1 2014-04-15 Richard Biener <rguenther@suse.de>
3 * alias.c (ncr_compar): New function.
4 (nonoverlapping_component_refs_p): Re-implement in O (n log n).
6 2014-04-15 Richard Biener <rguenther@suse.de>
8 * alias.c (record_component_aliases): Do not walk BINFOs.
10 2014-04-15 Richard Biener <rguenther@suse.de>
12 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13 Add struct function argument and adjust.
14 (find_func_aliases_for_call): Likewise.
15 (find_func_aliases): Likewise.
16 (find_func_clobbers): Likewise.
17 (intra_create_variable_infos): Likewise.
18 (compute_points_to_sets): Likewise.
19 (ipa_pta_execute): Adjust. Do not push/pop cfun.
21 2014-04-15 Richard Biener <rguenther@suse.de>
23 * tree.c (iterative_hash_expr): Use enum tree_code_class
24 to store TREE_CODE_CLASS.
25 (tree_block): Likewise.
26 (tree_set_block): Likewise.
27 * tree.h (fold_build_pointer_plus_loc): Use
28 convert_to_ptrofftype_loc.
30 2014-04-15 Jakub Jelinek <jakub@redhat.com>
33 * Makefile.in (PLUGIN_HEADERS): Add various headers that have been
36 2014-04-15 Eric Botcazou <ebotcazou@adacore.com>
38 * cfgloop.h (struct loop): Move force_vectorize down.
39 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Handle new kinds.
40 (gimplify_expr) <ANNOTATE_EXPR>: Minor tweak.
41 * lto-streamer-in.c (input_cfg): Read dont_vectorize field.
42 * lto-streamer-out.c (output_cfg): Write dont_vectorize field.
43 * tree-cfg.c (replace_loop_annotate): Revamp and handle new kinds.
44 * tree-core.h (enum annot_expr_kind): Add new kind values.
45 * tree-inline.c (copy_loops): Copy dont_vectorize field and reorder.
46 * tree-pretty-print.c (dump_generic_node) <ANNOTATE_EXPR>: Handle new
48 * tree.def (ANNOTATE_EXPR): Tweak comment.
50 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
52 * ipa-devirt.c (maybe_record_node): Ignore all non-methods (including
55 2014-04-14 Paolo Carlini <paolo.carlini@oracle.com>
57 * tree.h (TYPE_IDENTIFIER): Declare.
58 * tree.c (subrange_type_for_debug_p): Use it.
59 * godump.c (go_format_type): Likewise.
60 * dwarf2out.c (is_cxx_auto, modified_type_die,
61 gen_type_die_with_usage, gen_type_die_with_usage): Likewise.
62 * dbxout.c (dbxout_type, dbxout_symbol): Likewise.
64 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
67 * varpool.c (varpool_remove_node): Do not alter decls when streaming.
69 2014-04-14 Uros Bizjak <ubizjak@gmail.com>
71 * config/i386/i386.c (examine_argument): Return bool. Return true if
72 parameter should be passed in memory.
73 <case X86_64_COMPLEX_X87_CLASS>: Adjust.
74 (construct_container): Update calls to examine_argument.
75 (function_arg_advance_64): Ditto.
76 (return_in_memory_32): Merge with ix86_return_in_memory.
77 (return_in_memory_64): Ditto.
78 (return_in_memory_ms_64): Ditto.
80 2014-04-14 Jan Hubicka <hubicka@ucw.cz>
82 * ipa-utils.c (ipa_merge_profiles): Merge profile_id.
83 * coverage.c (coverage_compute_profile_id): Handle externally visible
86 2014-04-14 Martin Jambor <mjambor@suse.cz>
88 * tree-sra.c (ipa_sra_preliminary_function_checks): Skip
89 DECL_DISREGARD_INLINE_LIMITS functions.
91 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
94 * config/i386/i386.md (*fixuns_trunc<mode>_1): Revert the last change.
96 2014-04-14 H.J. Lu <hongjiu.lu@intel.com>
99 * config/i386/i386.md (*fixuns_trunc<mode>_1): Check
100 optimize_insn_for_speed_p instead of
101 optimize_function_for_speed_p.
103 2014-04-14 Yufeng Zhang <yufeng.zhang@arm.com>
105 * doc/invoke.texi (free): Document AArch64.
107 2014-04-14 Richard Biener <rguenther@suse.de>
109 PR tree-optimization/60042
110 * tree-ssa-pre.c (inhibit_phi_insertion): Remove.
111 (insert_into_preds_of_block): Do not prevent PHI insertion
112 for REFERENCE exprs here ...
113 (eliminate_dom_walker::before_dom_children): ... but prevent
114 their use here under similar conditions when applied to the
115 IL after PRE optimizations.
117 2014-04-14 Richard Biener <rguenther@suse.de>
119 * passes.def: Move early points-to after early SRA.
121 2014-04-14 Richard Biener <rguenther@suse.de>
123 * tree-ssa-forwprop.c (simplify_gimple_switch): Enhance
124 check for which sign-changes we allow when forwarding
125 a converted value into a switch.
127 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
129 * stor-layout.c (place_field): Finalize non-constant offset for the
132 2014-04-14 Richard Biener <rguenther@suse.de>
134 * tree-switch-conversion.c (lshift_cheap_p): Get speed_p
136 (expand_switch_using_bit_tests_p): Likewise.
137 (process_switch): Compute and pass on speed_p based on the
139 * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Use
140 optimize_bb_for_speed_p.
142 2014-04-14 Eric Botcazou <ebotcazou@adacore.com>
144 * cfgloop.h (struct loop): Rename force_vect into force_vectorize.
145 * function.h (struct function): Rename has_force_vect_loops into
146 has_force_vectorize_loops.
147 * lto-streamer-in.c (input_cfg): Adjust for renaming.
148 (input_struct_function_base): Likewise.
149 * lto-streamer-out.c (output_cfg): Likewise.
150 (output_struct_function_base): Likewise.
151 * omp-low.c (expand_omp_simd): Likewise.
152 * tree-cfg.c (move_sese_region_to_fn): Likewise.
153 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
154 (version_loop_for_if_conversion): Likewise.
155 (tree_if_conversion): Likewise.
156 (main_tree_if_conversion): Likewise.
157 (gate_tree_if_conversion): Likewise.
158 * tree-inline.c (copy_loops): Likewise.
159 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Likewise.
160 * tree-ssa-loop.c (tree_loop_vectorize): Likewise.
161 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Likewise.
162 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Likewise.
163 * tree-vectorizer.c (vectorize_loops): Likewise.
164 * tree-vectorizer.h (unlimited_cost_model): Likewise.
166 2014-04-14 Richard Biener <rguenther@suse.de>
169 * lto-streamer-out.c (wrap_refs): New function.
170 (lto_output): Wrap symbol references in global initializes in
171 type-preserving MEM_REFs.
173 2014-04-14 Christian Bruel <christian.bruel@st.com>
175 * config/sh/sh-mem.cc (sh_expand_strlen): Unroll last word.
177 2014-04-14 Christian Bruel <christian.bruel@st.com>
179 * config/sh/sh.md (setmemqi): New expand pattern.
180 * config/sh/sh.h (CLEAR_RATIO): Define.
181 * config/sh/sh-mem.cc (sh_expand_setmem): Define.
182 * config/sh/sh-protos.h (sh_expand_setmem): Declare.
184 2014-04-14 Richard Biener <rguenther@suse.de>
187 * fold-const.c (negate_expr_p): Don't negate directional rounding
189 (fold_negate_expr): Likewise.
191 2014-04-14 Richard Biener <rguenther@suse.de>
193 PR tree-optimization/59817
194 PR tree-optimization/60453
195 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
196 recursion to catch all CHRECs in the scalar evolution and restrict
197 the predicate for the remains appropriately.
199 2014-04-12 Catherine Moore <clm@codesourcery.com>
201 * config/mips/constraints.md: Add new register constraint "kb".
202 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
203 (*movhi_internal): Likewise.
204 (*movqi_internal): Likewise.
205 * config/mips/mips.h (M16_STORE_REGS): New register class.
206 (REG_CLASS_NAMES): Add M16_STORE_REGS.
207 (REG_CLASS_CONTENTS): Likewise.
208 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
210 2014-04-11 Tobias Burnus <burnus@net-b.de>
213 * doc/invoke.texi (-Wformat-signedness): Document it.
214 (Wformat=2): Mention that this enables -Wformat-signedness.
216 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
218 * common/config/epiphany/epiphany-common.c
219 (epiphany_option_optimization_table): Enable section anchors by
220 default at -O1 or higher.
221 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
222 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
223 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
224 carries no extra cost.
225 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
226 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
227 * config/epiphany/predicates.md (memclob_operand): New predicate.
228 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
229 Use memclob_operand predicate and X constraint for operand 3.
231 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
233 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
234 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
237 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
239 PR rtl-optimization/60651
240 * mode-switching.c (optimize_mode_switching): Make sure to emit
241 sets of a lower numbered entity before sets of a higher numbered
242 entity to a mode of the same or lower priority.
243 When creating a seginfo for a basic block that starts with a code
244 label, move the insertion point past the code label.
245 (new_seginfo): Document and enforce requirement that
246 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
247 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
248 * doc/tm.texi: Regenerate.
250 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
253 * config/arc/arc.c (arc_save_restore): Fix assert typo.
255 2013-04-11 Jakub Jelinek <jakub@redhat.com>
257 * BASE-VER: Set to 4.10.0.
259 2014-04-11 Tobias Burnus <burnus@net-b.de>
262 * doc/bugreport.texi (Bugs): Remove nodes pointing to the nirvana.
263 * doc/gcc.texi (Service): Update description in the @menu
264 * doc/invoke.texi (Option Summary): Remove misplaced and
267 2014-04-11 Steve Ellcey <sellcey@mips.com>
268 Jakub Jelinek <jakub@redhat.com>
271 * expr.c (convert_move): Use emit_store_flag_force instead of
272 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
275 2014-04-11 Richard Biener <rguenther@suse.de>
278 * varasm.c (assemble_alias): Avoid endless error reporting
279 recursion by setting TREE_ASM_WRITTEN.
281 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
283 * config/s390/s390.md: Add a splitter for NOT rtx.
285 2014-04-11 Jakub Jelinek <jakub@redhat.com>
287 PR rtl-optimization/60663
288 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in PARALLEL.
290 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
291 Jakub Jelinek <jakub@redhat.com>
294 * ipa.c (function_and_variable_visibility): Copy forced_by_abi
295 flag from decl_node to node.
297 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
300 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
301 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
302 ameliorating the cases where it can be.
304 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
307 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
309 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
310 (loadsync_<mode>): Change mode.
311 (load_quadpti, store_quadpti): New.
312 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
313 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
314 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
316 2014-04-09 Cong Hou <congh@google.com>
319 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
322 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
324 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
325 instead of vnor to exploit possible fusion opportunity in the
327 (altivec_expand_vec_perm_const_le): Likewise.
329 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
331 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
332 (loadsync_<mode>): Change mode.
333 (load_quadpti, store_quadpti): New.
334 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
335 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
337 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
340 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
341 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
342 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
344 2014-04-08 Richard Biener <rguenther@suse.de>
347 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
348 a 64bit widest int print double-int similar to on HWI64 hosts.
350 2014-04-08 Richard Biener <rguenther@suse.de>
352 PR tree-optimization/60785
353 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
354 default defs properly.
356 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
358 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
361 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
363 * ipa-devirt.c (maybe_record_node): When node is not recorded,
364 set completep to false rather than true.
366 2014-04-07 Douglas B Rupp <rupp@adacore.com>
369 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
370 ARM_TARGET2_DWARF_FORMAT.
372 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
375 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
376 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
379 2014-04-07 Richard Biener <rguenther@suse.de>
381 PR tree-optimization/60766
382 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
383 (may_eliminate_iv): Convert cand_value_at result to desired type.
385 2014-04-07 Jason Merrill <jason@redhat.com>
388 * common.opt (-fno-gnu-unique): Add.
389 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
391 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
393 * haifa-sched.c: Fix outdated function reference and minor
394 grammar errors in introductory comment.
396 2014-04-07 Richard Biener <rguenther@suse.de>
399 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
401 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
403 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
406 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
407 size accounting for thunks.
408 (pa_asm_output_mi_thunk): Use final_start_function() and
409 final_end_function() to output function start and end directives.
411 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
413 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have
414 device specific ISA/ feature information. Remove short_sp and
415 errata_skip ds. Add avr_device_specific_features enum to have device
417 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
418 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
419 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
420 updated device specific info.
421 * config/avr/avr-mcus.def: Merge device specific details to
423 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
425 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
426 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
427 assembler if RMW isa supported by current device.
428 * config/avr/genmultilib.awk: Update as device info structure changed.
429 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
431 2014-04-04 Cong Hou <congh@google.com>
433 PR tree-optimization/60656
434 * tree-vect-stmts.c (supportable_widening_operation):
435 Fix a bug that elements in a vector with vect_used_by_reduction
436 property are incorrectly reordered when the operation on it is not
437 consistant with the one in reduction operation.
439 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
441 PR rtl-optimization/60155
442 * gcse.c (record_set_data): New function.
443 (single_set_gcse): New function.
444 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
445 (hoist_code): Likewise.
446 (get_pressure_class_and_nregs): Likewise.
448 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
450 * explow.c (probe_stack_range): Emit a final optimization blockage.
452 2014-04-04 Anthony Green <green@moxielogic.com>
454 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
457 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
460 * lto-cgraph.c (input_overwrite_node): Check that partitioning
461 flags are set only during streaming.
462 * ipa.c (process_references, walk_polymorphic_call_targets,
463 symtab_remove_unreachable_nodes): Drop bodies of always inline
464 after early inlining.
465 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
467 2014-04-04 Jakub Jelinek <jakub@redhat.com>
468 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
471 * dwarf2out.c (const_ok_for_output_1): Reject expressions
474 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
477 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
479 (cortex_a53_fdivd): Likewise.
481 2014-04-04 Martin Jambor <mjambor@suse.cz>
484 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
486 * cgraph.c (clone_of_p): Also return true if thunks match.
487 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
488 cgraph_function_or_thunk_node and an obsolete comment.
489 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
491 (build_function_decl_skip_args): Likewise.
492 (set_new_clone_decl_and_node_flags): New function.
493 (duplicate_thunk_for_node): Likewise.
494 (redirect_edge_duplicating_thunks): Likewise.
495 (cgraph_clone_node): New parameter args_to_skip, pass it to
496 redirect_edge_duplicating_thunks which is called instead of
497 cgraph_redirect_edge_callee.
498 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
499 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
501 2014-04-04 Jeff Law <law@redhat.com>
504 * config/arm/predicates.md (const_int_I_operand): New predicate.
505 (const_int_M_operand): Similarly.
506 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
508 (insv_t2, extv_reg, extzv_t2): Likewise.
509 (load_multiple_with_writeback): Similarly for const_int_I_operand.
510 (pop_multiple_with_writeback_and_return): Likewise.
511 (vfp_pop_multiple_with_writeback): Likewise
513 2014-04-04 Richard Biener <rguenther@suse.de>
516 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
517 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
519 * gimplify.h (gimple_add_tmp_var_fn): Declare.
520 * gimplify.c (gimple_add_tmp_var_fn): New function.
521 * gimple-expr.h (create_tmp_reg_fn): Declare.
522 * gimple-expr.c (create_tmp_reg_fn): New function.
523 * gimple-low.c (record_vars_into): Don't change cfun.
524 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
525 code generation without cfun.
527 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
530 * Makefile.in (install-driver): Fix shell scripting.
532 2014-04-03 Cong Hou <congh@google.com>
534 PR tree-optimization/60505
535 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
536 threshold of number of iterations below which no vectorization
538 * tree-vect-loop.c (new_loop_vec_info):
539 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
540 * tree-vect-loop.c (vect_analyze_loop_operations):
541 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
542 * tree-vect-loop.c (vect_transform_loop):
543 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
544 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
545 of iterations of the loop and see if we should build the epilogue.
547 2014-04-03 Richard Biener <rguenther@suse.de>
549 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx member.
550 (streamer_tree_cache_create): Adjust.
551 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
552 to allow optional nodes array.
553 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
554 (streamer_tree_cache_append): Likewise.
555 (streamer_tree_cache_create): Create nodes array optionally
556 as specified by parameter.
557 * lto-streamer-out.c (create_output_block): Avoid maintaining
558 the node array in the writer cache.
559 (DFS_write_tree): Remove assertion.
560 (produce_asm_for_decls): Free the out decl state hash table early.
561 * lto-streamer-in.c (lto_data_in_create): Adjust for
562 streamer_tree_cache_create prototype change.
564 2014-04-03 Richard Biener <rguenther@suse.de>
566 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
567 set TREE_CHAIN to NULL_TREE.
569 2014-04-03 Richard Biener <rguenther@suse.de>
571 PR tree-optimization/60740
572 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
573 over all GIMPLE_COND operands.
575 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
577 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
578 (Weffc++): Remove Scott's numbering, merge lists and reference
584 * class.c (accessible_nvdtor_p): New.
585 (check_bases): Don't check base destructor here ...
586 (check_bases_and_members): ... check them here. Trigger on
587 Wnon-virtual-dtor flag.
588 (finish_struct_1): Use accessible_nvdtor_p.
591 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
592 * g++.dg/warn/Wnvdtor-2.C: New.
593 * g++.dg/warn/Wnvdtor-3.C: New.
594 * g++.dg/warn/Wnvdtor-4.C: New.
595 * g++.dg/warn/Weff1.C: Delete.
596 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
597 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
599 2014-04-03 Nick Clifton <nickc@redhat.com>
601 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
604 2014-04-03 Martin Jambor <mjambor@suse.cz>
606 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
607 mention gcc_unreachable before failing.
608 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
611 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
614 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type
615 inconsistent code and instead mark the context inconsistent.
616 (possible_polymorphic_call_targets): For inconsistent contexts
617 return empty complete list.
619 2014-04-02 Anthony Green <green@moxielogic.com>
621 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
622 (extendqisi2, extendhisi2): Define.
623 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
624 (WCHAR_TYPE): Change to unsigned int.
626 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
628 PR tree-optimization/60733
629 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
630 insertion point for PHI candidates to be the end of the feeding
631 block for the PHI argument.
633 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
635 PR rtl-optimization/60650
636 * lra-constraints.c (process_alt_operands): Decrease reject for
637 earlyclobber matching.
639 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
641 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
643 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
645 * config/spu/spu.c (pad_bb): Do not crash when the last
646 insn is CODE_FOR_blockage.
648 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
650 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
651 lies outside the target mode.
653 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
656 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
657 software floating point or no floating point registers, do not
658 allow any type in the FPRs. Eliminate a test for SPE SIMD types
659 in GPRs that occurs after we tested for GPRs that would never be
662 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
663 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
664 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
665 specifically allow DDmode, since that does not use the SPE SIMD
668 2014-04-02 Richard Biener <rguenther@suse.de>
671 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
672 MODE_INTs. Properly use negv_optab.
673 (expand_abs): Likewise.
675 2014-04-02 Richard Biener <rguenther@suse.de>
678 * Makefile.in (install-driver): Guard extra installs with special
681 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
683 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
686 2014-04-01 Richard Henderson <rth@redhat.com>
689 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
690 alternative enabled before register allocation.
692 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
694 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
695 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
697 (nios2_large_got_address): Remove unneeded 'sym' parameter.
698 (nios2_got_address): Update nios2_large_got_address call site.
699 (nios2_delegitimize_address): New function.
700 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
701 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
702 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
704 2014-04-01 Martin Husemann <martin@duskware.de>
706 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
709 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
711 PR rtl-optimization/60604
712 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
713 check from register_operand.
714 (register_operand): Redefine in terms of general_operand.
715 (nonmemory_operand): Use register_operand for the non-constant cases.
717 2014-04-01 Richard Biener <rguenther@suse.de>
719 * gimple.h (struct gimple_statement_base): Align subcode to 16 bits.
721 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
723 * doc/invoke.texi (mapp-regs): Clarify.
725 2014-03-31 Ulrich Drepper <drepper@gmail.com>
727 * config/i386/avx512fintrin.h (__v32hi): Define type.
729 (_mm512_set1_epi8): Define.
730 (_mm512_set1_epi16): Define.
731 (_mm512_set4_epi32): Define.
732 (_mm512_set4_epi64): Define.
733 (_mm512_set4_pd): Define.
734 (_mm512_set4_ps): Define.
735 (_mm512_setr4_epi64): Define.
736 (_mm512_setr4_epi32): Define.
737 (_mm512_setr4_pd): Define.
738 (_mm512_setr4_ps): Define.
739 (_mm512_setzero_epi32): Define.
741 2014-03-31 Martin Jambor <mjambor@suse.cz>
744 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
745 callsite_arguments_match_p. Updated all callers. Also check types of
746 corresponding formal parameters and actual arguments.
