
1981 (run_gcc): And pass it through to the link options.
1983 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1987 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1989 * cselib.c (remove_useless_values): Skip to avoid quadratic
1990 behavior if the condition moved from...
1991 (cselib_process_insn): ... here holds.
1993 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1996 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1997 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
2000 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
2001 (TM_H): Add x86-tune.def.
2003 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2005 * config/aarch64/aarch64.c (generic_tunings):
2006 Use cortexa57_extra_costs.
2008 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2011 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2012 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2013 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2016 2014-03-04 Heiher <r@hev.cc>
2018 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
2019 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
2021 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
2023 * config/i386/predicates.md (const2356_operand): Change to ...
2024 (const2367_operand): ... this.
2025 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
2027 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2028 (*avx512pf_scatterpf<mode>sf): Ditto.
2029 (avx512pf_scatterpf<mode>df): Ditto.
2030 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2031 (*avx512pf_scatterpf<mode>df): Ditto.
2032 * config/i386/i386.c (ix86_expand_builtin): Update
2033 incorrect hint operand error message.
2035 2014-03-04 Richard Biener <rguenther@suse.de>
2037 * lto-section-in.c (lto_get_section_data): Fix const cast.
2039 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2041 * tree-streamer.c (record_common_node): Assert we don't record
2042 nodes with type double.
2043 (preload_common_node): Skip type double, complex double and double
2044 pointer since it is now frontend dependent due to fshort-double option.
2046 2014-03-04 Richard Biener <rguenther@suse.de>
2049 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
2050 (lto_input_toplevel_asms): Likewise.
2051 * lto-section-in.c (lto_get_section_data): Instead do it here
2054 2014-03-04 Richard Biener <rguenther@suse.de>
2056 PR tree-optimization/60382
2057 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
2058 dead PHIs a reduction.
2060 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
2062 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2064 (_mm_prefetch): Move out of GCC target("sse") pragma.
2065 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2066 GCC target("prfchw") pragma.
2067 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2069 * config/i386/i386.c (ix86_option_override_internal): Enable
2070 -mprfchw with -mprefetchwt1.
2072 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2074 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2077 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2079 * opts.h (CL_PCH_IGNORE): Define.
2080 * targhooks.c (option_affects_pch_p):
2081 Return false for options that have CL_PCH_IGNORE set.
2082 * opt-functions.awk: Process PchIgnore.
2083 * doc/options.texi: Document PchIgnore.
2085 * config/arc/arc.opt (misize): Add PchIgnore property.
2087 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2089 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2090 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2091 constraint on constants to permit them being loaded into
2092 GENERAL_REGS or BASE_REGS.
2094 2014-03-03 Nick Clifton <nickc@redhat.com>
2096 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2097 anti-cacnonical alternatives.
2098 (negandhi3_real): New pattern.
2099 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2101 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2103 * config/avr/avr-mcus.def: Remove atxmega16x1.
2104 * config/avr/avr-tables.opt: Regenerate.
2105 * config/avr/t-multilib: Regenerate.
2106 * doc/avr-mmcu.texi: Regenerate.
2108 2014-03-03 Tobias Grosser <tobias@grosser.es>
2109 Mircea Namolaru <mircea.namolaru@inria.fr>
2111 PR tree-optimization/58028
2112 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2115 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2117 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2118 not handled by recognizers.
2120 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2123 * function.c (expand_function_end): Don't emit
2124 clobber_return_register sequence if clobber_after is a BARRIER.
2125 * cfgexpand.c (construct_exit_block): Append instructions before
2126 return_label to prev_bb.
2128 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2130 * config/rs6000/constraints.md: Document reserved use of "wc".
2132 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2135 * ipa.c (function_and_variable_visibility): When dissolving comdat
2136 group, also set all symbols to local.
2138 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2143 2013-12-14 Jan Hubicka <jh@suse.cz>
2145 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2147 2014-03-02 Jon Beniston <jon@beniston.com>
2153 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2154 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2155 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2156 (simple_return, *simple_return): New patterns
2157 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2158 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2160 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2162 * dwarf2out.c (gen_subprogram_die): Tidy.
2164 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2167 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2168 (*mov_t_msb_neg_negc): ... this new insn.
2170 2014-02-28 Jason Merrill <jason@redhat.com>
2173 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2176 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2179 * dwarf2out.c (decltype_auto_die): New static.
2180 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2181 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2182 (is_cxx_auto): Likewise.
2184 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2186 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2187 we are not using general regs only.
2189 2014-02-28 Richard Biener <rguenther@suse.de>
2192 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2193 previous fix and only allow to remove trivial pre-headers
2194 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2195 (remove_forwarder_block): Properly update the latch of a loop.
2197 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2200 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2201 (cselib_preserved_hash_table): New.
2202 (preserve_constants_and_equivs): Move preserved vals to it.
2203 (cselib_find_slot): Look it up first.
2204 (cselib_init): Initialize it.
2205 (cselib_finish): Release it.
2206 (dump_cselib_table): Dump it.
2208 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2211 * cselib.c (remove_useless_values): Skip to avoid quadratic
2212 behavior if the condition moved from...
2213 (cselib_process_insn): ... here holds.
2215 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2218 * var-tracking.c (vt_initialize): Apply the same condition to
2219 preserve the CFA base value.
2221 2014-02-28 Joey Ye <joey.ye@arm.com>
2224 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2225 if reload in progress or completed.
2227 2014-02-28 Tobias Burnus <burnus@net-b.de>
2230 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2233 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2235 * doc/tm.texi.in (Condition Code Status): Update documention for
2236 relative locations of cc0-setter and cc0-user.
2238 2014-02-27 Jeff Law <law@redhat.com>
2240 PR rtl-optimization/52714
2241 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2242 into two independent simple sets, if I3 is a jump, ensure the
2243 pattern we place into I3 is a (set (pc) ...).
2245 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2246 Jeff Law <law@redhat.com>
2248 PR rtl-optimization/49847
2249 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2250 are in different blocks.
2251 * doc/tm.texi (Condition Code Status): Update documention for
2252 relative locations of cc0-setter and cc0-user.
2254 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2257 * lra.c (lra_emit_add): Check SUBREG too.
2259 2014-02-27 Andreas Schwab <schwab@suse.de>
2261 * config/m68k/m68k.c (m68k_option_override): Disable
2262 -flive-range-shrinkage for classic m68k.
2263 (m68k_override_options_after_change): Likewise.
2265 2014-02-27 Marek Polacek <polacek@redhat.com>
2268 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2269 -Wmaybe-uninitialized.
2271 2014-02-27 Alan Modra <amodra@gmail.com>
2274 * reload1.c (emit_input_reload_insns): When reload_override_in,
2275 set old to rl->in_reg when rl->in_reg is a subreg.
2277 2014-02-26 Richard Biener <rguenther@suse.de>
2280 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2282 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2284 * common/config/i386/predicates.md (const1256_operand): Remove.
2285 (const2356_operand): New.
2286 (const_1_to_2_operand): Remove.
2287 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2288 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2289 (*avx512pf_gatherpf<mode>sf): Ditto.
2290 (avx512pf_gatherpf<mode>df): Ditto.
2291 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2292 (*avx512pf_gatherpf<mode>df): Ditto.
2293 (avx512pf_scatterpf<mode>sf): Ditto.
2294 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2295 (*avx512pf_scatterpf<mode>sf): Ditto.
2296 (avx512pf_scatterpf<mode>df): Ditto.
2297 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2298 (*avx512pf_scatterpf<mode>df): Ditto.
2299 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2301 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2303 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2304 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2305 (_mm512_mask_testn_epi64_mask): Move to ...
2306 * config/i386/avx512cdintrin.h: Here.
2307 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2308 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2309 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2310 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2311 TARGET_AVX512F from TARGET_AVX512CD.
2313 2014-02-26 Richard Biener <rguenther@suse.de>
2316 * ipa.c (walk_polymorphic_call_targets): Properly guard
2317 call to inline_update_overall_summary.
2319 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2322 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2323 and latches only if requested. Fix latch if it is removed.
2324 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2325 LOOPS_HAVE_PREHEADERS.
2327 2014-02-25 Andrew Pinski <apinski@cavium.com>
2329 * builtins.c (expand_builtin_thread_pointer): Create a new target
2330 when the target is NULL.
2332 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2334 PR rtl-optimization/60317
2335 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2336 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2337 * lra-assigns.c: Include params.h.
2338 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2339 other reload pseudos considerations.
2341 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2343 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2344 to use canonical form for nor<mode>3.
2346 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2349 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2352 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2354 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2355 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2356 (ix86_handle_option): Handle OPT_mprefetchwt1.
2357 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2358 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2360 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2361 OPTION_MASK_ISA_PREFETCHWT1.
2362 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2363 (PTA_PREFETCHWT1): New.
2364 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2365 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2366 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2367 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2368 (*prefetch_avx512pf_<mode>_: Change into ...
2369 (*prefetch_prefetchwt1_<mode>: This.
2370 * config/i386/i386.opt (mprefetchwt1): New.
2371 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2372 (_mm_prefetch): Handle intent to write.
2373 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2375 2014-02-25 Richard Biener <rguenther@suse.de>
2378 * emit-rtl.c (mem_attrs_htab): Remove.
2379 (mem_attrs_htab_hash): Likewise.
2380 (mem_attrs_htab_eq): Likewise.
2381 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2382 (init_emit_once): Do not allocate mem_attrs_htab.
2384 2014-02-25 Richard Biener <rguenther@suse.de>
2387 * lto-opts.c (lto_write_options): Output non-explicit conservative
2388 -fwrapv, -fno-trapv and -fno-strict-overflow.
2389 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2390 (run_gcc): And pass them through.
2392 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2394 * sel-sched.c (calculate_new_fences): New parameter ptime.
2395 Calculate it as a maximum over all fence cycles.
2396 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2397 Print the final schedule timing when sched_verbose.
2399 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2401 PR rtl-optimization/60292
2402 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2403 bit fot the fence instruction.
2405 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2407 * calls.h: Fix typo in comment.
2409 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2411 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2412 adjusting offsetable addresses.
2414 2014-02-24 Guozhi Wei <carrot@google.com>
2416 * sparseset.h (sparseset_pop): Fix the wrong index.
2418 2014-02-24 Walter Lee <walt@tilera.com>
2420 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2421 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2423 * common/config/tilegx/tilegx-common.c
2424 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2425 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2427 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2428 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2429 (tilegx_gimplify_va_arg_expr): Handle big endian.
2430 (tilegx_expand_unaligned_load): Ditto.
2431 (tilegx_expand_unaligned_store): Ditto.
2432 (TARGET_RETURN_IN_MSB): New.
2433 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2434 (TARGET_ENDIAN_DEFAULT): New.
2435 (TARGET_BIG_ENDIAN): Handle big endian.
2436 (BYTES_BIG_ENDIAN): Ditto.
2437 (WORDS_BIG_ENDIAN): Ditto.
2438 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2441 * config/tilegx/tilegx.md (extv): Handle big endian.
2443 (insn_st<n>): Ditto.
2444 (insn_st<n>_add<bitsuffix>): Ditto.
2445 (insn_stnt<n>): Ditto.
2446 (insn_stnt<n>_add<bitsuffix>):Ditto.
2447 (vec_interleave_highv8qi): Handle big endian.
2448 (vec_interleave_highv8qi_be): New.
2449 (vec_interleave_highv8qi_le): New.
2450 (insn_v1int_h): Handle big endian.
2451 (vec_interleave_lowv8qi): Handle big endian.
2452 (vec_interleave_lowv8qi_be): New.
2453 (vec_interleave_lowv8qi_le): New.
2454 (insn_v1int_l): Handle big endian.
2455 (vec_interleave_highv4hi): Handle big endian.
2456 (vec_interleave_highv4hi_be): New.
2457 (vec_interleave_highv4hi_le): New.
2458 (insn_v2int_h): Handle big endian.
2459 (vec_interleave_lowv4hi): Handle big endian.
2460 (vec_interleave_lowv4hi_be): New.
2461 (vec_interleave_lowv4hi_le): New.
2462 (insn_v2int_l): Handle big endian.
2463 (vec_interleave_highv2si): Handle big endian.
2464 (vec_interleave_highv2si_be): New.
2465 (vec_interleave_highv2si_le): New.
2466 (insn_v4int_h): Handle big endian.
2467 (vec_interleave_lowv2si): Handle big endian.
2468 (vec_interleave_lowv2si_be): New.
2469 (vec_interleave_lowv2si_le): New.
2470 (insn_v4int_l): Handle big endian.
2471 * config/tilegx/tilegx.opt (mbig-endian): New option.
2472 (mlittle-endian): New option.
2473 * doc/install.texi: Document tilegxbe-linux.
2474 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2476 2014-02-24 Martin Jambor <mjambor@suse.cz>
2479 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2480 there are no parameter descriptors.
2482 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2484 PR rtl-optimization/60268
2485 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2486 initialization to ...
2487 (sched_rgn_init): ... here.
2488 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2490 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2492 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2495 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2497 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2500 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2502 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2503 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2505 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2507 * config/microblaze/predicates.md: Add cmp_op predicate.
2508 * config/microblaze/microblaze.md: Add branch_compare instruction
2509 which uses cmp_op predicate and emits cmp insn before branch.
2510 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2511 to microblaze_expand_conditional_branch and consolidate logic.
2512 (microblaze_expand_conditional_branch): emit branch_compare
2513 insn instead of handling cmp op separate from branch insn.
2515 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2517 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2520 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2522 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2523 define_insn with define_expand and new define_insn
2524 *altivec_lve<VI_char>x_internal.
2525 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2526 and new define_insn *altivec_stve<VI_char>x_internal.
2527 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2529 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2531 (altivec_expand_stvex_be): New function.
2533 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2535 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2536 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2537 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2538 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2540 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2543 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2544 instead of emit_move_insn.
2546 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2548 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2550 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2551 gen_altivec_vsumsws.
2553 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2555 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2556 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2557 (altivec_lvxl_<mode>): New define_expand incorporating
2558 -maltivec=be semantics where needed.
2559 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2560 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2561 semantics where needed.
2562 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2563 (altivec_stvx_<mode>): New define_expand incorporating
2564 -maltivec=be semantics where needed.
2565 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2566 VM2 iterator instead of V4SI.
