1 2014-04-11 Tobias Burnus <burnus@net-b.de>
4 * doc/invoke.texi (-Wformat-signedness): Document it.
5 (Wformat=2): Mention that this enables -Wformat-signedness.
7 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
9 * common/config/epiphany/epiphany-common.c
10 (epiphany_option_optimization_table): Enable section anchors by
11 default at -O1 or higher.
12 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
13 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
14 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
15 carries no extra cost.
16 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
17 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
18 * config/epiphany/predicates.md (memclob_operand): New predicate.
19 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
20 Use memclob_operand predicate and X constraint for operand 3.
22 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
24 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
25 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
28 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
30 PR rtl-optimization/60651
31 * mode-switching.c (optimize_mode_switching): Make sure to emit
32 sets of a lower numbered entity before sets of a higher numbered
33 entity to a mode of the same or lower priority.
34 When creating a seginfo for a basic block that starts with a code
35 label, move the insertion point past the code label.
36 (new_seginfo): Document and enforce requirement that
37 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
38 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
39 * doc/tm.texi: Regenerate.
41 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
44 * config/arc/arc.c (arc_save_restore): Fix assert typo.
46 2013-04-11 Jakub Jelinek <jakub@redhat.com>
48 * BASE-VER: Set to 4.10.0.
50 2014-04-11 Tobias Burnus <burnus@net-b.de>
53 * doc/bugreport.texi (Bugs): Remove nodes pointing to the
55 * doc/gcc.texi (Service): Update description in the @menu
56 * doc/invoke.texi (Option Summary): Remove misplaced and
59 2014-04-11 Steve Ellcey <sellcey@mips.com>
60 Jakub Jelinek <jakub@redhat.com>
63 * expr.c (convert_move): Use emit_store_flag_force instead of
64 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
67 2014-04-11 Richard Biener <rguenther@suse.de>
70 * varasm.c (assemble_alias): Avoid endless error reporting
71 recursion by setting TREE_ASM_WRITTEN.
73 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
75 * config/s390/s390.md: Add a splitter for NOT rtx.
77 2014-04-11 Jakub Jelinek <jakub@redhat.com>
79 PR rtl-optimization/60663
80 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
83 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
84 Jakub Jelinek <jakub@redhat.com>
87 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
90 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
93 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
94 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
95 ameliorating the cases where it can be.
97 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
100 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
102 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
103 (loadsync_<mode>): Change mode.
104 (load_quadpti, store_quadpti): New.
105 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
106 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
107 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
109 2014-04-09 Cong Hou <congh@google.com>
112 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
115 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
117 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
118 instead of vnor to exploit possible fusion opportunity in the
120 (altivec_expand_vec_perm_const_le): Likewise.
122 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
124 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
125 (loadsync_<mode>): Change mode.
126 (load_quadpti, store_quadpti): New.
127 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
128 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
130 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
133 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
134 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
135 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
137 2014-04-08 Richard Biener <rguenther@suse.de>
140 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
141 a 64bit widest int print double-int similar to on HWI64 hosts.
143 2014-04-08 Richard Biener <rguenther@suse.de>
145 PR tree-optimization/60785
146 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
147 default defs properly.
149 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
151 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
154 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
156 * ipa-devirt.c (maybe_record_node): When node is not recorded,
157 set completep to false rather than true.
159 2014-04-07 Douglas B Rupp <rupp@adacore.com>
162 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
163 ARM_TARGET2_DWARF_FORMAT.
165 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
168 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
169 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
172 2014-04-07 Richard Biener <rguenther@suse.de>
174 PR tree-optimization/60766
175 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
176 (may_eliminate_iv): Convert cand_value_at result to desired type.
178 2014-04-07 Jason Merrill <jason@redhat.com>
181 * common.opt (-fno-gnu-unique): Add.
182 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
184 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
186 * haifa-sched.c: Fix outdated function reference and minor
187 grammar errors in introductory comment.
189 2014-04-07 Richard Biener <rguenther@suse.de>
192 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
194 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
196 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
199 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
200 size accounting for thunks.
201 (pa_asm_output_mi_thunk): Use final_start_function() and
202 final_end_function() to output function start and end directives.
204 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
206 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
207 specific ISA/ feature information. Remove short_sp and errata_skip ds.
208 Add avr_device_specific_features enum to have device specific info.
209 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
210 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
211 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
212 updated device specific info.
213 * config/avr/avr-mcus.def: Merge device specific details to
215 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
217 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
218 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
219 assembler if RMW isa supported by current device.
220 * config/avr/genmultilib.awk: Update as device info structure changed.
221 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
223 2014-04-04 Cong Hou <congh@google.com>
225 PR tree-optimization/60656
226 * tree-vect-stmts.c (supportable_widening_operation):
227 Fix a bug that elements in a vector with vect_used_by_reduction
228 property are incorrectly reordered when the operation on it is not
229 consistant with the one in reduction operation.
231 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
233 PR rtl-optimization/60155
234 * gcse.c (record_set_data): New function.
235 (single_set_gcse): New function.
236 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
237 (hoist_code): Likewise.
238 (get_pressure_class_and_nregs): Likewise.
240 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
242 * explow.c (probe_stack_range): Emit a final optimization blockage.
244 2014-04-04 Anthony Green <green@moxielogic.com>
246 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
249 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
252 * lto-cgraph.c (input_overwrite_node): Check that partitioning
253 flags are set only during streaming.
254 * ipa.c (process_references, walk_polymorphic_call_targets,
255 symtab_remove_unreachable_nodes): Drop bodies of always inline
256 after early inlining.
257 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
259 2014-04-04 Jakub Jelinek <jakub@redhat.com>
260 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
263 * dwarf2out.c (const_ok_for_output_1): Reject expressions
266 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
269 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
271 (cortex_a53_fdivd): Likewise.
273 2014-04-04 Martin Jambor <mjambor@suse.cz>
276 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
278 * cgraph.c (clone_of_p): Also return true if thunks match.
279 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
280 cgraph_function_or_thunk_node and an obsolete comment.
281 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
283 (build_function_decl_skip_args): Likewise.
284 (set_new_clone_decl_and_node_flags): New function.
285 (duplicate_thunk_for_node): Likewise.
286 (redirect_edge_duplicating_thunks): Likewise.
287 (cgraph_clone_node): New parameter args_to_skip, pass it to
288 redirect_edge_duplicating_thunks which is called instead of
289 cgraph_redirect_edge_callee.
290 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
291 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
293 2014-04-04 Jeff Law <law@redhat.com>
296 * config/arm/predicates.md (const_int_I_operand): New predicate.
297 (const_int_M_operand): Similarly.
298 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
300 (insv_t2, extv_reg, extzv_t2): Likewise.
301 (load_multiple_with_writeback): Similarly for const_int_I_operand.
302 (pop_multiple_with_writeback_and_return): Likewise.
303 (vfp_pop_multiple_with_writeback): Likewise
305 2014-04-04 Richard Biener <rguenther@suse.de>
308 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
309 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
311 * gimplify.h (gimple_add_tmp_var_fn): Declare.
312 * gimplify.c (gimple_add_tmp_var_fn): New function.
313 * gimple-expr.h (create_tmp_reg_fn): Declare.
314 * gimple-expr.c (create_tmp_reg_fn): New function.
315 * gimple-low.c (record_vars_into): Don't change cfun.
316 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
317 code generation without cfun.
319 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
322 * Makefile.in (install-driver): Fix shell scripting.
324 2014-04-03 Cong Hou <congh@google.com>
326 PR tree-optimization/60505
327 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
328 threshold of number of iterations below which no vectorization will be
330 * tree-vect-loop.c (new_loop_vec_info):
331 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
332 * tree-vect-loop.c (vect_analyze_loop_operations):
333 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
334 * tree-vect-loop.c (vect_transform_loop):
335 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
336 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
337 of iterations of the loop and see if we should build the epilogue.
339 2014-04-03 Richard Biener <rguenther@suse.de>
341 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
343 (streamer_tree_cache_create): Adjust.
344 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
345 to allow optional nodes array.
346 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
347 (streamer_tree_cache_append): Likewise.
348 (streamer_tree_cache_create): Create nodes array optionally
349 as specified by parameter.
350 * lto-streamer-out.c (create_output_block): Avoid maintaining
351 the node array in the writer cache.
352 (DFS_write_tree): Remove assertion.
353 (produce_asm_for_decls): Free the out decl state hash table
355 * lto-streamer-in.c (lto_data_in_create): Adjust for
356 streamer_tree_cache_create prototype change.
358 2014-04-03 Richard Biener <rguenther@suse.de>
360 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
361 set TREE_CHAIN to NULL_TREE.
363 2014-04-03 Richard Biener <rguenther@suse.de>
365 PR tree-optimization/60740
366 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
367 over all GIMPLE_COND operands.
369 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
371 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
372 (Weffc++): Remove Scott's numbering, merge lists and reference
378 * class.c (accessible_nvdtor_p): New.
379 (check_bases): Don't check base destructor here ...
380 (check_bases_and_members): ... check them here. Trigger on
381 Wnon-virtual-dtor flag.
382 (finish_struct_1): Use accessible_nvdtor_p.
385 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
386 * g++.dg/warn/Wnvdtor-2.C: New.
387 * g++.dg/warn/Wnvdtor-3.C: New.
388 * g++.dg/warn/Wnvdtor-4.C: New.
389 * g++.dg/warn/Weff1.C: Delete.
390 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
391 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
393 2014-04-03 Nick Clifton <nickc@redhat.com>
395 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
398 2014-04-03 Martin Jambor <mjambor@suse.cz>
400 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
401 mention gcc_unreachable before failing.
402 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
405 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
408 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
409 code and instead mark the context inconsistent.
410 (possible_polymorphic_call_targets): For inconsistent contexts
411 return empty complete list.
413 2014-04-02 Anthony Green <green@moxielogic.com>
415 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
416 (extendqisi2, extendhisi2): Define.
417 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
418 (WCHAR_TYPE): Change to unsigned int.
420 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
422 PR tree-optimization/60733
423 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
424 insertion point for PHI candidates to be the end of the feeding
425 block for the PHI argument.
427 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
429 PR rtl-optimization/60650
430 * lra-constraints.c (process_alt_operands): Decrease reject for
431 earlyclobber matching.
433 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
435 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
437 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
439 * config/spu/spu.c (pad_bb): Do not crash when the last
440 insn is CODE_FOR_blockage.
442 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
444 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
445 lies outside the target mode.
447 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
450 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
451 software floating point or no floating point registers, do not
452 allow any type in the FPRs. Eliminate a test for SPE SIMD types
453 in GPRs that occurs after we tested for GPRs that would never be
456 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
457 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
458 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
459 specifically allow DDmode, since that does not use the SPE SIMD
462 2014-04-02 Richard Biener <rguenther@suse.de>
465 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
466 MODE_INTs. Properly use negv_optab.
467 (expand_abs): Likewise.
469 2014-04-02 Richard Biener <rguenther@suse.de>
472 * Makefile.in (install-driver): Guard extra installs with special
475 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
477 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
480 2014-04-01 Richard Henderson <rth@redhat.com>
483 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
484 alternative enabled before register allocation.
486 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
488 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
489 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
491 (nios2_large_got_address): Remove unneeded 'sym' parameter.
492 (nios2_got_address): Update nios2_large_got_address call site.
493 (nios2_delegitimize_address): New function.
494 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
495 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
496 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
498 2014-04-01 Martin Husemann <martin@duskware.de>
500 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
503 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
505 PR rtl-optimization/60604
506 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
507 check from register_operand.
508 (register_operand): Redefine in terms of general_operand.
509 (nonmemory_operand): Use register_operand for the non-constant cases.
511 2014-04-01 Richard Biener <rguenther@suse.de>
513 * gimple.h (struct gimple_statement_base): Align subcode to
516 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
518 * doc/invoke.texi (mapp-regs): Clarify.
520 2014-03-31 Ulrich Drepper <drepper@gmail.com>
522 * config/i386/avx512fintrin.h (__v32hi): Define type.
524 (_mm512_set1_epi8): Define.
525 (_mm512_set1_epi16): Define.
526 (_mm512_set4_epi32): Define.
527 (_mm512_set4_epi64): Define.
528 (_mm512_set4_pd): Define.
529 (_mm512_set4_ps): Define.
530 (_mm512_setr4_epi64): Define.
531 (_mm512_setr4_epi32): Define.
532 (_mm512_setr4_pd): Define.
533 (_mm512_setr4_ps): Define.
534 (_mm512_setzero_epi32): Define.
536 2014-03-31 Martin Jambor <mjambor@suse.cz>
539 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
540 callsite_arguments_match_p. Updated all callers. Also check types of
541 corresponding formal parameters and actual arguments.
542 (not_all_callers_have_enough_arguments_p) Renamed to
543 some_callers_have_mismatched_arguments_p.
545 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
547 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
549 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
552 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
555 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
557 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
558 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
560 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
561 Use FMAMODE_NOVF512 mode iterator.
562 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
563 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
564 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
565 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
567 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
568 Use VF_128_256 mode iterator.
569 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
572 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
574 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
575 static chain if needed.
577 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
580 * lra-constraints.c (index_part_to_reg): New.