747 (not_all_callers_have_enough_arguments_p) Renamed to
748 some_callers_have_mismatched_arguments_p.
750 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
752 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
754 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
757 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
760 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
762 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
763 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
765 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
766 Use FMAMODE_NOVF512 mode iterator.
767 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
768 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
769 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
770 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
772 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
773 Use VF_128_256 mode iterator.
774 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
777 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
779 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
780 static chain if needed.
782 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
785 * lra-constraints.c (index_part_to_reg): New.
786 (process_address): Use it.
788 2014-03-27 Jeff Law <law@redhat.com>
789 Jakub Jelinek <jakub@redhat.com>
792 * expr.c (do_tablejump): Use simplify_gen_binary rather than
793 gen_rtx_{PLUS,MULT} to build up the address expression.
795 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
796 creating non-canonical RTL.
798 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
801 * ipa-inline.c (want_inline_small_function_p): Short circuit large
802 functions; reorganize to make cheap checks first.
803 (inline_small_functions): Do not estimate growth when dumping;
805 * ipa-inline.h (inline_summary): Add min_size.
806 (growth_likely_positive): New function.
807 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
808 (set_cond_stmt_execution_predicate): Cleanup.
809 (estimate_edge_size_and_time): Compute min_size.
810 (estimate_calls_size_and_time): Likewise.
811 (estimate_node_size_and_time): Likewise.
812 (inline_update_overall_summary): Update min_size.
813 (do_estimate_edge_time): Likewise.
814 (do_estimate_edge_size): Update.
815 (do_estimate_edge_hints): Update.
816 (growth_likely_positive): New function.
818 2014-03-28 Jakub Jelinek <jakub@redhat.com>
821 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
822 also if addr has VOIDmode.
824 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
826 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
827 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
829 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
830 instructions as well as AdvancedSIMD loads.
832 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
834 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
835 Use crypto_aese type.
836 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
837 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
838 crypto_aese, crypto_aesmc. Move to types.md.
839 * config/arm/types.md (crypto_aes): Split into crypto_aese,
841 * config/arm/iterators.md (crypto_type): Likewise.
843 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
845 * cgraph.c: Include expr.h and tree-dfa.h.
846 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
849 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
852 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
853 regs from checking multi-reg pseudos.
855 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
857 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
859 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
861 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
862 if it would clobber the stack pointer, even temporarily.
864 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
866 * mode-switching.c: Make small adjustments to the top comment.
868 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
870 * config/rs6000/constraints.md (wD constraint): New constraint to
871 match the constant integer to get the top DImode/DFmode out of a
872 vector in a VSX register.
874 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
875 match the constant integer to get the top DImode/DFmode out of a
876 vector in a VSX register.
878 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
881 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
884 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
885 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
887 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
888 Optimize vec_extract of 64-bit values, where the value being
889 extracted is in the top word, where we can use scalar
890 instructions. Add direct move and store support. Combine the big
891 endian/little endian vector select load support into a single insn.
892 (vsx_extract_<mode>_internal1): Likewise.
893 (vsx_extract_<mode>_internal2): Likewise.
894 (vsx_extract_<mode>_load): Likewise.
895 (vsx_extract_<mode>_store): Likewise.
896 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
897 combined into vsx_extract_<mode>_load.
898 (vsx_extract_<mode>_one_le): Likewise.
900 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
901 define the top 64-bit vector element.
903 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
906 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
907 Document vec_vbpermq builtin.
910 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
911 enable use of xxsldwi and xxpermdi builtin functions.
912 (vec_xxpermdi): Likewise.
914 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
915 Document use of vec_xxsldwi and vec_xxpermdi builtins.
917 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
919 PR rtl-optimization/60650
920 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
922 (find_spills_for): New.
923 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
924 Spill all pseudos on the second iteration.
926 2014-03-27 Marek Polacek <polacek@redhat.com>
929 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
932 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
934 * config/s390/s390.c (s390_can_use_return_insn): Check for
935 call-saved FPRs on 31 bit.
937 2014-03-27 Jakub Jelinek <jakub@redhat.com>
940 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
941 if they need regimplification, just drop them instead of
942 calling gimple_regimplify_operands on them.
944 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
947 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
948 (aarch64_frame_pointer_required): Adjust logic.
949 (aarch64_can_eliminate): Adjust logic.
950 (aarch64_override_options_after_change): Adjust logic.
952 2014-03-27 Dehao Chen <dehao@google.com>
954 * ipa-inline.c (early_inliner): Update node's inline info.
956 2014-03-26 Dehao Chen <dehao@google.com>
958 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
959 compiler inserted conditional jumps for NAN float check.
961 2014-03-26 Jakub Jelinek <jakub@redhat.com>
963 * ubsan.h (ubsan_create_data): Change second argument's type
964 to const location_t *.
965 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
967 (ubsan_create_data): Change second argument to const location_t *PLOC.
968 Create Loc field whenever PLOC is non-NULL.
969 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
970 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
974 * real.c (real_to_integer2): Change type of low to UHWI.
976 2014-03-26 Tobias Burnus <burnus@net-b.de>
978 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
979 (CILK_SELF_SPECS): New define.
980 (driver_self_specs): Use it.
982 2014-03-26 Richard Biener <rguenther@suse.de>
984 * tree-pretty-print.c (percent_K_format): Implement special
985 case for LTO and its stripped down BLOCK tree.
987 2014-03-26 Jakub Jelinek <jakub@redhat.com>
990 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
992 * tree-vrp.c (simplify_internal_call_using_ranges): If only
993 one range is range_int_cst_p, but not both, at least optimize
994 addition/subtraction of 0 and multiplication by 0 or 1.
995 * gimple-fold.c (gimple_fold_call): Fold
996 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
997 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
998 INTEGER_CSTs, try to fold at least x * 0 and y - y.
1000 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
1002 PR rtl-optimization/60452
1003 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
1004 <case REG>: Return 1 for invalid offsets from the frame pointer.
1006 2014-03-26 Marek Polacek <polacek@redhat.com>
1009 * doc/extend.texi (C Extensions): Mention variable-length arrays in
1012 2014-03-26 Marek Polacek <polacek@redhat.com>
1015 * doc/extend.texi (Designated Inits): Describe what happens to omitted
1018 2014-03-26 Marek Polacek <polacek@redhat.com>
1021 * ira-color.c (update_conflict_hard_regno_costs): Perform the
1022 multiplication in unsigned type.
1024 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1026 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
1028 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
1030 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
1032 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1035 * cif-code.def (UNREACHABLE) New code.
1036 * ipa-inline.c (inline_small_functions): Skip edges to
1037 __builtlin_unreachable.
1038 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
1039 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
1040 predicate to __bulitin_unreachable.
1041 (set_cond_stmt_execution_predicate): Fix issue when
1042 invert_tree_comparison returns ERROR_MARK.
1043 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
1044 propagate to inline clones.
1045 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
1047 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
1048 * cgraphclones.c (cgraph_clone_node): If call destination is already
1049 ureachable, do not redirect it back.
1050 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
1053 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
1055 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
1056 Do not modify inline clones.
1058 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1060 * config/i386/i386.md (general_sext_operand): New mode attr.
1061 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
1062 don't generate (sign_extend (const_int)).
1063 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
1064 operands[2]. Use We constraint instead of <i> and
1065 <general_sext_operand> predicate instead of <general_operand>.
1066 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
1067 * config/i386/constraints.md (We): New constraint.
1068 * config/i386/predicates.md (x86_64_sext_operand,
1069 sext_operand): New predicates.
1071 2014-03-25 Martin Jambor <mjambor@suse.cz>
1074 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
1075 inconsistent devirtualizations to __builtin_unreachable.
1077 2014-03-25 Marek Polacek <polacek@redhat.com>
1080 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
1082 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
1084 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
1085 order of elements for big-endian.
1087 2014-03-25 Richard Biener <rguenther@suse.de>
1090 * gimplify-me.c (gimple_regimplify_operands): Update the
1093 2014-03-25 Martin Jambor <mjambor@suse.cz>
1096 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
1097 (lto_output_varpool_node): Likewise.
1098 (input_overwrite_node): Likewise.
1099 (input_varpool_node): Likewise.
1101 2014-03-25 Richard Biener <rguenther@suse.de>
1103 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
1104 (run_gcc): Likewise.
1106 2014-03-25 Jakub Jelinek <jakub@redhat.com>
1108 * combine.c (simplify_compare_const): Add MODE argument.
1109 Handle mode_width 0 as very large mode_width.
1110 (try_combine, simplify_comparison): Adjust callers.
1112 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
1113 type to avoid signed integer overflow.
1114 * explow.c (plus_constant): Likewise.
1116 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1118 * doc/generic.texi: Correct typos.
1120 2014-03-24 Tobias Burnus <burnus@net-b.de>
1122 * doc/invoke.texi (-flto): Expand section about
1123 using static libraries with LTO.
1125 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
1127 PR rtl-optimization/60501
1128 * optabs.def (addptr3_optab): New optab.
1129 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
1130 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
1131 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
1133 * lra.c (emit_add3_insn): Use the addptr pattern if available.
1135 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
1137 2014-03-24 Ulrich Drepper <drepper@gmail.com>
1139 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
1142 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
1143 (_mm256_undefined_ps): Define.
1144 (_mm256_undefined_pd): Define.
1145 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
1146 (_mm_undefined_pd): Define.
1147 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
1148 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
1149 (_mm512_undefined_ps): Define.
1150 (_mm512_undefined_pd): Define.
1151 Use _mm*_undefined_*.
1152 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
1154 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
1156 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
1157 (lshr_simd): DI mode added.
1158 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
1159 (aarch64_ushr_simddi): Likewise.
1160 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
1161 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
1162 (vshrd_n_u64): Likewise.
1164 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1166 * Makefile.in (s-macro_list): Depend on cc1.
1168 2014-03-23 Teresa Johnson <tejohnson@google.com>
1170 * ipa-utils.c (ipa_print_order): Use specified dump file.
1172 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
1174 PR rtl-optimization/60601
1175 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
1177 * gcc.c (eval_spec_function): Initialize save_growing_value.
1179 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1182 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1183 code == MINUS_EXPR, never swap op0 with op1.
1185 * toplev.c (init_local_tick): Avoid signed integer multiplication
1187 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1188 shift by first operand's bitsize.
1190 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1193 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1195 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1196 TARGET_ISA_64BIT_P(x).
1198 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1200 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1201 pattern for vector nor instead of subtract from splat(-1).
1202 (altivec_expand_vec_perm_const_le): Likewise.
1204 2014-03-21 Richard Henderson <rth@twiddle.net>
1207 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1208 related insns after epilogue_completed.
1210 2014-03-21 Martin Jambor <mjambor@suse.cz>
1213 * cgraph.h (symtab_node): New flag body_removed.
1214 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1215 when removing bodies.
1216 * symtab.c (dump_symtab_base): Dump body_removed flag.
1217 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1218 had their bodies removed.
1220 2014-03-21 Martin Jambor <mjambor@suse.cz>
1223 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1226 2014-03-21 Richard Biener <rguenther@suse.de>
1228 PR tree-optimization/60577
1229 * tree-core.h (struct tree_base): Document nothrow_flag use
1230 in VAR_DECL_NONALIASED.
1231 * tree.h (VAR_DECL_NONALIASED): New.
1232 (may_be_aliased): Adjust.
1233 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1235 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1237 * expr.c (expand_expr_real_1): Remove outdated comment.
1239 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1242 * ira.c (adjust_cleared_regs): Call copy_rtx on
1243 *reg_equiv[REGNO (loc)].src_p before passing it to
1244 simplify_replace_fn_rtx.
1247 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1248 into CONST, put pic register as first operand of PLUS. Use
1249 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1251 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1253 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1255 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1257 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1258 around for store forwarding issue in the FPU on the UT699.
1259 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1260 loads and operations if -mfix-ut699 is specified.
1261 (divtf3_hq): Tweak attribute.
1262 (sqrttf2_hq): Likewise.
1264 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1266 * calls.c (store_one_arg): Remove incorrect const qualification on the
1267 type of the temporary.
1268 * cfgexpand.c (expand_return): Likewise.
1269 * expr.c (expand_constructor): Likewise.
1270 (expand_expr_real_1): Likewise.
1272 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1274 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1277 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1280 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1282 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1284 * config/arm/aarch-common-protos.h
1285 (alu_cost_table): Fix spelling of "extend".
1286 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1288 2014-03-19 Richard Biener <rguenther@suse.de>
1291 * tree-core.h (tree_type_common): Re-order pointer members
1292 to reduce recursion depth during GC walks.
1294 2014-03-19 Marek Polacek <polacek@redhat.com>
1297 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1298 before accessing it.
1300 2014-03-19 Richard Biener <rguenther@suse.de>
1303 * lto-streamer-in.c (input_function): In WPA stage do not drop
1306 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1308 PR tree-optimization/60559
1309 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1310 with build_zero_cst assignment.
1312 2014-03-18 Kai Tietz <ktietz@redhat.com>
1314 PR rtl-optimization/56356
1315 * sdbout.c (sdbout_parms): Verify that parms'
1316 incoming argument is valid.
1317 (sdbout_reg_parms): Likewise.
1319 2014-03-18 Richard Henderson <rth@redhat.com>
1322 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1323 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1324 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1326 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1328 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1329 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1330 Italicize plugin event names in description. Explain that
1331 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1332 Remind that no GCC functions should be called after PLUGIN_FINISH.
1333 Explain what pragmas with expansion are.
1335 2014-03-18 Martin Liska <mliska@suse.cz>
1337 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1338 gimple call statement is update.
1339 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1340 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1342 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1345 * ubsan.c (ubsan_instrument_unreachable): Call
1346 initialize_sanitizer_builtins.
1347 (ubsan_pass): Likewise.
1350 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1351 varpool_finalize_decl instead of rest_of_decl_compilation.
1353 2014-03-18 Richard Biener <rguenther@suse.de>
1355 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1356 by using bitmap_and_compl instead of bitmap_and_compl_into.
1357 (df_rd_transfer_function): Likewise.
1359 2014-03-18 Richard Biener <rguenther@suse.de>
1361 * doc/lto.texi (fresolution): Fix typo.
1363 2014-03-18 Richard Biener <rguenther@suse.de>
1365 * doc/invoke.texi (flto): Update for changes in 4.9.
1367 2014-03-18 Richard Biener <rguenther@suse.de>
1369 * doc/loop.texi: Remove section on the removed lambda framework.
1370 Update loop docs with recent changes in preserving loop structure.
1372 2014-03-18 Richard Biener <rguenther@suse.de>
1374 * doc/lto.texi (-fresolution): Document.
1376 2014-03-18 Richard Biener <rguenther@suse.de>
1378 * doc/contrib.texi: Adjust my name.
1380 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1383 * internal-fn.c: Include diagnostic-core.h.
1384 (expand_BUILTIN_EXPECT): New function.
1385 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1386 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1387 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1388 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1390 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1391 Revert 3 argument __builtin_expect code.
1392 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1393 * gimple-fold.c (gimple_fold_call): Likewise.
1394 * tree.h (fold_builtin_expect): New prototype.
1395 * builtins.c (build_builtin_expect_predicate): Add predictor
1396 argument, if non-NULL, create 3 argument __builtin_expect.
1397 (fold_builtin_expect): No longer static. Add ARG2 argument,
1398 pass it through to build_builtin_expect_predicate.
1399 (fold_builtin_2): Adjust caller.
1400 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1401 * internal-fn.def (BUILTIN_EXPECT): New.
1403 2014-03-18 Tobias Burnus <burnus@net-b.de>
1406 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1407 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1408 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1410 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1413 * predict.c (combine_predictions_for_bb): Fix up formatting.
1414 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1415 fill what it points to if non-NULL.
1416 (tree_predict_by_opcode): Adjust caller, use the predictor.
1417 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1419 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1421 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1422 proper constant for the store mode.
1424 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1426 * symtab.c (change_decl_assembler_name): Fix transparent alias
1429 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1431 * config/aarch64/aarch64.c: Correct the comments about the
1432 aarch64 stack layout.
1434 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1436 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1437 check for GF_OMP_FOR_KIND_FOR.
1439 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1441 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1442 ymm and zmm register names.
1444 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1447 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1448 note creation for the 2010-08-31 changes.
1450 2014-03-17 Marek Polacek <polacek@redhat.com>
1453 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1454 as -fno-tree-loop-vectorize.
1455 (expand_omp_simd): Likewise.
1457 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1459 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1460 (eligible_for_call_delay): New prototype.
1461 * config/sparc/sparc.c (tls_call_delay): Rename into...
1462 (eligible_for_call_delay): ...this. Return false if the instruction
1463 cannot be put in the delay slot of a branch.
1464 (eligible_for_restore_insn): Simplify.
1465 (eligible_for_return_delay): Return false if the instruction cannot be
1466 put in the delay slot of a branch and simplify.
1467 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1468 put in the delay slot of a branch.
1469 * config/sparc/sparc.md (fix_ut699): New attribute.
1470 (tls_call_delay): Delete.
1471 (in_call_delay): Reimplement.
1472 (eligible_for_sibcall_delay): Rename into...
1473 (in_sibcall_delay): ...this.
1474 (eligible_for_return_delay): Rename into...
1475 (in_return_delay): ...this.
1476 (in_branch_delay): Reimplement.
1477 (in_uncond_branch_delay): Delete.
1478 (in_annul_branch_delay): Delete.
1480 2014-03-14 Richard Henderson <rth@redhat.com>
1483 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1484 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1485 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1486 (floathi splitters): Remove.
1487 (float<SWI48x>xf2): New pattern.
1488 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1489 code that tried to handle DImode for 32-bit, but which was excluded
1490 by the pattern's condition. Drop allocation of stack temporary.
1491 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1492 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1493 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1494 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1495 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1496 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1497 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1498 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1499 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1500 (*float<SWI48x><X87MODEF>2_i387): Remove.
1501 (all float _with_temp splitters): Remove.
1502 (*float<SWI48x><MODEF>2_i387): New pattern.
1503 (*float<SWI48><MODEF>2_sse): New pattern.
1504 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1505 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1507 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1508 Marek Polacek <polacek@redhat.com>
1511 * common.opt (dump_base_name_prefixed): New Variable.
1512 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1513 if x_dump_base_name_prefixed is already set, set it at the end.
1515 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1517 PR rtl-optimization/60508
1518 * lra-constraints.c (get_reload_reg): Add new parameter
1520 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1521 Pass the new parameter values.
1523 2014-03-14 Richard Biener <rguenther@suse.de>
1525 * common.opt: Revert unintented changes from r205065.
1528 2014-03-14 Richard Biener <rguenther@suse.de>
1531 * cfghooks.c (split_block): Properly adjust all loops the
1532 block was a latch of.
1534 2014-03-14 Martin Jambor <mjambor@suse.cz>
1537 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1540 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1543 * config/avr/avr.c (avr_set_current_function): Pass function name
1544 through default_strip_name_encoding before sanity checking instead
1545 of skipping the first char of the assembler name.
1547 2014-03-13 Richard Henderson <rth@redhat.com>
1550 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1551 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1552 * config/i386/i386-protos.h: Likewise.
1553 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1554 in the expander instead of a splitter.
1555 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1556 any possibility of requiring a memory.
1557 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1558 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1559 (fp branch splitters): Update for ix86_split_fp_branch.
1560 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1561 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1562 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1563 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1564 (*fop_<MODEF>_3_i387): Likewise.
1565 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1566 (splitters for the fop_* register patterns): Remove.
1567 (fscalexf4_i387): Rename from *fscalexf4_i387.
1568 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1570 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1572 PR tree-optimization/59779
1573 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1574 type for bitsize and maxsize instead of HOST_WIDE_INT.
1576 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1578 PR rtl-optimization/57320
1579 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1580 the CFG after thread_prologue_and_epilogue_insns.
1582 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1584 PR rtl-optimization/57189
1585 * lra-constraints.c (process_alt_operands): Disfavor spilling
1588 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1590 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1592 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1594 PR tree-optimization/59025
1596 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1597 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1599 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1602 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1603 calls of avr_out_plus_1.
1605 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1607 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1608 BB's single pred and update the father loop's latch info later.
1610 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1612 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1616 (VEC_base): Likewise.
1617 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1618 registers, we need to swap double words in little endian mode.
1620 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1621 to be a container mode for 128-bit integer operations added in ISA
1622 2.07. Unlike TImode and PTImode, the preferred register set is
1623 the Altivec/VMX registers for the 128-bit operations.
1625 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1627 (rs6000_split_128bit_ok_p): Likewise.
1629 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1630 macros for creating ISA 2.07 normal and overloaded builtin
1631 functions with 3 arguments.
1632 (BU_P8V_OVERLOAD_3): Likewise.
1633 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1634 for use as overloaded functions.
1635 (VPERM_1TI_UNS): Likewise.
1636 (VSEL_1TI): Likewise.
1637 (VSEL_1TI_UNS): Likewise.