2567 (altivec_stvxl_<mode>): New define_expand incorporating
2568 -maltivec=be semantics where needed.
2569 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2570 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2571 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2572 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2573 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2574 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2575 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2576 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2577 ALTIVEC_BUILTIN_STVXL.
2578 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2579 (altivec_expand_stvx_be): Likewise.
2580 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2581 (altivec_expand_lvx_be): Likewise.
2582 (altivec_expand_stvx_be): Likewise.
2583 (altivec_expand_builtin): Add cases for
2584 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2585 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2586 (altivec_init_builtins): Add definitions for
2587 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2588 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2590 2014-02-21 Catherine Moore <clm@codesourcery.com>
2592 * doc/invoke.texi (mvirt, mno-virt): Document.
2593 * config/mips/mips.opt (mvirt): New option.
2594 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2596 2014-02-21 Richard Biener <rguenther@suse.de>
2598 PR tree-optimization/60276
2599 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2600 (STMT_VINFO_MIN_NEG_DIST): New macro.
2601 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2602 STMT_VINFO_MIN_NEG_DIST.
2603 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2604 made for negative dependence distances still hold.
2606 2014-02-21 Richard Biener <rguenther@suse.de>
2609 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2610 DECL_INITIAL for globals not in the current function context.
2612 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2614 PR tree-optimization/56490
2615 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2616 * tree-ssa-uninit.c: Include params.h.
2617 (compute_control_dep_chain): Add num_calls argument, return false
2618 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2619 num_calls to recursive call.
2620 (find_predicates): Change dep_chain into normal array,
2621 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2622 variable and adjust compute_control_dep_chain caller.
2623 (find_def_preds): Likewise.
2625 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2627 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2628 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2630 2014-02-21 Nick Clifton <nickc@redhat.com>
2632 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2633 (pushhi1): Likewise.
2634 (popqi1): Add mode to pre_dec.
2637 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2639 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2640 mode for mask of V8SFmode permutation.
2642 2014-02-20 Richard Henderson <rth@redhat.com>
2645 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2646 a new pseudo for OLDVAL.
2648 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2651 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2652 gen_reg_rtx if d->testing_p.
2653 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2654 if d->testing_p and we will certainly return true.
2655 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2658 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2660 * emit-rtl.c (gen_reg_rtx): Assert that
2661 crtl->emit.regno_pointer_align_length is non-zero.
2663 2014-02-20 Richard Henderson <rth@redhat.com>
2666 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2667 on failure the store back into EXPECT.
2669 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2670 Sandra Loosemore <sandra@codesourcery.com>
2672 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2673 * config/nios2/nios2.c (nios2_function_profiler): Add
2674 -fPIC (flag_pic == 2) support.
2675 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2676 (nios2_large_offset_p): New function.
2677 (nios2_unspec_reloc_p): Move up position, update to use
2678 nios2_large_offset_p.
2679 (nios2_unspec_address): Remove function.
2680 (nios2_unspec_offset): New function.
2681 (nios2_large_got_address): New function.
2682 (nios2_got_address): Add large offset support.
2683 (nios2_legitimize_tls_address): Update usage of removed and new
2685 (nios2_symbol_binds_local_p): New function.
2686 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2687 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2688 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2689 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2690 processing for (const (unspec ...)).
2691 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2693 2014-02-20 Richard Biener <rguenther@suse.de>
2695 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2696 doing the substitution.
2697 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2699 2014-02-20 Martin Jambor <mjambor@suse.cz>
2702 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2703 info when checking whether lattices are bottom.
2705 2014-02-20 Richard Biener <rguenther@suse.de>
2708 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2711 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2714 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2715 parameter specifying the scaling.
2716 (inline_call): Update.
2717 (want_inline_recursively): Guard division by zero.
2718 (recursive_inlining): Update.
2719 * ipa-inline.h (clone_inlined_nodes): Update.
2721 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2724 * config/i386/i386.c (classify_argument): Pass structures of size
2725 64 bytes or less in register.
2727 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2728 Kirill Yukhin <kirill.yukhin@intel.com>
2730 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2731 (_mm_rcp28_round_ss): Ditto.
2732 (_mm_rsqrt28_round_sd): Ditto.
2733 (_mm_rsqrt28_round_ss): Ditto.
2734 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2735 (_mm_rcp14_round_ss): Ditto.
2736 (_mm_rsqrt14_round_sd): Ditto.
2737 (_mm_rsqrt14_round_ss): Ditto.
2738 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2739 the first input operand, get rid of match_dup.
2740 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2742 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2744 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2745 operand as the first input operand, set type attribute.
2746 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2748 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2749 operand as the first input operand, set type attribute.
2751 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2753 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2756 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2759 * config/i386/i386.c (construct_container): Remove TFmode check
2760 for X86_64_INTEGER_CLASS.
2762 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2765 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2766 only when -Wpsabi is enabled.
2768 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2771 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2772 passing arrays in registers are the same as for structs, so remove the
2773 special case for them.
2775 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2777 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2778 destination type, extract only the valid bits if the source type is not
2779 integral and has a different mode.
2781 2014-02-19 Richard Biener <rguenther@suse.de>
2784 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2787 2014-02-19 Richard Biener <rguenther@suse.de>
2790 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2791 (ipa_modify_call_arguments): Emit an argument load explicitely and
2792 preserve virtual SSA form there and for the replacement call.
2793 Do not update SSA form nor free dominance info.
2795 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2797 * ipa.c (function_and_variable_visibility): Also clear WEAK
2798 flag when disolving COMDAT_GROUP.
2800 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2802 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2803 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2805 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2806 do more sanity checks.
2807 (detect_type_change): Return true when giving up early.
2808 (compute_complex_assign_jump_func): Fix type parameter of
2809 ipa_set_ancestor_jf.
2810 (compute_complex_ancestor_jump_func): Likewise.
2811 (update_jump_functions_after_inlining): Fix updating of
2813 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2815 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2817 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2818 inline clones when edge disappears.
2820 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2823 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2824 Split 64-bit moves into 2 patterns. Do not allow the use of
2825 direct move for TDmode in little endian, since the decimal value
2826 has little endian bytes within a word, but the 64-bit pieces are
2827 ordered in a big endian fashion, and normal subreg's of TDmode are
2829 (mov<mode>_64bit_dm): Likewise.
2830 (movtd_64bit_nodm): Likewise.
2832 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2834 PR tree-optimization/60174
2835 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2836 statement of an SSA_NAME that occurs in an abnormal PHI node.
2838 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2841 * final.c (SEEN_BB): Remove.
2842 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2843 (final_scan_insn): Don't force_source_line on second
2844 NOTE_INSN_BASIC_BLOCK.
2846 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2849 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2850 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2851 (type_natural_mode): Warn ABI change when %zmm register is not
2852 available for AVX512F vector value passing.
2854 2014-02-18 Kai Tietz <ktietz@redhat.com>
2857 * config/i386/i386.c (ix86_expand_prologue): Use value in
2858 rax register as displacement when restoring %r10 or %rax.
2859 Fix wrong offset when restoring both registers.
2861 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2863 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2864 assertion with conditional return.
2866 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2867 Uros Bizjak <ubizjak@gmail.com>
2870 * config/i386/driver-i386.c (host_detect_local_cpu): If
2871 YMM state is not saved by the OS, also clear has_f16c. Move
2872 CPUID 0x80000001 handling before YMM state saving checking.
2874 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2876 PR rtl-optimization/58960
2877 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2878 factored out from ...
2879 (sched_init): ... here.
2880 (free_global_sched_pressure_data): New, factored out from ...
2881 (sched_finish): ... here.
2882 * sched-int.h (free_global_sched_pressure_data): Declare.
2883 * sched-rgn.c (nr_regions_initial): New static global.
2884 (haifa_find_rgns): Initialize it.
2885 (schedule_region): Disable sched-pressure for the newly
2888 2014-02-17 Richard Biener <rguenther@suse.de>
2890 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2891 release SSA defs of pattern stmts.
2893 2014-02-17 Richard Biener <rguenther@suse.de>
2895 * tree-inline.c (expand_call_inline): Release the virtual
2896 operand defined by the call we are about to inline.
2898 2014-02-17 Richard Biener <rguenther@suse.de>
2900 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2902 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2903 Ilya Tocar <ilya.tocar@intel.com>
2905 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2906 arguments order in builtin.
2907 (_mm512_permutexvar_epi64): Ditto.
2908 (_mm512_mask_permutexvar_epi64): Ditto
2909 (_mm512_maskz_permutexvar_epi32): Ditto
2910 (_mm512_permutexvar_epi32): Ditto
2911 (_mm512_mask_permutexvar_epi32): Ditto
2913 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2915 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2916 (p8_vmrgow): Likewise.
2918 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2920 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2923 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2926 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2927 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2928 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2929 using direct move instructions on ISA 2.07. Also adjust
2930 instruction length for 64-bit.
2931 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2932 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2934 2014-02-15 Alan Modra <amodra@gmail.com>
2938 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2939 find_replacement on parts of insn rtl that might be reloaded.
2941 2014-02-15 Richard Biener <rguenther@suse.de>
2943 PR tree-optimization/60183
2944 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2945 (tree_ssa_phiprop): Calculate and free post-dominators.
2947 2014-02-14 Jeff Law <law@redhat.com>
2949 PR rtl-optimization/60131
2950 * ree.c (get_extended_src_reg): New function.
2951 (combine_reaching_defs): Use it rather than assuming location of REG.
2952 (find_and_remove_re): Verify first operand of extension is
2953 a REG before adding the insns to the copy list.
2955 2014-02-14 Roland McGrath <mcgrathr@google.com>
2957 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2958 * configure: Regenerated.
2959 * config.in: Regenerated.
2960 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2961 instead of ASM_SHORT.
2963 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2964 Richard Earnshaw <rearnsha@arm.com>
2966 PR rtl-optimization/59535
2967 * lra-constraints.c (process_alt_operands): Encourage alternative
2968 when unassigned pseudo class is superset of the alternative class.
2969 (inherit_reload_reg): Don't inherit when optimizing for code size.
2970 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2971 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2972 modes not less than 4 for Thumb1.
2974 2014-02-14 Kyle McMartin <kyle@redhat.com>
2977 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2979 2014-02-14 Richard Biener <rguenther@suse.de>
2981 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2982 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2983 Do not strip INDIRECT_REFs.
2985 2014-02-14 Richard Biener <rguenther@suse.de>
2988 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2989 DECL_FUNCTION_SPECIFIC_TARGET.
2990 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2991 * tree-streamer-out.c (pack_ts_target_option): Remove.
2992 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2993 (write_ts_function_decl_tree_pointers): Do not stream
2994 DECL_FUNCTION_SPECIFIC_TARGET.
2995 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2996 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2997 (lto_input_ts_function_decl_tree_pointers): Do not stream
2998 DECL_FUNCTION_SPECIFIC_TARGET.
3000 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3002 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
3003 (get_initial_def_for_induction, vectorizable_induction): Ignore
3004 debug stmts when looking for exit_phi.
3005 (vectorizable_live_operation): Fix up condition.
3007 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3009 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3010 nreverse() because it changes the content of original tree list.
3012 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3014 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3015 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3017 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3019 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
3020 GNU coding standards.
3022 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3025 * dwarf2out.c (gen_subprogram_die): Don't call
3026 add_calling_convention_attribute if subr_die is old_die.
3028 2014-02-13 Sharad Singhai <singhai@google.com>
3030 * doc/optinfo.texi: Fix order of nodes.
3032 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3034 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
3035 operands[2], not operands[3].
3037 2014-02-13 Richard Biener <rguenther@suse.de>
3040 * doc/install.texi (ISL): Update recommended version to 0.12.2,
3041 mention the possibility of an in-tree build.
3042 (CLooG): Update recommended version to 0.18.1, mention the
3043 possibility of an in-tree build and clarify that the ISL
3044 bundled with CLooG does not work.
3046 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3049 * expr.c (compress_float_constant): If x is a hard register,
3050 extend into a pseudo and then move to x.
3052 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
3054 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
3055 caused by bad second argument to warning_at() with -mhotpatch and
3056 nested functions (e.g. with gfortran).
3058 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
3060 * opts.c (option_name): Remove "enabled by default" rider.
3062 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3064 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3066 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3067 Uros Bizjak <ubizjak@gmail.com>
3070 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3071 * configure: Regenerated.
3073 2014-02-12 Richard Biener <rguenther@suse.de>
3075 * vec.c (vec_prefix::calculate_allocation): Move as
3076 inline variant to vec.h.
3077 (vec_prefix::calculate_allocation_1): New out-of-line version.
3078 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3079 (vec_prefix::m_has_auto_buf): Rename to ...
3080 (vec_prefix::m_using_auto_storage): ... this.
3081 (vec_prefix::calculate_allocation): Inline the easy cases
3082 and dispatch to calculate_allocation_1 which doesn't need the
3084 (va_heap::reserve): Use gcc_checking_assert.
3085 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3086 m_using_auto_storage.
3087 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3089 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3090 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3091 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3093 2014-02-12 Richard Biener <rguenther@suse.de>
3095 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3096 when we found a dependence.
3098 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3100 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3102 (maybe_fold_stmt): ... into this new function.
3103 * omp-low.c (lower_omp): Update comment.
3105 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3108 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3111 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3113 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3114 identifiers in comments.
3115 (cortexa53_extra_costs): Likewise.
3116 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3117 (cortexa7_extra_costs): Likewise.
3118 (cortexa12_extra_costs): Likewise.
3119 (cortexa15_extra_costs): Likewise.
3120 (v7m_extra_costs): Likewise.
3122 2014-02-12 Richard Biener <rguenther@suse.de>
3125 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3126 of posix_memalign being successful.
3127 (lower_stmt): Restrict lowering of posix_memalign to when
3128 -ftree-bit-ccp is enabled.
3130 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3132 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3134 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3136 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3138 PR rtl-optimization/60116
3139 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3140 other_insn once the combination has been validated.
3142 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3145 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3147 * ipa-devirt.c: Include demangle.h
3148 (odr_violation_reported): New static variable.
3149 (add_type_duplicate): Update odr_violations.
3150 (maybe_record_node): Add completep parameter; update it.
3151 (record_target_from_binfo): Add COMPLETEP parameter;
3152 update it as needed.