581 (process_address): Use it.
583 2014-03-27 Jeff Law <law@redhat.com>
584 Jakub Jelinek <jakub@redhat.com>
587 * expr.c (do_tablejump): Use simplify_gen_binary rather than
588 gen_rtx_{PLUS,MULT} to build up the address expression.
590 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
591 creating non-canonical RTL.
593 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
596 * ipa-inline.c (want_inline_small_function_p): Short circuit large
597 functions; reorganize to make cheap checks first.
598 (inline_small_functions): Do not estimate growth when dumping;
600 * ipa-inline.h (inline_summary): Add min_size.
601 (growth_likely_positive): New function.
602 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
603 (set_cond_stmt_execution_predicate): Cleanup.
604 (estimate_edge_size_and_time): Compute min_size.
605 (estimate_calls_size_and_time): Likewise.
606 (estimate_node_size_and_time): Likewise.
607 (inline_update_overall_summary): Update min_size.
608 (do_estimate_edge_time): Likewise.
609 (do_estimate_edge_size): Update.
610 (do_estimate_edge_hints): Update.
611 (growth_likely_positive): New function.
613 2014-03-28 Jakub Jelinek <jakub@redhat.com>
616 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
617 also if addr has VOIDmode.
619 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
621 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
622 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
624 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
625 instructions as well as AdvancedSIMD loads.
627 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
629 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
630 Use crypto_aese type.
631 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
632 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
633 crypto_aese, crypto_aesmc. Move to types.md.
634 * config/arm/types.md (crypto_aes): Split into crypto_aese,
636 * config/arm/iterators.md (crypto_type): Likewise.
638 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
640 * cgraph.c: Include expr.h and tree-dfa.h.
641 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
644 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
647 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
648 regs from checking multi-reg pseudos.
650 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
652 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
654 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
656 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
657 if it would clobber the stack pointer, even temporarily.
659 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
661 * mode-switching.c: Make small adjustments to the top comment.
663 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
665 * config/rs6000/constraints.md (wD constraint): New constraint to
666 match the constant integer to get the top DImode/DFmode out of a
667 vector in a VSX register.
669 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
670 match the constant integer to get the top DImode/DFmode out of a
671 vector in a VSX register.
673 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
676 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
679 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
680 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
682 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
683 Optimize vec_extract of 64-bit values, where the value being
684 extracted is in the top word, where we can use scalar
685 instructions. Add direct move and store support. Combine the big
686 endian/little endian vector select load support into a single insn.
687 (vsx_extract_<mode>_internal1): Likewise.
688 (vsx_extract_<mode>_internal2): Likewise.
689 (vsx_extract_<mode>_load): Likewise.
690 (vsx_extract_<mode>_store): Likewise.
691 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
692 combined into vsx_extract_<mode>_load.
693 (vsx_extract_<mode>_one_le): Likewise.
695 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
696 define the top 64-bit vector element.
698 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
701 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
702 Document vec_vbpermq builtin.
705 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
706 enable use of xxsldwi and xxpermdi builtin functions.
707 (vec_xxpermdi): Likewise.
709 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
710 Document use of vec_xxsldwi and vec_xxpermdi builtins.
712 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
714 PR rtl-optimization/60650
715 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
717 (find_spills_for): New.
718 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
719 Spill all pseudos on the second iteration.
721 2014-03-27 Marek Polacek <polacek@redhat.com>
724 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
727 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
729 * config/s390/s390.c (s390_can_use_return_insn): Check for
730 call-saved FPRs on 31 bit.
732 2014-03-27 Jakub Jelinek <jakub@redhat.com>
735 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
736 if they need regimplification, just drop them instead of
737 calling gimple_regimplify_operands on them.
739 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
742 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
743 (aarch64_frame_pointer_required): Adjust logic.
744 (aarch64_can_eliminate): Adjust logic.
745 (aarch64_override_options_after_change): Adjust logic.
747 2014-03-27 Dehao Chen <dehao@google.com>
749 * ipa-inline.c (early_inliner): Update node's inline info.
751 2014-03-26 Dehao Chen <dehao@google.com>
753 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
754 compiler inserted conditional jumps for NAN float check.
756 2014-03-26 Jakub Jelinek <jakub@redhat.com>
758 * ubsan.h (ubsan_create_data): Change second argument's type
759 to const location_t *.
760 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
762 (ubsan_create_data): Change second argument to const location_t *PLOC.
763 Create Loc field whenever PLOC is non-NULL.
764 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
765 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
769 * real.c (real_to_integer2): Change type of low to UHWI.
771 2014-03-26 Tobias Burnus <burnus@net-b.de>
773 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
774 (CILK_SELF_SPECS): New define.
775 (driver_self_specs): Use it.
777 2014-03-26 Richard Biener <rguenther@suse.de>
779 * tree-pretty-print.c (percent_K_format): Implement special
780 case for LTO and its stripped down BLOCK tree.
782 2014-03-26 Jakub Jelinek <jakub@redhat.com>
785 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
787 * tree-vrp.c (simplify_internal_call_using_ranges): If only
788 one range is range_int_cst_p, but not both, at least optimize
789 addition/subtraction of 0 and multiplication by 0 or 1.
790 * gimple-fold.c (gimple_fold_call): Fold
791 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
792 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
793 INTEGER_CSTs, try to fold at least x * 0 and y - y.
795 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
797 PR rtl-optimization/60452
798 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
799 <case REG>: Return 1 for invalid offsets from the frame pointer.
801 2014-03-26 Marek Polacek <polacek@redhat.com>
804 * doc/extend.texi (C Extensions): Mention variable-length arrays in
807 2014-03-26 Marek Polacek <polacek@redhat.com>
810 * doc/extend.texi (Designated Inits): Describe what happens to omitted
813 2014-03-26 Marek Polacek <polacek@redhat.com>
816 * ira-color.c (update_conflict_hard_regno_costs): Perform the
817 multiplication in unsigned type.
819 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
821 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
823 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
825 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
827 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
830 * cif-code.def (UNREACHABLE) New code.
831 * ipa-inline.c (inline_small_functions): Skip edges to
832 __builtlin_unreachable.
833 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
834 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
835 predicate to __bulitin_unreachable.
836 (set_cond_stmt_execution_predicate): Fix issue when
837 invert_tree_comparison returns ERROR_MARK.
838 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
839 propagate to inline clones.
840 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
842 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
843 * cgraphclones.c (cgraph_clone_node): If call destination is already
844 ureachable, do not redirect it back.
845 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
848 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
850 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
851 Do not modify inline clones.
853 2014-03-25 Jakub Jelinek <jakub@redhat.com>
855 * config/i386/i386.md (general_sext_operand): New mode attr.
856 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
857 don't generate (sign_extend (const_int)).
858 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
859 operands[2]. Use We constraint instead of <i> and
860 <general_sext_operand> predicate instead of <general_operand>.
861 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
862 * config/i386/constraints.md (We): New constraint.
863 * config/i386/predicates.md (x86_64_sext_operand,
864 sext_operand): New predicates.
866 2014-03-25 Martin Jambor <mjambor@suse.cz>
869 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
870 inconsistent devirtualizations to __builtin_unreachable.
872 2014-03-25 Marek Polacek <polacek@redhat.com>
875 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
877 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
879 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
880 order of elements for big-endian.
882 2014-03-25 Richard Biener <rguenther@suse.de>
885 * gimplify-me.c (gimple_regimplify_operands): Update the
888 2014-03-25 Martin Jambor <mjambor@suse.cz>
891 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
892 (lto_output_varpool_node): Likewise.
893 (input_overwrite_node): Likewise.
894 (input_varpool_node): Likewise.
896 2014-03-25 Richard Biener <rguenther@suse.de>
898 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
901 2014-03-25 Jakub Jelinek <jakub@redhat.com>
903 * combine.c (simplify_compare_const): Add MODE argument.
904 Handle mode_width 0 as very large mode_width.
905 (try_combine, simplify_comparison): Adjust callers.
907 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
908 type to avoid signed integer overflow.
909 * explow.c (plus_constant): Likewise.
911 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
913 * doc/generic.texi: Correct typos.
915 2014-03-24 Tobias Burnus <burnus@net-b.de>
917 * doc/invoke.texi (-flto): Expand section about
918 using static libraries with LTO.
920 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
922 PR rtl-optimization/60501
923 * optabs.def (addptr3_optab): New optab.
924 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
925 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
926 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
928 * lra.c (emit_add3_insn): Use the addptr pattern if available.
930 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
932 2014-03-24 Ulrich Drepper <drepper@gmail.com>
934 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
937 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
938 (_mm256_undefined_ps): Define.
939 (_mm256_undefined_pd): Define.
940 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
941 (_mm_undefined_pd): Define.
942 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
943 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
944 (_mm512_undefined_ps): Define.
945 (_mm512_undefined_pd): Define.
946 Use _mm*_undefined_*.
947 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
949 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
951 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
952 (lshr_simd): DI mode added.
953 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
954 (aarch64_ushr_simddi): Likewise.
955 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
956 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
957 (vshrd_n_u64): Likewise.
959 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
961 * Makefile.in (s-macro_list): Depend on cc1.
963 2014-03-23 Teresa Johnson <tejohnson@google.com>
965 * ipa-utils.c (ipa_print_order): Use specified dump file.
967 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
969 PR rtl-optimization/60601
970 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
972 * gcc.c (eval_spec_function): Initialize save_growing_value.
974 2014-03-22 Jakub Jelinek <jakub@redhat.com>
977 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
978 code == MINUS_EXPR, never swap op0 with op1.
980 * toplev.c (init_local_tick): Avoid signed integer multiplication
982 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
983 shift by first operand's bitsize.
985 2014-03-21 Jakub Jelinek <jakub@redhat.com>
988 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
990 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
991 TARGET_ISA_64BIT_P(x).
993 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
995 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
996 pattern for vector nor instead of subtract from splat(-1).
997 (altivec_expand_vec_perm_const_le): Likewise.
999 2014-03-21 Richard Henderson <rth@twiddle.net>
1002 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1003 related insns after epilogue_completed.
1005 2014-03-21 Martin Jambor <mjambor@suse.cz>
1008 * cgraph.h (symtab_node): New flag body_removed.
1009 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1010 when removing bodies.
1011 * symtab.c (dump_symtab_base): Dump body_removed flag.
1012 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1013 had their bodies removed.
1015 2014-03-21 Martin Jambor <mjambor@suse.cz>
1018 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1021 2014-03-21 Richard Biener <rguenther@suse.de>
1023 PR tree-optimization/60577
1024 * tree-core.h (struct tree_base): Document nothrow_flag use
1025 in VAR_DECL_NONALIASED.
1026 * tree.h (VAR_DECL_NONALIASED): New.
1027 (may_be_aliased): Adjust.
1028 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1030 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1032 * expr.c (expand_expr_real_1): Remove outdated comment.
1034 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1037 * ira.c (adjust_cleared_regs): Call copy_rtx on
1038 *reg_equiv[REGNO (loc)].src_p before passing it to
1039 simplify_replace_fn_rtx.
1042 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1043 into CONST, put pic register as first operand of PLUS. Use
1044 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1046 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1048 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1050 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1052 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1053 around for store forwarding issue in the FPU on the UT699.
1054 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1055 loads and operations if -mfix-ut699 is specified.
1056 (divtf3_hq): Tweak attribute.
1057 (sqrttf2_hq): Likewise.
1059 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1061 * calls.c (store_one_arg): Remove incorrect const qualification on the
1062 type of the temporary.
1063 * cfgexpand.c (expand_return): Likewise.
1064 * expr.c (expand_constructor): Likewise.
1065 (expand_expr_real_1): Likewise.
1067 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1069 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1072 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1075 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1077 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1079 * config/arm/aarch-common-protos.h
1080 (alu_cost_table): Fix spelling of "extend".
1081 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1083 2014-03-19 Richard Biener <rguenther@suse.de>
1086 * tree-core.h (tree_type_common): Re-order pointer members
1087 to reduce recursion depth during GC walks.
1089 2014-03-19 Marek Polacek <polacek@redhat.com>
1092 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1093 before accessing it.
1095 2014-03-19 Richard Biener <rguenther@suse.de>
1098 * lto-streamer-in.c (input_function): In WPA stage do not drop
1101 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1103 PR tree-optimization/60559
1104 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1105 with build_zero_cst assignment.
1107 2014-03-18 Kai Tietz <ktietz@redhat.com>
1109 PR rtl-optimization/56356
1110 * sdbout.c (sdbout_parms): Verify that parms'
1111 incoming argument is valid.
1112 (sdbout_reg_parms): Likewise.
1114 2014-03-18 Richard Henderson <rth@redhat.com>
1117 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1118 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1119 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1121 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1123 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1124 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1125 Italicize plugin event names in description. Explain that
1126 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1127 Remind that no GCC functions should be called after PLUGIN_FINISH.
1128 Explain what pragmas with expansion are.
1130 2014-03-18 Martin Liska <mliska@suse.cz>
1132 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1133 gimple call statement is update.
1134 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1135 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1137 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1140 * ubsan.c (ubsan_instrument_unreachable): Call
1141 initialize_sanitizer_builtins.
1142 (ubsan_pass): Likewise.