1638 (ST_INTERNAL_1ti): Likewise.
1639 (LD_INTERNAL_1ti): Likewise.
1640 (XXSEL_1TI): Likewise.
1641 (XXSEL_1TI_UNS): Likewise.
1642 (VPERM_1TI): Likewise.
1643 (VPERM_1TI_UNS): Likewise.
1644 (XXPERMDI_1TI): Likewise.
1645 (SET_1TI): Likewise.
1646 (LXVD2X_V1TI): Likewise.
1647 (STXVD2X_V1TI): Likewise.
1648 (VEC_INIT_V1TI): Likewise.
1649 (VEC_SET_V1TI): Likewise.
1650 (VEC_EXT_V1TI): Likewise.
1651 (EQV_V1TI): Likewise.
1652 (NAND_V1TI): Likewise.
1653 (ORC_V1TI): Likewise.
1654 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1655 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1657 (VADDUQM): Likewise.
1658 (VSUBCUQ): Likewise.
1659 (VADDEUQM): Likewise.
1660 (VADDECUQ): Likewise.
1661 (VSUBEUQM): Likewise.
1662 (VSUBECUQ): Likewise.
1664 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1665 __int128_t and __uint128_t types.
1666 (__uint128_type): Likewise.
1667 (altivec_categorize_keyword): Add support for vector __int128_t,
1668 vector __uint128_t, vector __int128, and vector unsigned __int128
1669 as a container type for TImode operations that need to be done in
1670 VSX/Altivec registers.
1671 (rs6000_macro_to_expand): Likewise.
1672 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1673 to support 128-bit integer instructions vaddcuq, vadduqm,
1674 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1675 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1677 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1678 for V1TImode, and set up preferences to use VSX/Altivec registers.
1679 Setup VSX reload handlers.
1680 (rs6000_debug_reg_global): Likewise.
1681 (rs6000_init_hard_regno_mode_ok): Likewise.
1682 (rs6000_preferred_simd_mode): Likewise.
1683 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1684 (easy_altivec_constant): Likewise.
1685 (output_vec_const_move): Likewise.
1686 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1688 (rs6000_expand_vector_extract): Likewise.
1689 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1691 (rs6000_const_vec): Add support for V1TImode.
1692 (rs6000_emit_le_vsx_load): Swap double words when loading or
1693 storing TImode/V1TImode.
1694 (rs6000_emit_le_vsx_store): Likewise.
1695 (rs6000_emit_le_vsx_move): Likewise.
1696 (rs6000_emit_move): Add support for V1TImode.
1697 (altivec_expand_ld_builtin): Likewise.
1698 (altivec_expand_st_builtin): Likewise.
1699 (altivec_expand_vec_init_builtin): Likewise.
1700 (altivec_expand_builtin): Likewise.
1701 (rs6000_init_builtins): Add support for V1TImode type. Add
1702 support for ISA 2.07 128-bit integer builtins. Define type names
1703 for the VSX/Altivec vector types.
1704 (altivec_init_builtins): Add support for overloaded vector
1705 functions with V1TImode type.
1706 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1707 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1709 (rs6000_split_128bit_ok_p): Likewise.
1710 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1711 __int128_t and vector __uint128_t.
1713 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1714 and mode attributes.
1721 (VS_scalar): Likewise.
1722 (VS_double): Likewise.
1723 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1725 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1726 we support the ISA 2.07 128-bit integer arithmetic instructions.
1727 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1728 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1729 and TImode types for use with the builtin functions.
1730 (V1TI_type_node): Likewise.
1731 (unsigned_V1TI_type_node): Likewise.
1732 (intTI_type_internal_node): Likewise.
1733 (uintTI_type_internal_node): Likewise.
1735 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1736 128-bit builtin functions.
1737 (UNSPEC_VADDEUQM): Likewise.
1738 (UNSPEC_VADDECUQ): Likewise.
1739 (UNSPEC_VSUBCUQ): Likewise.
1740 (UNSPEC_VSUBEUQM): Likewise.
1741 (UNSPEC_VSUBECUQ): Likewise.
1742 (VM): Add V1TImode to vector mode iterators.
1744 (VI_unit): Likewise.
1745 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1746 (altivec_vaddcuq): Likewise.
1747 (altivec_vsubuqm): Likewise.
1748 (altivec_vsubcuq): Likewise.
1749 (altivec_vaddeuqm): Likewise.
1750 (altivec_vaddecuq): Likewise.
1751 (altivec_vsubeuqm): Likewise.
1752 (altivec_vsubecuq): Likewise.
1754 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1756 (BOOL_128): Likewise.
1757 (BOOL_REGS_OUTPUT): Likewise.
1758 (BOOL_REGS_OP1): Likewise.
1759 (BOOL_REGS_OP2): Likewise.
1760 (BOOL_REGS_UNARY): Likewise.
1761 (BOOL_REGS_AND_CR0): Likewise.
1763 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1764 128-bit integer builtin support.
1765 (vec_vadduqm): Likewise.
1766 (vec_vaddecuq): Likewise.
1767 (vec_vaddeuqm): Likewise.
1768 (vec_vsubecuq): Likewise.
1769 (vec_vsubeuqm): Likewise.
1770 (vec_vsubcuq): Likewise.
1771 (vec_vsubuqm): Likewise.
1773 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1774 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1775 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1776 128-bit integer add/subtract to ISA 2.07.
1778 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1780 * config/arc/arc.c (arc_predicate_delay_insns):
1781 Fix third argument passed to conditionalize_nonjump.
1783 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1785 * config/aarch64/aarch64-builtins.c
1786 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1787 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1788 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1789 instead of __builtin_lfloor.
1790 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1792 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1794 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1795 (tree_ssa_ifcombine_bb_1): New function.
1796 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1797 is an empty forwarder block to then_bb or vice versa and then_bb
1798 and else_bb are effectively swapped.
1800 2014-03-12 Christian Bruel <christian.bruel@st.com>
1803 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1804 REG_CFA_DEF_CFA note.
1805 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1806 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1808 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1810 PR tree-optimization/60454
1811 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1813 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1815 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1816 Do not define target_cpu_default2 to generic.
1817 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1818 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1819 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1821 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1822 Marc Glisse <marc.glisse@inria.fr>
1824 PR tree-optimization/60502
1825 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1826 instead of build_low_bits_mask.
1828 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1831 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1832 if there are multiple uses, but op doesn't live on E edge.
1833 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1834 clobber stmts before __builtin_unreachable.
1836 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1838 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1839 hard_frame_pointer_rtx.
1840 * cse.c (cse_insn): Remove volatile check.
1841 * cselib.c (cselib_process_insn): Likewise.
1842 * dse.c (scan_insn): Likewise.
1844 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1846 * config/arc/arc.c (conditionalize_nonjump): New function,
1848 (arc_ifcvt): ... this.
1849 (arc_predicate_delay_insns): Use it.
1851 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1853 * config/arc/predicates.md (extend_operand): During/after reload,
1854 allow const_int_operand.
1855 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1856 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1857 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1859 (umulsi3_highpart_i): Likewise.
1861 2014-03-11 Richard Biener <rguenther@suse.de>
1863 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1864 Add asserts to guard possible wrong-code bugs.
1866 2014-03-11 Richard Biener <rguenther@suse.de>
1868 PR tree-optimization/60429
1869 PR tree-optimization/60485
1870 * tree-ssa-structalias.c (set_union_with_increment): Properly
1871 take into account all fields that overlap the shifted vars.
1872 (do_sd_constraint): Likewise.
1873 (do_ds_constraint): Likewise.
1874 (get_constraint_for_ptr_offset): Likewise.
1876 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1878 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1879 (nios2_compute_frame_layout):
1880 Add calculation of cfun->machine->fp_save_offset.
1881 (nios2_expand_prologue): Correct setting of frame pointer register
1883 (nios2_expand_epilogue): Update recovery of stack pointer from
1884 frame pointer accordingly.
1885 (nios2_initial_elimination_offset): Update calculation of offset
1886 for eliminating to HARD_FRAME_POINTER_REGNUM.
1888 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1891 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1892 cgraph_get_create_node on VAR_DECLs.
1894 2014-03-10 Richard Biener <rguenther@suse.de>
1897 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1899 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1901 * config/vms/vms.opt (vms_float_format): New variable.
1903 2014-03-08 Tobias Burnus <burnus@net-b.de>
1905 * doc/invoke.texi (-fcilkplus): Update implementation status.
1907 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1908 Richard Biener <rguenther@suse.de>
1910 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1911 consistently accross all TUs.
1912 (run_gcc): Enable -fshort-double automatically at link at link-time
1913 and disallow override.
1915 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1918 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1919 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1920 if they can't be used.
1922 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1924 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1925 for Solaris 11/x86 ld.
1926 * configure: Regenerate.
1928 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1930 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1931 (LIB_TLS_SPEC): Save as ld_tls_libs.
1932 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1933 (HAVE_AS_IX86_TLSLDM): New test.
1934 * configure, config.in: Regenerate.
1935 * config/i386/i386.c (legitimize_tls_address): Fall back to
1936 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1937 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1938 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1939 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1941 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1943 * common.opt (fira-loop-pressure): Mark as optimization.
1945 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1947 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1948 an OpenMP mappable type.
1950 2014-03-06 Matthias Klose <doko@ubuntu.com>
1952 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1953 MULTILIB_OSDIRNAMES is not defined.
1955 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1956 Meador Inge <meadori@codesourcery.com>
1959 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1960 (arm_legitimize_address): Call legitimize_tls_address for any
1961 arm_tls_referenced_p expression, handle constant addend. Call it
1962 before testing for !TARGET_ARM.
1963 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1965 2014-03-06 Richard Biener <rguenther@suse.de>
1971 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1973 * tree-streamer.c (record_common_node): Assert we don't record
1974 nodes with type double.
1975 (preload_common_node): Skip type double, complex double and double
1976 pointer since it is now frontend dependent due to fshort-double option.
1978 2014-03-06 Richard Biener <rguenther@suse.de>
1980 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1981 or -fno-lto is specified and the linker has full plugin support.
1982 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1983 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1984 * lto-wrapper.c (merge_and_complain): Merge compile-time
1985 optimization levels.
1986 (run_gcc): And pass it through to the link options.
1988 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1992 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1994 * cselib.c (remove_useless_values): Skip to avoid quadratic
1995 behavior if the condition moved from...
1996 (cselib_process_insn): ... here holds.
1998 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2001 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
2002 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
2005 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
2006 (TM_H): Add x86-tune.def.
2008 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2010 * config/aarch64/aarch64.c (generic_tunings):
2011 Use cortexa57_extra_costs.
2013 2014-03-05 Jakub Jelinek <jakub@redhat.com>
2016 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
2017 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
2018 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
2021 2014-03-04 Heiher <r@hev.cc>
2023 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
2024 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
2026 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
2028 * config/i386/predicates.md (const2356_operand): Change to ...
2029 (const2367_operand): ... this.
2030 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
2032 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2033 (*avx512pf_scatterpf<mode>sf): Ditto.
2034 (avx512pf_scatterpf<mode>df): Ditto.
2035 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2036 (*avx512pf_scatterpf<mode>df): Ditto.
2037 * config/i386/i386.c (ix86_expand_builtin): Update
2038 incorrect hint operand error message.
2040 2014-03-04 Richard Biener <rguenther@suse.de>
2042 * lto-section-in.c (lto_get_section_data): Fix const cast.
2044 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
2046 * tree-streamer.c (record_common_node): Assert we don't record
2047 nodes with type double.
2048 (preload_common_node): Skip type double, complex double and double
2049 pointer since it is now frontend dependent due to fshort-double option.
2051 2014-03-04 Richard Biener <rguenther@suse.de>
2054 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
2055 (lto_input_toplevel_asms): Likewise.
2056 * lto-section-in.c (lto_get_section_data): Instead do it here
2059 2014-03-04 Richard Biener <rguenther@suse.de>
2061 PR tree-optimization/60382
2062 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
2063 dead PHIs a reduction.
2065 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
2067 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
2069 (_mm_prefetch): Move out of GCC target("sse") pragma.
2070 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
2071 GCC target("prfchw") pragma.
2072 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
2074 * config/i386/i386.c (ix86_option_override_internal): Enable
2075 -mprfchw with -mprefetchwt1.
2077 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2079 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
2082 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
2084 * opts.h (CL_PCH_IGNORE): Define.
2085 * targhooks.c (option_affects_pch_p):
2086 Return false for options that have CL_PCH_IGNORE set.
2087 * opt-functions.awk: Process PchIgnore.
2088 * doc/options.texi: Document PchIgnore.
2090 * config/arc/arc.opt (misize): Add PchIgnore property.
2092 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2094 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
2095 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
2096 constraint on constants to permit them being loaded into
2097 GENERAL_REGS or BASE_REGS.
2099 2014-03-03 Nick Clifton <nickc@redhat.com>
2101 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
2102 anti-cacnonical alternatives.
2103 (negandhi3_real): New pattern.
2104 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
2106 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2108 * config/avr/avr-mcus.def: Remove atxmega16x1.
2109 * config/avr/avr-tables.opt: Regenerate.
2110 * config/avr/t-multilib: Regenerate.
2111 * doc/avr-mmcu.texi: Regenerate.
2113 2014-03-03 Tobias Grosser <tobias@grosser.es>
2114 Mircea Namolaru <mircea.namolaru@inria.fr>
2116 PR tree-optimization/58028
2117 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
2120 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2122 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
2123 not handled by recognizers.
2125 2014-03-03 Jakub Jelinek <jakub@redhat.com>
2128 * function.c (expand_function_end): Don't emit
2129 clobber_return_register sequence if clobber_after is a BARRIER.
2130 * cfgexpand.c (construct_exit_block): Append instructions before
2131 return_label to prev_bb.
2133 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2135 * config/rs6000/constraints.md: Document reserved use of "wc".
2137 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2140 * ipa.c (function_and_variable_visibility): When dissolving comdat
2141 group, also set all symbols to local.
2143 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
2148 2013-12-14 Jan Hubicka <jh@suse.cz>
2150 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
2152 2014-03-02 Jon Beniston <jon@beniston.com>
2158 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
2159 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
2160 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
2161 (simple_return, *simple_return): New patterns
2162 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
2163 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
2165 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
2167 * dwarf2out.c (gen_subprogram_die): Tidy.
2169 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
2172 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
2173 (*mov_t_msb_neg_negc): ... this new insn.
2175 2014-02-28 Jason Merrill <jason@redhat.com>
2178 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2181 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2184 * dwarf2out.c (decltype_auto_die): New static.
2185 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2186 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2187 (is_cxx_auto): Likewise.
2189 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2191 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2192 we are not using general regs only.
2194 2014-02-28 Richard Biener <rguenther@suse.de>
2197 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2198 previous fix and only allow to remove trivial pre-headers
2199 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2200 (remove_forwarder_block): Properly update the latch of a loop.
2202 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2205 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2206 (cselib_preserved_hash_table): New.
2207 (preserve_constants_and_equivs): Move preserved vals to it.
2208 (cselib_find_slot): Look it up first.
2209 (cselib_init): Initialize it.
2210 (cselib_finish): Release it.
2211 (dump_cselib_table): Dump it.
2213 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2216 * cselib.c (remove_useless_values): Skip to avoid quadratic
2217 behavior if the condition moved from...
2218 (cselib_process_insn): ... here holds.
2220 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2223 * var-tracking.c (vt_initialize): Apply the same condition to
2224 preserve the CFA base value.
2226 2014-02-28 Joey Ye <joey.ye@arm.com>
2229 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2230 if reload in progress or completed.
2232 2014-02-28 Tobias Burnus <burnus@net-b.de>
2235 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2238 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2240 * doc/tm.texi.in (Condition Code Status): Update documention for
2241 relative locations of cc0-setter and cc0-user.
2243 2014-02-27 Jeff Law <law@redhat.com>
2245 PR rtl-optimization/52714
2246 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2247 into two independent simple sets, if I3 is a jump, ensure the
2248 pattern we place into I3 is a (set (pc) ...).
2250 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2251 Jeff Law <law@redhat.com>
2253 PR rtl-optimization/49847
2254 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2255 are in different blocks.
2256 * doc/tm.texi (Condition Code Status): Update documention for
2257 relative locations of cc0-setter and cc0-user.
2259 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2262 * lra.c (lra_emit_add): Check SUBREG too.
2264 2014-02-27 Andreas Schwab <schwab@suse.de>
2266 * config/m68k/m68k.c (m68k_option_override): Disable
2267 -flive-range-shrinkage for classic m68k.
2268 (m68k_override_options_after_change): Likewise.
2270 2014-02-27 Marek Polacek <polacek@redhat.com>
2273 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2274 -Wmaybe-uninitialized.
2276 2014-02-27 Alan Modra <amodra@gmail.com>
2279 * reload1.c (emit_input_reload_insns): When reload_override_in,
2280 set old to rl->in_reg when rl->in_reg is a subreg.
2282 2014-02-26 Richard Biener <rguenther@suse.de>
2285 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2287 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2289 * common/config/i386/predicates.md (const1256_operand): Remove.
2290 (const2356_operand): New.
2291 (const_1_to_2_operand): Remove.
2292 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2293 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2294 (*avx512pf_gatherpf<mode>sf): Ditto.
2295 (avx512pf_gatherpf<mode>df): Ditto.
2296 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2297 (*avx512pf_gatherpf<mode>df): Ditto.
2298 (avx512pf_scatterpf<mode>sf): Ditto.
2299 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2300 (*avx512pf_scatterpf<mode>sf): Ditto.
2301 (avx512pf_scatterpf<mode>df): Ditto.
2302 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2303 (*avx512pf_scatterpf<mode>df): Ditto.
2304 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2306 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2308 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2309 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2310 (_mm512_mask_testn_epi64_mask): Move to ...
2311 * config/i386/avx512cdintrin.h: Here.
2312 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2313 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2314 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2315 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2316 TARGET_AVX512F from TARGET_AVX512CD.
2318 2014-02-26 Richard Biener <rguenther@suse.de>
2321 * ipa.c (walk_polymorphic_call_targets): Properly guard
2322 call to inline_update_overall_summary.
2324 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2327 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2328 and latches only if requested. Fix latch if it is removed.
2329 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2330 LOOPS_HAVE_PREHEADERS.
2332 2014-02-25 Andrew Pinski <apinski@cavium.com>
2334 * builtins.c (expand_builtin_thread_pointer): Create a new target
2335 when the target is NULL.
2337 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2339 PR rtl-optimization/60317
2340 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2341 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2342 * lra-assigns.c: Include params.h.
2343 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2344 other reload pseudos considerations.
2346 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2348 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2349 to use canonical form for nor<mode>3.
2351 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2354 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2357 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2360 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2361 (ix86_handle_option): Handle OPT_mprefetchwt1.
2362 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2363 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2365 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2366 OPTION_MASK_ISA_PREFETCHWT1.
2367 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2368 (PTA_PREFETCHWT1): New.
2369 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2370 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2371 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2372 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2373 (*prefetch_avx512pf_<mode>_: Change into ...
2374 (*prefetch_prefetchwt1_<mode>: This.
2375 * config/i386/i386.opt (mprefetchwt1): New.
2376 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2377 (_mm_prefetch): Handle intent to write.
2378 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2380 2014-02-25 Richard Biener <rguenther@suse.de>
2383 * emit-rtl.c (mem_attrs_htab): Remove.
2384 (mem_attrs_htab_hash): Likewise.
2385 (mem_attrs_htab_eq): Likewise.
2386 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2387 (init_emit_once): Do not allocate mem_attrs_htab.
2389 2014-02-25 Richard Biener <rguenther@suse.de>
2392 * lto-opts.c (lto_write_options): Output non-explicit conservative
2393 -fwrapv, -fno-trapv and -fno-strict-overflow.
2394 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2395 (run_gcc): And pass them through.
2397 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2399 * sel-sched.c (calculate_new_fences): New parameter ptime.
2400 Calculate it as a maximum over all fence cycles.
2401 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2402 Print the final schedule timing when sched_verbose.
2404 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2406 PR rtl-optimization/60292
2407 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2408 bit fot the fence instruction.
2410 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2412 * calls.h: Fix typo in comment.
2414 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2416 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2417 adjusting offsetable addresses.
2419 2014-02-24 Guozhi Wei <carrot@google.com>
2421 * sparseset.h (sparseset_pop): Fix the wrong index.
2423 2014-02-24 Walter Lee <walt@tilera.com>
2425 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2426 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2428 * common/config/tilegx/tilegx-common.c
2429 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2430 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2432 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2433 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2434 (tilegx_gimplify_va_arg_expr): Handle big endian.
2435 (tilegx_expand_unaligned_load): Ditto.
2436 (tilegx_expand_unaligned_store): Ditto.
2437 (TARGET_RETURN_IN_MSB): New.
2438 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2439 (TARGET_ENDIAN_DEFAULT): New.