3153 (possible_polymorphic_call_targets_1): Likewise.
3154 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3155 rename FINAL to COMPLETE.
3156 (record_targets_from_bases): Sanity check we found the binfo;
3157 fix COMPLETEP updating.
3158 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3159 parameter, fix computing of COMPLETEP.
3160 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3161 at LTO time do demangling.
3162 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3163 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3165 (gimple_get_virt_method_for_binfo): Likewise.
3166 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3167 gimple_get_virt_method_for_vtable): Update prototypes.
3169 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3172 * genautomata.c (add_presence_absence): Fix typo with
3173 {final_}presence_list.
3175 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3178 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3179 for VSX/Altivec vectors that land in GPR registers.
3181 2014-02-11 Richard Henderson <rth@redhat.com>
3182 Jakub Jelinek <jakub@redhat.com>
3185 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3186 around drhs if type conversion to lacc->type is not useless.
3188 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3190 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3192 (cortex-a57.cortex-a53): Likewise.
3193 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3195 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3197 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3200 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3202 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3203 add_options_for_arm_vfp3.
3205 2014-02-11 Jeff Law <law@redhat.com>
3208 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3209 object with an undesirable mode.
3211 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3214 * config/i386/sol2-9.h: New file.
3215 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3216 *-*-solaris2.9*): Use it.
3218 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3220 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3221 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3223 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3225 * config/microblaze/microblaze.c: Extend mcpu version format
3227 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3229 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3231 2014-02-10 Richard Henderson <rth@redhat.com>
3234 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3235 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3236 ms-abi vs -mno-accumulate-outgoing-args.
3237 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3238 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3241 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3244 * cfgexpand.c (expand_asm_operands): Attach source location to
3245 ASM_INPUT rtx objects.
3246 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3248 2014-02-10 Nick Clifton <nickc@redhat.com>
3250 * config/mn10300/mn10300.c (popcount): New function.
3251 (mn10300_expand_prologue): Include saved registers in stack usage
3254 2014-02-10 Jeff Law <law@redhat.com>
3257 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3258 when changing the SET_DEST of a prior insn to avoid an input reload.
3260 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3262 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3263 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3264 -mcall-openbsd, or -mcall-linux.
3265 (CC1_ENDIAN_BIG_SPEC): Remove.
3266 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3267 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3268 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3269 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3270 and %cc1_endian_default.
3271 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3273 2014-02-10 Richard Biener <rguenther@suse.de>
3275 PR tree-optimization/60115
3276 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3277 MEM_REF handling. Properly verify that the accesses are not
3278 out of the objects bound.
3280 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3282 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3285 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3287 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3288 proper constants and fix formatting.
3289 (possible_polymorphic_call_targets): Fix formatting.
3291 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3292 Ilya Tocar <ilya.tocar@intel.com>
3294 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3295 (_mm512_loadu_epi32): Renamed into...
3296 (_mm512_loadu_si512): This.
3297 (_mm512_storeu_epi32): Renamed into...
3298 (_mm512_storeu_si512): This.
3299 (_mm512_maskz_ceil_ps): Removed.
3300 (_mm512_maskz_ceil_pd): Ditto.
3301 (_mm512_maskz_floor_ps): Ditto.
3302 (_mm512_maskz_floor_pd): Ditto.
3303 (_mm512_floor_round_ps): Ditto.
3304 (_mm512_floor_round_pd): Ditto.
3305 (_mm512_ceil_round_ps): Ditto.
3306 (_mm512_ceil_round_pd): Ditto.
3307 (_mm512_mask_floor_round_ps): Ditto.
3308 (_mm512_mask_floor_round_pd): Ditto.
3309 (_mm512_mask_ceil_round_ps): Ditto.
3310 (_mm512_mask_ceil_round_pd): Ditto.
3311 (_mm512_maskz_floor_round_ps): Ditto.
3312 (_mm512_maskz_floor_round_pd): Ditto.
3313 (_mm512_maskz_ceil_round_ps): Ditto.
3314 (_mm512_maskz_ceil_round_pd): Ditto.
3315 (_mm512_expand_pd): Ditto.
3316 (_mm512_expand_ps): Ditto.
3317 * config/i386/i386.c (ix86_builtins): Remove
3318 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3319 (bdesc_args): Ditto.
3320 * config/i386/predicates.md (const1256_operand): New.
3321 (const_1_to_2_operand): Ditto.
3322 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3323 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3324 (*avx512pf_gatherpf<mode>sf): Ditto.
3325 (avx512pf_gatherpf<mode>df): Ditto.
3326 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3327 (*avx512pf_gatherpf<mode>df): Ditto.
3328 (avx512pf_scatterpf<mode>sf): Ditto.
3329 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3330 (*avx512pf_scatterpf<mode>sf): Ditto.
3331 (avx512pf_scatterpf<mode>df): Ditto.
3332 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3333 (*avx512pf_scatterpf<mode>df): Ditto.
3334 (avx512f_expand<mode>): Removed.
3335 (<shift_insn><mode>3<mask_name>): Change predicate type.
3337 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3339 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3340 not at the end of datarefs vector use ordered_remove to avoid
3341 reordering datarefs vector.
3344 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3345 mark local addressable non-static vars as GOVD_PRIVATE
3346 instead of GOVD_LOCAL.
3347 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3348 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3352 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3353 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3354 assume_aligned or alloc_align attributes.
3355 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3356 arguments. Handle also assume_aligned and alloc_align attributes.
3357 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3358 calls to functions with assume_aligned or alloc_align attributes.
3359 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3361 2014-02-08 Terry Guo <terry.guo@arm.com>
3363 * doc/invoke.texi: Document ARM -march=armv7e-m.
3365 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3367 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3368 flag on __cilkrts_rethrow builtin.
3371 * ipa-cp.c (determine_versionability): Fail at -O0
3372 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3373 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3376 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3379 * tree-inline.c (copy_forbidden): Fail for
3380 __attribute__((optimize (0))) functions.
3382 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3384 * varpool.c: Include pointer-set.h.
3385 (varpool_remove_unreferenced_decls): Variables in other partitions
3386 will not be output; be however careful to not lose information
3389 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3391 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3392 lookup in the vtable constructor.
3394 2014-02-07 Jeff Law <law@redhat.com>
3397 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3398 define_insn_and_split.
3400 * ipa-inline.c (inline_small_functions): Fix typos.
3402 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3404 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3405 (s390_can_use_return_insn): Declare.
3406 * config/s390/s390.h (EPILOGUE_USES): Define.
3407 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3409 (s390_chunkify_start): Handle return JUMP_LABELs.
3410 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3411 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3412 (s390_can_use_return_insn): New functions.
3413 (s390_fix_long_loop_prediction): Handle conditional returns.
3414 (TARGET_SET_UP_BY_PROLOGUE): Define.
3415 * config/s390/s390.md (ANY_RETURN): New code iterator.
3416 (*creturn, *csimple_return, return, simple_return): New patterns.
3418 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3420 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3421 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3422 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3423 REG_CFA_RESTORE list when deciding not to restore a register.
3425 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3427 * config/s390/s390.c: Include tree-pass.h and context.h.
3428 (s390_early_mach): New function, split out from...
3429 (s390_emit_prologue): ...here.
3430 (pass_data_s390_early_mach): New pass structure.
3431 (pass_s390_early_mach): New class.
3432 (s390_option_override): Create and register early_mach pass.
3433 Move to end of file.
3435 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3437 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3438 to match for the exit block.
3440 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3442 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3443 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3444 Reject misaligned operands.
3446 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3448 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3450 2014-02-07 Richard Biener <rguenther@suse.de>
3453 * gimple-low.c (lower_builtin_posix_memalign): New function.
3454 (lower_stmt): Call it to lower posix_memalign in a way
3455 to make alignment info accessible.
3457 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3460 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3461 __builtin_setjmp_receiver.
3463 2014-02-07 Richard Biener <rguenther@suse.de>
3466 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3467 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3468 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3469 Handle BUILT_IN_POSIX_MEMALIGN.
3470 (find_func_clobbers): Likewise.
3471 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3472 (call_may_clobber_ref_p_1): Likewise.
3474 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3477 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3480 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3483 * lto-cgraph.c (lto_output_node): Use
3484 symtab_get_symbol_partitioning_class.
3485 (lto_output_varpool_node): likewise.
3486 (symtab_get_symbol_partitioning_class): Move here from
3488 * cgraph.h (symbol_partitioning_class): Likewise.
3489 (symtab_get_symbol_partitioning_class): Declare.
3491 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3493 * ggc.h (ggc_internal_cleared_alloc): New macro.
3494 * vec.h (vec_safe_copy): Handle memory stats.
3495 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3496 * target-globals.c (save_target_globals): Likewise.
3498 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3501 * expr.c (emit_move_resolve_push): Export; be bit more selective
3502 on when to clear alias set.
3503 * expr.h (emit_move_resolve_push): Declare.
3504 * function.h (struct function): Add tail_call_marked.
3505 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3506 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3507 * config/i386/i386.md (TImode move expander): De not call
3509 (FP push expanders): Preserve memory attributes.
3510 * config/i386/sse.md (push<mode>1): Remove.
3511 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3512 (ix86_expand_push): Remove.
3513 * config/i386/mmx.md (push<mode>1): Remove.
3515 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3517 PR rtl-optimization/60030
3518 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3519 lopart with paradoxical subreg before shifting it up by hprec.
3521 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3523 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3524 Remove extra newline at end of file.
3525 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3526 (arm_issue_rate): Handle cortexa57.
3527 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3528 (cortex-a57.cortex-a53): Likewise.
3530 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3533 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3534 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3536 (arm_expand_prologue): Adjust all callers.
3537 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3538 info, registers also at the lowest numbered registers side. Use
3539 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3543 * var-tracking.c (adjust_mems): Before adding a SET to
3544 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3546 2014-02-06 Alan Modra <amodra@gmail.com>
3549 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3550 change SDmode to DDmode when lra_in_progress.
3552 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3555 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3556 free_data_ref on the dr first, and before goto again also set dr
3557 to the next dr. For simd_lane_access, free old datarefs[i] before
3558 overwriting it. For get_vectype_for_scalar_type failure, don't
3559 free_data_ref if simd_lane_access.
3561 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3564 * tree.h (opts_for_fn): New inline function.
3565 (opt_for_fn): Define.
3566 * config/i386/i386.c (ix86_function_regparm): Use
3567 opt_for_fn (decl, optimize) instead of optimize.
3569 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3571 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3572 for SYMBOL_REF in large memory model.
3574 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3576 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3578 (cortex-a57): Likewise.
3579 (cortex-a57.cortex-a53): Likewise.
3581 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3582 Kugan Vivekanandarajah <kuganv@linaro.org>
3584 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3586 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3588 2014-02-06 Richard Biener <rguenther@suse.de>
3590 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3591 set_loop_copy and initialize_original_copy_tables.
3593 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3595 * config/aarch64/aarch64-simd.md
3596 (aarch64_ashr_simddi): Change QI to SI.
3598 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3599 Jakub Jelinek <jakub@redhat.com>
3602 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3605 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3607 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3608 CODE_FOR_altivec_vpku[hw]um to
3609 CODE_FOR_altivec_vpku[hw]um_direct.
3610 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3611 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3612 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3613 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3615 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3617 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3618 generation for -maltivec=be.
3619 (altivec_vsumsws): Simplify redundant test.
3621 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3623 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3624 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3625 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3626 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3627 gen_altivec_vpkuwum.
3628 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3630 (altivec_vpks<VI_char>ss): Likewise.
3631 (altivec_vpks<VI_char>us): Likewise.
3632 (altivec_vpku<VI_char>us): Likewise.
3633 (altivec_vpku<VI_char>um): Likewise.
3634 (altivec_vpku<VI_char>um_direct): New (copy of
3635 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3637 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3638 target is little endian and -maltivec=be is not specified.
3639 (*altivec_vupkhs<VU_char>_direct): New (copy of
3640 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3641 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3642 target is little endian and -maltivec=be is not specified.
3643 (*altivec_vupkls<VU_char>_direct): New (copy of
3644 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3645 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3646 little endian and -maltivec=be is not specified.
3647 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3648 little endian and -maltivec=be is not specified.
3650 2014-02-05 Richard Henderson <rth@redhat.com>
3653 * combine-stack-adj.c: Revert r206943.
3654 * sched-int.h (struct deps_desc): Add last_args_size.
3655 * sched-deps.c (init_deps): Initialize it.
3656 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3657 contain REG_ARGS_SIZE notes.
3659 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3661 * lto-cgraph.c (asm_nodes_output): Make global.
3662 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3663 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3664 (driver_handle_option): Handle OPT_fwpa.
3666 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3669 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3670 a comment typo and formatting issue. If odr_hash hasn't been
3671 created, return vNULL and set *completep to false.
3674 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3675 bb with no successors.
3677 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3680 * doc/invoke.texi (-march): Clarify documentation for ARM.
3684 2014-02-05 Richard Biener <rguenther@suse.de>
3686 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3687 when not vectorizing because of too many alias checks.
3688 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3689 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3691 2014-02-05 Nick Clifton <nickc@redhat.com>
3693 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3694 accept extended registers in any mode when compiling for the MN10300.
3696 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3698 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3699 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3700 sanitization attributes.
3701 (can_inline_edge_p): Likewise.
3702 (sanitize_attrs_match_for_inline_p): New function.
3704 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3706 * ipa-prop.c (detect_type_change): Shor circuit testing of
3707 type changes on THIS pointer.
3709 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3712 * config/pa/pa.c (legitimize_tls_address): Return original address
3713 if not passed a SYMBOL_REF rtx.
3714 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3716 (pa_emit_move_sequence): Simplify TLS source operands.
3717 (pa_legitimate_constant_p): Reject all TLS constants.
3718 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3719 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3721 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3723 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3724 groups when we know they are controlled by LTO.
3725 * varasm.c (default_binds_local_p_1): If object is in other partition,
3726 it will be resolved locally.
3728 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3730 * config/host-linux.c (linux_gt_pch_use_address): Don't
3731 use SSIZE_MAX because it is not always defined.
3733 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3736 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3737 threshold for pseudo splitting.
3738 (update_ebb_live_info): Process call argument hard registers and
3739 hard registers from insn definition too.