1145 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1146 varpool_finalize_decl instead of rest_of_decl_compilation.
1148 2014-03-18 Richard Biener <rguenther@suse.de>
1150 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1151 by using bitmap_and_compl instead of bitmap_and_compl_into.
1152 (df_rd_transfer_function): Likewise.
1154 2014-03-18 Richard Biener <rguenther@suse.de>
1156 * doc/lto.texi (fresolution): Fix typo.
1158 2014-03-18 Richard Biener <rguenther@suse.de>
1160 * doc/invoke.texi (flto): Update for changes in 4.9.
1162 2014-03-18 Richard Biener <rguenther@suse.de>
1164 * doc/loop.texi: Remove section on the removed lambda framework.
1165 Update loop docs with recent changes in preserving loop structure.
1167 2014-03-18 Richard Biener <rguenther@suse.de>
1169 * doc/lto.texi (-fresolution): Document.
1171 2014-03-18 Richard Biener <rguenther@suse.de>
1173 * doc/contrib.texi: Adjust my name.
1175 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1178 * internal-fn.c: Include diagnostic-core.h.
1179 (expand_BUILTIN_EXPECT): New function.
1180 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1181 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1182 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1183 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1185 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1186 Revert 3 argument __builtin_expect code.
1187 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1188 * gimple-fold.c (gimple_fold_call): Likewise.
1189 * tree.h (fold_builtin_expect): New prototype.
1190 * builtins.c (build_builtin_expect_predicate): Add predictor
1191 argument, if non-NULL, create 3 argument __builtin_expect.
1192 (fold_builtin_expect): No longer static. Add ARG2 argument,
1193 pass it through to build_builtin_expect_predicate.
1194 (fold_builtin_2): Adjust caller.
1195 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1196 * internal-fn.def (BUILTIN_EXPECT): New.
1198 2014-03-18 Tobias Burnus <burnus@net-b.de>
1201 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1202 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1203 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1205 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1208 * predict.c (combine_predictions_for_bb): Fix up formatting.
1209 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1210 fill what it points to if non-NULL.
1211 (tree_predict_by_opcode): Adjust caller, use the predictor.
1212 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1214 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1216 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1217 proper constant for the store mode.
1219 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1221 * symtab.c (change_decl_assembler_name): Fix transparent alias
1224 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1226 * config/aarch64/aarch64.c: Correct the comments about the
1227 aarch64 stack layout.
1229 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1231 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1232 check for GF_OMP_FOR_KIND_FOR.
1234 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1236 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1237 ymm and zmm register names.
1239 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1242 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1243 note creation for the 2010-08-31 changes.
1245 2014-03-17 Marek Polacek <polacek@redhat.com>
1248 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1249 as -fno-tree-loop-vectorize.
1250 (expand_omp_simd): Likewise.
1252 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1254 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1255 (eligible_for_call_delay): New prototype.
1256 * config/sparc/sparc.c (tls_call_delay): Rename into...
1257 (eligible_for_call_delay): ...this. Return false if the instruction
1258 cannot be put in the delay slot of a branch.
1259 (eligible_for_restore_insn): Simplify.
1260 (eligible_for_return_delay): Return false if the instruction cannot be
1261 put in the delay slot of a branch and simplify.
1262 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1263 put in the delay slot of a branch.
1264 * config/sparc/sparc.md (fix_ut699): New attribute.
1265 (tls_call_delay): Delete.
1266 (in_call_delay): Reimplement.
1267 (eligible_for_sibcall_delay): Rename into...
1268 (in_sibcall_delay): ...this.
1269 (eligible_for_return_delay): Rename into...
1270 (in_return_delay): ...this.
1271 (in_branch_delay): Reimplement.
1272 (in_uncond_branch_delay): Delete.
1273 (in_annul_branch_delay): Delete.
1275 2014-03-14 Richard Henderson <rth@redhat.com>
1278 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1279 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1280 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1281 (floathi splitters): Remove.
1282 (float<SWI48x>xf2): New pattern.
1283 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1284 code that tried to handle DImode for 32-bit, but which was excluded
1285 by the pattern's condition. Drop allocation of stack temporary.
1286 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1287 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1288 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1289 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1290 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1291 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1292 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1293 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1294 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1295 (*float<SWI48x><X87MODEF>2_i387): Remove.
1296 (all float _with_temp splitters): Remove.
1297 (*float<SWI48x><MODEF>2_i387): New pattern.
1298 (*float<SWI48><MODEF>2_sse): New pattern.
1299 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1300 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1302 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1303 Marek Polacek <polacek@redhat.com>
1306 * common.opt (dump_base_name_prefixed): New Variable.
1307 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1308 if x_dump_base_name_prefixed is already set, set it at the end.
1310 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1312 PR rtl-optimization/60508
1313 * lra-constraints.c (get_reload_reg): Add new parameter
1315 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1316 Pass the new parameter values.
1318 2014-03-14 Richard Biener <rguenther@suse.de>
1320 * common.opt: Revert unintented changes from r205065.
1323 2014-03-14 Richard Biener <rguenther@suse.de>
1326 * cfghooks.c (split_block): Properly adjust all loops the
1327 block was a latch of.
1329 2014-03-14 Martin Jambor <mjambor@suse.cz>
1332 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1335 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1338 * config/avr/avr.c (avr_set_current_function): Pass function name
1339 through default_strip_name_encoding before sanity checking instead
1340 of skipping the first char of the assembler name.
1342 2014-03-13 Richard Henderson <rth@redhat.com>
1345 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1346 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1347 * config/i386/i386-protos.h: Likewise.
1348 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1349 in the expander instead of a splitter.
1350 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1351 any possibility of requiring a memory.
1352 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1353 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1354 (fp branch splitters): Update for ix86_split_fp_branch.
1355 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1356 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1357 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1358 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1359 (*fop_<MODEF>_3_i387): Likewise.
1360 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1361 (splitters for the fop_* register patterns): Remove.
1362 (fscalexf4_i387): Rename from *fscalexf4_i387.
1363 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1365 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1367 PR tree-optimization/59779
1368 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1369 type for bitsize and maxsize instead of HOST_WIDE_INT.
1371 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1373 PR rtl-optimization/57320
1374 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1375 the CFG after thread_prologue_and_epilogue_insns.
1377 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1379 PR rtl-optimization/57189
1380 * lra-constraints.c (process_alt_operands): Disfavor spilling
1383 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1385 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1387 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1389 PR tree-optimization/59025
1391 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1392 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1394 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1397 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1398 calls of avr_out_plus_1.
1400 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1402 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1403 BB's single pred and update the father loop's latch info later.
1405 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1407 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1411 (VEC_base): Likewise.
1412 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1413 registers, we need to swap double words in little endian mode.
1415 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1416 to be a container mode for 128-bit integer operations added in ISA
1417 2.07. Unlike TImode and PTImode, the preferred register set is
1418 the Altivec/VMX registers for the 128-bit operations.
1420 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1422 (rs6000_split_128bit_ok_p): Likewise.
1424 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1425 macros for creating ISA 2.07 normal and overloaded builtin
1426 functions with 3 arguments.
1427 (BU_P8V_OVERLOAD_3): Likewise.
1428 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1429 for use as overloaded functions.
1430 (VPERM_1TI_UNS): Likewise.
1431 (VSEL_1TI): Likewise.
1432 (VSEL_1TI_UNS): Likewise.
1433 (ST_INTERNAL_1ti): Likewise.
1434 (LD_INTERNAL_1ti): Likewise.
1435 (XXSEL_1TI): Likewise.
1436 (XXSEL_1TI_UNS): Likewise.
1437 (VPERM_1TI): Likewise.
1438 (VPERM_1TI_UNS): Likewise.
1439 (XXPERMDI_1TI): Likewise.
1440 (SET_1TI): Likewise.
1441 (LXVD2X_V1TI): Likewise.
1442 (STXVD2X_V1TI): Likewise.
1443 (VEC_INIT_V1TI): Likewise.
1444 (VEC_SET_V1TI): Likewise.
1445 (VEC_EXT_V1TI): Likewise.
1446 (EQV_V1TI): Likewise.
1447 (NAND_V1TI): Likewise.
1448 (ORC_V1TI): Likewise.
1449 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1450 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1452 (VADDUQM): Likewise.
1453 (VSUBCUQ): Likewise.
1454 (VADDEUQM): Likewise.
1455 (VADDECUQ): Likewise.
1456 (VSUBEUQM): Likewise.
1457 (VSUBECUQ): Likewise.
1459 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1460 __int128_t and __uint128_t types.
1461 (__uint128_type): Likewise.
1462 (altivec_categorize_keyword): Add support for vector __int128_t,
1463 vector __uint128_t, vector __int128, and vector unsigned __int128
1464 as a container type for TImode operations that need to be done in
1465 VSX/Altivec registers.
1466 (rs6000_macro_to_expand): Likewise.
1467 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1468 to support 128-bit integer instructions vaddcuq, vadduqm,
1469 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1470 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1472 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1473 for V1TImode, and set up preferences to use VSX/Altivec registers.
1474 Setup VSX reload handlers.
1475 (rs6000_debug_reg_global): Likewise.
1476 (rs6000_init_hard_regno_mode_ok): Likewise.
1477 (rs6000_preferred_simd_mode): Likewise.
1478 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1479 (easy_altivec_constant): Likewise.
1480 (output_vec_const_move): Likewise.
1481 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1483 (rs6000_expand_vector_extract): Likewise.
1484 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1486 (rs6000_const_vec): Add support for V1TImode.
1487 (rs6000_emit_le_vsx_load): Swap double words when loading or
1488 storing TImode/V1TImode.
1489 (rs6000_emit_le_vsx_store): Likewise.
1490 (rs6000_emit_le_vsx_move): Likewise.
1491 (rs6000_emit_move): Add support for V1TImode.
1492 (altivec_expand_ld_builtin): Likewise.
1493 (altivec_expand_st_builtin): Likewise.
1494 (altivec_expand_vec_init_builtin): Likewise.
1495 (altivec_expand_builtin): Likewise.
1496 (rs6000_init_builtins): Add support for V1TImode type. Add
1497 support for ISA 2.07 128-bit integer builtins. Define type names
1498 for the VSX/Altivec vector types.
1499 (altivec_init_builtins): Add support for overloaded vector
1500 functions with V1TImode type.
1501 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1502 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1504 (rs6000_split_128bit_ok_p): Likewise.
1505 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1506 __int128_t and vector __uint128_t.
1508 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1509 and mode attributes.
1516 (VS_scalar): Likewise.
1517 (VS_double): Likewise.
1518 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1520 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1521 we support the ISA 2.07 128-bit integer arithmetic instructions.
1522 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1523 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1524 and TImode types for use with the builtin functions.
1525 (V1TI_type_node): Likewise.
1526 (unsigned_V1TI_type_node): Likewise.
1527 (intTI_type_internal_node): Likewise.
1528 (uintTI_type_internal_node): Likewise.
1530 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1531 128-bit builtin functions.
1532 (UNSPEC_VADDEUQM): Likewise.
1533 (UNSPEC_VADDECUQ): Likewise.
1534 (UNSPEC_VSUBCUQ): Likewise.
1535 (UNSPEC_VSUBEUQM): Likewise.
1536 (UNSPEC_VSUBECUQ): Likewise.
1537 (VM): Add V1TImode to vector mode iterators.
1539 (VI_unit): Likewise.
1540 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1541 (altivec_vaddcuq): Likewise.
1542 (altivec_vsubuqm): Likewise.
1543 (altivec_vsubcuq): Likewise.
1544 (altivec_vaddeuqm): Likewise.
1545 (altivec_vaddecuq): Likewise.
1546 (altivec_vsubeuqm): Likewise.
1547 (altivec_vsubecuq): Likewise.
1549 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1551 (BOOL_128): Likewise.
1552 (BOOL_REGS_OUTPUT): Likewise.
1553 (BOOL_REGS_OP1): Likewise.
1554 (BOOL_REGS_OP2): Likewise.
1555 (BOOL_REGS_UNARY): Likewise.
1556 (BOOL_REGS_AND_CR0): Likewise.
1558 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1559 128-bit integer builtin support.
1560 (vec_vadduqm): Likewise.
1561 (vec_vaddecuq): Likewise.
1562 (vec_vaddeuqm): Likewise.
1563 (vec_vsubecuq): Likewise.
1564 (vec_vsubeuqm): Likewise.
1565 (vec_vsubcuq): Likewise.
1566 (vec_vsubuqm): Likewise.
1568 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1569 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1570 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1571 128-bit integer add/subtract to ISA 2.07.
1573 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1575 * config/arc/arc.c (arc_predicate_delay_insns):
1576 Fix third argument passed to conditionalize_nonjump.
1578 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1580 * config/aarch64/aarch64-builtins.c
1581 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1582 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1583 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1584 instead of __builtin_lfloor.
1585 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1587 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1589 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1590 (tree_ssa_ifcombine_bb_1): New function.
1591 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1592 is an empty forwarder block to then_bb or vice versa and then_bb
1593 and else_bb are effectively swapped.
1595 2014-03-12 Christian Bruel <christian.bruel@st.com>
1598 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1599 REG_CFA_DEF_CFA note.