2440 (TARGET_BIG_ENDIAN): Handle big endian.
2441 (BYTES_BIG_ENDIAN): Ditto.
2442 (WORDS_BIG_ENDIAN): Ditto.
2443 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2446 * config/tilegx/tilegx.md (extv): Handle big endian.
2448 (insn_st<n>): Ditto.
2449 (insn_st<n>_add<bitsuffix>): Ditto.
2450 (insn_stnt<n>): Ditto.
2451 (insn_stnt<n>_add<bitsuffix>):Ditto.
2452 (vec_interleave_highv8qi): Handle big endian.
2453 (vec_interleave_highv8qi_be): New.
2454 (vec_interleave_highv8qi_le): New.
2455 (insn_v1int_h): Handle big endian.
2456 (vec_interleave_lowv8qi): Handle big endian.
2457 (vec_interleave_lowv8qi_be): New.
2458 (vec_interleave_lowv8qi_le): New.
2459 (insn_v1int_l): Handle big endian.
2460 (vec_interleave_highv4hi): Handle big endian.
2461 (vec_interleave_highv4hi_be): New.
2462 (vec_interleave_highv4hi_le): New.
2463 (insn_v2int_h): Handle big endian.
2464 (vec_interleave_lowv4hi): Handle big endian.
2465 (vec_interleave_lowv4hi_be): New.
2466 (vec_interleave_lowv4hi_le): New.
2467 (insn_v2int_l): Handle big endian.
2468 (vec_interleave_highv2si): Handle big endian.
2469 (vec_interleave_highv2si_be): New.
2470 (vec_interleave_highv2si_le): New.
2471 (insn_v4int_h): Handle big endian.
2472 (vec_interleave_lowv2si): Handle big endian.
2473 (vec_interleave_lowv2si_be): New.
2474 (vec_interleave_lowv2si_le): New.
2475 (insn_v4int_l): Handle big endian.
2476 * config/tilegx/tilegx.opt (mbig-endian): New option.
2477 (mlittle-endian): New option.
2478 * doc/install.texi: Document tilegxbe-linux.
2479 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2481 2014-02-24 Martin Jambor <mjambor@suse.cz>
2484 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2485 there are no parameter descriptors.
2487 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2489 PR rtl-optimization/60268
2490 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2491 initialization to ...
2492 (sched_rgn_init): ... here.
2493 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2495 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2497 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2500 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2502 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2505 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2507 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2508 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2510 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2512 * config/microblaze/predicates.md: Add cmp_op predicate.
2513 * config/microblaze/microblaze.md: Add branch_compare instruction
2514 which uses cmp_op predicate and emits cmp insn before branch.
2515 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2516 to microblaze_expand_conditional_branch and consolidate logic.
2517 (microblaze_expand_conditional_branch): emit branch_compare
2518 insn instead of handling cmp op separate from branch insn.
2520 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2522 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2525 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2527 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2528 define_insn with define_expand and new define_insn
2529 *altivec_lve<VI_char>x_internal.
2530 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2531 and new define_insn *altivec_stve<VI_char>x_internal.
2532 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2534 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2536 (altivec_expand_stvex_be): New function.
2538 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2540 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2541 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2542 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2543 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2545 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2548 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2549 instead of emit_move_insn.
2551 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2553 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2555 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2556 gen_altivec_vsumsws.
2558 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2560 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2561 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2562 (altivec_lvxl_<mode>): New define_expand incorporating
2563 -maltivec=be semantics where needed.
2564 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2565 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2566 semantics where needed.
2567 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2568 (altivec_stvx_<mode>): New define_expand incorporating
2569 -maltivec=be semantics where needed.
2570 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2571 VM2 iterator instead of V4SI.
2572 (altivec_stvxl_<mode>): New define_expand incorporating
2573 -maltivec=be semantics where needed.
2574 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2575 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2576 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2577 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2578 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2579 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2580 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2581 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2582 ALTIVEC_BUILTIN_STVXL.
2583 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2584 (altivec_expand_stvx_be): Likewise.
2585 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2586 (altivec_expand_lvx_be): Likewise.
2587 (altivec_expand_stvx_be): Likewise.
2588 (altivec_expand_builtin): Add cases for
2589 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2590 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2591 (altivec_init_builtins): Add definitions for
2592 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2593 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2595 2014-02-21 Catherine Moore <clm@codesourcery.com>
2597 * doc/invoke.texi (mvirt, mno-virt): Document.
2598 * config/mips/mips.opt (mvirt): New option.
2599 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2601 2014-02-21 Richard Biener <rguenther@suse.de>
2603 PR tree-optimization/60276
2604 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2605 (STMT_VINFO_MIN_NEG_DIST): New macro.
2606 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2607 STMT_VINFO_MIN_NEG_DIST.
2608 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2609 made for negative dependence distances still hold.
2611 2014-02-21 Richard Biener <rguenther@suse.de>
2614 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2615 DECL_INITIAL for globals not in the current function context.
2617 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2619 PR tree-optimization/56490
2620 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2621 * tree-ssa-uninit.c: Include params.h.
2622 (compute_control_dep_chain): Add num_calls argument, return false
2623 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2624 num_calls to recursive call.
2625 (find_predicates): Change dep_chain into normal array,
2626 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2627 variable and adjust compute_control_dep_chain caller.
2628 (find_def_preds): Likewise.
2630 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2632 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2633 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2635 2014-02-21 Nick Clifton <nickc@redhat.com>
2637 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2638 (pushhi1): Likewise.
2639 (popqi1): Add mode to pre_dec.
2642 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2644 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2645 mode for mask of V8SFmode permutation.
2647 2014-02-20 Richard Henderson <rth@redhat.com>
2650 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2651 a new pseudo for OLDVAL.
2653 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2656 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2657 gen_reg_rtx if d->testing_p.
2658 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2659 if d->testing_p and we will certainly return true.
2660 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2663 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2665 * emit-rtl.c (gen_reg_rtx): Assert that
2666 crtl->emit.regno_pointer_align_length is non-zero.
2668 2014-02-20 Richard Henderson <rth@redhat.com>
2671 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2672 on failure the store back into EXPECT.
2674 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2675 Sandra Loosemore <sandra@codesourcery.com>
2677 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2678 * config/nios2/nios2.c (nios2_function_profiler): Add
2679 -fPIC (flag_pic == 2) support.
2680 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2681 (nios2_large_offset_p): New function.
2682 (nios2_unspec_reloc_p): Move up position, update to use
2683 nios2_large_offset_p.
2684 (nios2_unspec_address): Remove function.
2685 (nios2_unspec_offset): New function.
2686 (nios2_large_got_address): New function.
2687 (nios2_got_address): Add large offset support.
2688 (nios2_legitimize_tls_address): Update usage of removed and new
2690 (nios2_symbol_binds_local_p): New function.
2691 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2692 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2693 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2694 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2695 processing for (const (unspec ...)).
2696 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2698 2014-02-20 Richard Biener <rguenther@suse.de>
2700 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2701 doing the substitution.
2702 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2704 2014-02-20 Martin Jambor <mjambor@suse.cz>
2707 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2708 info when checking whether lattices are bottom.
2710 2014-02-20 Richard Biener <rguenther@suse.de>
2713 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2716 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2719 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2720 parameter specifying the scaling.
2721 (inline_call): Update.
2722 (want_inline_recursively): Guard division by zero.
2723 (recursive_inlining): Update.
2724 * ipa-inline.h (clone_inlined_nodes): Update.
2726 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2729 * config/i386/i386.c (classify_argument): Pass structures of size
2730 64 bytes or less in register.
2732 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2733 Kirill Yukhin <kirill.yukhin@intel.com>
2735 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2736 (_mm_rcp28_round_ss): Ditto.
2737 (_mm_rsqrt28_round_sd): Ditto.
2738 (_mm_rsqrt28_round_ss): Ditto.
2739 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2740 (_mm_rcp14_round_ss): Ditto.
2741 (_mm_rsqrt14_round_sd): Ditto.
2742 (_mm_rsqrt14_round_ss): Ditto.
2743 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2744 the first input operand, get rid of match_dup.
2745 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2747 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2749 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2750 operand as the first input operand, set type attribute.
2751 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2753 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2754 operand as the first input operand, set type attribute.
2756 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2758 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2761 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2764 * config/i386/i386.c (construct_container): Remove TFmode check
2765 for X86_64_INTEGER_CLASS.
2767 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2770 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2771 only when -Wpsabi is enabled.
2773 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2776 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2777 passing arrays in registers are the same as for structs, so remove the
2778 special case for them.
2780 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2782 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2783 destination type, extract only the valid bits if the source type is not
2784 integral and has a different mode.
2786 2014-02-19 Richard Biener <rguenther@suse.de>
2789 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2792 2014-02-19 Richard Biener <rguenther@suse.de>
2795 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2796 (ipa_modify_call_arguments): Emit an argument load explicitely and
2797 preserve virtual SSA form there and for the replacement call.
2798 Do not update SSA form nor free dominance info.
2800 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2802 * ipa.c (function_and_variable_visibility): Also clear WEAK
2803 flag when disolving COMDAT_GROUP.
2805 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2807 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2808 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2810 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2811 do more sanity checks.
2812 (detect_type_change): Return true when giving up early.
2813 (compute_complex_assign_jump_func): Fix type parameter of
2814 ipa_set_ancestor_jf.
2815 (compute_complex_ancestor_jump_func): Likewise.
2816 (update_jump_functions_after_inlining): Fix updating of
2818 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2820 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2822 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2823 inline clones when edge disappears.
2825 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2828 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2829 Split 64-bit moves into 2 patterns. Do not allow the use of
2830 direct move for TDmode in little endian, since the decimal value
2831 has little endian bytes within a word, but the 64-bit pieces are
2832 ordered in a big endian fashion, and normal subreg's of TDmode are
2834 (mov<mode>_64bit_dm): Likewise.
2835 (movtd_64bit_nodm): Likewise.
2837 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2839 PR tree-optimization/60174
2840 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2841 statement of an SSA_NAME that occurs in an abnormal PHI node.
2843 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2846 * final.c (SEEN_BB): Remove.
2847 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2848 (final_scan_insn): Don't force_source_line on second
2849 NOTE_INSN_BASIC_BLOCK.
2851 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2854 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2855 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2856 (type_natural_mode): Warn ABI change when %zmm register is not
2857 available for AVX512F vector value passing.
2859 2014-02-18 Kai Tietz <ktietz@redhat.com>
2862 * config/i386/i386.c (ix86_expand_prologue): Use value in
2863 rax register as displacement when restoring %r10 or %rax.
2864 Fix wrong offset when restoring both registers.
2866 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2868 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2869 assertion with conditional return.
2871 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2872 Uros Bizjak <ubizjak@gmail.com>
2875 * config/i386/driver-i386.c (host_detect_local_cpu): If
2876 YMM state is not saved by the OS, also clear has_f16c. Move
2877 CPUID 0x80000001 handling before YMM state saving checking.
2879 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2881 PR rtl-optimization/58960
2882 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2883 factored out from ...
2884 (sched_init): ... here.
2885 (free_global_sched_pressure_data): New, factored out from ...
2886 (sched_finish): ... here.
2887 * sched-int.h (free_global_sched_pressure_data): Declare.
2888 * sched-rgn.c (nr_regions_initial): New static global.
2889 (haifa_find_rgns): Initialize it.
2890 (schedule_region): Disable sched-pressure for the newly
2893 2014-02-17 Richard Biener <rguenther@suse.de>
2895 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2896 release SSA defs of pattern stmts.
2898 2014-02-17 Richard Biener <rguenther@suse.de>
2900 * tree-inline.c (expand_call_inline): Release the virtual
2901 operand defined by the call we are about to inline.
2903 2014-02-17 Richard Biener <rguenther@suse.de>
2905 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2907 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2908 Ilya Tocar <ilya.tocar@intel.com>
2910 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2911 arguments order in builtin.
2912 (_mm512_permutexvar_epi64): Ditto.
2913 (_mm512_mask_permutexvar_epi64): Ditto
2914 (_mm512_maskz_permutexvar_epi32): Ditto
2915 (_mm512_permutexvar_epi32): Ditto
2916 (_mm512_mask_permutexvar_epi32): Ditto
2918 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2920 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2921 (p8_vmrgow): Likewise.
2923 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2925 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2928 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2931 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2932 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2933 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2934 using direct move instructions on ISA 2.07. Also adjust
2935 instruction length for 64-bit.
2936 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2937 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2939 2014-02-15 Alan Modra <amodra@gmail.com>
2943 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2944 find_replacement on parts of insn rtl that might be reloaded.
2946 2014-02-15 Richard Biener <rguenther@suse.de>
2948 PR tree-optimization/60183
2949 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2950 (tree_ssa_phiprop): Calculate and free post-dominators.
2952 2014-02-14 Jeff Law <law@redhat.com>
2954 PR rtl-optimization/60131
2955 * ree.c (get_extended_src_reg): New function.
2956 (combine_reaching_defs): Use it rather than assuming location of REG.
2957 (find_and_remove_re): Verify first operand of extension is
2958 a REG before adding the insns to the copy list.
2960 2014-02-14 Roland McGrath <mcgrathr@google.com>
2962 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2963 * configure: Regenerated.
2964 * config.in: Regenerated.
2965 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2966 instead of ASM_SHORT.
2968 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2969 Richard Earnshaw <rearnsha@arm.com>
2971 PR rtl-optimization/59535
2972 * lra-constraints.c (process_alt_operands): Encourage alternative
2973 when unassigned pseudo class is superset of the alternative class.
2974 (inherit_reload_reg): Don't inherit when optimizing for code size.
2975 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2976 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2977 modes not less than 4 for Thumb1.
2979 2014-02-14 Kyle McMartin <kyle@redhat.com>
2982 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2984 2014-02-14 Richard Biener <rguenther@suse.de>
2986 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2987 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2988 Do not strip INDIRECT_REFs.
2990 2014-02-14 Richard Biener <rguenther@suse.de>
2993 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2994 DECL_FUNCTION_SPECIFIC_TARGET.
2995 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2996 * tree-streamer-out.c (pack_ts_target_option): Remove.
2997 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2998 (write_ts_function_decl_tree_pointers): Do not stream
2999 DECL_FUNCTION_SPECIFIC_TARGET.
3000 * tree-streamer-in.c (unpack_ts_target_option): Remove.
3001 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
3002 (lto_input_ts_function_decl_tree_pointers): Do not stream
3003 DECL_FUNCTION_SPECIFIC_TARGET.
3005 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3007 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
3008 (get_initial_def_for_induction, vectorizable_induction): Ignore
3009 debug stmts when looking for exit_phi.
3010 (vectorizable_live_operation): Fix up condition.
3012 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3014 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
3015 nreverse() because it changes the content of original tree list.
3017 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3019 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
3020 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
3022 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
3024 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
3025 GNU coding standards.
3027 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3030 * dwarf2out.c (gen_subprogram_die): Don't call
3031 add_calling_convention_attribute if subr_die is old_die.
3033 2014-02-13 Sharad Singhai <singhai@google.com>
3035 * doc/optinfo.texi: Fix order of nodes.
3037 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3039 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
3040 operands[2], not operands[3].
3042 2014-02-13 Richard Biener <rguenther@suse.de>
3045 * doc/install.texi (ISL): Update recommended version to 0.12.2,
3046 mention the possibility of an in-tree build.
3047 (CLooG): Update recommended version to 0.18.1, mention the
3048 possibility of an in-tree build and clarify that the ISL
3049 bundled with CLooG does not work.
3051 2014-02-13 Jakub Jelinek <jakub@redhat.com>
3054 * expr.c (compress_float_constant): If x is a hard register,
3055 extend into a pseudo and then move to x.
3057 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
3059 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
3060 caused by bad second argument to warning_at() with -mhotpatch and
3061 nested functions (e.g. with gfortran).
3063 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
3065 * opts.c (option_name): Remove "enabled by default" rider.
3067 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
3069 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
3071 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
3072 Uros Bizjak <ubizjak@gmail.com>
3075 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
3076 * configure: Regenerated.
3078 2014-02-12 Richard Biener <rguenther@suse.de>
3080 * vec.c (vec_prefix::calculate_allocation): Move as
3081 inline variant to vec.h.
3082 (vec_prefix::calculate_allocation_1): New out-of-line version.
3083 * vec.h (vec_prefix::calculate_allocation_1): Declare.
3084 (vec_prefix::m_has_auto_buf): Rename to ...
3085 (vec_prefix::m_using_auto_storage): ... this.
3086 (vec_prefix::calculate_allocation): Inline the easy cases
3087 and dispatch to calculate_allocation_1 which doesn't need the
3089 (va_heap::reserve): Use gcc_checking_assert.
3090 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
3091 m_using_auto_storage.
3092 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
3094 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
3095 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
3096 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
3098 2014-02-12 Richard Biener <rguenther@suse.de>
3100 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
3101 when we found a dependence.
3103 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
3105 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
3107 (maybe_fold_stmt): ... into this new function.
3108 * omp-low.c (lower_omp): Update comment.
3110 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
3113 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
3116 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
3118 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
3119 identifiers in comments.
3120 (cortexa53_extra_costs): Likewise.
3121 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
3122 (cortexa7_extra_costs): Likewise.
3123 (cortexa12_extra_costs): Likewise.
3124 (cortexa15_extra_costs): Likewise.
3125 (v7m_extra_costs): Likewise.
3127 2014-02-12 Richard Biener <rguenther@suse.de>
3130 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
3131 of posix_memalign being successful.
3132 (lower_stmt): Restrict lowering of posix_memalign to when
3133 -ftree-bit-ccp is enabled.
3135 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
3137 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
3139 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
3141 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
3143 PR rtl-optimization/60116
3144 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
3145 other_insn once the combination has been validated.
3147 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
3150 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
3152 * ipa-devirt.c: Include demangle.h
3153 (odr_violation_reported): New static variable.
3154 (add_type_duplicate): Update odr_violations.
3155 (maybe_record_node): Add completep parameter; update it.
3156 (record_target_from_binfo): Add COMPLETEP parameter;
3157 update it as needed.
3158 (possible_polymorphic_call_targets_1): Likewise.
3159 (struct polymorphic_call_target_d): Add nonconstruction_targets;
3160 rename FINAL to COMPLETE.
3161 (record_targets_from_bases): Sanity check we found the binfo;
3162 fix COMPLETEP updating.
3163 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
3164 parameter, fix computing of COMPLETEP.
3165 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
3166 at LTO time do demangling.
3167 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
3168 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
3170 (gimple_get_virt_method_for_binfo): Likewise.
3171 * gimple-fold.h (gimple_get_virt_method_for_binfo,
3172 gimple_get_virt_method_for_vtable): Update prototypes.
3174 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
3177 * genautomata.c (add_presence_absence): Fix typo with
3178 {final_}presence_list.
3180 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3183 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3184 for VSX/Altivec vectors that land in GPR registers.
3186 2014-02-11 Richard Henderson <rth@redhat.com>
3187 Jakub Jelinek <jakub@redhat.com>
3190 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3191 around drhs if type conversion to lacc->type is not useless.
3193 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3195 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3197 (cortex-a57.cortex-a53): Likewise.
3198 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3200 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3202 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3205 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3207 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3208 add_options_for_arm_vfp3.
3210 2014-02-11 Jeff Law <law@redhat.com>
3213 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3214 object with an undesirable mode.
3216 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3219 * config/i386/sol2-9.h: New file.
3220 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3221 *-*-solaris2.9*): Use it.
3223 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3225 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3226 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3228 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3230 * config/microblaze/microblaze.c: Extend mcpu version format
3232 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3234 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3236 2014-02-10 Richard Henderson <rth@redhat.com>
3239 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3240 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3241 ms-abi vs -mno-accumulate-outgoing-args.
3242 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3243 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3246 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3249 * cfgexpand.c (expand_asm_operands): Attach source location to
3250 ASM_INPUT rtx objects.
3251 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3253 2014-02-10 Nick Clifton <nickc@redhat.com>
3255 * config/mn10300/mn10300.c (popcount): New function.
3256 (mn10300_expand_prologue): Include saved registers in stack usage
3259 2014-02-10 Jeff Law <law@redhat.com>
3262 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3263 when changing the SET_DEST of a prior insn to avoid an input reload.
3265 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3267 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3268 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3269 -mcall-openbsd, or -mcall-linux.
3270 (CC1_ENDIAN_BIG_SPEC): Remove.
3271 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3272 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3273 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3274 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3275 and %cc1_endian_default.
3276 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3278 2014-02-10 Richard Biener <rguenther@suse.de>
3280 PR tree-optimization/60115
3281 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3282 MEM_REF handling. Properly verify that the accesses are not
3283 out of the objects bound.
3285 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3287 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3290 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3292 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3293 proper constants and fix formatting.
3294 (possible_polymorphic_call_targets): Fix formatting.
3296 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3297 Ilya Tocar <ilya.tocar@intel.com>
3299 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3300 (_mm512_loadu_epi32): Renamed into...