3740 (max_small_class_regs_num): New constant.
3741 (inherit_in_ebb): Update live hard regs through EBBs. Update
3742 reloads_num only for small register classes. Don't split for
3745 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3748 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3751 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3753 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3756 2014-02-04 Marek Polacek <polacek@redhat.com>
3758 * gdbinit.in (pel): Define.
3760 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3762 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3765 2014-02-04 Richard Biener <rguenther@suse.de>
3768 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3769 in function context local.
3770 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3771 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3772 similar to LTO_imported_decl_ref.
3774 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3776 PR tree-optimization/60002
3777 * cgraphclones.c (build_function_decl_skip_args): Clear
3780 PR tree-optimization/60023
3781 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3782 false to gsi_replace.
3783 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3784 has been in some EH region and vec_stmt could throw, add
3785 vec_stmt into the same EH region.
3786 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3787 has no lhs, ignore it.
3788 * internal-fn.c (expand_MASK_LOAD): Likewise.
3791 * tree-inline.c (copy_forbidden): Fail for
3792 __attribute__((optimize (0))) functions.
3795 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3796 is set, copy one less argument.
3797 (expand_simd_clones): Don't subtract clone_info->inbranch
3798 from simd_clone_struct_alloc argument.
3800 PR rtl-optimization/57915
3801 * recog.c (simplify_while_replacing): If all unary/binary/relational
3802 operation arguments are constant, attempt to simplify those.
3805 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3806 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3808 2014-02-04 Richard Biener <rguenther@suse.de>
3810 PR tree-optimization/60012
3811 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3812 TBAA disambiguation to all DDRs.
3814 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3817 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3818 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3820 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3823 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3825 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3827 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3828 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3830 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3833 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3834 to figure out targets of polymorphic calls with known decl.
3835 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3836 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3837 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3838 (get_polymorphic_call_info): ... here.
3839 (get_polymorphic_call_info_from_invariant): New function.
3841 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3843 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3844 lookup via vtable pointer; check for type consistency
3845 and turn inconsitent facts into UNREACHABLE.
3846 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3847 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3848 type inconsistent querries; return UNREACHABLE instead.
3850 2014-02-03 Richard Henderson <rth@twiddle.net>
3853 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3854 already processed this node.
3855 (normalize_one_pred_1): Pass along mark_set.
3856 (normalize_one_pred): Create and destroy a pointer_set_t.
3857 (normalize_one_pred_chain): Likewise.
3859 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3861 PR gcov-profile/58602
3862 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3864 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3867 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3868 -fno-devirtualize; try to devirtualize by the knowledge of
3869 virtual table pointer given by aggregate propagation.
3870 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3871 (ipa_print_node_jump_functions): Dump also offset that
3872 is relevant for polymorphic calls.
3873 (determine_known_aggregate_parts): Add arg_type parameter; use it
3874 instead of determining the type from pointer type.
3875 (ipa_compute_jump_functions_for_edge): Update call of
3876 determine_known_aggregate_parts.
3877 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3878 (gimple_get_virt_method_for_binfo): ... here; simplify using
3879 vtable_pointer_value_to_vtable.
3880 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3881 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3882 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3883 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3885 (vtable_pointer_value_to_binfo): ... here.
3886 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3888 2014-02-03 Teresa Johnson <tejohnson@google.com>
3890 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3891 redef of outer loop index variable.
3893 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3897 * doc/extend.texi (Function Attributes): Typo.
3899 2014-02-03 Cong Hou <congh@google.com>
3901 PR tree-optimization/60000
3902 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3903 if the vectorized statement is a store. A store statement can only
3904 appear at the end of pattern statements.
3906 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3908 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3909 (ix86_option_override_internal): Default long double to 64-bit for
3910 32-bit Bionic and to 128-bit for 64-bit Bionic.
3912 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3913 TARGET_LONG_DOUBLE_128 is true.
3914 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3916 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3917 (mlong-double-64): Negate -mlong-double-128.
3918 (mlong-double-128): New option.
3920 * config/i386/i386-c.c (ix86_target_macros): Define
3921 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3923 * doc/invoke.texi: Document -mlong-double-128.
3925 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3927 PR rtl-optimization/60024
3928 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3930 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3932 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3934 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3936 PR rtl-optimization/57662
3937 * sel-sched.c (code_motion_path_driver): Do not mark already not
3938 existing blocks in the visiting bitmap.
3940 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3942 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3943 on the insn being emitted.
3945 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3946 Will Deacon <will.deacon@arm.com>
3948 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3950 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3952 * config/arm/arm-tables.opt: Regenerate.
3954 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3956 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3957 for vector types other than V16QImode.
3958 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3959 define_expand, and call altivec_expand_vec_perm_le when producing
3960 code with little endian element order.
3961 (*altivec_vperm_<mode>_internal): New insn having previous
3962 behavior of altivec_vperm_<mode>.
3963 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3964 altivec_expand_vec_perm_le when producing code with little endian
3966 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3967 behavior of altivec_vperm_<mode>_uns.
3969 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3971 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3972 (altivec_vsumsws): Add handling for -maltivec=be with a little
3974 (altivec_vsumsws_direct): New.
3975 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3976 gen_altivec_vsumsws.
3978 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3980 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3981 vtable_pointer_value_to_binfo): New functions.
3982 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3983 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3985 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3987 * config/nios2/nios2.md (load_got_register): Initialize GOT
3988 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3989 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3991 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3993 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3994 preserverd by passthrough, do not propagate the type.
3996 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3998 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3999 (mips_atomic_assign_expand_fenv): New function.
4000 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
4002 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4004 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4005 (__builtin_mips_set_fcsr): Likewise.
4006 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4007 MIPS_USI_FTYPE_VOID.
4008 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4009 (mips16_expand_set_fcsr): Likewise.
4010 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4011 (mips16_set_fcsr_stub): Likewise.
4012 (mips16_get_fcsr_one_only_stub): New class.
4013 (mips16_set_fcsr_one_only_stub): Likewise.
4014 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4015 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4016 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4017 (hard_float): New availability predicate.
4018 (mips_builtins): Add get_fcsr and set_fcsr.
4019 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
4020 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
4021 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
4022 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
4023 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
4026 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4028 * config/mips/mips.c (mips_one_only_stub): New class.
4029 (mips_need_mips16_rdhwr_p): Replace with...
4030 (mips16_rdhwr_stub): ...this new variable.
4031 (mips16_stub_call_address): New function.
4032 (mips16_rdhwr_one_only_stub): New class.
4033 (mips_expand_thread_pointer): Use mips16_stub_call_address.
4034 (mips_output_mips16_rdhwr): Delete.
4035 (mips_finish_stub): New function.
4036 (mips_code_end): Use it to handle rdhwr stubs.
4038 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
4041 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
4042 when calculating size of integer atomic types.
4044 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
4046 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
4048 2014-02-01 Jakub Jelinek <jakub@redhat.com>
4050 PR tree-optimization/60003
4051 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
4052 * profile.c (branch_prob): Use gimple_call_builtin_p
4053 to check for BUILT_IN_SETJMP_RECEIVER.
4054 * tree-inline.c (copy_bb): Call notice_special_calls.
4056 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
4059 * lra-constraints.c (process_alt_operands): Update reload_sum only
4062 2014-01-31 Richard Henderson <rth@redhat.com>
4065 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4066 until after else_eh is processed.
4068 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4070 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4071 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4072 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4073 in smmintrin.h, remove them.
4074 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4075 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4076 * config/i386/i386.md (ROUND_SAE): Fix value.
4077 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4078 (const48_operand): New.
4079 * config/i386/subst.md (round), (round_expand): Use
4080 const_4_or_8_to_11_operand.
4081 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4083 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4085 * config/i386/constraints.md (Yk): Swap meaning with k.
4086 * config/i386/i386.md (movhi_internal): Change Yk to k.
4087 (movqi_internal): Ditto.
4088 (*k<logic><mode>): Ditto.
4091 (kandn<mode>): Ditto.
4092 (*<code>hi_1): Ditto.
4093 (*<code>qi_1): Ditto.
4094 (kxnor<mode>): Ditto.
4095 (kortestzhi): Ditto.
4096 (kortestchi): Ditto.
4098 (*one_cmplhi2_1): Ditto.
4099 (*one_cmplqi2_1): Ditto.
4100 * config/i386/sse.md (): Change k to Yk.
4101 (avx512f_load<mode>_mask): Ditto.
4102 (avx512f_blendm<mode>): Ditto.
4103 (avx512f_store<mode>_mask): Ditto.
4104 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4105 (avx512f_storedqu<mode>_mask): Ditto.
4106 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4108 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4109 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4110 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4111 (avx512f_maskcmp<mode>3): Ditto.
4112 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4113 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4114 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4115 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4116 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4117 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4118 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4119 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4120 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4121 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4122 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4123 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4124 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4125 (vec_extract_lo_<mode>_maskm): Ditto.
4126 (vec_extract_hi_<mode>_maskm): Ditto.
4127 (avx512f_vternlog<mode>_mask): Ditto.
4128 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4129 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4130 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4131 (avx512f_<code>v8div16qi2_mask): Ditto.
4132 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4133 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4134 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4135 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4136 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4137 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4138 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4139 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4140 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4141 (avx512cd_maskb_vec_dupv8di): Ditto.
4142 (avx512cd_maskw_vec_dupv16si): Ditto.
4143 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4144 (avx512f_vpermi2var<mode>3_mask): Ditto.
4145 (avx512f_vpermi2var<mode>3_mask): Ditto.
4146 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4147 (*avx512f_gathersi<mode>): Ditto.
4148 (*avx512f_gathersi<mode>_2): Ditto.
4149 (*avx512f_gatherdi<mode>): Ditto.
4150 (*avx512f_gatherdi<mode>_2): Ditto.
4151 (*avx512f_scattersi<mode>): Ditto.
4152 (*avx512f_scatterdi<mode>): Ditto.
4153 (avx512f_compress<mode>_mask): Ditto.
4154 (avx512f_compressstore<mode>_mask): Ditto.
4155 (avx512f_expand<mode>_mask): Ditto.
4156 * config/i386/subst.md (mask): Change k to Yk.
4157 (mask_scalar_merge): Ditto.
4160 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4162 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4164 2014-01-31 Richard Biener <rguenther@suse.de>
4167 * builtins.c (fold_builtin_memory_op): Make sure to not
4168 use a floating-point mode or a boolean or enumeral type for
4171 2014-01-30 DJ Delorie <dj@redhat.com>
4173 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4174 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4175 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4176 whenever main() has an epilogue.
4178 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4180 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4181 unused variable "field".
4182 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4183 (vsx_mergeh_<mode>): Likewise.
4184 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4185 (altivec_vmrghh): Likewise.
4186 (altivec_vmrghw): Likewise.
4187 (altivec_vmrglb): Likewise.
4188 (altivec_vmrglh): Likewise.
4189 (altivec_vmrglw): Likewise.
4190 (altivec_vspltb): Add missing uses.
4191 (altivec_vsplth): Likewise.
4192 (altivec_vspltw): Likewise.
4193 (altivec_vspltsf): Likewise.
4195 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4198 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4199 frame related instructions.
4201 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4203 PR rtl-optimization/59959
4204 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4205 any reload of register whose subreg is invalid.
4207 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4209 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4210 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4211 Add missing return type - void.
4213 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4215 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4216 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4217 remove element index adjustment for endian (now handled in vsx.md
4219 (altivec_expand_vec_perm_const): Use
4220 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4221 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4222 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4223 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4224 define_expand and a new define_insn *altivec_vspltb_internal;
4225 adjust for -maltivec=be on a little endian target.
4226 (altivec_vspltb_direct): New.
4227 (altivec_vsplth): Divide into a define_expand and a new
4228 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4229 little endian target.
4230 (altivec_vsplth_direct): New.
4231 (altivec_vspltw): Divide into a define_expand and a new
4232 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4233 little endian target.
4234 (altivec_vspltw_direct): New.
4235 (altivec_vspltsf): Divide into a define_expand and a new
4236 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4237 a little endian target.
4239 2014-01-30 Richard Biener <rguenther@suse.de>
4241 PR tree-optimization/59993
4242 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4243 can propagate form the earlier stmt and avoid the transform
4244 when the intermediate result is needed.
4246 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4248 * README.Portability: Fix typo.
4250 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4252 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4253 comparison_operator with ordered_comparison_operator.
4255 2014-01-30 Nick Clifton <nickc@redhat.com>
4257 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4258 Rename to mn10300_store_multiple_regs.
4259 * config/mn10300/mn10300.c: Likewise.
4260 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4261 store_multiple_regs.
4262 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4263 Call mn10300_store_multiple_regs.
4265 2014-01-30 Nick Clifton <nickc@redhat.com>
4266 DJ Delorie <dj@redhat.com>
4268 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4269 %fp 2 to keep registers after it properly word-aligned.
4270 (rl78_alloc_physical_registers_umul): Handle the case where both
4271 input operands are the same.
4273 2014-01-30 Richard Biener <rguenther@suse.de>
4275 PR tree-optimization/59903
4276 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4279 2014-01-30 Jason Merrill <jason@redhat.com>
4282 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4285 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4287 2014-01-30 Richard Biener <rguenther@suse.de>
4289 PR tree-optimization/59951
4290 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4292 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4295 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4296 SFmode to DFmode case.
4298 2014-01-29 DJ Delorie <dj@redhat.com>
4300 * config/msp430/msp430.opt (-minrt): New.
4301 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4303 (ENDFILE_SPEC): Likewise.
4305 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4307 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4308 (estimate_function_body_sizes): Use it.
4310 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4313 * dwarf2out.c (is_cxx_auto): New.
4314 (is_base_type): Use it.
4315 (gen_type_die_with_usage): Likewise.
4317 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4319 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4320 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4321 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4322 -maltivec=be with LE targets.
4323 (vsx_mergeh_<mode>): Likewise.
4324 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4325 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4326 (altivec_vmrghb): Replace with define_expand and new
4327 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4328 (altivec_vmrghb_direct): New define_insn.
4329 (altivec_vmrghh): Replace with define_expand and new
4330 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4331 (altivec_vmrghh_direct): New define_insn.
4332 (altivec_vmrghw): Replace with define_expand and new
4333 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4334 (altivec_vmrghw_direct): New define_insn.