1600 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1601 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1603 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1605 PR tree-optimization/60454
1606 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1608 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1610 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1611 Do not define target_cpu_default2 to generic.
1612 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1613 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1614 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1616 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1617 Marc Glisse <marc.glisse@inria.fr>
1619 PR tree-optimization/60502
1620 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1621 instead of build_low_bits_mask.
1623 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1626 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1627 if there are multiple uses, but op doesn't live on E edge.
1628 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1629 clobber stmts before __builtin_unreachable.
1631 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1633 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1634 hard_frame_pointer_rtx.
1635 * cse.c (cse_insn): Remove volatile check.
1636 * cselib.c (cselib_process_insn): Likewise.
1637 * dse.c (scan_insn): Likewise.
1639 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1641 * config/arc/arc.c (conditionalize_nonjump): New function,
1643 (arc_ifcvt): ... this.
1644 (arc_predicate_delay_insns): Use it.
1646 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1648 * config/arc/predicates.md (extend_operand): During/after reload,
1649 allow const_int_operand.
1650 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1651 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1652 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1654 (umulsi3_highpart_i): Likewise.
1656 2014-03-11 Richard Biener <rguenther@suse.de>
1658 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1659 Add asserts to guard possible wrong-code bugs.
1661 2014-03-11 Richard Biener <rguenther@suse.de>
1663 PR tree-optimization/60429
1664 PR tree-optimization/60485
1665 * tree-ssa-structalias.c (set_union_with_increment): Properly
1666 take into account all fields that overlap the shifted vars.
1667 (do_sd_constraint): Likewise.
1668 (do_ds_constraint): Likewise.
1669 (get_constraint_for_ptr_offset): Likewise.
1671 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1673 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1674 (nios2_compute_frame_layout):
1675 Add calculation of cfun->machine->fp_save_offset.
1676 (nios2_expand_prologue): Correct setting of frame pointer register
1678 (nios2_expand_epilogue): Update recovery of stack pointer from
1679 frame pointer accordingly.
1680 (nios2_initial_elimination_offset): Update calculation of offset
1681 for eliminating to HARD_FRAME_POINTER_REGNUM.
1683 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1686 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1687 cgraph_get_create_node on VAR_DECLs.
1689 2014-03-10 Richard Biener <rguenther@suse.de>
1692 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1694 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1696 * config/vms/vms.opt (vms_float_format): New variable.
1698 2014-03-08 Tobias Burnus <burnus@net-b.de>
1700 * doc/invoke.texi (-fcilkplus): Update implementation status.
1702 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1703 Richard Biener <rguenther@suse.de>
1705 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1706 consistently accross all TUs.
1707 (run_gcc): Enable -fshort-double automatically at link at link-time
1708 and disallow override.
1710 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1713 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1714 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1715 if they can't be used.
1717 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1719 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1720 for Solaris 11/x86 ld.
1721 * configure: Regenerate.
1723 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1725 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1726 (LIB_TLS_SPEC): Save as ld_tls_libs.
1727 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1728 (HAVE_AS_IX86_TLSLDM): New test.
1729 * configure, config.in: Regenerate.
1730 * config/i386/i386.c (legitimize_tls_address): Fall back to
1731 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1732 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1733 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1734 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1736 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1738 * common.opt (fira-loop-pressure): Mark as optimization.
1740 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1742 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1743 an OpenMP mappable type.
1745 2014-03-06 Matthias Klose <doko@ubuntu.com>
1747 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1748 MULTILIB_OSDIRNAMES is not defined.
1750 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1751 Meador Inge <meadori@codesourcery.com>
1754 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1755 (arm_legitimize_address): Call legitimize_tls_address for any
1756 arm_tls_referenced_p expression, handle constant addend. Call it
1757 before testing for !TARGET_ARM.
1758 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1760 2014-03-06 Richard Biener <rguenther@suse.de>
1766 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1768 * tree-streamer.c (record_common_node): Assert we don't record
1769 nodes with type double.
1770 (preload_common_node): Skip type double, complex double and double
1771 pointer since it is now frontend dependent due to fshort-double option.
1773 2014-03-06 Richard Biener <rguenther@suse.de>
1775 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1776 or -fno-lto is specified and the linker has full plugin support.
1777 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1778 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1779 * lto-wrapper.c (merge_and_complain): Merge compile-time
1780 optimization levels.
1781 (run_gcc): And pass it through to the link options.
1783 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1787 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1789 * cselib.c (remove_useless_values): Skip to avoid quadratic
1790 behavior if the condition moved from...
1791 (cselib_process_insn): ... here holds.
1793 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1796 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1797 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1800 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1801 (TM_H): Add x86-tune.def.
1803 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1805 * config/aarch64/aarch64.c (generic_tunings):
1806 Use cortexa57_extra_costs.
1808 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1811 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1812 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1813 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1816 2014-03-04 Heiher <r@hev.cc>
1818 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1819 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1821 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1823 * config/i386/predicates.md (const2356_operand): Change to ...
1824 (const2367_operand): ... this.
1825 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1827 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1828 (*avx512pf_scatterpf<mode>sf): Ditto.
1829 (avx512pf_scatterpf<mode>df): Ditto.
1830 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1831 (*avx512pf_scatterpf<mode>df): Ditto.
1832 * config/i386/i386.c (ix86_expand_builtin): Update
1833 incorrect hint operand error message.
1835 2014-03-04 Richard Biener <rguenther@suse.de>
1837 * lto-section-in.c (lto_get_section_data): Fix const cast.
1839 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1841 * tree-streamer.c (record_common_node): Assert we don't record
1842 nodes with type double.
1843 (preload_common_node): Skip type double, complex double and double
1844 pointer since it is now frontend dependent due to fshort-double option.
1846 2014-03-04 Richard Biener <rguenther@suse.de>
1849 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1850 (lto_input_toplevel_asms): Likewise.
1851 * lto-section-in.c (lto_get_section_data): Instead do it here
1854 2014-03-04 Richard Biener <rguenther@suse.de>
1856 PR tree-optimization/60382
1857 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1858 dead PHIs a reduction.
1860 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1862 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1864 (_mm_prefetch): Move out of GCC target("sse") pragma.
1865 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1866 GCC target("prfchw") pragma.
1867 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1869 * config/i386/i386.c (ix86_option_override_internal): Enable
1870 -mprfchw with -mprefetchwt1.
1872 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1874 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1877 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1879 * opts.h (CL_PCH_IGNORE): Define.
1880 * targhooks.c (option_affects_pch_p):
1881 Return false for options that have CL_PCH_IGNORE set.
1882 * opt-functions.awk: Process PchIgnore.
1883 * doc/options.texi: Document PchIgnore.
1885 * config/arc/arc.opt (misize): Add PchIgnore property.
1887 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1889 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1890 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1891 constraint on constants to permit them being loaded into
1892 GENERAL_REGS or BASE_REGS.
1894 2014-03-03 Nick Clifton <nickc@redhat.com>
1896 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1897 anti-cacnonical alternatives.
1898 (negandhi3_real): New pattern.
1899 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1901 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1903 * config/avr/avr-mcus.def: Remove atxmega16x1.
1904 * config/avr/avr-tables.opt: Regenerate.
1905 * config/avr/t-multilib: Regenerate.
1906 * doc/avr-mmcu.texi: Regenerate.
1908 2014-03-03 Tobias Grosser <tobias@grosser.es>
1909 Mircea Namolaru <mircea.namolaru@inria.fr>
1911 PR tree-optimization/58028
1912 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1915 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1917 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1918 not handled by recognizers.
1920 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1923 * function.c (expand_function_end): Don't emit
1924 clobber_return_register sequence if clobber_after is a BARRIER.
1925 * cfgexpand.c (construct_exit_block): Append instructions before
1926 return_label to prev_bb.
1928 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1930 * config/rs6000/constraints.md: Document reserved use of "wc".
1932 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1935 * ipa.c (function_and_variable_visibility): When dissolving comdat
1936 group, also set all symbols to local.
1938 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1943 2013-12-14 Jan Hubicka <jh@suse.cz>
1945 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1947 2014-03-02 Jon Beniston <jon@beniston.com>
1953 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1954 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1955 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1956 (simple_return, *simple_return): New patterns
1957 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1958 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1960 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1962 * dwarf2out.c (gen_subprogram_die): Tidy.
1964 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1967 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1968 (*mov_t_msb_neg_negc): ... this new insn.
1970 2014-02-28 Jason Merrill <jason@redhat.com>
1973 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1976 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1979 * dwarf2out.c (decltype_auto_die): New static.
1980 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1981 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1982 (is_cxx_auto): Likewise.
1984 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1986 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1987 we are not using general regs only.
1989 2014-02-28 Richard Biener <rguenther@suse.de>
1992 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1993 previous fix and only allow to remove trivial pre-headers
1994 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1995 (remove_forwarder_block): Properly update the latch of a loop.
1997 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2000 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2001 (cselib_preserved_hash_table): New.
2002 (preserve_constants_and_equivs): Move preserved vals to it.
2003 (cselib_find_slot): Look it up first.
2004 (cselib_init): Initialize it.
2005 (cselib_finish): Release it.
2006 (dump_cselib_table): Dump it.
2008 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2011 * cselib.c (remove_useless_values): Skip to avoid quadratic
2012 behavior if the condition moved from...
2013 (cselib_process_insn): ... here holds.
2015 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2018 * var-tracking.c (vt_initialize): Apply the same condition to
2019 preserve the CFA base value.
2021 2014-02-28 Joey Ye <joey.ye@arm.com>
2024 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2025 if reload in progress or completed.
2027 2014-02-28 Tobias Burnus <burnus@net-b.de>
2030 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2033 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2035 * doc/tm.texi.in (Condition Code Status): Update documention for
2036 relative locations of cc0-setter and cc0-user.
2038 2014-02-27 Jeff Law <law@redhat.com>
2040 PR rtl-optimization/52714
2041 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2042 into two independent simple sets, if I3 is a jump, ensure the
2043 pattern we place into I3 is a (set (pc) ...).
2045 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2046 Jeff Law <law@redhat.com>
2048 PR rtl-optimization/49847
2049 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2050 are in different blocks.
2051 * doc/tm.texi (Condition Code Status): Update documention for
2052 relative locations of cc0-setter and cc0-user.
2054 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2057 * lra.c (lra_emit_add): Check SUBREG too.
2059 2014-02-27 Andreas Schwab <schwab@suse.de>
2061 * config/m68k/m68k.c (m68k_option_override): Disable
2062 -flive-range-shrinkage for classic m68k.
2063 (m68k_override_options_after_change): Likewise.
2065 2014-02-27 Marek Polacek <polacek@redhat.com>
2068 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2069 -Wmaybe-uninitialized.
2071 2014-02-27 Alan Modra <amodra@gmail.com>
2074 * reload1.c (emit_input_reload_insns): When reload_override_in,
2075 set old to rl->in_reg when rl->in_reg is a subreg.
2077 2014-02-26 Richard Biener <rguenther@suse.de>
2080 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2082 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2084 * common/config/i386/predicates.md (const1256_operand): Remove.
2085 (const2356_operand): New.
2086 (const_1_to_2_operand): Remove.
2087 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2088 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2089 (*avx512pf_gatherpf<mode>sf): Ditto.
2090 (avx512pf_gatherpf<mode>df): Ditto.
2091 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2092 (*avx512pf_gatherpf<mode>df): Ditto.
2093 (avx512pf_scatterpf<mode>sf): Ditto.
2094 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2095 (*avx512pf_scatterpf<mode>sf): Ditto.
2096 (avx512pf_scatterpf<mode>df): Ditto.
2097 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2098 (*avx512pf_scatterpf<mode>df): Ditto.
2099 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2101 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2103 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2104 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2105 (_mm512_mask_testn_epi64_mask): Move to ...
2106 * config/i386/avx512cdintrin.h: Here.
2107 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2108 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2109 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2110 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2111 TARGET_AVX512F from TARGET_AVX512CD.
2113 2014-02-26 Richard Biener <rguenther@suse.de>
2116 * ipa.c (walk_polymorphic_call_targets): Properly guard
2117 call to inline_update_overall_summary.
2119 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2122 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2123 and latches only if requested. Fix latch if it is removed.
2124 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2125 LOOPS_HAVE_PREHEADERS.
2127 2014-02-25 Andrew Pinski <apinski@cavium.com>
2129 * builtins.c (expand_builtin_thread_pointer): Create a new target
2130 when the target is NULL.
2132 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2134 PR rtl-optimization/60317
2135 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2136 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2137 * lra-assigns.c: Include params.h.
2138 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2139 other reload pseudos considerations.
2141 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2143 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2144 to use canonical form for nor<mode>3.
2146 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2149 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2152 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2154 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2155 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2156 (ix86_handle_option): Handle OPT_mprefetchwt1.
2157 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2158 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2160 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2161 OPTION_MASK_ISA_PREFETCHWT1.
2162 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2163 (PTA_PREFETCHWT1): New.
2164 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2165 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2166 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2167 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2168 (*prefetch_avx512pf_<mode>_: Change into ...
2169 (*prefetch_prefetchwt1_<mode>: This.
2170 * config/i386/i386.opt (mprefetchwt1): New.