3301 (_mm512_loadu_si512): This.
3302 (_mm512_storeu_epi32): Renamed into...
3303 (_mm512_storeu_si512): This.
3304 (_mm512_maskz_ceil_ps): Removed.
3305 (_mm512_maskz_ceil_pd): Ditto.
3306 (_mm512_maskz_floor_ps): Ditto.
3307 (_mm512_maskz_floor_pd): Ditto.
3308 (_mm512_floor_round_ps): Ditto.
3309 (_mm512_floor_round_pd): Ditto.
3310 (_mm512_ceil_round_ps): Ditto.
3311 (_mm512_ceil_round_pd): Ditto.
3312 (_mm512_mask_floor_round_ps): Ditto.
3313 (_mm512_mask_floor_round_pd): Ditto.
3314 (_mm512_mask_ceil_round_ps): Ditto.
3315 (_mm512_mask_ceil_round_pd): Ditto.
3316 (_mm512_maskz_floor_round_ps): Ditto.
3317 (_mm512_maskz_floor_round_pd): Ditto.
3318 (_mm512_maskz_ceil_round_ps): Ditto.
3319 (_mm512_maskz_ceil_round_pd): Ditto.
3320 (_mm512_expand_pd): Ditto.
3321 (_mm512_expand_ps): Ditto.
3322 * config/i386/i386.c (ix86_builtins): Remove
3323 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3324 (bdesc_args): Ditto.
3325 * config/i386/predicates.md (const1256_operand): New.
3326 (const_1_to_2_operand): Ditto.
3327 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3328 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3329 (*avx512pf_gatherpf<mode>sf): Ditto.
3330 (avx512pf_gatherpf<mode>df): Ditto.
3331 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3332 (*avx512pf_gatherpf<mode>df): Ditto.
3333 (avx512pf_scatterpf<mode>sf): Ditto.
3334 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3335 (*avx512pf_scatterpf<mode>sf): Ditto.
3336 (avx512pf_scatterpf<mode>df): Ditto.
3337 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3338 (*avx512pf_scatterpf<mode>df): Ditto.
3339 (avx512f_expand<mode>): Removed.
3340 (<shift_insn><mode>3<mask_name>): Change predicate type.
3342 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3344 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3345 not at the end of datarefs vector use ordered_remove to avoid
3346 reordering datarefs vector.
3349 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3350 mark local addressable non-static vars as GOVD_PRIVATE
3351 instead of GOVD_LOCAL.
3352 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3353 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3357 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3358 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3359 assume_aligned or alloc_align attributes.
3360 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3361 arguments. Handle also assume_aligned and alloc_align attributes.
3362 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3363 calls to functions with assume_aligned or alloc_align attributes.
3364 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3366 2014-02-08 Terry Guo <terry.guo@arm.com>
3368 * doc/invoke.texi: Document ARM -march=armv7e-m.
3370 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3372 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3373 flag on __cilkrts_rethrow builtin.
3376 * ipa-cp.c (determine_versionability): Fail at -O0
3377 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3378 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3381 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3384 * tree-inline.c (copy_forbidden): Fail for
3385 __attribute__((optimize (0))) functions.
3387 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3389 * varpool.c: Include pointer-set.h.
3390 (varpool_remove_unreferenced_decls): Variables in other partitions
3391 will not be output; be however careful to not lose information
3394 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3396 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3397 lookup in the vtable constructor.
3399 2014-02-07 Jeff Law <law@redhat.com>
3402 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3403 define_insn_and_split.
3405 * ipa-inline.c (inline_small_functions): Fix typos.
3407 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3409 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3410 (s390_can_use_return_insn): Declare.
3411 * config/s390/s390.h (EPILOGUE_USES): Define.
3412 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3414 (s390_chunkify_start): Handle return JUMP_LABELs.
3415 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3416 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3417 (s390_can_use_return_insn): New functions.
3418 (s390_fix_long_loop_prediction): Handle conditional returns.
3419 (TARGET_SET_UP_BY_PROLOGUE): Define.
3420 * config/s390/s390.md (ANY_RETURN): New code iterator.
3421 (*creturn, *csimple_return, return, simple_return): New patterns.
3423 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3425 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3426 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3427 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3428 REG_CFA_RESTORE list when deciding not to restore a register.
3430 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3432 * config/s390/s390.c: Include tree-pass.h and context.h.
3433 (s390_early_mach): New function, split out from...
3434 (s390_emit_prologue): ...here.
3435 (pass_data_s390_early_mach): New pass structure.
3436 (pass_s390_early_mach): New class.
3437 (s390_option_override): Create and register early_mach pass.
3438 Move to end of file.
3440 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3442 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3443 to match for the exit block.
3445 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3447 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3448 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3449 Reject misaligned operands.
3451 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3453 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3455 2014-02-07 Richard Biener <rguenther@suse.de>
3458 * gimple-low.c (lower_builtin_posix_memalign): New function.
3459 (lower_stmt): Call it to lower posix_memalign in a way
3460 to make alignment info accessible.
3462 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3465 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3466 __builtin_setjmp_receiver.
3468 2014-02-07 Richard Biener <rguenther@suse.de>
3471 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3472 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3473 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3474 Handle BUILT_IN_POSIX_MEMALIGN.
3475 (find_func_clobbers): Likewise.
3476 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3477 (call_may_clobber_ref_p_1): Likewise.
3479 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3482 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3485 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3488 * lto-cgraph.c (lto_output_node): Use
3489 symtab_get_symbol_partitioning_class.
3490 (lto_output_varpool_node): likewise.
3491 (symtab_get_symbol_partitioning_class): Move here from
3493 * cgraph.h (symbol_partitioning_class): Likewise.
3494 (symtab_get_symbol_partitioning_class): Declare.
3496 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3498 * ggc.h (ggc_internal_cleared_alloc): New macro.
3499 * vec.h (vec_safe_copy): Handle memory stats.
3500 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3501 * target-globals.c (save_target_globals): Likewise.
3503 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3506 * expr.c (emit_move_resolve_push): Export; be bit more selective
3507 on when to clear alias set.
3508 * expr.h (emit_move_resolve_push): Declare.
3509 * function.h (struct function): Add tail_call_marked.
3510 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3511 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3512 * config/i386/i386.md (TImode move expander): De not call
3514 (FP push expanders): Preserve memory attributes.
3515 * config/i386/sse.md (push<mode>1): Remove.
3516 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3517 (ix86_expand_push): Remove.
3518 * config/i386/mmx.md (push<mode>1): Remove.
3520 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3522 PR rtl-optimization/60030
3523 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3524 lopart with paradoxical subreg before shifting it up by hprec.
3526 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3528 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3529 Remove extra newline at end of file.
3530 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3531 (arm_issue_rate): Handle cortexa57.
3532 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3533 (cortex-a57.cortex-a53): Likewise.
3535 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3538 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3539 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3541 (arm_expand_prologue): Adjust all callers.
3542 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3543 info, registers also at the lowest numbered registers side. Use
3544 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3548 * var-tracking.c (adjust_mems): Before adding a SET to
3549 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3551 2014-02-06 Alan Modra <amodra@gmail.com>
3554 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3555 change SDmode to DDmode when lra_in_progress.
3557 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3560 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3561 free_data_ref on the dr first, and before goto again also set dr
3562 to the next dr. For simd_lane_access, free old datarefs[i] before
3563 overwriting it. For get_vectype_for_scalar_type failure, don't
3564 free_data_ref if simd_lane_access.
3566 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3569 * tree.h (opts_for_fn): New inline function.
3570 (opt_for_fn): Define.
3571 * config/i386/i386.c (ix86_function_regparm): Use
3572 opt_for_fn (decl, optimize) instead of optimize.
3574 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3576 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3577 for SYMBOL_REF in large memory model.
3579 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3581 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3583 (cortex-a57): Likewise.
3584 (cortex-a57.cortex-a53): Likewise.
3586 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3587 Kugan Vivekanandarajah <kuganv@linaro.org>
3589 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3591 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3593 2014-02-06 Richard Biener <rguenther@suse.de>
3595 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3596 set_loop_copy and initialize_original_copy_tables.
3598 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3600 * config/aarch64/aarch64-simd.md
3601 (aarch64_ashr_simddi): Change QI to SI.
3603 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3604 Jakub Jelinek <jakub@redhat.com>
3607 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3610 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3612 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3613 CODE_FOR_altivec_vpku[hw]um to
3614 CODE_FOR_altivec_vpku[hw]um_direct.
3615 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3616 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3617 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3618 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3620 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3622 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3623 generation for -maltivec=be.
3624 (altivec_vsumsws): Simplify redundant test.
3626 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3628 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3629 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3630 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3631 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3632 gen_altivec_vpkuwum.
3633 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3635 (altivec_vpks<VI_char>ss): Likewise.
3636 (altivec_vpks<VI_char>us): Likewise.
3637 (altivec_vpku<VI_char>us): Likewise.
3638 (altivec_vpku<VI_char>um): Likewise.
3639 (altivec_vpku<VI_char>um_direct): New (copy of
3640 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3642 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3643 target is little endian and -maltivec=be is not specified.
3644 (*altivec_vupkhs<VU_char>_direct): New (copy of
3645 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3646 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3647 target is little endian and -maltivec=be is not specified.
3648 (*altivec_vupkls<VU_char>_direct): New (copy of
3649 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3650 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3651 little endian and -maltivec=be is not specified.
3652 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3653 little endian and -maltivec=be is not specified.
3655 2014-02-05 Richard Henderson <rth@redhat.com>
3658 * combine-stack-adj.c: Revert r206943.
3659 * sched-int.h (struct deps_desc): Add last_args_size.
3660 * sched-deps.c (init_deps): Initialize it.
3661 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3662 contain REG_ARGS_SIZE notes.
3664 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3666 * lto-cgraph.c (asm_nodes_output): Make global.
3667 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3668 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3669 (driver_handle_option): Handle OPT_fwpa.
3671 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3674 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3675 a comment typo and formatting issue. If odr_hash hasn't been
3676 created, return vNULL and set *completep to false.
3679 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3680 bb with no successors.
3682 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3685 * doc/invoke.texi (-march): Clarify documentation for ARM.
3689 2014-02-05 Richard Biener <rguenther@suse.de>
3691 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3692 when not vectorizing because of too many alias checks.
3693 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3694 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3696 2014-02-05 Nick Clifton <nickc@redhat.com>
3698 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3699 accept extended registers in any mode when compiling for the MN10300.
3701 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3703 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3704 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3705 sanitization attributes.
3706 (can_inline_edge_p): Likewise.
3707 (sanitize_attrs_match_for_inline_p): New function.
3709 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3711 * ipa-prop.c (detect_type_change): Shor circuit testing of
3712 type changes on THIS pointer.
3714 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3717 * config/pa/pa.c (legitimize_tls_address): Return original address
3718 if not passed a SYMBOL_REF rtx.
3719 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3721 (pa_emit_move_sequence): Simplify TLS source operands.
3722 (pa_legitimate_constant_p): Reject all TLS constants.
3723 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3724 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3726 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3728 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3729 groups when we know they are controlled by LTO.
3730 * varasm.c (default_binds_local_p_1): If object is in other partition,
3731 it will be resolved locally.
3733 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3735 * config/host-linux.c (linux_gt_pch_use_address): Don't
3736 use SSIZE_MAX because it is not always defined.
3738 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3741 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3742 threshold for pseudo splitting.
3743 (update_ebb_live_info): Process call argument hard registers and
3744 hard registers from insn definition too.
3745 (max_small_class_regs_num): New constant.
3746 (inherit_in_ebb): Update live hard regs through EBBs. Update
3747 reloads_num only for small register classes. Don't split for
3750 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3753 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3756 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3758 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3761 2014-02-04 Marek Polacek <polacek@redhat.com>
3763 * gdbinit.in (pel): Define.
3765 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3767 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3770 2014-02-04 Richard Biener <rguenther@suse.de>
3773 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3774 in function context local.
3775 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3776 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3777 similar to LTO_imported_decl_ref.
3779 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3781 PR tree-optimization/60002
3782 * cgraphclones.c (build_function_decl_skip_args): Clear
3785 PR tree-optimization/60023
3786 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3787 false to gsi_replace.
3788 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3789 has been in some EH region and vec_stmt could throw, add
3790 vec_stmt into the same EH region.
3791 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3792 has no lhs, ignore it.
3793 * internal-fn.c (expand_MASK_LOAD): Likewise.
3796 * tree-inline.c (copy_forbidden): Fail for
3797 __attribute__((optimize (0))) functions.
3800 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3801 is set, copy one less argument.
3802 (expand_simd_clones): Don't subtract clone_info->inbranch
3803 from simd_clone_struct_alloc argument.
3805 PR rtl-optimization/57915
3806 * recog.c (simplify_while_replacing): If all unary/binary/relational
3807 operation arguments are constant, attempt to simplify those.
3810 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3811 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3813 2014-02-04 Richard Biener <rguenther@suse.de>
3815 PR tree-optimization/60012
3816 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3817 TBAA disambiguation to all DDRs.
3819 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3822 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3823 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3825 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3828 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3830 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3832 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3833 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3835 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3838 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3839 to figure out targets of polymorphic calls with known decl.
3840 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3841 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3842 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3843 (get_polymorphic_call_info): ... here.
3844 (get_polymorphic_call_info_from_invariant): New function.
3846 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3848 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3849 lookup via vtable pointer; check for type consistency
3850 and turn inconsitent facts into UNREACHABLE.
3851 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3852 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3853 type inconsistent querries; return UNREACHABLE instead.
3855 2014-02-03 Richard Henderson <rth@twiddle.net>
3858 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3859 already processed this node.
3860 (normalize_one_pred_1): Pass along mark_set.
3861 (normalize_one_pred): Create and destroy a pointer_set_t.
3862 (normalize_one_pred_chain): Likewise.
3864 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3866 PR gcov-profile/58602
3867 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3869 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3872 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3873 -fno-devirtualize; try to devirtualize by the knowledge of
3874 virtual table pointer given by aggregate propagation.
3875 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3876 (ipa_print_node_jump_functions): Dump also offset that
3877 is relevant for polymorphic calls.
3878 (determine_known_aggregate_parts): Add arg_type parameter; use it
3879 instead of determining the type from pointer type.
3880 (ipa_compute_jump_functions_for_edge): Update call of
3881 determine_known_aggregate_parts.
3882 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3883 (gimple_get_virt_method_for_binfo): ... here; simplify using
3884 vtable_pointer_value_to_vtable.
3885 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3886 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3887 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3888 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3890 (vtable_pointer_value_to_binfo): ... here.
3891 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3893 2014-02-03 Teresa Johnson <tejohnson@google.com>
3895 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3896 redef of outer loop index variable.
3898 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3902 * doc/extend.texi (Function Attributes): Typo.
3904 2014-02-03 Cong Hou <congh@google.com>
3906 PR tree-optimization/60000
3907 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3908 if the vectorized statement is a store. A store statement can only
3909 appear at the end of pattern statements.
3911 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3913 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3914 (ix86_option_override_internal): Default long double to 64-bit for
3915 32-bit Bionic and to 128-bit for 64-bit Bionic.
3917 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3918 TARGET_LONG_DOUBLE_128 is true.
3919 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3921 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3922 (mlong-double-64): Negate -mlong-double-128.
3923 (mlong-double-128): New option.
3925 * config/i386/i386-c.c (ix86_target_macros): Define
3926 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3928 * doc/invoke.texi: Document -mlong-double-128.
3930 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3932 PR rtl-optimization/60024
3933 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3935 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3937 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3939 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3941 PR rtl-optimization/57662
3942 * sel-sched.c (code_motion_path_driver): Do not mark already not
3943 existing blocks in the visiting bitmap.
3945 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3947 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3948 on the insn being emitted.
3950 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3951 Will Deacon <will.deacon@arm.com>
3953 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3955 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3957 * config/arm/arm-tables.opt: Regenerate.
3959 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3961 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3962 for vector types other than V16QImode.
3963 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3964 define_expand, and call altivec_expand_vec_perm_le when producing
3965 code with little endian element order.
3966 (*altivec_vperm_<mode>_internal): New insn having previous
3967 behavior of altivec_vperm_<mode>.
3968 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3969 altivec_expand_vec_perm_le when producing code with little endian
3971 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3972 behavior of altivec_vperm_<mode>_uns.
3974 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3976 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3977 (altivec_vsumsws): Add handling for -maltivec=be with a little
3979 (altivec_vsumsws_direct): New.
3980 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3981 gen_altivec_vsumsws.
3983 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3985 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3986 vtable_pointer_value_to_binfo): New functions.
3987 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3988 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3990 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3992 * config/nios2/nios2.md (load_got_register): Initialize GOT
3993 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3994 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3996 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3998 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3999 preserverd by passthrough, do not propagate the type.
4001 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4003 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
4004 (mips_atomic_assign_expand_fenv): New function.
4005 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
4007 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4009 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
4010 (__builtin_mips_set_fcsr): Likewise.
4011 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
4012 MIPS_USI_FTYPE_VOID.
4013 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
4014 (mips16_expand_set_fcsr): Likewise.
4015 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
4016 (mips16_set_fcsr_stub): Likewise.
4017 (mips16_get_fcsr_one_only_stub): New class.
4018 (mips16_set_fcsr_one_only_stub): Likewise.
4019 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
4020 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
4021 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
4022 (hard_float): New availability predicate.
4023 (mips_builtins): Add get_fcsr and set_fcsr.
4024 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
4025 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
4026 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
4027 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
4028 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
4031 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
4033 * config/mips/mips.c (mips_one_only_stub): New class.
4034 (mips_need_mips16_rdhwr_p): Replace with...
4035 (mips16_rdhwr_stub): ...this new variable.
4036 (mips16_stub_call_address): New function.
4037 (mips16_rdhwr_one_only_stub): New class.
4038 (mips_expand_thread_pointer): Use mips16_stub_call_address.
4039 (mips_output_mips16_rdhwr): Delete.
4040 (mips_finish_stub): New function.
4041 (mips_code_end): Use it to handle rdhwr stubs.
4043 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
4046 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
4047 when calculating size of integer atomic types.
4049 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
4051 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
4053 2014-02-01 Jakub Jelinek <jakub@redhat.com>
4055 PR tree-optimization/60003
4056 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
4057 * profile.c (branch_prob): Use gimple_call_builtin_p
4058 to check for BUILT_IN_SETJMP_RECEIVER.
4059 * tree-inline.c (copy_bb): Call notice_special_calls.
4061 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
4064 * lra-constraints.c (process_alt_operands): Update reload_sum only
4067 2014-01-31 Richard Henderson <rth@redhat.com>
4070 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
4071 until after else_eh is processed.
4073 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4075 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
4076 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
4077 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
4078 in smmintrin.h, remove them.
4079 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
4080 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
4081 * config/i386/i386.md (ROUND_SAE): Fix value.
4082 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
4083 (const48_operand): New.
4084 * config/i386/subst.md (round), (round_expand): Use
4085 const_4_or_8_to_11_operand.
4086 (round_saeonly), (round_saeonly_expand): Use const48_operand.
4088 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
4090 * config/i386/constraints.md (Yk): Swap meaning with k.
4091 * config/i386/i386.md (movhi_internal): Change Yk to k.
4092 (movqi_internal): Ditto.
4093 (*k<logic><mode>): Ditto.
4096 (kandn<mode>): Ditto.
4097 (*<code>hi_1): Ditto.
4098 (*<code>qi_1): Ditto.
4099 (kxnor<mode>): Ditto.
4100 (kortestzhi): Ditto.
4101 (kortestchi): Ditto.
4103 (*one_cmplhi2_1): Ditto.
4104 (*one_cmplqi2_1): Ditto.
4105 * config/i386/sse.md (): Change k to Yk.
4106 (avx512f_load<mode>_mask): Ditto.
4107 (avx512f_blendm<mode>): Ditto.
4108 (avx512f_store<mode>_mask): Ditto.
4109 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
4110 (avx512f_storedqu<mode>_mask): Ditto.
4111 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
4113 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
4114 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
4115 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
4116 (avx512f_maskcmp<mode>3): Ditto.
4117 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
4118 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
4119 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
4120 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
4121 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
4122 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
4123 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
4124 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
4125 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
4126 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
4127 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
4128 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
4129 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
4130 (vec_extract_lo_<mode>_maskm): Ditto.
4131 (vec_extract_hi_<mode>_maskm): Ditto.
4132 (avx512f_vternlog<mode>_mask): Ditto.
4133 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
4134 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
4135 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
4136 (avx512f_<code>v8div16qi2_mask): Ditto.
4137 (avx512f_<code>v8div16qi2_mask_store): Ditto.
4138 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
4139 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
4140 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
4141 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
4142 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
4143 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4144 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
4145 (*avx512pf_scatterpf<mode>df_mask): Ditto.
4146 (avx512cd_maskb_vec_dupv8di): Ditto.