4335 (*altivec_vmrghsf): Adjust for endianness.
4336 (altivec_vmrglb): Replace with define_expand and new
4337 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4338 (altivec_vmrglb_direct): New define_insn.
4339 (altivec_vmrglh): Replace with define_expand and new
4340 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4341 (altivec_vmrglh_direct): New define_insn.
4342 (altivec_vmrglw): Replace with define_expand and new
4343 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4344 (altivec_vmrglw_direct): New define_insn.
4345 (*altivec_vmrglsf): Adjust for endianness.
4346 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4347 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4348 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4349 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4350 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4351 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4352 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4353 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4355 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4357 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4358 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4361 2014-01-29 Richard Biener <rguenther@suse.de>
4363 PR tree-optimization/58742
4364 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4365 associate_pointerplus_align.
4366 (associate_pointerplus_diff): New function.
4367 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4368 and associate_pointerplus_diff.
4370 2014-01-29 Richard Biener <rguenther@suse.de>
4372 * lto-streamer.h (LTO_major_version): Bump to 3.
4373 (LTO_minor_version): Reset to 0.
4375 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4377 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4378 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4379 (arm_file_start): Generate correct asm header for armv7ve.
4380 * config/arm/bpabi.h: Add multilib support for armv7ve.
4381 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4382 and cortex-a15 to armv7ve.
4383 * config/arm/t-aprofile: Add multilib support for armv7ve.
4384 * doc/invoke.texi: Document -march=armv7ve.
4386 2014-01-29 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/58742
4389 * tree-ssa-forwprop.c (associate_plusminus): Return true
4390 if we changed sth, defer EH cleanup to ...
4391 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4392 (simplify_mult): New function.
4394 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4397 PR tree-optimization/59920
4398 * tree.c (build_common_builtin_nodes): Remove
4399 __builtin_setjmp_dispatcher initialization.
4400 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4401 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4402 instead of gsi_after_labels + manually skipping debug stmts.
4403 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4404 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4405 * tree-inline.c (copy_edges_for_bb): Remove
4406 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4407 argument. Ignore computed_goto_p stmts. Don't call
4408 make_abnormal_goto_edges. If a call might need abnormal edges
4409 for non-local gotos, see if it already has an edge to
4410 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4411 with true argument, don't do anything then, otherwise add
4412 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4413 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4415 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4416 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4417 (lower_stmt): Don't set data->calls_builtin_setjmp.
4418 (lower_builtin_setjmp): Adjust comment.
4419 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4420 * tree-cfg.c (found_computed_goto): Remove.
4421 (factor_computed_gotos): Remove.
4422 (make_goto_expr_edges): Return bool, true for computed gotos.
4423 Don't call make_abnormal_goto_edges.
4424 (build_gimple_cfg): Don't set found_computed_goto, don't call
4425 factor_computed_gotos.
4426 (computed_goto_p): No longer static.
4427 (make_blocks): Don't set found_computed_goto.
4428 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4429 (make_edges): If make_goto_expr_edges returns true, push bb
4430 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4431 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4432 vector. Record mapping between bbs and OpenMP regions if there
4433 are any, adjust make_gimple_omp_edges caller. Call
4434 handle_abnormal_edges.
4435 (make_abnormal_goto_edges): Remove.
4436 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4437 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4438 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4439 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4440 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4441 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4442 filling *region also set *region_idx to (*region)->entry->index.
4445 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4446 For REGs set ORIGINAL_REGNO.
4448 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4450 * doc/md.texi: Mention that a target shouldn't implement
4451 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4454 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4456 PR tree-optimization/59594
4457 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4458 a copy of the datarefs vector rather than the vector itself.
4460 2014-01-28 Jason Merrill <jason@redhat.com>
4463 * dwarf2out.c (auto_die): New static.
4464 (gen_type_die_with_usage): Handle C++1y 'auto'.
4465 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4468 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4471 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4472 (SPEC_X32): Likewise.
4473 (SPEC_64): Likewise.
4474 * config/i386/i386.c (ix86_option_override_internal): Turn off
4475 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4477 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4478 * config/i386/i386.h (TARGET_16BIT): New macro.
4479 (TARGET_16BIT_P): Likewise.
4480 * config/i386/i386.opt: Add m16.
4481 * doc/invoke.texi: Document -m16.
4483 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4485 PR preprocessor/59935
4486 * input.c (location_get_source_line): Bail out on when line number
4487 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4489 2014-01-28 Richard Biener <rguenther@suse.de>
4491 PR tree-optimization/58742
4492 * tree-ssa-forwprop.c (associate_plusminus): Handle
4493 pointer subtraction of the form (T)(P + A) - (T)P.
4495 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4497 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4500 2014-01-28 Richard Biener <rguenther@suse.de>
4503 2014-01-28 Richard Biener <rguenther@suse.de>
4505 PR rtl-optimization/45364
4506 PR rtl-optimization/59890
4507 * var-tracking.c (local_get_addr_clear_given_value): Handle
4508 already cleared slot.
4509 (val_reset): Handle not allocated local_get_addr_cache.
4510 (vt_find_locations): Use post-order on the inverted CFG.
4512 2014-01-28 Richard Biener <rguenther@suse.de>
4514 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4516 2014-01-28 Richard Biener <rguenther@suse.de>
4518 PR rtl-optimization/45364
4519 PR rtl-optimization/59890
4520 * var-tracking.c (local_get_addr_clear_given_value): Handle
4521 already cleared slot.
4522 (val_reset): Handle not allocated local_get_addr_cache.
4523 (vt_find_locations): Use post-order on the inverted CFG.
4525 2014-01-28 Alan Modra <amodra@gmail.com>
4527 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4528 * configure.ac <recursive call for build != host>: Define
4529 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4530 and LD_FOR_BUILD too.
4531 * configure: Regenerate.
4533 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4535 * config/i386/i386.c (get_builtin_code_for_version): Separate
4536 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4537 Broadwell from Haswell.
4539 2014-01-27 Steve Ellcey <sellcey@mips.com>
4541 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4542 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4543 * config/mips/mips.c (mips_option_override): Change setting
4545 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4546 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4547 Change from Mask to Var.
4549 2014-01-27 Jeff Law <law@redhat.com>
4551 * ipa-inline.c (inline_small_functions): Fix typo.
4553 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4555 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4556 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4557 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4558 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4559 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4560 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4561 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4562 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4563 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4564 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4565 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4566 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4567 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4568 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4569 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4570 (_mm512_storeu_epi64): Ditto.
4571 (_mm512_cmpge_epi32_mask): Ditto.
4572 (_mm512_cmpge_epu32_mask): Ditto.
4573 (_mm512_cmpge_epi64_mask): Ditto.
4574 (_mm512_cmpge_epu64_mask): Ditto.
4575 (_mm512_cmple_epi32_mask): Ditto.
4576 (_mm512_cmple_epu32_mask): Ditto.
4577 (_mm512_cmple_epi64_mask): Ditto.
4578 (_mm512_cmple_epu64_mask): Ditto.
4579 (_mm512_cmplt_epi32_mask): Ditto.
4580 (_mm512_cmplt_epu32_mask): Ditto.
4581 (_mm512_cmplt_epi64_mask): Ditto.
4582 (_mm512_cmplt_epu64_mask): Ditto.
4583 (_mm512_cmpneq_epi32_mask): Ditto.
4584 (_mm512_cmpneq_epu32_mask): Ditto.
4585 (_mm512_cmpneq_epi64_mask): Ditto.
4586 (_mm512_cmpneq_epu64_mask): Ditto.
4587 (_mm512_expand_pd): Ditto.
4588 (_mm512_expand_ps): Ditto.
4589 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4590 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4591 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4592 * config/i386/i386.c (ix86_builtins): Add
4593 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4594 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4595 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4596 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4597 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4598 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4599 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4600 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4601 IX86_BUILTIN_PMOVUSQW512_MEM.
4602 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4603 __builtin_ia32_pmovsqd512mem_mask,
4604 __builtin_ia32_pmovqd512mem_mask,
4605 __builtin_ia32_pmovusqw512mem_mask,
4606 __builtin_ia32_pmovsqw512mem_mask,
4607 __builtin_ia32_pmovqw512mem_mask,
4608 __builtin_ia32_pmovusdw512mem_mask,
4609 __builtin_ia32_pmovsdw512mem_mask,
4610 __builtin_ia32_pmovdw512mem_mask,
4611 __builtin_ia32_pmovqb512mem_mask,
4612 __builtin_ia32_pmovusqb512mem_mask,
4613 __builtin_ia32_pmovsqb512mem_mask,
4614 __builtin_ia32_pmovusdb512mem_mask,
4615 __builtin_ia32_pmovsdb512mem_mask,
4616 __builtin_ia32_pmovdb512mem_mask.
4617 (bdesc_args): Add __builtin_ia32_expanddf512,
4618 __builtin_ia32_expandsf512.
4619 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4620 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4621 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4622 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4623 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4624 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4625 (avx512f_<code>v8div16qi2_mask_store): This.
4626 (avx512f_expand<mode>): New.
4628 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4630 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4632 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4633 (_mm512_prefetch_i32scatter_pd): Ditto.
4634 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4635 (_mm512_prefetch_i64scatter_pd): Ditto.
4636 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4637 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4638 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4639 (_mm512_prefetch_i32scatter_ps): Ditto.
4640 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4641 (_mm512_prefetch_i64scatter_ps): Ditto.
4642 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4643 * config/i386/i386-builtin-types.def: Define
4644 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4645 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4646 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4647 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4648 IX86_BUILTIN_SCATTERPFQPD.
4649 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4650 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4651 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4652 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4653 __builtin_ia32_scatterpfqps.
4654 (ix86_expand_builtin): Expand new built-ins.
4655 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4656 fix memory access data type.
4657 (*avx512pf_gatherpf<mode>_mask): Ditto.
4658 (*avx512pf_gatherpf<mode>): Ditto.
4659 (avx512pf_scatterpf<mode>): Ditto.
4660 (*avx512pf_scatterpf<mode>_mask): Ditto.
4661 (*avx512pf_scatterpf<mode>): Ditto.
4662 (GATHER_SCATTER_SF_MEM_MODE): New.
4663 (avx512pf_gatherpf<mode>df): Ditto.
4664 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4665 (*avx512pf_scatterpf<mode>df): Ditto.
4667 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4670 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4671 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4674 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4676 * common/config/arm/arm-common.c
4677 (arm_rewrite_mcpu): Handle multiple names.
4679 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4681 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4683 * gimple-builder.h (create_gimple_tmp): Delete.
4685 2014-01-27 Christian Bruel <christian.bruel@st.com>
4687 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4690 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4692 * config/pa/pa.md (call): Generate indirect long calls to non-local
4693 functions when outputing 32-bit code.
4694 (call_value): Likewise except for special call to buggy powf function.
4696 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4697 portable runtime and PIC indirect calls.
4698 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4699 and PIC call sequences. Use ldo instead of blr to set return register
4700 in PIC call sequence.
4702 2014-01-25 Walter Lee <walt@tilera.com>
4704 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4705 avoid clobbering a live register.
4707 2014-01-25 Walter Lee <walt@tilera.com>
4709 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4710 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4711 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4712 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4714 2014-01-25 Walter Lee <walt@tilera.com>
4716 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4717 arguments on even registers.
4718 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4720 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4721 (BIGGEST_ALIGNMENT): Ditto.
4722 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4724 2014-01-25 Walter Lee <walt@tilera.com>
4726 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4727 insns before bundling.
4728 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4730 2014-01-25 Walter Lee <walt@tilera.com>
4732 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4733 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4734 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4736 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4738 * config/mips/constraints.md (kl): Delete.
4739 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4740 define expands, using...
4741 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4742 instructions for MIPS16.
4743 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4744 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4746 2014-01-25 Walter Lee <walt@tilera.com>
4748 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4752 2014-01-25 Walter Lee <walt@tilera.com>
4754 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4755 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4757 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4759 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4762 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4764 * print-rtl.c (in_call_function_usage): New var.
4765 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4766 EXPR_LIST mode as mode and not as reg note name.
4769 * cfgloopmanip.c (copy_loop_info): If
4770 loop->warned_aggressive_loop_optimizations, make sure
4771 the flag is set in target loop too.
4773 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4775 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4777 * builtins.def: Likewise.
4778 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4779 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4780 * ira.c (ira_setup_eliminable_regset): Likewise.
4781 * omp-low.c (gate_expand_omp): Likewise.
4782 (execute_lower_omp): Likewise.
4783 (diagnose_sb_0): Likewise.
4784 (gate_diagnose_omp_blocks): Likewise.
4785 (simd_clone_clauses_extract): Likewise.
4788 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4790 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4791 correction for little endian...
4792 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4795 2014-01-24 Jeff Law <law@redhat.com>
4797 PR tree-optimization/59919
4798 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4799 for non-returning calls.
4801 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4803 * common/config/aarch64/aarch64-common.c
4804 (aarch64_rewrite_mcpu): Handle multiple names.
4805 * config/aarch64/aarch64.h
4806 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4808 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4810 * input.c (add_file_to_cache_tab): Handle the case where fopen
4813 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4816 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4817 from push operand if code of push isn't PRE_DEC.
4819 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4822 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4823 -mquad-memory-atomic. Update -mquad-memory documentation to say
4824 it is only used for non-atomic loads/stores.
4826 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4827 -mquad-memory or -mquad-memory-atomic switches.
4829 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4830 -mquad-memory-atomic to ISA 2.07 support.
4832 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4833 to separate support of normal quad word memory operations (ldq, stq)
4834 from the atomic quad word memory operations.
4836 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4837 support to separate non-atomic quad word operations from atomic
4838 quad word operations. Disable non-atomic quad word operations in
4839 little endian mode so that we don't have to swap words after the
4840 load and before the store.
4841 (quad_load_store_p): Add comment about atomic quad word support.
4842 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4843 options printed with -mdebug=reg.
4845 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4846 -mquad-memory-atomic as the test for whether we have quad word
4847 atomic instructions.
4848 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4849 or -mp8-vector are used, allow byte/half-word atomic operations.
4851 * config/rs6000/sync.md (load_lockedti): Insure that the address
4852 is a proper indexed or indirect address for the lqarx instruction.