2171 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2172 (_mm_prefetch): Handle intent to write.
2173 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2175 2014-02-25 Richard Biener <rguenther@suse.de>
2178 * emit-rtl.c (mem_attrs_htab): Remove.
2179 (mem_attrs_htab_hash): Likewise.
2180 (mem_attrs_htab_eq): Likewise.
2181 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2182 (init_emit_once): Do not allocate mem_attrs_htab.
2184 2014-02-25 Richard Biener <rguenther@suse.de>
2187 * lto-opts.c (lto_write_options): Output non-explicit conservative
2188 -fwrapv, -fno-trapv and -fno-strict-overflow.
2189 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2190 (run_gcc): And pass them through.
2192 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2194 * sel-sched.c (calculate_new_fences): New parameter ptime.
2195 Calculate it as a maximum over all fence cycles.
2196 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2197 Print the final schedule timing when sched_verbose.
2199 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2201 PR rtl-optimization/60292
2202 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2203 bit fot the fence instruction.
2205 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2207 * calls.h: Fix typo in comment.
2209 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2211 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2212 adjusting offsetable addresses.
2214 2014-02-24 Guozhi Wei <carrot@google.com>
2216 * sparseset.h (sparseset_pop): Fix the wrong index.
2218 2014-02-24 Walter Lee <walt@tilera.com>
2220 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2221 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2223 * common/config/tilegx/tilegx-common.c
2224 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2225 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2227 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2228 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2229 (tilegx_gimplify_va_arg_expr): Handle big endian.
2230 (tilegx_expand_unaligned_load): Ditto.
2231 (tilegx_expand_unaligned_store): Ditto.
2232 (TARGET_RETURN_IN_MSB): New.
2233 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2234 (TARGET_ENDIAN_DEFAULT): New.
2235 (TARGET_BIG_ENDIAN): Handle big endian.
2236 (BYTES_BIG_ENDIAN): Ditto.
2237 (WORDS_BIG_ENDIAN): Ditto.
2238 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2241 * config/tilegx/tilegx.md (extv): Handle big endian.
2243 (insn_st<n>): Ditto.
2244 (insn_st<n>_add<bitsuffix>): Ditto.
2245 (insn_stnt<n>): Ditto.
2246 (insn_stnt<n>_add<bitsuffix>):Ditto.
2247 (vec_interleave_highv8qi): Handle big endian.
2248 (vec_interleave_highv8qi_be): New.
2249 (vec_interleave_highv8qi_le): New.
2250 (insn_v1int_h): Handle big endian.
2251 (vec_interleave_lowv8qi): Handle big endian.
2252 (vec_interleave_lowv8qi_be): New.
2253 (vec_interleave_lowv8qi_le): New.
2254 (insn_v1int_l): Handle big endian.
2255 (vec_interleave_highv4hi): Handle big endian.
2256 (vec_interleave_highv4hi_be): New.
2257 (vec_interleave_highv4hi_le): New.
2258 (insn_v2int_h): Handle big endian.
2259 (vec_interleave_lowv4hi): Handle big endian.
2260 (vec_interleave_lowv4hi_be): New.
2261 (vec_interleave_lowv4hi_le): New.
2262 (insn_v2int_l): Handle big endian.
2263 (vec_interleave_highv2si): Handle big endian.
2264 (vec_interleave_highv2si_be): New.
2265 (vec_interleave_highv2si_le): New.
2266 (insn_v4int_h): Handle big endian.
2267 (vec_interleave_lowv2si): Handle big endian.
2268 (vec_interleave_lowv2si_be): New.
2269 (vec_interleave_lowv2si_le): New.
2270 (insn_v4int_l): Handle big endian.
2271 * config/tilegx/tilegx.opt (mbig-endian): New option.
2272 (mlittle-endian): New option.
2273 * doc/install.texi: Document tilegxbe-linux.
2274 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2276 2014-02-24 Martin Jambor <mjambor@suse.cz>
2279 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2280 there are no parameter descriptors.
2282 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2284 PR rtl-optimization/60268
2285 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2286 initialization to ...
2287 (sched_rgn_init): ... here.
2288 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2290 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2292 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2295 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2297 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2300 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2302 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2303 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2305 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2307 * config/microblaze/predicates.md: Add cmp_op predicate.
2308 * config/microblaze/microblaze.md: Add branch_compare instruction
2309 which uses cmp_op predicate and emits cmp insn before branch.
2310 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2311 to microblaze_expand_conditional_branch and consolidate logic.
2312 (microblaze_expand_conditional_branch): emit branch_compare
2313 insn instead of handling cmp op separate from branch insn.
2315 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2317 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2320 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2322 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2323 define_insn with define_expand and new define_insn
2324 *altivec_lve<VI_char>x_internal.
2325 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2326 and new define_insn *altivec_stve<VI_char>x_internal.
2327 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2329 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2331 (altivec_expand_stvex_be): New function.
2333 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2335 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2336 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2337 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2338 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2340 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2343 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2344 instead of emit_move_insn.
2346 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2348 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2350 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2351 gen_altivec_vsumsws.
2353 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2355 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2356 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2357 (altivec_lvxl_<mode>): New define_expand incorporating
2358 -maltivec=be semantics where needed.
2359 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2360 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2361 semantics where needed.
2362 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2363 (altivec_stvx_<mode>): New define_expand incorporating
2364 -maltivec=be semantics where needed.
2365 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2366 VM2 iterator instead of V4SI.
2367 (altivec_stvxl_<mode>): New define_expand incorporating
2368 -maltivec=be semantics where needed.
2369 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2370 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2371 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2372 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2373 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2374 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2375 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2376 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2377 ALTIVEC_BUILTIN_STVXL.
2378 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2379 (altivec_expand_stvx_be): Likewise.
2380 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2381 (altivec_expand_lvx_be): Likewise.
2382 (altivec_expand_stvx_be): Likewise.
2383 (altivec_expand_builtin): Add cases for
2384 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2385 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2386 (altivec_init_builtins): Add definitions for
2387 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2388 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2390 2014-02-21 Catherine Moore <clm@codesourcery.com>
2392 * doc/invoke.texi (mvirt, mno-virt): Document.
2393 * config/mips/mips.opt (mvirt): New option.
2394 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2396 2014-02-21 Richard Biener <rguenther@suse.de>
2398 PR tree-optimization/60276
2399 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2400 (STMT_VINFO_MIN_NEG_DIST): New macro.
2401 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2402 STMT_VINFO_MIN_NEG_DIST.
2403 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2404 made for negative dependence distances still hold.
2406 2014-02-21 Richard Biener <rguenther@suse.de>
2409 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2410 DECL_INITIAL for globals not in the current function context.
2412 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2414 PR tree-optimization/56490
2415 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2416 * tree-ssa-uninit.c: Include params.h.
2417 (compute_control_dep_chain): Add num_calls argument, return false
2418 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2419 num_calls to recursive call.
2420 (find_predicates): Change dep_chain into normal array,
2421 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2422 variable and adjust compute_control_dep_chain caller.
2423 (find_def_preds): Likewise.
2425 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2427 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2428 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2430 2014-02-21 Nick Clifton <nickc@redhat.com>
2432 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2433 (pushhi1): Likewise.
2434 (popqi1): Add mode to pre_dec.
2437 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2439 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2440 mode for mask of V8SFmode permutation.
2442 2014-02-20 Richard Henderson <rth@redhat.com>
2445 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2446 a new pseudo for OLDVAL.
2448 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2451 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2452 gen_reg_rtx if d->testing_p.
2453 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2454 if d->testing_p and we will certainly return true.
2455 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2458 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2460 * emit-rtl.c (gen_reg_rtx): Assert that
2461 crtl->emit.regno_pointer_align_length is non-zero.
2463 2014-02-20 Richard Henderson <rth@redhat.com>
2466 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2467 on failure the store back into EXPECT.
2469 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2470 Sandra Loosemore <sandra@codesourcery.com>
2472 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2473 * config/nios2/nios2.c (nios2_function_profiler): Add
2474 -fPIC (flag_pic == 2) support.
2475 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2476 (nios2_large_offset_p): New function.
2477 (nios2_unspec_reloc_p): Move up position, update to use
2478 nios2_large_offset_p.
2479 (nios2_unspec_address): Remove function.
2480 (nios2_unspec_offset): New function.
2481 (nios2_large_got_address): New function.
2482 (nios2_got_address): Add large offset support.
2483 (nios2_legitimize_tls_address): Update usage of removed and new
2485 (nios2_symbol_binds_local_p): New function.
2486 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2487 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2488 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2489 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2490 processing for (const (unspec ...)).
2491 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2493 2014-02-20 Richard Biener <rguenther@suse.de>
2495 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2496 doing the substitution.
2497 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2499 2014-02-20 Martin Jambor <mjambor@suse.cz>
2502 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2503 info when checking whether lattices are bottom.
2505 2014-02-20 Richard Biener <rguenther@suse.de>
2508 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2511 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2514 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2515 parameter specifying the scaling.
2516 (inline_call): Update.
2517 (want_inline_recursively): Guard division by zero.
2518 (recursive_inlining): Update.
2519 * ipa-inline.h (clone_inlined_nodes): Update.
2521 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2524 * config/i386/i386.c (classify_argument): Pass structures of size
2525 64 bytes or less in register.
2527 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2528 Kirill Yukhin <kirill.yukhin@intel.com>
2530 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2531 (_mm_rcp28_round_ss): Ditto.
2532 (_mm_rsqrt28_round_sd): Ditto.
2533 (_mm_rsqrt28_round_ss): Ditto.
2534 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2535 (_mm_rcp14_round_ss): Ditto.
2536 (_mm_rsqrt14_round_sd): Ditto.
2537 (_mm_rsqrt14_round_ss): Ditto.
2538 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2539 the first input operand, get rid of match_dup.
2540 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2542 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2544 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2545 operand as the first input operand, set type attribute.
2546 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2548 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2549 operand as the first input operand, set type attribute.
2551 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2553 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2556 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2559 * config/i386/i386.c (construct_container): Remove TFmode check
2560 for X86_64_INTEGER_CLASS.
2562 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2565 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2566 only when -Wpsabi is enabled.
2568 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2571 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2572 passing arrays in registers are the same as for structs, so remove the
2573 special case for them.
2575 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2577 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2578 destination type, extract only the valid bits if the source type is not
2579 integral and has a different mode.
2581 2014-02-19 Richard Biener <rguenther@suse.de>
2584 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2587 2014-02-19 Richard Biener <rguenther@suse.de>
2590 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2591 (ipa_modify_call_arguments): Emit an argument load explicitely and
2592 preserve virtual SSA form there and for the replacement call.
2593 Do not update SSA form nor free dominance info.
2595 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2597 * ipa.c (function_and_variable_visibility): Also clear WEAK
2598 flag when disolving COMDAT_GROUP.
2600 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2602 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2603 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2605 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2606 do more sanity checks.
2607 (detect_type_change): Return true when giving up early.
2608 (compute_complex_assign_jump_func): Fix type parameter of
2609 ipa_set_ancestor_jf.
2610 (compute_complex_ancestor_jump_func): Likewise.
2611 (update_jump_functions_after_inlining): Fix updating of
2613 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2615 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2617 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2618 inline clones when edge disappears.
2620 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2623 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2624 Split 64-bit moves into 2 patterns. Do not allow the use of
2625 direct move for TDmode in little endian, since the decimal value
2626 has little endian bytes within a word, but the 64-bit pieces are
2627 ordered in a big endian fashion, and normal subreg's of TDmode are
2629 (mov<mode>_64bit_dm): Likewise.
2630 (movtd_64bit_nodm): Likewise.
2632 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2634 PR tree-optimization/60174
2635 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2636 statement of an SSA_NAME that occurs in an abnormal PHI node.
2638 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2641 * final.c (SEEN_BB): Remove.
2642 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2643 (final_scan_insn): Don't force_source_line on second
2644 NOTE_INSN_BASIC_BLOCK.
2646 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2649 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2650 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2651 (type_natural_mode): Warn ABI change when %zmm register is not
2652 available for AVX512F vector value passing.
2654 2014-02-18 Kai Tietz <ktietz@redhat.com>
2657 * config/i386/i386.c (ix86_expand_prologue): Use value in
2658 rax register as displacement when restoring %r10 or %rax.
2659 Fix wrong offset when restoring both registers.
2661 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2663 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2664 assertion with conditional return.
2666 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2667 Uros Bizjak <ubizjak@gmail.com>
2670 * config/i386/driver-i386.c (host_detect_local_cpu): If
2671 YMM state is not saved by the OS, also clear has_f16c. Move
2672 CPUID 0x80000001 handling before YMM state saving checking.
2674 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2676 PR rtl-optimization/58960
2677 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2678 factored out from ...
2679 (sched_init): ... here.
2680 (free_global_sched_pressure_data): New, factored out from ...
2681 (sched_finish): ... here.
2682 * sched-int.h (free_global_sched_pressure_data): Declare.
2683 * sched-rgn.c (nr_regions_initial): New static global.
2684 (haifa_find_rgns): Initialize it.
2685 (schedule_region): Disable sched-pressure for the newly
2688 2014-02-17 Richard Biener <rguenther@suse.de>
2690 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2691 release SSA defs of pattern stmts.