4147 (avx512cd_maskw_vec_dupv16si): Ditto.
4148 (avx512f_vpermi2var<mode>3_maskz): Ditto.
4149 (avx512f_vpermi2var<mode>3_mask): Ditto.
4150 (avx512f_vpermi2var<mode>3_mask): Ditto.
4151 (avx512f_vpermt2var<mode>3_maskz): Ditto.
4152 (*avx512f_gathersi<mode>): Ditto.
4153 (*avx512f_gathersi<mode>_2): Ditto.
4154 (*avx512f_gatherdi<mode>): Ditto.
4155 (*avx512f_gatherdi<mode>_2): Ditto.
4156 (*avx512f_scattersi<mode>): Ditto.
4157 (*avx512f_scatterdi<mode>): Ditto.
4158 (avx512f_compress<mode>_mask): Ditto.
4159 (avx512f_compressstore<mode>_mask): Ditto.
4160 (avx512f_expand<mode>_mask): Ditto.
4161 * config/i386/subst.md (mask): Change k to Yk.
4162 (mask_scalar_merge): Ditto.
4165 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
4167 * doc/extend.texi (Vector Extensions): Document ?: in C++.
4169 2014-01-31 Richard Biener <rguenther@suse.de>
4172 * builtins.c (fold_builtin_memory_op): Make sure to not
4173 use a floating-point mode or a boolean or enumeral type for
4176 2014-01-30 DJ Delorie <dj@redhat.com>
4178 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4179 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4180 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4181 whenever main() has an epilogue.
4183 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4185 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4186 unused variable "field".
4187 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4188 (vsx_mergeh_<mode>): Likewise.
4189 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4190 (altivec_vmrghh): Likewise.
4191 (altivec_vmrghw): Likewise.
4192 (altivec_vmrglb): Likewise.
4193 (altivec_vmrglh): Likewise.
4194 (altivec_vmrglw): Likewise.
4195 (altivec_vspltb): Add missing uses.
4196 (altivec_vsplth): Likewise.
4197 (altivec_vspltw): Likewise.
4198 (altivec_vspltsf): Likewise.
4200 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4203 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4204 frame related instructions.
4206 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4208 PR rtl-optimization/59959
4209 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4210 any reload of register whose subreg is invalid.
4212 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4214 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4215 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4216 Add missing return type - void.
4218 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4220 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4221 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4222 remove element index adjustment for endian (now handled in vsx.md
4224 (altivec_expand_vec_perm_const): Use
4225 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4226 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4227 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4228 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4229 define_expand and a new define_insn *altivec_vspltb_internal;
4230 adjust for -maltivec=be on a little endian target.
4231 (altivec_vspltb_direct): New.
4232 (altivec_vsplth): Divide into a define_expand and a new
4233 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4234 little endian target.
4235 (altivec_vsplth_direct): New.
4236 (altivec_vspltw): Divide into a define_expand and a new
4237 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4238 little endian target.
4239 (altivec_vspltw_direct): New.
4240 (altivec_vspltsf): Divide into a define_expand and a new
4241 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4242 a little endian target.
4244 2014-01-30 Richard Biener <rguenther@suse.de>
4246 PR tree-optimization/59993
4247 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4248 can propagate form the earlier stmt and avoid the transform
4249 when the intermediate result is needed.
4251 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4253 * README.Portability: Fix typo.
4255 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4257 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4258 comparison_operator with ordered_comparison_operator.
4260 2014-01-30 Nick Clifton <nickc@redhat.com>
4262 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4263 Rename to mn10300_store_multiple_regs.
4264 * config/mn10300/mn10300.c: Likewise.
4265 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4266 store_multiple_regs.
4267 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4268 Call mn10300_store_multiple_regs.
4270 2014-01-30 Nick Clifton <nickc@redhat.com>
4271 DJ Delorie <dj@redhat.com>
4273 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4274 %fp 2 to keep registers after it properly word-aligned.
4275 (rl78_alloc_physical_registers_umul): Handle the case where both
4276 input operands are the same.
4278 2014-01-30 Richard Biener <rguenther@suse.de>
4280 PR tree-optimization/59903
4281 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4284 2014-01-30 Jason Merrill <jason@redhat.com>
4287 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4290 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4292 2014-01-30 Richard Biener <rguenther@suse.de>
4294 PR tree-optimization/59951
4295 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4297 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4300 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4301 SFmode to DFmode case.
4303 2014-01-29 DJ Delorie <dj@redhat.com>
4305 * config/msp430/msp430.opt (-minrt): New.
4306 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4308 (ENDFILE_SPEC): Likewise.
4310 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4312 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4313 (estimate_function_body_sizes): Use it.
4315 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4318 * dwarf2out.c (is_cxx_auto): New.
4319 (is_base_type): Use it.
4320 (gen_type_die_with_usage): Likewise.
4322 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4324 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4325 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4326 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4327 -maltivec=be with LE targets.
4328 (vsx_mergeh_<mode>): Likewise.
4329 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4330 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4331 (altivec_vmrghb): Replace with define_expand and new
4332 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4333 (altivec_vmrghb_direct): New define_insn.
4334 (altivec_vmrghh): Replace with define_expand and new
4335 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4336 (altivec_vmrghh_direct): New define_insn.
4337 (altivec_vmrghw): Replace with define_expand and new
4338 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4339 (altivec_vmrghw_direct): New define_insn.
4340 (*altivec_vmrghsf): Adjust for endianness.
4341 (altivec_vmrglb): Replace with define_expand and new
4342 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4343 (altivec_vmrglb_direct): New define_insn.
4344 (altivec_vmrglh): Replace with define_expand and new
4345 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4346 (altivec_vmrglh_direct): New define_insn.
4347 (altivec_vmrglw): Replace with define_expand and new
4348 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4349 (altivec_vmrglw_direct): New define_insn.
4350 (*altivec_vmrglsf): Adjust for endianness.
4351 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4352 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4353 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4354 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4355 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4356 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4357 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4358 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4360 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4362 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4363 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4366 2014-01-29 Richard Biener <rguenther@suse.de>
4368 PR tree-optimization/58742
4369 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4370 associate_pointerplus_align.
4371 (associate_pointerplus_diff): New function.
4372 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4373 and associate_pointerplus_diff.
4375 2014-01-29 Richard Biener <rguenther@suse.de>
4377 * lto-streamer.h (LTO_major_version): Bump to 3.
4378 (LTO_minor_version): Reset to 0.
4380 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4382 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4383 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4384 (arm_file_start): Generate correct asm header for armv7ve.
4385 * config/arm/bpabi.h: Add multilib support for armv7ve.
4386 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4387 and cortex-a15 to armv7ve.
4388 * config/arm/t-aprofile: Add multilib support for armv7ve.
4389 * doc/invoke.texi: Document -march=armv7ve.
4391 2014-01-29 Richard Biener <rguenther@suse.de>
4393 PR tree-optimization/58742
4394 * tree-ssa-forwprop.c (associate_plusminus): Return true
4395 if we changed sth, defer EH cleanup to ...
4396 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4397 (simplify_mult): New function.
4399 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4402 PR tree-optimization/59920
4403 * tree.c (build_common_builtin_nodes): Remove
4404 __builtin_setjmp_dispatcher initialization.
4405 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4406 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4407 instead of gsi_after_labels + manually skipping debug stmts.
4408 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4409 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4410 * tree-inline.c (copy_edges_for_bb): Remove
4411 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4412 argument. Ignore computed_goto_p stmts. Don't call
4413 make_abnormal_goto_edges. If a call might need abnormal edges
4414 for non-local gotos, see if it already has an edge to
4415 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4416 with true argument, don't do anything then, otherwise add
4417 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4418 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4420 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4421 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4422 (lower_stmt): Don't set data->calls_builtin_setjmp.
4423 (lower_builtin_setjmp): Adjust comment.
4424 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4425 * tree-cfg.c (found_computed_goto): Remove.
4426 (factor_computed_gotos): Remove.
4427 (make_goto_expr_edges): Return bool, true for computed gotos.
4428 Don't call make_abnormal_goto_edges.
4429 (build_gimple_cfg): Don't set found_computed_goto, don't call
4430 factor_computed_gotos.
4431 (computed_goto_p): No longer static.
4432 (make_blocks): Don't set found_computed_goto.
4433 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4434 (make_edges): If make_goto_expr_edges returns true, push bb
4435 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4436 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4437 vector. Record mapping between bbs and OpenMP regions if there
4438 are any, adjust make_gimple_omp_edges caller. Call
4439 handle_abnormal_edges.
4440 (make_abnormal_goto_edges): Remove.
4441 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4442 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4443 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4444 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4445 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4446 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4447 filling *region also set *region_idx to (*region)->entry->index.
4450 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4451 For REGs set ORIGINAL_REGNO.
4453 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4455 * doc/md.texi: Mention that a target shouldn't implement
4456 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4459 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4461 PR tree-optimization/59594
4462 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4463 a copy of the datarefs vector rather than the vector itself.
4465 2014-01-28 Jason Merrill <jason@redhat.com>
4468 * dwarf2out.c (auto_die): New static.
4469 (gen_type_die_with_usage): Handle C++1y 'auto'.
4470 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4473 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4476 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4477 (SPEC_X32): Likewise.
4478 (SPEC_64): Likewise.
4479 * config/i386/i386.c (ix86_option_override_internal): Turn off
4480 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4482 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4483 * config/i386/i386.h (TARGET_16BIT): New macro.
4484 (TARGET_16BIT_P): Likewise.
4485 * config/i386/i386.opt: Add m16.
4486 * doc/invoke.texi: Document -m16.
4488 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4490 PR preprocessor/59935
4491 * input.c (location_get_source_line): Bail out on when line number
4492 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4494 2014-01-28 Richard Biener <rguenther@suse.de>
4496 PR tree-optimization/58742
4497 * tree-ssa-forwprop.c (associate_plusminus): Handle
4498 pointer subtraction of the form (T)(P + A) - (T)P.
4500 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4502 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4505 2014-01-28 Richard Biener <rguenther@suse.de>
4508 2014-01-28 Richard Biener <rguenther@suse.de>
4510 PR rtl-optimization/45364
4511 PR rtl-optimization/59890
4512 * var-tracking.c (local_get_addr_clear_given_value): Handle
4513 already cleared slot.
4514 (val_reset): Handle not allocated local_get_addr_cache.
4515 (vt_find_locations): Use post-order on the inverted CFG.
4517 2014-01-28 Richard Biener <rguenther@suse.de>
4519 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4521 2014-01-28 Richard Biener <rguenther@suse.de>
4523 PR rtl-optimization/45364
4524 PR rtl-optimization/59890
4525 * var-tracking.c (local_get_addr_clear_given_value): Handle
4526 already cleared slot.
4527 (val_reset): Handle not allocated local_get_addr_cache.
4528 (vt_find_locations): Use post-order on the inverted CFG.
4530 2014-01-28 Alan Modra <amodra@gmail.com>
4532 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4533 * configure.ac <recursive call for build != host>: Define
4534 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4535 and LD_FOR_BUILD too.
4536 * configure: Regenerate.
4538 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4540 * config/i386/i386.c (get_builtin_code_for_version): Separate
4541 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4542 Broadwell from Haswell.
4544 2014-01-27 Steve Ellcey <sellcey@mips.com>
4546 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4547 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4548 * config/mips/mips.c (mips_option_override): Change setting
4550 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4551 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4552 Change from Mask to Var.
4554 2014-01-27 Jeff Law <law@redhat.com>
4556 * ipa-inline.c (inline_small_functions): Fix typo.
4558 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4560 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4561 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4562 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4563 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4564 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4565 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4566 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4567 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4568 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4569 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4570 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4571 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4572 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4573 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4574 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4575 (_mm512_storeu_epi64): Ditto.
4576 (_mm512_cmpge_epi32_mask): Ditto.
4577 (_mm512_cmpge_epu32_mask): Ditto.
4578 (_mm512_cmpge_epi64_mask): Ditto.
4579 (_mm512_cmpge_epu64_mask): Ditto.
4580 (_mm512_cmple_epi32_mask): Ditto.
4581 (_mm512_cmple_epu32_mask): Ditto.
4582 (_mm512_cmple_epi64_mask): Ditto.
4583 (_mm512_cmple_epu64_mask): Ditto.
4584 (_mm512_cmplt_epi32_mask): Ditto.
4585 (_mm512_cmplt_epu32_mask): Ditto.
4586 (_mm512_cmplt_epi64_mask): Ditto.
4587 (_mm512_cmplt_epu64_mask): Ditto.
4588 (_mm512_cmpneq_epi32_mask): Ditto.
4589 (_mm512_cmpneq_epu32_mask): Ditto.
4590 (_mm512_cmpneq_epi64_mask): Ditto.
4591 (_mm512_cmpneq_epu64_mask): Ditto.
4592 (_mm512_expand_pd): Ditto.
4593 (_mm512_expand_ps): Ditto.
4594 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4595 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4596 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4597 * config/i386/i386.c (ix86_builtins): Add
4598 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4599 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4600 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4601 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4602 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4603 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4604 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4605 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4606 IX86_BUILTIN_PMOVUSQW512_MEM.
4607 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4608 __builtin_ia32_pmovsqd512mem_mask,
4609 __builtin_ia32_pmovqd512mem_mask,
4610 __builtin_ia32_pmovusqw512mem_mask,
4611 __builtin_ia32_pmovsqw512mem_mask,
4612 __builtin_ia32_pmovqw512mem_mask,
4613 __builtin_ia32_pmovusdw512mem_mask,
4614 __builtin_ia32_pmovsdw512mem_mask,
4615 __builtin_ia32_pmovdw512mem_mask,
4616 __builtin_ia32_pmovqb512mem_mask,
4617 __builtin_ia32_pmovusqb512mem_mask,
4618 __builtin_ia32_pmovsqb512mem_mask,
4619 __builtin_ia32_pmovusdb512mem_mask,
4620 __builtin_ia32_pmovsdb512mem_mask,
4621 __builtin_ia32_pmovdb512mem_mask.
4622 (bdesc_args): Add __builtin_ia32_expanddf512,
4623 __builtin_ia32_expandsf512.
4624 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4625 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4626 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4627 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4628 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4629 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4630 (avx512f_<code>v8div16qi2_mask_store): This.
4631 (avx512f_expand<mode>): New.
4633 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4635 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4637 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4638 (_mm512_prefetch_i32scatter_pd): Ditto.
4639 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4640 (_mm512_prefetch_i64scatter_pd): Ditto.
4641 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4642 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4643 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4644 (_mm512_prefetch_i32scatter_ps): Ditto.
4645 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4646 (_mm512_prefetch_i64scatter_ps): Ditto.
4647 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4648 * config/i386/i386-builtin-types.def: Define
4649 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4650 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4651 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4652 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4653 IX86_BUILTIN_SCATTERPFQPD.
4654 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4655 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4656 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4657 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4658 __builtin_ia32_scatterpfqps.
4659 (ix86_expand_builtin): Expand new built-ins.
4660 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4661 fix memory access data type.
4662 (*avx512pf_gatherpf<mode>_mask): Ditto.
4663 (*avx512pf_gatherpf<mode>): Ditto.
4664 (avx512pf_scatterpf<mode>): Ditto.
4665 (*avx512pf_scatterpf<mode>_mask): Ditto.
4666 (*avx512pf_scatterpf<mode>): Ditto.
4667 (GATHER_SCATTER_SF_MEM_MODE): New.
4668 (avx512pf_gatherpf<mode>df): Ditto.
4669 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4670 (*avx512pf_scatterpf<mode>df): Ditto.
4672 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4675 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4676 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4679 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4681 * common/config/arm/arm-common.c
4682 (arm_rewrite_mcpu): Handle multiple names.
4684 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4686 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4688 * gimple-builder.h (create_gimple_tmp): Delete.
4690 2014-01-27 Christian Bruel <christian.bruel@st.com>
4692 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4695 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4697 * config/pa/pa.md (call): Generate indirect long calls to non-local
4698 functions when outputing 32-bit code.
4699 (call_value): Likewise except for special call to buggy powf function.
4701 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4702 portable runtime and PIC indirect calls.
4703 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4704 and PIC call sequences. Use ldo instead of blr to set return register
4705 in PIC call sequence.
4707 2014-01-25 Walter Lee <walt@tilera.com>
4709 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4710 avoid clobbering a live register.
4712 2014-01-25 Walter Lee <walt@tilera.com>
4714 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4715 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4716 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4717 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4719 2014-01-25 Walter Lee <walt@tilera.com>
4721 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4722 arguments on even registers.
4723 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4725 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4726 (BIGGEST_ALIGNMENT): Ditto.
4727 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4729 2014-01-25 Walter Lee <walt@tilera.com>
4731 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4732 insns before bundling.
4733 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4735 2014-01-25 Walter Lee <walt@tilera.com>
4737 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4738 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4739 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4741 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4743 * config/mips/constraints.md (kl): Delete.
4744 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4745 define expands, using...
4746 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4747 instructions for MIPS16.
4748 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4749 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4751 2014-01-25 Walter Lee <walt@tilera.com>
4753 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4757 2014-01-25 Walter Lee <walt@tilera.com>
4759 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4760 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4762 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4764 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4767 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4769 * print-rtl.c (in_call_function_usage): New var.
4770 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4771 EXPR_LIST mode as mode and not as reg note name.
4774 * cfgloopmanip.c (copy_loop_info): If
4775 loop->warned_aggressive_loop_optimizations, make sure
4776 the flag is set in target loop too.
4778 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4780 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4782 * builtins.def: Likewise.
4783 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4784 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4785 * ira.c (ira_setup_eliminable_regset): Likewise.
4786 * omp-low.c (gate_expand_omp): Likewise.
4787 (execute_lower_omp): Likewise.
4788 (diagnose_sb_0): Likewise.
4789 (gate_diagnose_omp_blocks): Likewise.
4790 (simd_clone_clauses_extract): Likewise.
4793 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4795 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4796 correction for little endian...
4797 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4800 2014-01-24 Jeff Law <law@redhat.com>
4802 PR tree-optimization/59919
4803 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4804 for non-returning calls.
4806 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4808 * common/config/aarch64/aarch64-common.c
4809 (aarch64_rewrite_mcpu): Handle multiple names.
4810 * config/aarch64/aarch64.h
4811 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4813 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4815 * input.c (add_file_to_cache_tab): Handle the case where fopen
4818 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4821 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4822 from push operand if code of push isn't PRE_DEC.
4824 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4827 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4828 -mquad-memory-atomic. Update -mquad-memory documentation to say
4829 it is only used for non-atomic loads/stores.
4831 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4832 -mquad-memory or -mquad-memory-atomic switches.
4834 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4835 -mquad-memory-atomic to ISA 2.07 support.
4837 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4838 to separate support of normal quad word memory operations (ldq, stq)
4839 from the atomic quad word memory operations.
4841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4842 support to separate non-atomic quad word operations from atomic
4843 quad word operations. Disable non-atomic quad word operations in
4844 little endian mode so that we don't have to swap words after the
4845 load and before the store.
4846 (quad_load_store_p): Add comment about atomic quad word support.
4847 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4848 options printed with -mdebug=reg.
4850 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4851 -mquad-memory-atomic as the test for whether we have quad word
4852 atomic instructions.
4853 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4854 or -mp8-vector are used, allow byte/half-word atomic operations.
4856 * config/rs6000/sync.md (load_lockedti): Insure that the address
4857 is a proper indexed or indirect address for the lqarx instruction.
4858 On little endian systems, swap the hi/lo registers after the lqarx
4860 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4861 insure the address is valid for the lqarx instruction.
4862 (store_conditionalti): Insure that the address is a proper indexed
4863 or indirect address for the stqcrx. instruction. On little endian
4864 systems, swap the hi/lo registers before doing the stqcrx.
4866 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4867 insure the address is valid for the stqcrx. instruction.
4869 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4870 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4871 type of quad memory support is available.
4873 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4876 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4877 there is a danger of looping.
4879 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4881 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4882 force flag_ira_loop_pressure if set via command line.
4884 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4886 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4887 (ashr_simd): New builtin handling DI mode.
4888 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4889 (aarch64_sshr_simddi): New match pattern.
4890 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4891 (vshrd_n_s64): Likewise.
4892 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4894 2014-01-23 Nick Clifton <nickc@redhat.com>
4896 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4897 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4898 favour of mcu specific scripts.
4899 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4902 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4903 Alex Velenko <Alex.Velenko@arm.com>
4905 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4906 (vaddv_s16): Likewise.
4907 (vaddv_s32): Likewise.
4908 (vaddv_u8): Likewise.
4909 (vaddv_u16): Likewise.
4910 (vaddv_u32): Likewise.
4911 (vaddvq_s8): Likewise.
4912 (vaddvq_s16): Likewise.
4913 (vaddvq_s32): Likewise.
4914 (vaddvq_s64): Likewise.
4915 (vaddvq_u8): Likewise.