4853 On little endian systems, swap the hi/lo registers after the lqarx
4855 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4856 insure the address is valid for the lqarx instruction.
4857 (store_conditionalti): Insure that the address is a proper indexed
4858 or indirect address for the stqcrx. instruction. On little endian
4859 systems, swap the hi/lo registers before doing the stqcrx.
4861 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4862 insure the address is valid for the stqcrx. instruction.
4864 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4865 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4866 type of quad memory support is available.
4868 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4871 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4872 there is a danger of looping.
4874 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4876 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4877 force flag_ira_loop_pressure if set via command line.
4879 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4881 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4882 (ashr_simd): New builtin handling DI mode.
4883 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4884 (aarch64_sshr_simddi): New match pattern.
4885 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4886 (vshrd_n_s64): Likewise.
4887 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4889 2014-01-23 Nick Clifton <nickc@redhat.com>
4891 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4892 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4893 favour of mcu specific scripts.
4894 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4897 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4898 Alex Velenko <Alex.Velenko@arm.com>
4900 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4901 (vaddv_s16): Likewise.
4902 (vaddv_s32): Likewise.
4903 (vaddv_u8): Likewise.
4904 (vaddv_u16): Likewise.
4905 (vaddv_u32): Likewise.
4906 (vaddvq_s8): Likewise.
4907 (vaddvq_s16): Likewise.
4908 (vaddvq_s32): Likewise.
4909 (vaddvq_s64): Likewise.
4910 (vaddvq_u8): Likewise.
4911 (vaddvq_u16): Likewise.
4912 (vaddvq_u32): Likewise.
4913 (vaddvq_u64): Likewise.
4914 (vaddv_f32): Likewise.
4915 (vaddvq_f32): Likewise.
4916 (vaddvq_f64): Likewise.
4917 (vmaxv_f32): Likewise.
4918 (vmaxv_s8): Likewise.
4919 (vmaxv_s16): Likewise.
4920 (vmaxv_s32): Likewise.
4921 (vmaxv_u8): Likewise.
4922 (vmaxv_u16): Likewise.
4923 (vmaxv_u32): Likewise.
4924 (vmaxvq_f32): Likewise.
4925 (vmaxvq_f64): Likewise.
4926 (vmaxvq_s8): Likewise.
4927 (vmaxvq_s16): Likewise.
4928 (vmaxvq_s32): Likewise.
4929 (vmaxvq_u8): Likewise.
4930 (vmaxvq_u16): Likewise.
4931 (vmaxvq_u32): Likewise.
4932 (vmaxnmv_f32): Likewise.
4933 (vmaxnmvq_f32): Likewise.
4934 (vmaxnmvq_f64): Likewise.
4935 (vminv_f32): Likewise.
4936 (vminv_s8): Likewise.
4937 (vminv_s16): Likewise.
4938 (vminv_s32): Likewise.
4939 (vminv_u8): Likewise.
4940 (vminv_u16): Likewise.
4941 (vminv_u32): Likewise.
4942 (vminvq_f32): Likewise.
4943 (vminvq_f64): Likewise.
4944 (vminvq_s8): Likewise.
4945 (vminvq_s16): Likewise.
4946 (vminvq_s32): Likewise.
4947 (vminvq_u8): Likewise.
4948 (vminvq_u16): Likewise.
4949 (vminvq_u32): Likewise.
4950 (vminnmv_f32): Likewise.
4951 (vminnmvq_f32): Likewise.
4952 (vminnmvq_f64): Likewise.
4954 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4956 * config/aarch64/aarch64-simd.md
4957 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4958 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4959 (*aarch64_mul3_elt<mode>): Likewise.
4960 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4961 (*aarch64_mul3_elt_to_64v2df): Likewise.
4962 (*aarch64_mla_elt<mode>): Likewise.
4963 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4964 (*aarch64_mls_elt<mode>): Likewise.
4965 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4966 (*aarch64_fma4_elt<mode>): Likewise.
4967 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4968 (*aarch64_fma4_elt_to_64v2df): Likewise.
4969 (*aarch64_fnma4_elt<mode>): Likewise.
4970 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4971 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4972 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4973 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4974 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4975 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4976 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4977 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4978 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4980 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4982 * config/aarch64/aarch64-simd.md
4983 (aarch64_be_checked_get_lane<mode>): New define_expand.
4984 * config/aarch64/aarch64-simd-builtins.def
4985 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4986 New builtin definition.
4987 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4988 Use new safe be builtin.
4990 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4992 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4994 (aarch64_be_st1<mode>): Likewise.
4995 (aarch_ld1<VALL:mode>): Define_expand modified.
4996 (aarch_st1<VALL:mode>): Likewise.
4997 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4998 (UNSPEC_ST1): Likewise.
5000 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
5002 * config/microblaze/microblaze.md: Add trap insn and attribute
5004 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5006 PR preprocessor/58580
5007 * input.h (location_get_source_line): Take an additional line_size
5009 (void diagnostics_file_cache_fini): Declare new function.
5010 * input.c (struct fcache): New type.
5011 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5012 New static constants.
5013 (diagnostic_file_cache_init, total_lines_num)
5014 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5015 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5016 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5017 (get_next_line, read_next_line, goto_next_line, read_line_num):
5018 New static function definitions.
5019 (diagnostic_file_cache_fini): New function.
5020 (location_get_source_line): Take an additional output line_len
5021 parameter. Re-write using lookup_or_add_file_to_cache_tab and
5023 * diagnostic.c (diagnostic_finish): Call
5024 diagnostic_file_cache_fini.
5025 (adjust_line): Take an additional input parameter for the length
5026 of the line, rather than calculating it with strlen.
5027 (diagnostic_show_locus): Adjust the use of
5028 location_get_source_line and adjust_line with respect to their new
5029 signature. While displaying a line now, do not stop at the first
5030 null byte. Rather, display the zero byte as a space and keep
5031 going until we reach the size of the line.
5032 * Makefile.in: Add vec.o to OBJS-libcommon
5034 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5035 Ilya Tocar <ilya.tocar@intel.com>
5037 * config/i386/avx512fintrin.h (_mm512_kmov): New.
5038 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
5039 (__builtin_ia32_kmov16): Ditto.
5040 * config/i386/i386.md (UNSPEC_KMOV): New.
5043 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5045 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
5046 (_mm512_storeu_si512): Ditto.
5048 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
5051 * rtl.h (get_referenced_operands): Declare.
5052 * recog.c (get_referenced_operands): New function.
5053 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
5054 operands have been referenced when recording LO_SUM references.
5056 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
5058 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
5060 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5062 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5063 Enable for generic and recent AMD targets.
5065 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5067 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5068 ARG_SIZE note when adjustment was eliminated.
5070 2014-01-22 Jeff Law <law@redhat.com>
5072 PR tree-optimization/59597
5073 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5074 in file. Accept new argument REGISTERING and use it to modify
5075 dump output appropriately.
5076 (register_jump_thread): Corresponding changes.
5077 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5078 thread paths involving joiner blocks. Add code to dump cancelled
5079 jump threading paths.
5081 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5083 PR rtl-optimization/59477
5084 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5085 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5087 2014-01-22 Tom Tromey <tromey@redhat.com>
5089 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5091 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5093 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5095 PR rtl-optimization/59896
5096 * lra-constraints.c (process_alt_operands): Check unused note for
5097 matched operands of insn with no output reloads.
5099 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5101 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5102 (mips_move_from_gpr_cost): Likewise.
5104 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5106 PR rtl-optimization/59858
5107 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5108 ira_class_hard_regs_num.
5109 (process_alt_operands): Increase reject for dying matched operand.
5111 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5114 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5115 smaller than size, perform several stores or loads and stores
5116 at dst + count - size to store or copy all of size bytes, rather
5117 than just last modesize bytes.
5119 2014-01-20 DJ Delorie <dj@redhat.com>
5121 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5122 that CLOBBERs are REGs before propogating their values.
5124 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5127 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5128 (cgraph_inline_failed_type): New function.
5129 * cgraph.h (DEFCIFCODE): Add type.
5130 (cgraph_inline_failed_type_t): New enum.
5131 (cgraph_inline_failed_type): New prototype.
5132 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5133 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5134 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5135 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5136 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5137 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5138 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5139 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5140 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5141 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5142 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5143 OPTIMIZATION_MISMATCH.
5144 * tree-inline.c (expand_call_inline): Emit errors during
5145 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5147 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5150 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5151 mode attribute in insn output.
5153 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5155 * output.h (output_constant): Delete.
5156 * varasm.c (output_constant): Make private.
5158 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5160 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5162 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5165 * tree.h (fold_builtin_strcat): New prototype.
5166 * builtins.c (fold_builtin_strcat): No longer static. Add len
5167 argument, if non-NULL, don't call c_strlen. Optimize
5168 directly into __builtin_memcpy instead of __builtin_strcpy.
5169 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5170 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5172 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5174 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5175 for SImode_address_operand operands, having only a REG argument.
5177 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5179 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5180 loader name using mbig-endian.
5181 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5183 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5185 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5189 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5191 * config/aarch64/aarch64-protos.h
5192 (aarch64_cannot_change_mode_class_ptr): Declare.
5193 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5194 aarch64_cannot_change_mode_class_ptr): New.
5195 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5196 backend hook aarch64_cannot_change_mode_class.
5198 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5200 * common/config/aarch64/aarch64-common.c
5201 (aarch64_handle_option): Don't handle any option order logic here.
5202 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5203 selected_cpu, warn on architecture version mismatch.
5204 (aarch64_override_options): Fix parsing order for option strings.
5206 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5207 Iain Sandoe <iain@codesourcery.com>
5210 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5211 warning. Amend comment to reflect current functionality.
5213 2014-01-20 Richard Biener <rguenther@suse.de>
5216 * builtins.c (fold_builtin_strcat): Remove case better handled
5217 by tree-ssa-strlen.c.
5219 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5221 * config/aarch64/aarch64.opt
5222 (mcpu, march, mtune): Make case-insensitive.
5224 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5227 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5228 if operands[1] is a REG or ZERO_EXTEND of a REG.
5230 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5232 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5234 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5236 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5237 long non-pic millicode calls.
5239 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5241 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5243 2014-01-19 Kito Cheng <kito@0xlab.org>
5245 * builtins.c (expand_movstr): Check movstr expand done or fail.
5247 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5248 H.J. Lu <hongjiu.lu@intel.com>
5251 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5252 to DImode for zero-extended addresses.
5254 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5256 PR rtl-optimization/57763
5257 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5258 on the new indirect jump_insn and increment LABEL_NUSES (label).
5260 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5264 * config.gcc (x86_archs): New variable.
5265 (x86_64_archs): Likewise.
5266 (x86_cpus): Likewise.
5267 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5268 --with-arch/--with-cpu= options.
5269 Support --with-arch=/--with-cpu={nehalem,westmere,
5270 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5272 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5274 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5275 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5277 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5279 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5281 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5284 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5285 clear cpp_get_options (parse_in)->warn_unused_macros for
5286 ix86_target_macros_internal with cpp_define.
5288 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5290 * jump.c (delete_related_insns): Keep (use (insn))s.
5291 * reorg.c (redundant_insn): Check for barriers too.
5293 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5295 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5297 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5299 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5300 call to $$dyncall when TARGET_LONG_CALLS is true.
5302 2014-01-17 Jeff Law <law@redhat.com>
5304 * ree.c (combine_set_extension): Temporarily disable test for
5305 changing number of hard registers.
5307 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5310 * ipa-inline-analysis.c (inline_free_summary):
5311 Do not free summary of aliases.
5313 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5316 * gimplify.c (gimplify_expr): Use create_tmp_var
5317 instead of create_tmp_var_raw. If cond doesn't have
5318 integral type, don't add the IFN_ANNOTATE builtin at all.
5320 2014-01-17 Martin Jambor <mjambor@suse.cz>
5323 * ipa-cp.c (prev_edge_clone): New variable.
5324 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5325 Also resize prev_edge_clone vector.
5326 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5327 (ipcp_edge_removal_hook): New function.
5328 (ipcp_driver): Register ipcp_edge_removal_hook.
5330 2014-01-17 Andrew Pinski <apinski@cavium.com>
5331 Steve Ellcey <sellcey@mips.com>
5334 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5337 2014-01-17 Jeff Law <law@redhat.com>
5340 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5341 so that pass_ccp runs first.
5343 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5345 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5346 (ix86_adjust_cost): Use !TARGET_XXX.
5347 (do_reorder_for_imul): Likewise.
5348 (swap_top_of_ready_list): Likewise.
5349 (ix86_sched_reorder): Likewise.
5351 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5353 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5354 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5355 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5356 (intel_memset): New. Duplicate slm_memset.
5357 (intel_cost): New. Duplicate slm_cost.
5358 (m_INTEL): New macro.
5359 (processor_target_table): Add "intel".
5360 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5361 with PROCESSOR_INTEL for "intel".
5362 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5363 PROCESSOR_SILVERMONT.
5364 (ix86_issue_rate): Likewise.
5365 (ix86_adjust_cost): Likewise.
5366 (ia32_multipass_dfa_lookahead): Likewise.
5367 (swap_top_of_ready_list): Likewise.
5368 (ix86_sched_reorder): Likewise.
5369 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5370 instead of TARGET_OPT_AGU.
5371 * config/i386/i386.h (TARGET_INTEL): New.
5372 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5373 (processor_type): Add PROCESSOR_INTEL.
5374 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5375 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5377 2014-01-17 Marek Polacek <polacek@redhat.com>
5380 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5383 2014-01-17 Richard Biener <rguenther@suse.de>
5385 PR tree-optimization/46590
5386 * opts.c (default_options_table): Add entries for
5387 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5388 all enabled at -O1 but not for -Og.
5389 * common.opt (fbranch-count-reg): Remove Init(1).
5390 (fmove-loop-invariants): Likewise.
5391 (ftree-pta): Likewise.
5393 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5395 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5396 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5397 decls to at least the GCC 4.8 used alignments.
5400 * tree-nested.c (convert_nonlocal_reference_stmt,
5401 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5402 of GIMPLE_BIND stmts, adjust associated decls.
5404 2014-01-17 Richard Biener <rguenther@suse.de>
5406 PR tree-optimization/46590
5407 * vec.h (vec<>::bseach): New member function implementing
5408 binary search according to C89 bsearch.