2693 2014-02-17 Richard Biener <rguenther@suse.de>
2695 * tree-inline.c (expand_call_inline): Release the virtual
2696 operand defined by the call we are about to inline.
2698 2014-02-17 Richard Biener <rguenther@suse.de>
2700 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2702 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2703 Ilya Tocar <ilya.tocar@intel.com>
2705 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2706 arguments order in builtin.
2707 (_mm512_permutexvar_epi64): Ditto.
2708 (_mm512_mask_permutexvar_epi64): Ditto
2709 (_mm512_maskz_permutexvar_epi32): Ditto
2710 (_mm512_permutexvar_epi32): Ditto
2711 (_mm512_mask_permutexvar_epi32): Ditto
2713 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2715 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2716 (p8_vmrgow): Likewise.
2718 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2720 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2723 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2726 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2727 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2728 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2729 using direct move instructions on ISA 2.07. Also adjust
2730 instruction length for 64-bit.
2731 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2732 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2734 2014-02-15 Alan Modra <amodra@gmail.com>
2738 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2739 find_replacement on parts of insn rtl that might be reloaded.
2741 2014-02-15 Richard Biener <rguenther@suse.de>
2743 PR tree-optimization/60183
2744 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2745 (tree_ssa_phiprop): Calculate and free post-dominators.
2747 2014-02-14 Jeff Law <law@redhat.com>
2749 PR rtl-optimization/60131
2750 * ree.c (get_extended_src_reg): New function.
2751 (combine_reaching_defs): Use it rather than assuming location of REG.
2752 (find_and_remove_re): Verify first operand of extension is
2753 a REG before adding the insns to the copy list.
2755 2014-02-14 Roland McGrath <mcgrathr@google.com>
2757 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2758 * configure: Regenerated.
2759 * config.in: Regenerated.
2760 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2761 instead of ASM_SHORT.
2763 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2764 Richard Earnshaw <rearnsha@arm.com>
2766 PR rtl-optimization/59535
2767 * lra-constraints.c (process_alt_operands): Encourage alternative
2768 when unassigned pseudo class is superset of the alternative class.
2769 (inherit_reload_reg): Don't inherit when optimizing for code size.
2770 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2771 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2772 modes not less than 4 for Thumb1.
2774 2014-02-14 Kyle McMartin <kyle@redhat.com>
2777 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2779 2014-02-14 Richard Biener <rguenther@suse.de>
2781 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2782 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2783 Do not strip INDIRECT_REFs.
2785 2014-02-14 Richard Biener <rguenther@suse.de>
2788 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2789 DECL_FUNCTION_SPECIFIC_TARGET.
2790 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2791 * tree-streamer-out.c (pack_ts_target_option): Remove.
2792 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2793 (write_ts_function_decl_tree_pointers): Do not stream
2794 DECL_FUNCTION_SPECIFIC_TARGET.
2795 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2796 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2797 (lto_input_ts_function_decl_tree_pointers): Do not stream
2798 DECL_FUNCTION_SPECIFIC_TARGET.
2800 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2802 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2803 (get_initial_def_for_induction, vectorizable_induction): Ignore
2804 debug stmts when looking for exit_phi.
2805 (vectorizable_live_operation): Fix up condition.
2807 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2809 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2810 nreverse() because it changes the content of original tree list.
2812 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2814 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2815 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2817 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2819 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2820 GNU coding standards.
2822 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2825 * dwarf2out.c (gen_subprogram_die): Don't call
2826 add_calling_convention_attribute if subr_die is old_die.
2828 2014-02-13 Sharad Singhai <singhai@google.com>
2830 * doc/optinfo.texi: Fix order of nodes.
2832 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2834 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2835 operands[2], not operands[3].
2837 2014-02-13 Richard Biener <rguenther@suse.de>
2840 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2841 mention the possibility of an in-tree build.
2842 (CLooG): Update recommended version to 0.18.1, mention the
2843 possibility of an in-tree build and clarify that the ISL
2844 bundled with CLooG does not work.
2846 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2849 * expr.c (compress_float_constant): If x is a hard register,
2850 extend into a pseudo and then move to x.
2852 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2854 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2855 caused by bad second argument to warning_at() with -mhotpatch and
2856 nested functions (e.g. with gfortran).
2858 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2860 * opts.c (option_name): Remove "enabled by default" rider.
2862 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2864 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2866 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2867 Uros Bizjak <ubizjak@gmail.com>
2870 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2871 * configure: Regenerated.
2873 2014-02-12 Richard Biener <rguenther@suse.de>
2875 * vec.c (vec_prefix::calculate_allocation): Move as
2876 inline variant to vec.h.
2877 (vec_prefix::calculate_allocation_1): New out-of-line version.
2878 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2879 (vec_prefix::m_has_auto_buf): Rename to ...
2880 (vec_prefix::m_using_auto_storage): ... this.
2881 (vec_prefix::calculate_allocation): Inline the easy cases
2882 and dispatch to calculate_allocation_1 which doesn't need the
2884 (va_heap::reserve): Use gcc_checking_assert.
2885 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2886 m_using_auto_storage.
2887 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2889 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2890 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2891 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2893 2014-02-12 Richard Biener <rguenther@suse.de>
2895 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2896 when we found a dependence.
2898 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2900 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2902 (maybe_fold_stmt): ... into this new function.
2903 * omp-low.c (lower_omp): Update comment.
2905 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2908 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2911 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2913 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2914 identifiers in comments.
2915 (cortexa53_extra_costs): Likewise.
2916 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2917 (cortexa7_extra_costs): Likewise.
2918 (cortexa12_extra_costs): Likewise.
2919 (cortexa15_extra_costs): Likewise.
2920 (v7m_extra_costs): Likewise.
2922 2014-02-12 Richard Biener <rguenther@suse.de>
2925 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2926 of posix_memalign being successful.
2927 (lower_stmt): Restrict lowering of posix_memalign to when
2928 -ftree-bit-ccp is enabled.
2930 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2932 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2934 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2936 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2938 PR rtl-optimization/60116
2939 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2940 other_insn once the combination has been validated.
2942 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2945 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2947 * ipa-devirt.c: Include demangle.h
2948 (odr_violation_reported): New static variable.
2949 (add_type_duplicate): Update odr_violations.
2950 (maybe_record_node): Add completep parameter; update it.
2951 (record_target_from_binfo): Add COMPLETEP parameter;
2952 update it as needed.
2953 (possible_polymorphic_call_targets_1): Likewise.
2954 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2955 rename FINAL to COMPLETE.
2956 (record_targets_from_bases): Sanity check we found the binfo;
2957 fix COMPLETEP updating.
2958 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2959 parameter, fix computing of COMPLETEP.
2960 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2961 at LTO time do demangling.
2962 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2963 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2965 (gimple_get_virt_method_for_binfo): Likewise.
2966 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2967 gimple_get_virt_method_for_vtable): Update prototypes.
2969 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2972 * genautomata.c (add_presence_absence): Fix typo with
2973 {final_}presence_list.
2975 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2978 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2979 for VSX/Altivec vectors that land in GPR registers.
2981 2014-02-11 Richard Henderson <rth@redhat.com>
2982 Jakub Jelinek <jakub@redhat.com>
2985 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2986 around drhs if type conversion to lacc->type is not useless.
2988 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2990 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2992 (cortex-a57.cortex-a53): Likewise.
2993 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2995 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2997 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3000 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3002 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3003 add_options_for_arm_vfp3.
3005 2014-02-11 Jeff Law <law@redhat.com>
3008 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3009 object with an undesirable mode.
3011 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3014 * config/i386/sol2-9.h: New file.
3015 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3016 *-*-solaris2.9*): Use it.
3018 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3020 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3021 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3023 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3025 * config/microblaze/microblaze.c: Extend mcpu version format
3027 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3029 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3031 2014-02-10 Richard Henderson <rth@redhat.com>
3034 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3035 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3036 ms-abi vs -mno-accumulate-outgoing-args.
3037 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3038 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3041 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3044 * cfgexpand.c (expand_asm_operands): Attach source location to
3045 ASM_INPUT rtx objects.
3046 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3048 2014-02-10 Nick Clifton <nickc@redhat.com>
3050 * config/mn10300/mn10300.c (popcount): New function.
3051 (mn10300_expand_prologue): Include saved registers in stack usage
3054 2014-02-10 Jeff Law <law@redhat.com>
3057 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3058 when changing the SET_DEST of a prior insn to avoid an input reload.
3060 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3062 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3063 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3064 -mcall-openbsd, or -mcall-linux.
3065 (CC1_ENDIAN_BIG_SPEC): Remove.
3066 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3067 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3068 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3069 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3070 and %cc1_endian_default.
3071 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3073 2014-02-10 Richard Biener <rguenther@suse.de>
3075 PR tree-optimization/60115
3076 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3077 MEM_REF handling. Properly verify that the accesses are not
3078 out of the objects bound.
3080 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3082 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3085 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3087 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3088 proper constants and fix formatting.
3089 (possible_polymorphic_call_targets): Fix formatting.
3091 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3092 Ilya Tocar <ilya.tocar@intel.com>
3094 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3095 (_mm512_loadu_epi32): Renamed into...
3096 (_mm512_loadu_si512): This.
3097 (_mm512_storeu_epi32): Renamed into...
3098 (_mm512_storeu_si512): This.
3099 (_mm512_maskz_ceil_ps): Removed.
3100 (_mm512_maskz_ceil_pd): Ditto.
3101 (_mm512_maskz_floor_ps): Ditto.
3102 (_mm512_maskz_floor_pd): Ditto.
3103 (_mm512_floor_round_ps): Ditto.
3104 (_mm512_floor_round_pd): Ditto.
3105 (_mm512_ceil_round_ps): Ditto.
3106 (_mm512_ceil_round_pd): Ditto.
3107 (_mm512_mask_floor_round_ps): Ditto.
3108 (_mm512_mask_floor_round_pd): Ditto.
3109 (_mm512_mask_ceil_round_ps): Ditto.
3110 (_mm512_mask_ceil_round_pd): Ditto.
3111 (_mm512_maskz_floor_round_ps): Ditto.
3112 (_mm512_maskz_floor_round_pd): Ditto.
3113 (_mm512_maskz_ceil_round_ps): Ditto.
3114 (_mm512_maskz_ceil_round_pd): Ditto.
3115 (_mm512_expand_pd): Ditto.
3116 (_mm512_expand_ps): Ditto.
3117 * config/i386/i386.c (ix86_builtins): Remove
3118 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3119 (bdesc_args): Ditto.
3120 * config/i386/predicates.md (const1256_operand): New.
3121 (const_1_to_2_operand): Ditto.
3122 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3123 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3124 (*avx512pf_gatherpf<mode>sf): Ditto.
3125 (avx512pf_gatherpf<mode>df): Ditto.
3126 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3127 (*avx512pf_gatherpf<mode>df): Ditto.
3128 (avx512pf_scatterpf<mode>sf): Ditto.
3129 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3130 (*avx512pf_scatterpf<mode>sf): Ditto.
3131 (avx512pf_scatterpf<mode>df): Ditto.
3132 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3133 (*avx512pf_scatterpf<mode>df): Ditto.
3134 (avx512f_expand<mode>): Removed.
3135 (<shift_insn><mode>3<mask_name>): Change predicate type.
3137 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3139 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3140 not at the end of datarefs vector use ordered_remove to avoid
3141 reordering datarefs vector.
3144 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3145 mark local addressable non-static vars as GOVD_PRIVATE
3146 instead of GOVD_LOCAL.
3147 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3148 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3152 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3153 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3154 assume_aligned or alloc_align attributes.
3155 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3156 arguments. Handle also assume_aligned and alloc_align attributes.
3157 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3158 calls to functions with assume_aligned or alloc_align attributes.
3159 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3161 2014-02-08 Terry Guo <terry.guo@arm.com>
3163 * doc/invoke.texi: Document ARM -march=armv7e-m.
3165 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3167 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3168 flag on __cilkrts_rethrow builtin.
3171 * ipa-cp.c (determine_versionability): Fail at -O0
3172 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3173 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3176 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3179 * tree-inline.c (copy_forbidden): Fail for
3180 __attribute__((optimize (0))) functions.
3182 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3184 * varpool.c: Include pointer-set.h.
3185 (varpool_remove_unreferenced_decls): Variables in other partitions
3186 will not be output; be however careful to not lose information
3189 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3191 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3192 lookup in the vtable constructor.
3194 2014-02-07 Jeff Law <law@redhat.com>
3197 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3198 define_insn_and_split.
3200 * ipa-inline.c (inline_small_functions): Fix typos.
3202 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3204 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3205 (s390_can_use_return_insn): Declare.
3206 * config/s390/s390.h (EPILOGUE_USES): Define.
3207 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3209 (s390_chunkify_start): Handle return JUMP_LABELs.
3210 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3211 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3212 (s390_can_use_return_insn): New functions.
3213 (s390_fix_long_loop_prediction): Handle conditional returns.
3214 (TARGET_SET_UP_BY_PROLOGUE): Define.
3215 * config/s390/s390.md (ANY_RETURN): New code iterator.