4916 (vaddvq_u16): Likewise.
4917 (vaddvq_u32): Likewise.
4918 (vaddvq_u64): Likewise.
4919 (vaddv_f32): Likewise.
4920 (vaddvq_f32): Likewise.
4921 (vaddvq_f64): Likewise.
4922 (vmaxv_f32): Likewise.
4923 (vmaxv_s8): Likewise.
4924 (vmaxv_s16): Likewise.
4925 (vmaxv_s32): Likewise.
4926 (vmaxv_u8): Likewise.
4927 (vmaxv_u16): Likewise.
4928 (vmaxv_u32): Likewise.
4929 (vmaxvq_f32): Likewise.
4930 (vmaxvq_f64): Likewise.
4931 (vmaxvq_s8): Likewise.
4932 (vmaxvq_s16): Likewise.
4933 (vmaxvq_s32): Likewise.
4934 (vmaxvq_u8): Likewise.
4935 (vmaxvq_u16): Likewise.
4936 (vmaxvq_u32): Likewise.
4937 (vmaxnmv_f32): Likewise.
4938 (vmaxnmvq_f32): Likewise.
4939 (vmaxnmvq_f64): Likewise.
4940 (vminv_f32): Likewise.
4941 (vminv_s8): Likewise.
4942 (vminv_s16): Likewise.
4943 (vminv_s32): Likewise.
4944 (vminv_u8): Likewise.
4945 (vminv_u16): Likewise.
4946 (vminv_u32): Likewise.
4947 (vminvq_f32): Likewise.
4948 (vminvq_f64): Likewise.
4949 (vminvq_s8): Likewise.
4950 (vminvq_s16): Likewise.
4951 (vminvq_s32): Likewise.
4952 (vminvq_u8): Likewise.
4953 (vminvq_u16): Likewise.
4954 (vminvq_u32): Likewise.
4955 (vminnmv_f32): Likewise.
4956 (vminnmvq_f32): Likewise.
4957 (vminnmvq_f64): Likewise.
4959 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4961 * config/aarch64/aarch64-simd.md
4962 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4963 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4964 (*aarch64_mul3_elt<mode>): Likewise.
4965 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4966 (*aarch64_mul3_elt_to_64v2df): Likewise.
4967 (*aarch64_mla_elt<mode>): Likewise.
4968 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4969 (*aarch64_mls_elt<mode>): Likewise.
4970 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4971 (*aarch64_fma4_elt<mode>): Likewise.
4972 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4973 (*aarch64_fma4_elt_to_64v2df): Likewise.
4974 (*aarch64_fnma4_elt<mode>): Likewise.
4975 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4976 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4977 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4978 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4979 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4980 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4981 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4982 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4983 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4985 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4987 * config/aarch64/aarch64-simd.md
4988 (aarch64_be_checked_get_lane<mode>): New define_expand.
4989 * config/aarch64/aarch64-simd-builtins.def
4990 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4991 New builtin definition.
4992 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4993 Use new safe be builtin.
4995 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4997 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4999 (aarch64_be_st1<mode>): Likewise.
5000 (aarch_ld1<VALL:mode>): Define_expand modified.
5001 (aarch_st1<VALL:mode>): Likewise.
5002 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
5003 (UNSPEC_ST1): Likewise.
5005 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
5007 * config/microblaze/microblaze.md: Add trap insn and attribute
5009 2014-01-23 Dodji Seketeli <dodji@redhat.com>
5011 PR preprocessor/58580
5012 * input.h (location_get_source_line): Take an additional line_size
5014 (void diagnostics_file_cache_fini): Declare new function.
5015 * input.c (struct fcache): New type.
5016 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
5017 New static constants.
5018 (diagnostic_file_cache_init, total_lines_num)
5019 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
5020 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
5021 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
5022 (get_next_line, read_next_line, goto_next_line, read_line_num):
5023 New static function definitions.
5024 (diagnostic_file_cache_fini): New function.
5025 (location_get_source_line): Take an additional output line_len
5026 parameter. Re-write using lookup_or_add_file_to_cache_tab and
5028 * diagnostic.c (diagnostic_finish): Call
5029 diagnostic_file_cache_fini.
5030 (adjust_line): Take an additional input parameter for the length
5031 of the line, rather than calculating it with strlen.
5032 (diagnostic_show_locus): Adjust the use of
5033 location_get_source_line and adjust_line with respect to their new
5034 signature. While displaying a line now, do not stop at the first
5035 null byte. Rather, display the zero byte as a space and keep
5036 going until we reach the size of the line.
5037 * Makefile.in: Add vec.o to OBJS-libcommon
5039 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5040 Ilya Tocar <ilya.tocar@intel.com>
5042 * config/i386/avx512fintrin.h (_mm512_kmov): New.
5043 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
5044 (__builtin_ia32_kmov16): Ditto.
5045 * config/i386/i386.md (UNSPEC_KMOV): New.
5048 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
5050 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
5051 (_mm512_storeu_si512): Ditto.
5053 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
5056 * rtl.h (get_referenced_operands): Declare.
5057 * recog.c (get_referenced_operands): New function.
5058 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
5059 operands have been referenced when recording LO_SUM references.
5061 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
5063 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
5065 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5067 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
5068 Enable for generic and recent AMD targets.
5070 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
5072 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
5073 ARG_SIZE note when adjustment was eliminated.
5075 2014-01-22 Jeff Law <law@redhat.com>
5077 PR tree-optimization/59597
5078 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
5079 in file. Accept new argument REGISTERING and use it to modify
5080 dump output appropriately.
5081 (register_jump_thread): Corresponding changes.
5082 (mark_threaded_blocks): Reinstate code to cancel unprofitable
5083 thread paths involving joiner blocks. Add code to dump cancelled
5084 jump threading paths.
5086 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
5088 PR rtl-optimization/59477
5089 * lra-constraints.c (inherit_in_ebb): Process call for living hard
5090 regs. Update reloads_num and potential_reload_hard_regs for all insns.
5092 2014-01-22 Tom Tromey <tromey@redhat.com>
5094 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
5096 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
5098 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5100 PR rtl-optimization/59896
5101 * lra-constraints.c (process_alt_operands): Check unused note for
5102 matched operands of insn with no output reloads.
5104 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
5106 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
5107 (mips_move_from_gpr_cost): Likewise.
5109 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
5111 PR rtl-optimization/59858
5112 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
5113 ira_class_hard_regs_num.
5114 (process_alt_operands): Increase reject for dying matched operand.
5116 2014-01-21 Jakub Jelinek <jakub@redhat.com>
5119 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
5120 smaller than size, perform several stores or loads and stores
5121 at dst + count - size to store or copy all of size bytes, rather
5122 than just last modesize bytes.
5124 2014-01-20 DJ Delorie <dj@redhat.com>
5126 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
5127 that CLOBBERs are REGs before propogating their values.
5129 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
5132 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
5133 (cgraph_inline_failed_type): New function.
5134 * cgraph.h (DEFCIFCODE): Add type.
5135 (cgraph_inline_failed_type_t): New enum.
5136 (cgraph_inline_failed_type): New prototype.
5137 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
5138 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
5139 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
5140 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
5141 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
5142 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
5143 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
5144 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
5145 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
5146 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
5147 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
5148 OPTIMIZATION_MISMATCH.
5149 * tree-inline.c (expand_call_inline): Emit errors during
5150 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
5152 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5155 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
5156 mode attribute in insn output.
5158 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
5160 * output.h (output_constant): Delete.
5161 * varasm.c (output_constant): Make private.
5163 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
5165 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
5167 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5170 * tree.h (fold_builtin_strcat): New prototype.
5171 * builtins.c (fold_builtin_strcat): No longer static. Add len
5172 argument, if non-NULL, don't call c_strlen. Optimize
5173 directly into __builtin_memcpy instead of __builtin_strcpy.
5174 (fold_builtin_2): Adjust fold_builtin_strcat caller.
5175 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
5177 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5179 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5180 for SImode_address_operand operands, having only a REG argument.
5182 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5184 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5185 loader name using mbig-endian.
5186 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5188 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5190 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5194 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5196 * config/aarch64/aarch64-protos.h
5197 (aarch64_cannot_change_mode_class_ptr): Declare.
5198 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5199 aarch64_cannot_change_mode_class_ptr): New.
5200 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5201 backend hook aarch64_cannot_change_mode_class.
5203 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5205 * common/config/aarch64/aarch64-common.c
5206 (aarch64_handle_option): Don't handle any option order logic here.
5207 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5208 selected_cpu, warn on architecture version mismatch.
5209 (aarch64_override_options): Fix parsing order for option strings.
5211 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5212 Iain Sandoe <iain@codesourcery.com>
5215 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5216 warning. Amend comment to reflect current functionality.
5218 2014-01-20 Richard Biener <rguenther@suse.de>
5221 * builtins.c (fold_builtin_strcat): Remove case better handled
5222 by tree-ssa-strlen.c.
5224 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5226 * config/aarch64/aarch64.opt
5227 (mcpu, march, mtune): Make case-insensitive.
5229 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5232 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5233 if operands[1] is a REG or ZERO_EXTEND of a REG.
5235 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5237 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5239 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5241 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5242 long non-pic millicode calls.
5244 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5246 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5248 2014-01-19 Kito Cheng <kito@0xlab.org>
5250 * builtins.c (expand_movstr): Check movstr expand done or fail.
5252 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5253 H.J. Lu <hongjiu.lu@intel.com>
5256 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5257 to DImode for zero-extended addresses.
5259 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5261 PR rtl-optimization/57763
5262 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5263 on the new indirect jump_insn and increment LABEL_NUSES (label).
5265 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5269 * config.gcc (x86_archs): New variable.
5270 (x86_64_archs): Likewise.
5271 (x86_cpus): Likewise.
5272 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5273 --with-arch/--with-cpu= options.
5274 Support --with-arch=/--with-cpu={nehalem,westmere,
5275 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5277 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5279 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5280 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5282 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5284 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5286 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5289 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5290 clear cpp_get_options (parse_in)->warn_unused_macros for
5291 ix86_target_macros_internal with cpp_define.
5293 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5295 * jump.c (delete_related_insns): Keep (use (insn))s.
5296 * reorg.c (redundant_insn): Check for barriers too.
5298 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5300 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5302 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5304 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5305 call to $$dyncall when TARGET_LONG_CALLS is true.
5307 2014-01-17 Jeff Law <law@redhat.com>
5309 * ree.c (combine_set_extension): Temporarily disable test for
5310 changing number of hard registers.
5312 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5315 * ipa-inline-analysis.c (inline_free_summary):
5316 Do not free summary of aliases.
5318 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5321 * gimplify.c (gimplify_expr): Use create_tmp_var
5322 instead of create_tmp_var_raw. If cond doesn't have
5323 integral type, don't add the IFN_ANNOTATE builtin at all.
5325 2014-01-17 Martin Jambor <mjambor@suse.cz>
5328 * ipa-cp.c (prev_edge_clone): New variable.
5329 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5330 Also resize prev_edge_clone vector.
5331 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5332 (ipcp_edge_removal_hook): New function.
5333 (ipcp_driver): Register ipcp_edge_removal_hook.
5335 2014-01-17 Andrew Pinski <apinski@cavium.com>
5336 Steve Ellcey <sellcey@mips.com>
5339 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5342 2014-01-17 Jeff Law <law@redhat.com>
5345 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5346 so that pass_ccp runs first.
5348 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5350 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5351 (ix86_adjust_cost): Use !TARGET_XXX.
5352 (do_reorder_for_imul): Likewise.
5353 (swap_top_of_ready_list): Likewise.
5354 (ix86_sched_reorder): Likewise.
5356 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5358 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5359 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5360 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5361 (intel_memset): New. Duplicate slm_memset.
5362 (intel_cost): New. Duplicate slm_cost.
5363 (m_INTEL): New macro.
5364 (processor_target_table): Add "intel".
5365 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5366 with PROCESSOR_INTEL for "intel".
5367 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5368 PROCESSOR_SILVERMONT.
5369 (ix86_issue_rate): Likewise.
5370 (ix86_adjust_cost): Likewise.
5371 (ia32_multipass_dfa_lookahead): Likewise.
5372 (swap_top_of_ready_list): Likewise.
5373 (ix86_sched_reorder): Likewise.
5374 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5375 instead of TARGET_OPT_AGU.
5376 * config/i386/i386.h (TARGET_INTEL): New.
5377 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5378 (processor_type): Add PROCESSOR_INTEL.
5379 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5380 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5382 2014-01-17 Marek Polacek <polacek@redhat.com>
5385 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5388 2014-01-17 Richard Biener <rguenther@suse.de>
5390 PR tree-optimization/46590
5391 * opts.c (default_options_table): Add entries for
5392 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5393 all enabled at -O1 but not for -Og.
5394 * common.opt (fbranch-count-reg): Remove Init(1).
5395 (fmove-loop-invariants): Likewise.
5396 (ftree-pta): Likewise.
5398 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5400 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5401 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5402 decls to at least the GCC 4.8 used alignments.
5405 * tree-nested.c (convert_nonlocal_reference_stmt,
5406 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5407 of GIMPLE_BIND stmts, adjust associated decls.
5409 2014-01-17 Richard Biener <rguenther@suse.de>
5411 PR tree-optimization/46590
5412 * vec.h (vec<>::bseach): New member function implementing
5413 binary search according to C89 bsearch.
5414 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5415 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5416 bitmap pointer again. Make accesses_in_loop a flat array.
5417 (mem_ref_obstack): New global.
5418 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5419 (mark_ref_stored): Likewise.
5420 (ref_indep_loop_p_2): Likewise.
5421 (set_ref_stored_in_loop): New helper function.
5422 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5423 (memref_free): Adjust.
5424 (record_mem_ref_loc): Simplify.
5425 (gather_mem_refs_stmt): Adjust.
5426 (sort_locs_in_loop_postorder_cmp): New function.
5427 (analyze_memory_references): Sort accesses_in_loop after
5428 loop postorder number.
5429 (find_ref_loc_in_loop_cmp): New function.
5430 (for_all_locs_in_loop): Find relevant cluster of locs in
5431 accesses_in_loop and iterate without recursion.
5432 (execute_sm): Avoid uninit warning.
5433 (struct ref_always_accessed): Simplify.
5434 (ref_always_accessed::operator ()): Likewise.
5435 (ref_always_accessed_p): Likewise.
5436 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5437 loop postorder numbers here.
5438 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5441 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5444 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5445 on decls for which assemble_alias has been called.
5447 2014-01-17 Nick Clifton <nickc@redhat.com>
5449 * config/msp430/msp430.opt: (mcpu): New option.
5450 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5451 (msp430_option_override): Parse target_cpu. If the MCU name
5452 matches a generic string, clear target_mcu.
5453 (msp430_attr): Allow numeric interrupt values up to 63.
5454 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5455 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5457 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5459 * config/msp430/msp430.md (popm): Use %J rather than %I.
5460 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5461 (addhi_cy_i): Use immediate_operand for operand 2.
5462 * doc/invoke.texi: Document -mcpu option.
5464 2014-01-17 Richard Biener <rguenther@suse.de>
5466 PR rtl-optimization/38518
5467 * df.h (df_analyze_loop): Declare.
5468 * df-core.c: Include cfgloop.h.
5469 (df_analyze_1): Split out main part of df_analyze.
5470 (df_analyze): Adjust.
5471 (loop_inverted_post_order_compute): New function.
5472 (loop_post_order_compute): Likewise.
5473 (df_analyze_loop): New function avoiding whole-function
5475 * loop-invariant.c (find_defs): Use df_analyze_loop.
5476 (find_invariants): Adjust.
5477 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5479 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5481 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5482 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5484 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5486 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5487 traversal when removing references.
5489 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5492 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5494 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5497 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5498 pushing a reload for an autoinc when we had previously reloaded an
5499 inner part of the address.
5501 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5503 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5505 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5506 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5507 when not giving up or versioning for alias only because of
5509 (vect_analyze_data_ref_dependences): Set to true.
5510 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5512 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5513 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5517 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5520 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5521 operand 0 predicate for gathers, use a new pseudo as subtarget.
5523 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5526 * lra-constraints.c (process_alt_operands): Add printing debug info.
5527 Check absence of input/output reloads for matched operands too.
5529 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5531 PR rtl-optimization/59835
5532 * ira.c (ira_init_register_move_cost): Increase cost for
5535 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5537 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5539 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5542 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5543 non-register objects. Use gen_(high/low)part more consistently.
5546 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5549 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5550 endian support, remove tests for WORDS_BIG_ENDIAN.
5551 (p8_mfvsrd_3_<mode>): Likewise.
5552 (reload_gpr_from_vsx<mode>): Likewise.
5553 (reload_gpr_from_vsxsf): Likewise.
5554 (p8_mfvsrd_4_disf): Likewise.
5556 2014-01-16 Richard Biener <rguenther@suse.de>
5558 PR rtl-optimization/46590
5559 * lcm.c (compute_antinout_edge): Use postorder iteration.
5560 (compute_laterin): Use inverted postorder iteration.
5562 2014-01-16 Nick Clifton <nickc@redhat.com>
5565 * varasm.c (output_constant): Return the number of bytes actually
5567 (output_constructor_array_range): Update the field size with the
5568 number of bytes emitted by output_constant.
5569 (output_constructor_regular_field): Likewise. Also do not
5570 complain if the total number of bytes emitted is now greater
5571 than the expected fieldpos.
5572 * output.h (output_constant): Update prototype and descriptive comment.
5574 2014-01-16 Marek Polacek <polacek@redhat.com>
5577 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5578 it is error_mark_node.
5580 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5582 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5583 VALID_AVX256_REG_OR_OI_MODE.
5585 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5587 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5588 current procedure should be profiled.
5590 2014-01-15 Andrew Pinski <apinski@cavium.com>
5592 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5593 of moving from/to the STACK_REG register class.
5595 2014-01-15 Richard Henderson <rth@redhat.com>
5598 * reginfo.c (global_regs_decl): Globalize.
5599 * rtl.h (global_regs_decl): Declare.
5600 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5601 reserved via global_regs.
5603 2014-01-15 Teresa Johnson <tejohnson@google.com>
5605 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5607 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5609 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5610 and vmulosh rather than call gen_vec_widen_smult_*.
5611 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5612 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5613 (vec_widen_smult_even_v16qi): Likewise.
5614 (vec_widen_umult_even_v8hi): Likewise.
5615 (vec_widen_smult_even_v8hi): Likewise.
5616 (vec_widen_umult_odd_v16qi): Likewise.
5617 (vec_widen_smult_odd_v16qi): Likewise.
5618 (vec_widen_umult_odd_v8hi): Likewise.
5619 (vec_widen_smult_odd_v8hi): Likewise.
5620 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5621 vmuloub rather than call gen_vec_widen_umult_*.
5622 (vec_widen_umult_lo_v16qi): Likewise.
5623 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5624 vmulosb rather than call gen_vec_widen_smult_*.
5625 (vec_widen_smult_lo_v16qi): Likewise.
5626 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5627 rather than call gen_vec_widen_umult_*.
5628 (vec_widen_umult_lo_v8hi): Likewise.
5629 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5630 rather than call gen_vec_widen_smult_*.
5631 (vec_widen_smult_lo_v8hi): Likewise.
5633 2014-01-15 Jeff Law <law@redhat.com>
5635 PR tree-optimization/59747
5636 * ree.c (find_and_remove_re): Properly handle case where a second
5637 eliminated extension requires widening a copy created for elimination
5638 of a prior extension.
5639 (combine_set_extension): Ensure that the number of hard regs needed
5640 for a destination register does not change when we widen it.
5642 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5644 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5645 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5646 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5647 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5648 (avr-*-rtems*): Likewise.
5649 (bfin*-rtems*): Likewise.
5650 (moxie-*-rtems*): Likewise.
5651 (h8300-*-rtems*): Likewise.
5652 (i[34567]86-*-rtems*): Likewise.
5653 (lm32-*-rtems*): Likewise.
5654 (m32r-*-rtems*): Likewise.
5655 (m68k-*-rtems*): Likewise.
5656 (microblaze*-*-rtems*): Likewise.
5657 (mips*-*-rtems*): Likewise.
5658 (powerpc-*-rtems*): Likewise.
5659 (sh-*-rtems*): Likewise.
5660 (sparc-*-rtems*): Likewise.
5661 (sparc64-*-rtems*): Likewise.
5662 (v850-*-rtems*): Likewise.
5663 (m32c-*-rtems*): Likewise.
5665 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5667 PR rtl-optimization/59511
5668 * ira.c (ira_init_register_move_cost): Use memory costs for some
5669 cases of register move cost calculations.
5670 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5671 instead of BB frequency.
5672 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5673 * lra-assigns.c (find_hard_regno_for): Ditto.
5675 2014-01-15 Richard Biener <rguenther@suse.de>
5677 PR tree-optimization/59822
5678 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5679 (vectorizable_load): Use it to hoist defs of uses of invariant
5680 loads out of the loop.