5409 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5410 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5411 bitmap pointer again. Make accesses_in_loop a flat array.
5412 (mem_ref_obstack): New global.
5413 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5414 (mark_ref_stored): Likewise.
5415 (ref_indep_loop_p_2): Likewise.
5416 (set_ref_stored_in_loop): New helper function.
5417 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5418 (memref_free): Adjust.
5419 (record_mem_ref_loc): Simplify.
5420 (gather_mem_refs_stmt): Adjust.
5421 (sort_locs_in_loop_postorder_cmp): New function.
5422 (analyze_memory_references): Sort accesses_in_loop after
5423 loop postorder number.
5424 (find_ref_loc_in_loop_cmp): New function.
5425 (for_all_locs_in_loop): Find relevant cluster of locs in
5426 accesses_in_loop and iterate without recursion.
5427 (execute_sm): Avoid uninit warning.
5428 (struct ref_always_accessed): Simplify.
5429 (ref_always_accessed::operator ()): Likewise.
5430 (ref_always_accessed_p): Likewise.
5431 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5432 loop postorder numbers here.
5433 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5436 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5439 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5440 on decls for which assemble_alias has been called.
5442 2014-01-17 Nick Clifton <nickc@redhat.com>
5444 * config/msp430/msp430.opt: (mcpu): New option.
5445 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5446 (msp430_option_override): Parse target_cpu. If the MCU name
5447 matches a generic string, clear target_mcu.
5448 (msp430_attr): Allow numeric interrupt values up to 63.
5449 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5450 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5452 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5454 * config/msp430/msp430.md (popm): Use %J rather than %I.
5455 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5456 (addhi_cy_i): Use immediate_operand for operand 2.
5457 * doc/invoke.texi: Document -mcpu option.
5459 2014-01-17 Richard Biener <rguenther@suse.de>
5461 PR rtl-optimization/38518
5462 * df.h (df_analyze_loop): Declare.
5463 * df-core.c: Include cfgloop.h.
5464 (df_analyze_1): Split out main part of df_analyze.
5465 (df_analyze): Adjust.
5466 (loop_inverted_post_order_compute): New function.
5467 (loop_post_order_compute): Likewise.
5468 (df_analyze_loop): New function avoiding whole-function
5470 * loop-invariant.c (find_defs): Use df_analyze_loop.
5471 (find_invariants): Adjust.
5472 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5474 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5476 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5477 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5479 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5481 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5482 traversal when removing references.
5484 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5487 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5489 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5492 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5493 pushing a reload for an autoinc when we had previously reloaded an
5494 inner part of the address.
5496 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5498 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5500 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5501 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5502 when not giving up or versioning for alias only because of
5504 (vect_analyze_data_ref_dependences): Set to true.
5505 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5507 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5508 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5512 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5515 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5516 operand 0 predicate for gathers, use a new pseudo as subtarget.
5518 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5521 * lra-constraints.c (process_alt_operands): Add printing debug info.
5522 Check absence of input/output reloads for matched operands too.
5524 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5526 PR rtl-optimization/59835
5527 * ira.c (ira_init_register_move_cost): Increase cost for
5530 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5532 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5534 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5537 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5538 non-register objects. Use gen_(high/low)part more consistently.
5541 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5544 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5545 endian support, remove tests for WORDS_BIG_ENDIAN.
5546 (p8_mfvsrd_3_<mode>): Likewise.
5547 (reload_gpr_from_vsx<mode>): Likewise.
5548 (reload_gpr_from_vsxsf): Likewise.
5549 (p8_mfvsrd_4_disf): Likewise.
5551 2014-01-16 Richard Biener <rguenther@suse.de>
5553 PR rtl-optimization/46590
5554 * lcm.c (compute_antinout_edge): Use postorder iteration.
5555 (compute_laterin): Use inverted postorder iteration.
5557 2014-01-16 Nick Clifton <nickc@redhat.com>
5560 * varasm.c (output_constant): Return the number of bytes actually
5562 (output_constructor_array_range): Update the field size with the
5563 number of bytes emitted by output_constant.
5564 (output_constructor_regular_field): Likewise. Also do not
5565 complain if the total number of bytes emitted is now greater
5566 than the expected fieldpos.
5567 * output.h (output_constant): Update prototype and descriptive comment.
5569 2014-01-16 Marek Polacek <polacek@redhat.com>
5572 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5573 it is error_mark_node.
5575 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5577 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5578 VALID_AVX256_REG_OR_OI_MODE.
5580 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5582 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5583 current procedure should be profiled.
5585 2014-01-15 Andrew Pinski <apinski@cavium.com>
5587 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5588 of moving from/to the STACK_REG register class.
5590 2014-01-15 Richard Henderson <rth@redhat.com>
5593 * reginfo.c (global_regs_decl): Globalize.
5594 * rtl.h (global_regs_decl): Declare.
5595 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5596 reserved via global_regs.
5598 2014-01-15 Teresa Johnson <tejohnson@google.com>
5600 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5602 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5604 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5605 and vmulosh rather than call gen_vec_widen_smult_*.
5606 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5607 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5608 (vec_widen_smult_even_v16qi): Likewise.
5609 (vec_widen_umult_even_v8hi): Likewise.
5610 (vec_widen_smult_even_v8hi): Likewise.
5611 (vec_widen_umult_odd_v16qi): Likewise.
5612 (vec_widen_smult_odd_v16qi): Likewise.
5613 (vec_widen_umult_odd_v8hi): Likewise.
5614 (vec_widen_smult_odd_v8hi): Likewise.
5615 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5616 vmuloub rather than call gen_vec_widen_umult_*.
5617 (vec_widen_umult_lo_v16qi): Likewise.
5618 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5619 vmulosb rather than call gen_vec_widen_smult_*.
5620 (vec_widen_smult_lo_v16qi): Likewise.
5621 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5622 rather than call gen_vec_widen_umult_*.
5623 (vec_widen_umult_lo_v8hi): Likewise.
5624 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5625 rather than call gen_vec_widen_smult_*.
5626 (vec_widen_smult_lo_v8hi): Likewise.
5628 2014-01-15 Jeff Law <law@redhat.com>
5630 PR tree-optimization/59747
5631 * ree.c (find_and_remove_re): Properly handle case where a second
5632 eliminated extension requires widening a copy created for elimination
5633 of a prior extension.
5634 (combine_set_extension): Ensure that the number of hard regs needed
5635 for a destination register does not change when we widen it.
5637 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5639 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5640 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5641 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5642 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5643 (avr-*-rtems*): Likewise.
5644 (bfin*-rtems*): Likewise.
5645 (moxie-*-rtems*): Likewise.
5646 (h8300-*-rtems*): Likewise.
5647 (i[34567]86-*-rtems*): Likewise.
5648 (lm32-*-rtems*): Likewise.
5649 (m32r-*-rtems*): Likewise.
5650 (m68k-*-rtems*): Likewise.
5651 (microblaze*-*-rtems*): Likewise.
5652 (mips*-*-rtems*): Likewise.
5653 (powerpc-*-rtems*): Likewise.
5654 (sh-*-rtems*): Likewise.
5655 (sparc-*-rtems*): Likewise.
5656 (sparc64-*-rtems*): Likewise.
5657 (v850-*-rtems*): Likewise.
5658 (m32c-*-rtems*): Likewise.
5660 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5662 PR rtl-optimization/59511
5663 * ira.c (ira_init_register_move_cost): Use memory costs for some
5664 cases of register move cost calculations.
5665 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5666 instead of BB frequency.
5667 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5668 * lra-assigns.c (find_hard_regno_for): Ditto.
5670 2014-01-15 Richard Biener <rguenther@suse.de>
5672 PR tree-optimization/59822
5673 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5674 (vectorizable_load): Use it to hoist defs of uses of invariant
5675 loads out of the loop.
5677 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5678 Kugan Vivekanandarajah <kuganv@linaro.org>
5681 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5684 2014-01-15 Richard Biener <rguenther@suse.de>
5686 PR rtl-optimization/59802
5687 * lcm.c (compute_available): Use inverted postorder to seed
5688 the initial worklist.
5690 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5693 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5694 ADDR_REGS for invalid symrefs in non-PIC code.
5696 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5699 * builtins.c (determine_block_size): Initialize *probable_max_size
5700 even if len_rtx is CONST_INT.
5702 2014-01-14 Andrew Pinski <apinski@cavium.com>
5704 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5705 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5706 (cortexa53_tunings): Likewise.
5707 (aarch64_sched_issue_rate): New function.
5708 (TARGET_SCHED_ISSUE_RATE): Define.
5710 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5712 * ira-costs.c (find_costs_and_classes): Add missed
5713 ira_init_register_move_cost_if_necessary.
5715 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5718 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5720 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5723 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5724 to indicate if type is used for function return value. Warn ABI
5725 change if the vector mode isn't available for function return value.
5726 (ix86_function_arg_advance): Pass false to type_natural_mode.
5727 (ix86_function_arg): Likewise.
5728 (ix86_gimplify_va_arg): Likewise.
5729 (function_arg_32): Don't warn ABI change.
5730 (ix86_function_value): Pass true to type_natural_mode.
5731 (ix86_return_in_memory): Likewise.
5732 (ix86_struct_value_rtx): Removed.
5733 (TARGET_STRUCT_VALUE_RTX): Likewise.
5735 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5737 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5738 converting a conditional jump into a conditional return.
5740 2014-01-14 Richard Biener <rguenther@suse.de>
5742 PR tree-optimization/58921
5743 PR tree-optimization/59006
5744 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5745 hoisting invariant stmts.
5746 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5747 invariant loads on the preheader edge if possible.
5749 2014-01-14 Joey Ye <joey.ye@arm.com>
5751 * doc/plugin.texi (Building GCC plugins): Update to C++.
5753 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5755 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5756 (_mm_rcp28_round_ss): Ditto.
5757 (_mm_rsqrt28_round_sd): Ditto.
5758 (_mm_rsqrt28_round_ss): Ditto.
5759 (_mm_rcp28_sd): Ditto.
5760 (_mm_rcp28_ss): Ditto.
5761 (_mm_rsqrt28_sd): Ditto.
5762 (_mm_rsqrt28_ss): Ditto.
5763 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5764 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5765 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5766 (IX86_BUILTIN_RCP28SD): Ditto.
5767 (IX86_BUILTIN_RCP28SS): Ditto.
5768 (IX86_BUILTIN_RSQRT28SD): Ditto.
5769 (IX86_BUILTIN_RSQRT28SS): Ditto.
5770 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5771 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5772 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5773 (ix86_expand_special_args_builtin): Expand new FTYPE.
5774 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5775 (srcp14<mode>): Make insn unary.
5776 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5777 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5778 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5779 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5780 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5781 Fix rounding: make it SAE only.
5782 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5784 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5786 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5787 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5788 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5789 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5790 (round_saeonly_mask_scalar_operand4): Ditto.
5791 (round_saeonly_mask_scalar_op3): Ditto.
5792 (round_saeonly_mask_scalar_op4): Ditto.
5794 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5796 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5797 Implement -maltivec=be for vec_insert and vec_extract.
5799 2014-01-10 DJ Delorie <dj@redhat.com>
5801 * config/msp430/msp430.md (call_internal): Don't allow memory
5802 references with SP as the base register.
5803 (call_value_internal): Likewise.
5804 * config/msp430/constraints.md (Yc): New. For memory references
5805 that don't use SP as a base register.
5807 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5808 "an integer without a # prefix"
5809 * config/msp430/msp430.md (epilogue_helper): Use it.
5811 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5814 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5815 AVX512F gather builtins.
5816 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5817 on gather decls with INTEGER_TYPE masktype.
5818 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5819 directly into the builtin rather than hoisting it before loop.
5821 PR tree-optimization/59387
5822 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5823 (scev_const_prop): If folded_casts and type has undefined overflow,
5824 use force_gimple_operand instead of force_gimple_operand_gsi and
5825 for each added stmt if it is assign with
5826 arith_code_with_undefined_signed_overflow, call
5827 rewrite_to_defined_overflow.
5828 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5829 gimple-fold.h instead.
5830 (arith_code_with_undefined_signed_overflow,
5831 rewrite_to_defined_overflow): Moved to ...
5832 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5833 rewrite_to_defined_overflow): ... here. No longer static.
5834 Include gimplify-me.h.
5835 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5836 rewrite_to_defined_overflow): New prototypes.
5838 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5840 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5842 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5844 * builtins.c (get_object_alignment_2): Minor tweak.
5845 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5847 2014-01-13 Christian Bruel <christian.bruel@st.com>
5849 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5850 optimized non constant lengths.
5852 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5855 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5856 load as __atomic_load_N if possible.
5858 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5860 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5862 (rs6000_expand_builtin): Adjust call.
5864 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5867 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5868 * config/rs6000/rs6000.c: Include target-globals.h.
5869 (rs6000_set_current_function): Instead of doing target_reinit
5870 unconditionally, use save_target_globals_default_opts and
5871 restore_target_globals.
5873 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5875 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5876 (rs6000_expand_builtin): Handle mffs and mtfsf.
5877 (rs6000_init_builtins): Define mffs and mtfsf.
5878 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5879 (rs6000_mffs): New pattern.
5880 (rs6000_mtfsf): New pattern.
5882 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5884 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5885 Start narrowing with START. Apply candidate-use pair
5886 and check overall cost in narrowing.
5887 (iv_ca_prune): Pass new argument.
5889 2014-01-10 Jeff Law <law@redhat.com>
5892 * ree.c (combine_reaching_defs): Ensure the defining statement
5893 occurs before the extension when optimizing extensions with
5894 different source and destination hard registers.
5896 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5899 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5900 vtables into the type inheritance graph.
5902 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5904 PR rtl-optimization/59754
5905 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5906 modes in the REGNO != REGNO case.
5908 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5910 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5912 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5914 PR tree-optimization/59745
5915 * tree-predcom.c (tree_predictive_commoning_loop): Call
5916 free_affine_expand_cache if giving up because components is NULL.
5918 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5919 GC in payload of target_globals struct instead of allocating them on
5920 the heap and the larger structs separately using GC.
5921 * target-globals.h (struct target_globals): Make regs, hard_regs,
5922 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5923 of GTY((skip)) and change type to void *.