3216 (*creturn, *csimple_return, return, simple_return): New patterns.
3218 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3220 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3221 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3222 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3223 REG_CFA_RESTORE list when deciding not to restore a register.
3225 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3227 * config/s390/s390.c: Include tree-pass.h and context.h.
3228 (s390_early_mach): New function, split out from...
3229 (s390_emit_prologue): ...here.
3230 (pass_data_s390_early_mach): New pass structure.
3231 (pass_s390_early_mach): New class.
3232 (s390_option_override): Create and register early_mach pass.
3233 Move to end of file.
3235 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3237 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3238 to match for the exit block.
3240 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3242 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3243 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3244 Reject misaligned operands.
3246 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3248 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3250 2014-02-07 Richard Biener <rguenther@suse.de>
3253 * gimple-low.c (lower_builtin_posix_memalign): New function.
3254 (lower_stmt): Call it to lower posix_memalign in a way
3255 to make alignment info accessible.
3257 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3260 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3261 __builtin_setjmp_receiver.
3263 2014-02-07 Richard Biener <rguenther@suse.de>
3266 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3267 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3268 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3269 Handle BUILT_IN_POSIX_MEMALIGN.
3270 (find_func_clobbers): Likewise.
3271 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3272 (call_may_clobber_ref_p_1): Likewise.
3274 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3277 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3280 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3283 * lto-cgraph.c (lto_output_node): Use
3284 symtab_get_symbol_partitioning_class.
3285 (lto_output_varpool_node): likewise.
3286 (symtab_get_symbol_partitioning_class): Move here from
3288 * cgraph.h (symbol_partitioning_class): Likewise.
3289 (symtab_get_symbol_partitioning_class): Declare.
3291 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3293 * ggc.h (ggc_internal_cleared_alloc): New macro.
3294 * vec.h (vec_safe_copy): Handle memory stats.
3295 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3296 * target-globals.c (save_target_globals): Likewise.
3298 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3301 * expr.c (emit_move_resolve_push): Export; be bit more selective
3302 on when to clear alias set.
3303 * expr.h (emit_move_resolve_push): Declare.
3304 * function.h (struct function): Add tail_call_marked.
3305 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3306 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3307 * config/i386/i386.md (TImode move expander): De not call
3309 (FP push expanders): Preserve memory attributes.
3310 * config/i386/sse.md (push<mode>1): Remove.
3311 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3312 (ix86_expand_push): Remove.
3313 * config/i386/mmx.md (push<mode>1): Remove.
3315 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3317 PR rtl-optimization/60030
3318 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3319 lopart with paradoxical subreg before shifting it up by hprec.
3321 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3323 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3324 Remove extra newline at end of file.
3325 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3326 (arm_issue_rate): Handle cortexa57.
3327 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3328 (cortex-a57.cortex-a53): Likewise.
3330 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3333 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3334 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3336 (arm_expand_prologue): Adjust all callers.
3337 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3338 info, registers also at the lowest numbered registers side. Use
3339 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3343 * var-tracking.c (adjust_mems): Before adding a SET to
3344 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3346 2014-02-06 Alan Modra <amodra@gmail.com>
3349 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3350 change SDmode to DDmode when lra_in_progress.
3352 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3355 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3356 free_data_ref on the dr first, and before goto again also set dr
3357 to the next dr. For simd_lane_access, free old datarefs[i] before
3358 overwriting it. For get_vectype_for_scalar_type failure, don't
3359 free_data_ref if simd_lane_access.
3361 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3364 * tree.h (opts_for_fn): New inline function.
3365 (opt_for_fn): Define.
3366 * config/i386/i386.c (ix86_function_regparm): Use
3367 opt_for_fn (decl, optimize) instead of optimize.
3369 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3371 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3372 for SYMBOL_REF in large memory model.
3374 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3376 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3378 (cortex-a57): Likewise.
3379 (cortex-a57.cortex-a53): Likewise.
3381 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3382 Kugan Vivekanandarajah <kuganv@linaro.org>
3384 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3386 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3388 2014-02-06 Richard Biener <rguenther@suse.de>
3390 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3391 set_loop_copy and initialize_original_copy_tables.
3393 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3395 * config/aarch64/aarch64-simd.md
3396 (aarch64_ashr_simddi): Change QI to SI.
3398 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3399 Jakub Jelinek <jakub@redhat.com>
3402 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3405 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3407 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3408 CODE_FOR_altivec_vpku[hw]um to
3409 CODE_FOR_altivec_vpku[hw]um_direct.
3410 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3411 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3412 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3413 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3415 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3417 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3418 generation for -maltivec=be.
3419 (altivec_vsumsws): Simplify redundant test.
3421 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3423 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3424 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3425 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3426 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3427 gen_altivec_vpkuwum.
3428 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3430 (altivec_vpks<VI_char>ss): Likewise.
3431 (altivec_vpks<VI_char>us): Likewise.
3432 (altivec_vpku<VI_char>us): Likewise.
3433 (altivec_vpku<VI_char>um): Likewise.
3434 (altivec_vpku<VI_char>um_direct): New (copy of
3435 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3437 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3438 target is little endian and -maltivec=be is not specified.
3439 (*altivec_vupkhs<VU_char>_direct): New (copy of
3440 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3441 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3442 target is little endian and -maltivec=be is not specified.
3443 (*altivec_vupkls<VU_char>_direct): New (copy of
3444 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3445 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3446 little endian and -maltivec=be is not specified.
3447 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3448 little endian and -maltivec=be is not specified.
3450 2014-02-05 Richard Henderson <rth@redhat.com>
3453 * combine-stack-adj.c: Revert r206943.
3454 * sched-int.h (struct deps_desc): Add last_args_size.
3455 * sched-deps.c (init_deps): Initialize it.
3456 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3457 contain REG_ARGS_SIZE notes.
3459 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3461 * lto-cgraph.c (asm_nodes_output): Make global.
3462 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3463 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3464 (driver_handle_option): Handle OPT_fwpa.
3466 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3469 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3470 a comment typo and formatting issue. If odr_hash hasn't been
3471 created, return vNULL and set *completep to false.
3474 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3475 bb with no successors.
3477 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3480 * doc/invoke.texi (-march): Clarify documentation for ARM.
3484 2014-02-05 Richard Biener <rguenther@suse.de>
3486 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3487 when not vectorizing because of too many alias checks.
3488 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3489 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3491 2014-02-05 Nick Clifton <nickc@redhat.com>
3493 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3494 accept extended registers in any mode when compiling for the MN10300.
3496 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3498 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3499 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3500 sanitization attributes.
3501 (can_inline_edge_p): Likewise.
3502 (sanitize_attrs_match_for_inline_p): New function.
3504 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3506 * ipa-prop.c (detect_type_change): Shor circuit testing of
3507 type changes on THIS pointer.
3509 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3512 * config/pa/pa.c (legitimize_tls_address): Return original address
3513 if not passed a SYMBOL_REF rtx.
3514 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3516 (pa_emit_move_sequence): Simplify TLS source operands.
3517 (pa_legitimate_constant_p): Reject all TLS constants.
3518 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3519 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3521 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3523 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3524 groups when we know they are controlled by LTO.
3525 * varasm.c (default_binds_local_p_1): If object is in other partition,
3526 it will be resolved locally.
3528 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3530 * config/host-linux.c (linux_gt_pch_use_address): Don't
3531 use SSIZE_MAX because it is not always defined.
3533 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3536 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3537 threshold for pseudo splitting.
3538 (update_ebb_live_info): Process call argument hard registers and
3539 hard registers from insn definition too.
3540 (max_small_class_regs_num): New constant.
3541 (inherit_in_ebb): Update live hard regs through EBBs. Update
3542 reloads_num only for small register classes. Don't split for
3545 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3548 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3551 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3553 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3556 2014-02-04 Marek Polacek <polacek@redhat.com>
3558 * gdbinit.in (pel): Define.
3560 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3562 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3565 2014-02-04 Richard Biener <rguenther@suse.de>
3568 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3569 in function context local.
3570 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3571 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3572 similar to LTO_imported_decl_ref.
3574 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3576 PR tree-optimization/60002
3577 * cgraphclones.c (build_function_decl_skip_args): Clear
3580 PR tree-optimization/60023
3581 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3582 false to gsi_replace.
3583 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3584 has been in some EH region and vec_stmt could throw, add
3585 vec_stmt into the same EH region.
3586 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3587 has no lhs, ignore it.
3588 * internal-fn.c (expand_MASK_LOAD): Likewise.
3591 * tree-inline.c (copy_forbidden): Fail for
3592 __attribute__((optimize (0))) functions.
3595 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3596 is set, copy one less argument.
3597 (expand_simd_clones): Don't subtract clone_info->inbranch
3598 from simd_clone_struct_alloc argument.
3600 PR rtl-optimization/57915
3601 * recog.c (simplify_while_replacing): If all unary/binary/relational
3602 operation arguments are constant, attempt to simplify those.
3605 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3606 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3608 2014-02-04 Richard Biener <rguenther@suse.de>
3610 PR tree-optimization/60012
3611 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3612 TBAA disambiguation to all DDRs.
3614 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3617 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3618 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3620 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3623 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3625 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3627 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3628 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3630 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3633 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3634 to figure out targets of polymorphic calls with known decl.
3635 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3636 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3637 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3638 (get_polymorphic_call_info): ... here.
3639 (get_polymorphic_call_info_from_invariant): New function.
3641 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3643 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3644 lookup via vtable pointer; check for type consistency
3645 and turn inconsitent facts into UNREACHABLE.
3646 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3647 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3648 type inconsistent querries; return UNREACHABLE instead.
3650 2014-02-03 Richard Henderson <rth@twiddle.net>
3653 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3654 already processed this node.
3655 (normalize_one_pred_1): Pass along mark_set.
3656 (normalize_one_pred): Create and destroy a pointer_set_t.
3657 (normalize_one_pred_chain): Likewise.
3659 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3661 PR gcov-profile/58602
3662 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3664 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3667 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3668 -fno-devirtualize; try to devirtualize by the knowledge of
3669 virtual table pointer given by aggregate propagation.
3670 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3671 (ipa_print_node_jump_functions): Dump also offset that
3672 is relevant for polymorphic calls.
3673 (determine_known_aggregate_parts): Add arg_type parameter; use it
3674 instead of determining the type from pointer type.
3675 (ipa_compute_jump_functions_for_edge): Update call of
3676 determine_known_aggregate_parts.
3677 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3678 (gimple_get_virt_method_for_binfo): ... here; simplify using
3679 vtable_pointer_value_to_vtable.
3680 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3681 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3682 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3683 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3685 (vtable_pointer_value_to_binfo): ... here.
3686 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3688 2014-02-03 Teresa Johnson <tejohnson@google.com>
3690 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3691 redef of outer loop index variable.
3693 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3697 * doc/extend.texi (Function Attributes): Typo.
3699 2014-02-03 Cong Hou <congh@google.com>
3701 PR tree-optimization/60000
3702 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3703 if the vectorized statement is a store. A store statement can only
3704 appear at the end of pattern statements.
3706 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3708 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3709 (ix86_option_override_internal): Default long double to 64-bit for
3710 32-bit Bionic and to 128-bit for 64-bit Bionic.
3712 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3713 TARGET_LONG_DOUBLE_128 is true.
3714 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3716 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3717 (mlong-double-64): Negate -mlong-double-128.
3718 (mlong-double-128): New option.
3720 * config/i386/i386-c.c (ix86_target_macros): Define
3721 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3723 * doc/invoke.texi: Document -mlong-double-128.
3725 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3727 PR rtl-optimization/60024
3728 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3730 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3732 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3734 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3736 PR rtl-optimization/57662
3737 * sel-sched.c (code_motion_path_driver): Do not mark already not
3738 existing blocks in the visiting bitmap.
3740 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3742 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3743 on the insn being emitted.
3745 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3746 Will Deacon <will.deacon@arm.com>
3748 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3750 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3752 * config/arm/arm-tables.opt: Regenerate.
3754 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3756 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3757 for vector types other than V16QImode.
3758 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3759 define_expand, and call altivec_expand_vec_perm_le when producing
3760 code with little endian element order.
3761 (*altivec_vperm_<mode>_internal): New insn having previous
3762 behavior of altivec_vperm_<mode>.
3763 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3764 altivec_expand_vec_perm_le when producing code with little endian
3766 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3767 behavior of altivec_vperm_<mode>_uns.
3769 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3771 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3772 (altivec_vsumsws): Add handling for -maltivec=be with a little
3774 (altivec_vsumsws_direct): New.
3775 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3776 gen_altivec_vsumsws.
3778 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3780 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3781 vtable_pointer_value_to_binfo): New functions.
3782 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3783 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3785 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3787 * config/nios2/nios2.md (load_got_register): Initialize GOT
3788 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3789 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3791 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3793 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3794 preserverd by passthrough, do not propagate the type.
3796 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3798 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3799 (mips_atomic_assign_expand_fenv): New function.
3800 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3802 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3804 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3805 (__builtin_mips_set_fcsr): Likewise.
3806 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3807 MIPS_USI_FTYPE_VOID.
3808 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3809 (mips16_expand_set_fcsr): Likewise.
3810 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3811 (mips16_set_fcsr_stub): Likewise.