5682 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5683 Kugan Vivekanandarajah <kuganv@linaro.org>
5686 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5689 2014-01-15 Richard Biener <rguenther@suse.de>
5691 PR rtl-optimization/59802
5692 * lcm.c (compute_available): Use inverted postorder to seed
5693 the initial worklist.
5695 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5698 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5699 ADDR_REGS for invalid symrefs in non-PIC code.
5701 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5704 * builtins.c (determine_block_size): Initialize *probable_max_size
5705 even if len_rtx is CONST_INT.
5707 2014-01-14 Andrew Pinski <apinski@cavium.com>
5709 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5710 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5711 (cortexa53_tunings): Likewise.
5712 (aarch64_sched_issue_rate): New function.
5713 (TARGET_SCHED_ISSUE_RATE): Define.
5715 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5717 * ira-costs.c (find_costs_and_classes): Add missed
5718 ira_init_register_move_cost_if_necessary.
5720 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5723 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5725 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5728 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5729 to indicate if type is used for function return value. Warn ABI
5730 change if the vector mode isn't available for function return value.
5731 (ix86_function_arg_advance): Pass false to type_natural_mode.
5732 (ix86_function_arg): Likewise.
5733 (ix86_gimplify_va_arg): Likewise.
5734 (function_arg_32): Don't warn ABI change.
5735 (ix86_function_value): Pass true to type_natural_mode.
5736 (ix86_return_in_memory): Likewise.
5737 (ix86_struct_value_rtx): Removed.
5738 (TARGET_STRUCT_VALUE_RTX): Likewise.
5740 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5742 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5743 converting a conditional jump into a conditional return.
5745 2014-01-14 Richard Biener <rguenther@suse.de>
5747 PR tree-optimization/58921
5748 PR tree-optimization/59006
5749 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5750 hoisting invariant stmts.
5751 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5752 invariant loads on the preheader edge if possible.
5754 2014-01-14 Joey Ye <joey.ye@arm.com>
5756 * doc/plugin.texi (Building GCC plugins): Update to C++.
5758 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5760 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5761 (_mm_rcp28_round_ss): Ditto.
5762 (_mm_rsqrt28_round_sd): Ditto.
5763 (_mm_rsqrt28_round_ss): Ditto.
5764 (_mm_rcp28_sd): Ditto.
5765 (_mm_rcp28_ss): Ditto.
5766 (_mm_rsqrt28_sd): Ditto.
5767 (_mm_rsqrt28_ss): Ditto.
5768 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5769 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5770 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5771 (IX86_BUILTIN_RCP28SD): Ditto.
5772 (IX86_BUILTIN_RCP28SS): Ditto.
5773 (IX86_BUILTIN_RSQRT28SD): Ditto.
5774 (IX86_BUILTIN_RSQRT28SS): Ditto.
5775 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5776 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5777 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5778 (ix86_expand_special_args_builtin): Expand new FTYPE.
5779 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5780 (srcp14<mode>): Make insn unary.
5781 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5782 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5783 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5784 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5785 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5786 Fix rounding: make it SAE only.
5787 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5789 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5791 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5792 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5793 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5794 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5795 (round_saeonly_mask_scalar_operand4): Ditto.
5796 (round_saeonly_mask_scalar_op3): Ditto.
5797 (round_saeonly_mask_scalar_op4): Ditto.
5799 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5801 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5802 Implement -maltivec=be for vec_insert and vec_extract.
5804 2014-01-10 DJ Delorie <dj@redhat.com>
5806 * config/msp430/msp430.md (call_internal): Don't allow memory
5807 references with SP as the base register.
5808 (call_value_internal): Likewise.
5809 * config/msp430/constraints.md (Yc): New. For memory references
5810 that don't use SP as a base register.
5812 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5813 "an integer without a # prefix"
5814 * config/msp430/msp430.md (epilogue_helper): Use it.
5816 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5819 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5820 AVX512F gather builtins.
5821 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5822 on gather decls with INTEGER_TYPE masktype.
5823 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5824 directly into the builtin rather than hoisting it before loop.
5826 PR tree-optimization/59387
5827 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5828 (scev_const_prop): If folded_casts and type has undefined overflow,
5829 use force_gimple_operand instead of force_gimple_operand_gsi and
5830 for each added stmt if it is assign with
5831 arith_code_with_undefined_signed_overflow, call
5832 rewrite_to_defined_overflow.
5833 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5834 gimple-fold.h instead.
5835 (arith_code_with_undefined_signed_overflow,
5836 rewrite_to_defined_overflow): Moved to ...
5837 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5838 rewrite_to_defined_overflow): ... here. No longer static.
5839 Include gimplify-me.h.
5840 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5841 rewrite_to_defined_overflow): New prototypes.
5843 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5845 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5847 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5849 * builtins.c (get_object_alignment_2): Minor tweak.
5850 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5852 2014-01-13 Christian Bruel <christian.bruel@st.com>
5854 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5855 optimized non constant lengths.
5857 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5860 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5861 load as __atomic_load_N if possible.
5863 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5865 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5867 (rs6000_expand_builtin): Adjust call.
5869 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5872 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5873 * config/rs6000/rs6000.c: Include target-globals.h.
5874 (rs6000_set_current_function): Instead of doing target_reinit
5875 unconditionally, use save_target_globals_default_opts and
5876 restore_target_globals.
5878 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5880 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5881 (rs6000_expand_builtin): Handle mffs and mtfsf.
5882 (rs6000_init_builtins): Define mffs and mtfsf.
5883 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5884 (rs6000_mffs): New pattern.
5885 (rs6000_mtfsf): New pattern.
5887 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5889 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5890 Start narrowing with START. Apply candidate-use pair
5891 and check overall cost in narrowing.
5892 (iv_ca_prune): Pass new argument.
5894 2014-01-10 Jeff Law <law@redhat.com>
5897 * ree.c (combine_reaching_defs): Ensure the defining statement
5898 occurs before the extension when optimizing extensions with
5899 different source and destination hard registers.
5901 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5904 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5905 vtables into the type inheritance graph.
5907 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5909 PR rtl-optimization/59754
5910 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5911 modes in the REGNO != REGNO case.
5913 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5915 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5917 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5919 PR tree-optimization/59745
5920 * tree-predcom.c (tree_predictive_commoning_loop): Call
5921 free_affine_expand_cache if giving up because components is NULL.
5923 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5924 GC in payload of target_globals struct instead of allocating them on
5925 the heap and the larger structs separately using GC.
5926 * target-globals.h (struct target_globals): Make regs, hard_regs,
5927 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5928 of GTY((skip)) and change type to void *.
5929 (reset_target_globals): Cast loads from those fields to corresponding
5932 2014-01-10 Steve Ellcey <sellcey@mips.com>
5935 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5936 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5937 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5939 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5942 * aarch64-modes.def (CC_Zmode): New flags mode.
5943 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5944 represents an equality.
5945 (aarch64_get_condition_code): Handle CC_Zmode.
5946 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5948 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5950 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5951 extraction in good case.
5953 2014-01-10 Richard Biener <rguenther@suse.de>
5955 PR tree-optimization/59374
5956 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5957 checking after SLP discovery. Mark stmts not participating
5958 in any SLP instance properly.
5960 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5962 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5963 when handling a SET rtx.
5965 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5967 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5968 (cortex-a57): Likewise.
5969 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5971 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5973 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5974 non-iwmmxt builtins.
5976 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5980 * ipa-devirt.c record_target_from_binfo): Take as argument
5981 stack of binfos and lookup matching one for virtual inheritance.
5982 (possible_polymorphic_call_targets_1): Update.
5984 2014-01-10 Huacai Chen <chenhc@lemote.com>
5986 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5987 kernel strings for Loongson-2E/2F/3A.
5989 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5992 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5993 is_gimple_call before calling gimple_call_internal_p.
5995 2014-01-09 Steve Ellcey <sellcey@mips.com>
5997 * Makefile.in (TREE_FLOW_H): Remove.
5998 (TREE_SSA_H): Add file names from tree-flow.h.
5999 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
6000 * tree.h: Remove tree-flow.h reference.
6001 * hash-table.h: Remove tree-flow.h reference.
6002 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
6003 reference with tree-ssa-loop.h.
6005 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6007 * doc/invoke.texi: Add -maltivec={be,le} options, and document
6008 default element-order behavior for -maltivec.
6009 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
6010 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
6011 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
6012 when targeting big endian, at least for now.
6013 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
6015 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6018 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
6019 var satisfies use_register_for_decl, just take into account type
6020 alignment, rather than decl alignment.
6022 PR tree-optimization/59622
6023 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
6024 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
6025 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
6026 Don't devirtualize for inplace at all. For targets.length () == 1,
6027 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
6029 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6031 * config/i386/i386.md (cpu): Remove the unused btver1.
6033 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
6035 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
6037 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6040 * tree-core.h (struct target_globals): New forward declaration.
6041 (struct tree_target_option): Add globals field.
6042 * tree.h (TREE_TARGET_GLOBALS): Define.
6043 (prepare_target_option_nodes_for_pch): New prototype.
6044 * target-globals.h (struct target_globals): Define even if
6046 * tree.c (prepare_target_option_node_for_pch,
6047 prepare_target_option_nodes_for_pch): New functions.
6048 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
6049 * config/i386/i386.c: Include target-globals.h.
6050 (ix86_set_current_function): Instead of doing target_reinit
6051 unconditionally, use save_target_globals_default_opts and
6052 restore_target_globals.
6054 2014-01-09 Richard Biener <rguenther@suse.de>
6056 PR tree-optimization/59715
6057 * tree-cfg.h (split_critical_edges): Declare.
6058 * tree-cfg.c (split_critical_edges): Export.
6059 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
6061 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
6063 * cfgexpand.c (expand_stack_vars): Optionally disable
6064 asan stack protection.
6065 (expand_used_vars): Likewise.
6066 (partition_stack_vars): Likewise.
6067 * asan.c (asan_emit_stack_protection): Optionally disable
6068 after return stack usage.
6069 (instrument_derefs): Optionally disable memory access instrumentation.
6070 (instrument_builtin_call): Likewise.
6071 (instrument_strlen_call): Likewise.
6072 (asan_protect_global): Optionally disable global variables protection.
6073 * doc/invoke.texi: Added doc for new options.
6074 * params.def: Added new options.
6075 * params.h: Likewise.
6077 2014-01-09 Jakub Jelinek <jakub@redhat.com>
6079 PR rtl-optimization/59724
6080 * ifcvt.c (cond_exec_process_if_block): Don't call
6081 flow_find_head_matching_sequence with 0 longest_match.
6082 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
6083 non-active insns if !stop_after.
6084 (try_head_merge_bb): Revert 2014-01-07 changes.
6086 2014-01-08 Jeff Law <law@redhat.com>
6088 * ree.c (get_sub_rtx): New function, extracted from...
6089 (merge_def_and_ext): Here.
6090 (combine_reaching_defs): Use get_sub_rtx.
6092 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
6094 * cgraph.h (varpool_variable_node): Do not choke on null node.
6096 2014-01-08 Catherine Moore <clm@codesourcery.com>
6098 * config/mips/mips.md (simple_return): Attempt to use JRC
6100 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
6102 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6104 PR rtl-optimization/59137
6105 * reorg.c (steal_delay_list_from_target): Call update_block for
6107 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
6109 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6111 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
6112 two duplicate entries.
6114 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
6117 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
6119 * config/mips/mips.c (mips_truncated_op_cost): New function.
6120 (mips_rtx_costs): Adjust test for BADDU.
6121 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
6123 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
6125 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
6126 (*baddu_si): ...this new pattern.
6128 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6131 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
6133 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
6136 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
6138 (expand_expr, expand_normal): Adjust.
6139 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
6140 inner_reference_p. Use inner_reference_p to expand inner references.
6141 (store_expr): Adjust.
6142 * cfgexpand.c (expand_call_stmt): Adjust.
6144 2014-01-08 Rong Xu <xur@google.com>
6146 * gcov-io.c (gcov_var): Move from gcov-io.h.
6147 (gcov_position): Ditto.
6148 (gcov_is_error): Ditto.
6149 (gcov_rewrite): Ditto.
6150 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
6151 only part to libgcc/libgcov.h.
6153 2014-01-08 Marek Polacek <polacek@redhat.com>
6156 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
6158 2014-01-08 Marek Polacek <polacek@redhat.com>
6161 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
6163 2014-01-08 Jakub Jelinek <jakub@redhat.com>
6165 PR rtl-optimization/59649
6166 * stor-layout.c (get_mode_bounds): For BImode return
6167 0 and STORE_FLAG_VALUE.
6169 2014-01-08 Richard Biener <rguenther@suse.de>
6172 * gimple.h (is_gimple_builtin_call): Remove.
6173 (gimple_builtin_call_types_compatible_p): New.
6174 (gimple_call_builtin_p): New overload.
6175 * gimple.c (is_gimple_builtin_call): Remove.
6176 (validate_call): Rename to ...
6177 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6179 (validate_type): New static function.
6180 (gimple_call_builtin_p): New overload and adjust.
6181 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6182 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6183 (gimple_fold_stmt_to_constant_1): Likewise.
6184 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6186 2014-01-08 Richard Biener <rguenther@suse.de>
6189 * gimplify.c (gimplify_expr): Gimplify register-register type
6190 VIEW_CONVERT_EXPRs to separate stmts.
6192 2014-01-07 Jeff Law <law@redhat.com>
6195 * ree.c (combine_set_extension): Handle case where source
6196 and destination registers in an extension insn are different.
6197 (combine_reaching_defs): Allow source and destination registers
6198 in extension to be different under limited circumstances.
6199 (add_removable_extension): Remove restriction that the
6200 source and destination registers in the extension are the same.
6201 (find_and_remove_re): Emit a copy from the extension's
6202 destination to its source after the defining insn if
6203 the source and destination registers are different.
6206 * ifcvt.c (merge_if_block): If we are merging a block with more than
6207 one successor with a block with no successors, remove any BARRIER
6208 after the second block.
6210 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6212 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6214 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6217 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6218 for 14-bit register offsets when INT14_OK_STRICT is false.
6220 2014-01-07 Roland Stigge <stigge@antcom.de>
6221 Michael Meissner <meissner@linux.vnet.ibm.com>
6224 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6225 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6227 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6229 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6232 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6234 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6235 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6236 rtx is const0_rtx or not.
6238 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6241 * target-globals.c (save_target_globals): Remove this_fn_optab
6243 * toplev.c: Include optabs.h.
6244 (target_reinit): Temporarily restore the global options if another
6245 set of options are in force.
6247 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6249 PR rtl-optimization/58668
6250 * cfgcleanup.c (flow_find_cross_jump): Don't count
6251 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6252 to determine what is counted.
6253 (flow_find_head_matching_sequence): Use active_insn_p to determine
6255 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6257 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6258 determine what is counted.
6260 PR tree-optimization/59643
6261 * tree-predcom.c (split_data_refs_to_components): If one dr is
6262 read and one write, determine_offset fails and the write isn't
6263 in the bad component, just put the read into the bad component.
6265 2014-01-07 Mike Stump <mikestump@comcast.net>
6266 Jakub Jelinek <jakub@redhat.com>
6269 * tree-core.h (struct tree_optimization_option): Change optabs
6270 type from unsigned char * to void *.
6271 * optabs.c (init_tree_optimization_optabs): Adjust
6272 TREE_OPTIMIZATION_OPTABS initialization.
6274 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6277 * config/i386/i386.h (struct machine_function): Add
6278 no_drap_save_restore field.
6279 * config/i386/i386.c (ix86_save_reg): Use
6280 !cfun->machine->no_drap_save_restore instead of
6281 crtl->stack_realign_needed.
6282 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6283 this function clears frame_pointer_needed. Set
6284 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6285 and DRAP reg is needed.
6287 2014-01-06 Marek Polacek <polacek@redhat.com>
6290 * doc/implement-c.texi: Mention that other integer types are
6291 permitted as bit-field types in strictly conforming mode.
6293 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6295 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6298 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6300 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6302 2014-01-06 Martin Jambor <mjambor@suse.cz>
6305 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6307 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6309 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6313 * var-tracking.c (add_stores): Preserve the value of the source even if
6314 we don't record the store.
6316 2014-01-06 Terry Guo <terry.guo@arm.com>
6318 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6320 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6323 * config/darwin.c (darwin_function_section): Adjust return values to
6324 correspond to optimisation changes made in r206070.
6326 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6328 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6329 from prefetch_block tune setting.
6330 (nocona_cost): Correct size of prefetch block to 64.
6332 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6334 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6335 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6336 used to save the static chain register in the computation of the offset
6337 from which the FP registers need to be restored.
6339 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6341 PR tree-optimization/59519
6342 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6343 ICE if get_current_def (current_new_name) is already non-NULL, as long
6344 as it is a phi result of some other phi in *new_exit_bb that has
6347 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6348 or vmovdqu* for misaligned_operand.
6349 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6350 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6351 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6352 aligned_mem for AVX512F masked aligned load and store builtins and for
6355 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6357 PR tree-optimization/59651
6358 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6359 Address range for negative step should be added by TYPE_SIZE_UNIT.
6361 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6363 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6364 overlapping registers also for registers other than the stack pointer.
6366 2014-01-03 Marek Polacek <polacek@redhat.com>
6369 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6372 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6375 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6378 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6379 (push splitter): Use <P:MODE_SIZE> instead of
6380 GET_MODE_SIZE (<P:MODE>mode).
6381 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6382 (mov -1, reg peephole2): Likewise.
6383 * config/i386/sse.md (*mov<mode>_internal,
6384 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6385 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6386 *<code><mode>3, *andnot<mode>3<mask_name>,
6387 <mask_codefor><code><mode>3<mask_name>): Likewise.
6388 * config/i386/subst.md (mask_mode512bit_condition,
6389 sd_mask_mode512bit_condition): Likewise.
6391 2014-01-02 Xinliang David Li <davidxl@google.com>
6393 PR tree-optimization/59303
6394 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6395 (dump_predicates): Better output format.
6396 (pred_equal_p): New function.
6397 (is_neq_relop_p): Ditto.
6398 (is_neq_zero_form_p): Ditto.
6399 (pred_expr_equal_p): Ditto.
6400 (pred_neg_p): Ditto.
6401 (simplify_pred): Ditto.
6402 (simplify_preds_2): Ditto.
6403 (simplify_preds_3): Ditto.
6404 (simplify_preds_4): Ditto.
6405 (simplify_preds): Ditto.
6407 (push_to_worklist): Ditto.
6408 (get_pred_info_from_cmp): Ditto.
6409 (is_degenerated_phi): Ditto.
6410 (normalize_one_pred_1): Ditto.
6411 (normalize_one_pred): Ditto.
6412 (normalize_one_pred_chain): Ditto.
6413 (normalize_preds): Ditto.
6414 (normalize_cond_1): Remove function.
6415 (normalize_cond): Ditto.
6416 (is_gcond_subset_of): Ditto.
6417 (is_subset_of_any): Ditto.
6418 (is_or_set_subset_of): Ditto.
6419 (is_and_set_subset_of): Ditto.
6420 (is_norm_cond_subset_of): Ditto.
6421 (pred_chain_length_cmp): Ditto.
6422 (convert_control_dep_chain_into_preds): Type change.
6423 (find_predicates): Ditto.
6424 (find_def_preds): Ditto.
6425 (destroy_predicates_vecs): Ditto.
6426 (find_matching_predicates_in_rest_chains): Ditto.
6427 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6428 (is_pred_expr_subset): Ditto.
6429 (is_pred_chain_subset_of): Ditto.
6430 (is_included_in): Ditto.
6431 (is_superset_of): Ditto.
6433 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6435 Update copyright years.
6437 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6439 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6440 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6441 config/arc/arc.md, config/arc/arc.opt,
6442 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6443 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6444 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6445 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6446 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6447 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6448 vtable-verify.c, vtable-verify.h: Use the standard form for the
6451 2014-01-02 Tobias Burnus <burnus@net-b.de>
6453 * gcc.c (process_command): Update copyright notice dates.
6454 * gcov-dump.c: Ditto.
6456 * doc/cpp.texi: Bump @copying's copyright year.
6457 * doc/cppinternals.texi: Ditto.
6458 * doc/gcc.texi: Ditto.
6459 * doc/gccint.texi: Ditto.
6460 * doc/gcov.texi: Ditto.
6461 * doc/install.texi: Ditto.
6462 * doc/invoke.texi: Ditto.
6464 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6466 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6468 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6470 * config/i386/sse.md (*mov<mode>_internal): Guard
6471 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6473 PR rtl-optimization/59647
6474 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6475 new_rtx into UNSIGNED_FLOAT rtxes.
6477 Copyright (C) 2014 Free Software Foundation, Inc.
6479 Copying and distribution of this file, with or without modification,
6480 are permitted in any medium without royalty provided the copyright
6481 notice and this notice are preserved.