5924 (reset_target_globals): Cast loads from those fields to corresponding
5927 2014-01-10 Steve Ellcey <sellcey@mips.com>
5930 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5931 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5932 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5934 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5937 * aarch64-modes.def (CC_Zmode): New flags mode.
5938 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5939 represents an equality.
5940 (aarch64_get_condition_code): Handle CC_Zmode.
5941 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5943 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5945 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5946 extraction in good case.
5948 2014-01-10 Richard Biener <rguenther@suse.de>
5950 PR tree-optimization/59374
5951 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5952 checking after SLP discovery. Mark stmts not participating
5953 in any SLP instance properly.
5955 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5957 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5958 when handling a SET rtx.
5960 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5962 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5963 (cortex-a57): Likewise.
5964 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5966 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5968 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5969 non-iwmmxt builtins.
5971 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5975 * ipa-devirt.c record_target_from_binfo): Take as argument
5976 stack of binfos and lookup matching one for virtual inheritance.
5977 (possible_polymorphic_call_targets_1): Update.
5979 2014-01-10 Huacai Chen <chenhc@lemote.com>
5981 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5982 kernel strings for Loongson-2E/2F/3A.
5984 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5987 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5988 is_gimple_call before calling gimple_call_internal_p.
5990 2014-01-09 Steve Ellcey <sellcey@mips.com>
5992 * Makefile.in (TREE_FLOW_H): Remove.
5993 (TREE_SSA_H): Add file names from tree-flow.h.
5994 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5995 * tree.h: Remove tree-flow.h reference.
5996 * hash-table.h: Remove tree-flow.h reference.
5997 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5998 reference with tree-ssa-loop.h.
6000 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6002 * doc/invoke.texi: Add -maltivec={be,le} options, and document
6003 default element-order behavior for -maltivec.
6004 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6005 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6006 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6007 when targeting big endian, at least for now.
6008 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6010 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6013 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6014 var satisfies use_register_for_decl, just take into account type
6015 alignment, rather than decl alignment.
6017 PR tree-optimization/59622
6018 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
6019 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
6020 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
6021 Don't devirtualize for inplace at all. For targets.length () == 1,
6022 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
6024 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6026 * config/i386/i386.md (cpu): Remove the unused btver1.
6028 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6030 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
6032 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6035 * tree-core.h (struct target_globals): New forward declaration.
6036 (struct tree_target_option): Add globals field.
6037 * tree.h (TREE_TARGET_GLOBALS): Define.
6038 (prepare_target_option_nodes_for_pch): New prototype.
6039 * target-globals.h (struct target_globals): Define even if
6041 * tree.c (prepare_target_option_node_for_pch,
6042 prepare_target_option_nodes_for_pch): New functions.
6043 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
6044 * config/i386/i386.c: Include target-globals.h.
6045 (ix86_set_current_function): Instead of doing target_reinit
6046 unconditionally, use save_target_globals_default_opts and
6047 restore_target_globals.
6049 2014-01-09 Richard Biener <rguenther@suse.de>
6051 PR tree-optimization/59715
6052 * tree-cfg.h (split_critical_edges): Declare.
6053 * tree-cfg.c (split_critical_edges): Export.
6054 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
6056 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
6058 * cfgexpand.c (expand_stack_vars): Optionally disable
6059 asan stack protection.
6060 (expand_used_vars): Likewise.
6061 (partition_stack_vars): Likewise.
6062 * asan.c (asan_emit_stack_protection): Optionally disable
6063 after return stack usage.
6064 (instrument_derefs): Optionally disable memory access instrumentation.
6065 (instrument_builtin_call): Likewise.
6066 (instrument_strlen_call): Likewise.
6067 (asan_protect_global): Optionally disable global variables protection.
6068 * doc/invoke.texi: Added doc for new options.
6069 * params.def: Added new options.
6070 * params.h: Likewise.
6072 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6074 PR rtl-optimization/59724
6075 * ifcvt.c (cond_exec_process_if_block): Don't call
6076 flow_find_head_matching_sequence with 0 longest_match.
6077 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6078 non-active insns if !stop_after.
6079 (try_head_merge_bb): Revert 2014-01-07 changes.
6081 2014-01-08 Jeff Law <law@redhat.com>
6083 * ree.c (get_sub_rtx): New function, extracted from...
6084 (merge_def_and_ext): Here.
6085 (combine_reaching_defs): Use get_sub_rtx.
6087 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6089 * cgraph.h (varpool_variable_node): Do not choke on null node.
6091 2014-01-08 Catherine Moore <clm@codesourcery.com>
6093 * config/mips/mips.md (simple_return): Attempt to use JRC
6095 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6097 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6099 PR rtl-optimization/59137
6100 * reorg.c (steal_delay_list_from_target): Call update_block for
6102 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6104 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6106 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6107 two duplicate entries.
6109 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6112 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6114 * config/mips/mips.c (mips_truncated_op_cost): New function.
6115 (mips_rtx_costs): Adjust test for BADDU.
6116 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6118 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6120 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6121 (*baddu_si): ...this new pattern.
6123 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6126 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6128 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6131 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6133 (expand_expr, expand_normal): Adjust.
6134 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6135 inner_reference_p. Use inner_reference_p to expand inner references.
6136 (store_expr): Adjust.
6137 * cfgexpand.c (expand_call_stmt): Adjust.
6139 2014-01-08 Rong Xu <xur@google.com>
6141 * gcov-io.c (gcov_var): Move from gcov-io.h.
6142 (gcov_position): Ditto.
6143 (gcov_is_error): Ditto.
6144 (gcov_rewrite): Ditto.
6145 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6146 only part to libgcc/libgcov.h.
6148 2014-01-08 Marek Polacek <polacek@redhat.com>
6151 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6153 2014-01-08 Marek Polacek <polacek@redhat.com>
6156 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6158 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6160 PR rtl-optimization/59649
6161 * stor-layout.c (get_mode_bounds): For BImode return
6162 0 and STORE_FLAG_VALUE.
6164 2014-01-08 Richard Biener <rguenther@suse.de>
6167 * gimple.h (is_gimple_builtin_call): Remove.
6168 (gimple_builtin_call_types_compatible_p): New.
6169 (gimple_call_builtin_p): New overload.
6170 * gimple.c (is_gimple_builtin_call): Remove.
6171 (validate_call): Rename to ...
6172 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6174 (validate_type): New static function.
6175 (gimple_call_builtin_p): New overload and adjust.
6176 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6177 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6178 (gimple_fold_stmt_to_constant_1): Likewise.
6179 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6181 2014-01-08 Richard Biener <rguenther@suse.de>
6184 * gimplify.c (gimplify_expr): Gimplify register-register type
6185 VIEW_CONVERT_EXPRs to separate stmts.
6187 2014-01-07 Jeff Law <law@redhat.com>
6190 * ree.c (combine_set_extension): Handle case where source
6191 and destination registers in an extension insn are different.
6192 (combine_reaching_defs): Allow source and destination registers
6193 in extension to be different under limited circumstances.
6194 (add_removable_extension): Remove restriction that the
6195 source and destination registers in the extension are the same.
6196 (find_and_remove_re): Emit a copy from the extension's
6197 destination to its source after the defining insn if
6198 the source and destination registers are different.
6201 * ifcvt.c (merge_if_block): If we are merging a block with more than
6202 one successor with a block with no successors, remove any BARRIER
6203 after the second block.
6205 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6207 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6209 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6212 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6213 for 14-bit register offsets when INT14_OK_STRICT is false.
6215 2014-01-07 Roland Stigge <stigge@antcom.de>
6216 Michael Meissner <meissner@linux.vnet.ibm.com>
6219 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6220 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6222 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6224 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6227 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6229 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6230 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6231 rtx is const0_rtx or not.
6233 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6236 * target-globals.c (save_target_globals): Remove this_fn_optab
6238 * toplev.c: Include optabs.h.
6239 (target_reinit): Temporarily restore the global options if another
6240 set of options are in force.
6242 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6244 PR rtl-optimization/58668
6245 * cfgcleanup.c (flow_find_cross_jump): Don't count
6246 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6247 to determine what is counted.
6248 (flow_find_head_matching_sequence): Use active_insn_p to determine
6250 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6252 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6253 determine what is counted.
6255 PR tree-optimization/59643
6256 * tree-predcom.c (split_data_refs_to_components): If one dr is
6257 read and one write, determine_offset fails and the write isn't
6258 in the bad component, just put the read into the bad component.
6260 2014-01-07 Mike Stump <mikestump@comcast.net>
6261 Jakub Jelinek <jakub@redhat.com>
6264 * tree-core.h (struct tree_optimization_option): Change optabs
6265 type from unsigned char * to void *.
6266 * optabs.c (init_tree_optimization_optabs): Adjust
6267 TREE_OPTIMIZATION_OPTABS initialization.
6269 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6272 * config/i386/i386.h (struct machine_function): Add
6273 no_drap_save_restore field.
6274 * config/i386/i386.c (ix86_save_reg): Use
6275 !cfun->machine->no_drap_save_restore instead of
6276 crtl->stack_realign_needed.
6277 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6278 this function clears frame_pointer_needed. Set
6279 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6280 and DRAP reg is needed.
6282 2014-01-06 Marek Polacek <polacek@redhat.com>
6285 * doc/implement-c.texi: Mention that other integer types are
6286 permitted as bit-field types in strictly conforming mode.
6288 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6290 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6293 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6295 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6297 2014-01-06 Martin Jambor <mjambor@suse.cz>
6300 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6302 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6304 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6308 * var-tracking.c (add_stores): Preserve the value of the source even if
6309 we don't record the store.
6311 2014-01-06 Terry Guo <terry.guo@arm.com>
6313 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6315 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6318 * config/darwin.c (darwin_function_section): Adjust return values to
6319 correspond to optimisation changes made in r206070.
6321 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6323 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6324 from prefetch_block tune setting.
6325 (nocona_cost): Correct size of prefetch block to 64.
6327 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6329 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6330 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6331 used to save the static chain register in the computation of the offset
6332 from which the FP registers need to be restored.
6334 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6336 PR tree-optimization/59519
6337 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6338 ICE if get_current_def (current_new_name) is already non-NULL, as long
6339 as it is a phi result of some other phi in *new_exit_bb that has
6342 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6343 or vmovdqu* for misaligned_operand.
6344 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6345 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6346 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6347 aligned_mem for AVX512F masked aligned load and store builtins and for
6350 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6352 PR tree-optimization/59651
6353 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6354 Address range for negative step should be added by TYPE_SIZE_UNIT.
6356 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6358 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6359 overlapping registers also for registers other than the stack pointer.
6361 2014-01-03 Marek Polacek <polacek@redhat.com>
6364 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6367 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6370 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6373 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6374 (push splitter): Use <P:MODE_SIZE> instead of
6375 GET_MODE_SIZE (<P:MODE>mode).
6376 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6377 (mov -1, reg peephole2): Likewise.
6378 * config/i386/sse.md (*mov<mode>_internal,
6379 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6380 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6381 *<code><mode>3, *andnot<mode>3<mask_name>,
6382 <mask_codefor><code><mode>3<mask_name>): Likewise.
6383 * config/i386/subst.md (mask_mode512bit_condition,
6384 sd_mask_mode512bit_condition): Likewise.
6386 2014-01-02 Xinliang David Li <davidxl@google.com>
6388 PR tree-optimization/59303
6389 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6390 (dump_predicates): Better output format.
6391 (pred_equal_p): New function.
6392 (is_neq_relop_p): Ditto.
6393 (is_neq_zero_form_p): Ditto.
6394 (pred_expr_equal_p): Ditto.
6395 (pred_neg_p): Ditto.
6396 (simplify_pred): Ditto.
6397 (simplify_preds_2): Ditto.
6398 (simplify_preds_3): Ditto.
6399 (simplify_preds_4): Ditto.
6400 (simplify_preds): Ditto.
6402 (push_to_worklist): Ditto.
6403 (get_pred_info_from_cmp): Ditto.
6404 (is_degenerated_phi): Ditto.
6405 (normalize_one_pred_1): Ditto.
6406 (normalize_one_pred): Ditto.
6407 (normalize_one_pred_chain): Ditto.
6408 (normalize_preds): Ditto.
6409 (normalize_cond_1): Remove function.
6410 (normalize_cond): Ditto.
6411 (is_gcond_subset_of): Ditto.
6412 (is_subset_of_any): Ditto.
6413 (is_or_set_subset_of): Ditto.
6414 (is_and_set_subset_of): Ditto.
6415 (is_norm_cond_subset_of): Ditto.
6416 (pred_chain_length_cmp): Ditto.
6417 (convert_control_dep_chain_into_preds): Type change.
6418 (find_predicates): Ditto.
6419 (find_def_preds): Ditto.
6420 (destroy_predicates_vecs): Ditto.
6421 (find_matching_predicates_in_rest_chains): Ditto.
6422 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6423 (is_pred_expr_subset): Ditto.
6424 (is_pred_chain_subset_of): Ditto.
6425 (is_included_in): Ditto.
6426 (is_superset_of): Ditto.
6428 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6430 Update copyright years.
6432 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6434 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6435 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6436 config/arc/arc.md, config/arc/arc.opt,
6437 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6438 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6439 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6440 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6441 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6442 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6443 vtable-verify.c, vtable-verify.h: Use the standard form for the
6446 2014-01-02 Tobias Burnus <burnus@net-b.de>
6448 * gcc.c (process_command): Update copyright notice dates.
6449 * gcov-dump.c: Ditto.
6451 * doc/cpp.texi: Bump @copying's copyright year.
6452 * doc/cppinternals.texi: Ditto.
6453 * doc/gcc.texi: Ditto.
6454 * doc/gccint.texi: Ditto.
6455 * doc/gcov.texi: Ditto.
6456 * doc/install.texi: Ditto.
6457 * doc/invoke.texi: Ditto.
6459 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6461 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6463 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6465 * config/i386/sse.md (*mov<mode>_internal): Guard
6466 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6468 PR rtl-optimization/59647
6469 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6470 new_rtx into UNSIGNED_FLOAT rtxes.
6472 Copyright (C) 2014 Free Software Foundation, Inc.
6474 Copying and distribution of this file, with or without modification,
6475 are permitted in any medium without royalty provided the copyright
6476 notice and this notice are preserved.