3812 (mips16_get_fcsr_one_only_stub): New class.
3813 (mips16_set_fcsr_one_only_stub): Likewise.
3814 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3815 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3816 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3817 (hard_float): New availability predicate.
3818 (mips_builtins): Add get_fcsr and set_fcsr.
3819 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3820 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3821 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3822 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3823 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3826 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3828 * config/mips/mips.c (mips_one_only_stub): New class.
3829 (mips_need_mips16_rdhwr_p): Replace with...
3830 (mips16_rdhwr_stub): ...this new variable.
3831 (mips16_stub_call_address): New function.
3832 (mips16_rdhwr_one_only_stub): New class.
3833 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3834 (mips_output_mips16_rdhwr): Delete.
3835 (mips_finish_stub): New function.
3836 (mips_code_end): Use it to handle rdhwr stubs.
3838 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3841 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3842 when calculating size of integer atomic types.
3844 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3846 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3848 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3850 PR tree-optimization/60003
3851 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3852 * profile.c (branch_prob): Use gimple_call_builtin_p
3853 to check for BUILT_IN_SETJMP_RECEIVER.
3854 * tree-inline.c (copy_bb): Call notice_special_calls.
3856 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3859 * lra-constraints.c (process_alt_operands): Update reload_sum only
3862 2014-01-31 Richard Henderson <rth@redhat.com>
3865 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3866 until after else_eh is processed.
3868 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3870 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3871 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3872 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3873 in smmintrin.h, remove them.
3874 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3875 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3876 * config/i386/i386.md (ROUND_SAE): Fix value.
3877 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3878 (const48_operand): New.
3879 * config/i386/subst.md (round), (round_expand): Use
3880 const_4_or_8_to_11_operand.
3881 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3883 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3885 * config/i386/constraints.md (Yk): Swap meaning with k.
3886 * config/i386/i386.md (movhi_internal): Change Yk to k.
3887 (movqi_internal): Ditto.
3888 (*k<logic><mode>): Ditto.
3891 (kandn<mode>): Ditto.
3892 (*<code>hi_1): Ditto.
3893 (*<code>qi_1): Ditto.
3894 (kxnor<mode>): Ditto.
3895 (kortestzhi): Ditto.
3896 (kortestchi): Ditto.
3898 (*one_cmplhi2_1): Ditto.
3899 (*one_cmplqi2_1): Ditto.
3900 * config/i386/sse.md (): Change k to Yk.
3901 (avx512f_load<mode>_mask): Ditto.
3902 (avx512f_blendm<mode>): Ditto.
3903 (avx512f_store<mode>_mask): Ditto.
3904 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3905 (avx512f_storedqu<mode>_mask): Ditto.
3906 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3908 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3909 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3910 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3911 (avx512f_maskcmp<mode>3): Ditto.
3912 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3913 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3914 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3915 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3916 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3917 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3918 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3919 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3920 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3921 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3922 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3923 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3924 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3925 (vec_extract_lo_<mode>_maskm): Ditto.
3926 (vec_extract_hi_<mode>_maskm): Ditto.
3927 (avx512f_vternlog<mode>_mask): Ditto.
3928 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3929 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3930 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3931 (avx512f_<code>v8div16qi2_mask): Ditto.
3932 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3933 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3934 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3935 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3936 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3937 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3938 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3939 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3940 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3941 (avx512cd_maskb_vec_dupv8di): Ditto.
3942 (avx512cd_maskw_vec_dupv16si): Ditto.
3943 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3944 (avx512f_vpermi2var<mode>3_mask): Ditto.
3945 (avx512f_vpermi2var<mode>3_mask): Ditto.
3946 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3947 (*avx512f_gathersi<mode>): Ditto.
3948 (*avx512f_gathersi<mode>_2): Ditto.
3949 (*avx512f_gatherdi<mode>): Ditto.
3950 (*avx512f_gatherdi<mode>_2): Ditto.
3951 (*avx512f_scattersi<mode>): Ditto.
3952 (*avx512f_scatterdi<mode>): Ditto.
3953 (avx512f_compress<mode>_mask): Ditto.
3954 (avx512f_compressstore<mode>_mask): Ditto.
3955 (avx512f_expand<mode>_mask): Ditto.
3956 * config/i386/subst.md (mask): Change k to Yk.
3957 (mask_scalar_merge): Ditto.
3960 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3962 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3964 2014-01-31 Richard Biener <rguenther@suse.de>
3967 * builtins.c (fold_builtin_memory_op): Make sure to not
3968 use a floating-point mode or a boolean or enumeral type for
3971 2014-01-30 DJ Delorie <dj@redhat.com>
3973 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3974 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3975 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3976 whenever main() has an epilogue.
3978 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3980 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3981 unused variable "field".
3982 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3983 (vsx_mergeh_<mode>): Likewise.
3984 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3985 (altivec_vmrghh): Likewise.
3986 (altivec_vmrghw): Likewise.
3987 (altivec_vmrglb): Likewise.
3988 (altivec_vmrglh): Likewise.
3989 (altivec_vmrglw): Likewise.
3990 (altivec_vspltb): Add missing uses.
3991 (altivec_vsplth): Likewise.
3992 (altivec_vspltw): Likewise.
3993 (altivec_vspltsf): Likewise.
3995 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3998 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3999 frame related instructions.
4001 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4003 PR rtl-optimization/59959
4004 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4005 any reload of register whose subreg is invalid.
4007 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4009 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4010 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4011 Add missing return type - void.
4013 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4015 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4016 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4017 remove element index adjustment for endian (now handled in vsx.md
4019 (altivec_expand_vec_perm_const): Use
4020 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4021 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4022 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4023 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4024 define_expand and a new define_insn *altivec_vspltb_internal;
4025 adjust for -maltivec=be on a little endian target.
4026 (altivec_vspltb_direct): New.
4027 (altivec_vsplth): Divide into a define_expand and a new
4028 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4029 little endian target.
4030 (altivec_vsplth_direct): New.
4031 (altivec_vspltw): Divide into a define_expand and a new
4032 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4033 little endian target.
4034 (altivec_vspltw_direct): New.
4035 (altivec_vspltsf): Divide into a define_expand and a new
4036 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4037 a little endian target.
4039 2014-01-30 Richard Biener <rguenther@suse.de>
4041 PR tree-optimization/59993
4042 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4043 can propagate form the earlier stmt and avoid the transform
4044 when the intermediate result is needed.
4046 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4048 * README.Portability: Fix typo.
4050 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4052 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4053 comparison_operator with ordered_comparison_operator.
4055 2014-01-30 Nick Clifton <nickc@redhat.com>
4057 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4058 Rename to mn10300_store_multiple_regs.
4059 * config/mn10300/mn10300.c: Likewise.
4060 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4061 store_multiple_regs.
4062 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4063 Call mn10300_store_multiple_regs.
4065 2014-01-30 Nick Clifton <nickc@redhat.com>
4066 DJ Delorie <dj@redhat.com>
4068 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4069 %fp 2 to keep registers after it properly word-aligned.
4070 (rl78_alloc_physical_registers_umul): Handle the case where both
4071 input operands are the same.
4073 2014-01-30 Richard Biener <rguenther@suse.de>
4075 PR tree-optimization/59903
4076 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4079 2014-01-30 Jason Merrill <jason@redhat.com>
4082 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4085 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4087 2014-01-30 Richard Biener <rguenther@suse.de>
4089 PR tree-optimization/59951
4090 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4092 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4095 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4096 SFmode to DFmode case.
4098 2014-01-29 DJ Delorie <dj@redhat.com>
4100 * config/msp430/msp430.opt (-minrt): New.
4101 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4103 (ENDFILE_SPEC): Likewise.
4105 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4107 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4108 (estimate_function_body_sizes): Use it.
4110 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4113 * dwarf2out.c (is_cxx_auto): New.
4114 (is_base_type): Use it.
4115 (gen_type_die_with_usage): Likewise.
4117 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4119 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4120 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4121 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4122 -maltivec=be with LE targets.
4123 (vsx_mergeh_<mode>): Likewise.
4124 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4125 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4126 (altivec_vmrghb): Replace with define_expand and new
4127 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4128 (altivec_vmrghb_direct): New define_insn.
4129 (altivec_vmrghh): Replace with define_expand and new
4130 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4131 (altivec_vmrghh_direct): New define_insn.
4132 (altivec_vmrghw): Replace with define_expand and new
4133 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4134 (altivec_vmrghw_direct): New define_insn.
4135 (*altivec_vmrghsf): Adjust for endianness.
4136 (altivec_vmrglb): Replace with define_expand and new
4137 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4138 (altivec_vmrglb_direct): New define_insn.
4139 (altivec_vmrglh): Replace with define_expand and new
4140 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4141 (altivec_vmrglh_direct): New define_insn.
4142 (altivec_vmrglw): Replace with define_expand and new
4143 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4144 (altivec_vmrglw_direct): New define_insn.
4145 (*altivec_vmrglsf): Adjust for endianness.
4146 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4147 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4148 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4149 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4150 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4151 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4152 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4153 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4155 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4157 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4158 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4161 2014-01-29 Richard Biener <rguenther@suse.de>
4163 PR tree-optimization/58742
4164 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4165 associate_pointerplus_align.
4166 (associate_pointerplus_diff): New function.
4167 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4168 and associate_pointerplus_diff.
4170 2014-01-29 Richard Biener <rguenther@suse.de>
4172 * lto-streamer.h (LTO_major_version): Bump to 3.
4173 (LTO_minor_version): Reset to 0.
4175 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4177 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4178 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4179 (arm_file_start): Generate correct asm header for armv7ve.
4180 * config/arm/bpabi.h: Add multilib support for armv7ve.
4181 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4182 and cortex-a15 to armv7ve.
4183 * config/arm/t-aprofile: Add multilib support for armv7ve.
4184 * doc/invoke.texi: Document -march=armv7ve.
4186 2014-01-29 Richard Biener <rguenther@suse.de>
4188 PR tree-optimization/58742
4189 * tree-ssa-forwprop.c (associate_plusminus): Return true
4190 if we changed sth, defer EH cleanup to ...
4191 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4192 (simplify_mult): New function.
4194 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4197 PR tree-optimization/59920
4198 * tree.c (build_common_builtin_nodes): Remove
4199 __builtin_setjmp_dispatcher initialization.
4200 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4201 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4202 instead of gsi_after_labels + manually skipping debug stmts.
4203 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4204 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4205 * tree-inline.c (copy_edges_for_bb): Remove
4206 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4207 argument. Ignore computed_goto_p stmts. Don't call
4208 make_abnormal_goto_edges. If a call might need abnormal edges
4209 for non-local gotos, see if it already has an edge to
4210 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4211 with true argument, don't do anything then, otherwise add
4212 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4213 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4215 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4216 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4217 (lower_stmt): Don't set data->calls_builtin_setjmp.
4218 (lower_builtin_setjmp): Adjust comment.
4219 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4220 * tree-cfg.c (found_computed_goto): Remove.
4221 (factor_computed_gotos): Remove.
4222 (make_goto_expr_edges): Return bool, true for computed gotos.
4223 Don't call make_abnormal_goto_edges.
4224 (build_gimple_cfg): Don't set found_computed_goto, don't call
4225 factor_computed_gotos.
4226 (computed_goto_p): No longer static.
4227 (make_blocks): Don't set found_computed_goto.
4228 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4229 (make_edges): If make_goto_expr_edges returns true, push bb
4230 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4231 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4232 vector. Record mapping between bbs and OpenMP regions if there
4233 are any, adjust make_gimple_omp_edges caller. Call
4234 handle_abnormal_edges.
4235 (make_abnormal_goto_edges): Remove.
4236 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4237 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4238 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4239 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4240 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4241 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4242 filling *region also set *region_idx to (*region)->entry->index.
4245 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4246 For REGs set ORIGINAL_REGNO.
4248 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4250 * doc/md.texi: Mention that a target shouldn't implement
4251 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4254 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4256 PR tree-optimization/59594
4257 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4258 a copy of the datarefs vector rather than the vector itself.
4260 2014-01-28 Jason Merrill <jason@redhat.com>
4263 * dwarf2out.c (auto_die): New static.
4264 (gen_type_die_with_usage): Handle C++1y 'auto'.
4265 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4268 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4271 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4272 (SPEC_X32): Likewise.
4273 (SPEC_64): Likewise.
4274 * config/i386/i386.c (ix86_option_override_internal): Turn off
4275 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4277 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4278 * config/i386/i386.h (TARGET_16BIT): New macro.
4279 (TARGET_16BIT_P): Likewise.
4280 * config/i386/i386.opt: Add m16.
4281 * doc/invoke.texi: Document -m16.
4283 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4285 PR preprocessor/59935
4286 * input.c (location_get_source_line): Bail out on when line number
4287 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4289 2014-01-28 Richard Biener <rguenther@suse.de>
4291 PR tree-optimization/58742
4292 * tree-ssa-forwprop.c (associate_plusminus): Handle
4293 pointer subtraction of the form (T)(P + A) - (T)P.
4295 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4297 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4300 2014-01-28 Richard Biener <rguenther@suse.de>
4303 2014-01-28 Richard Biener <rguenther@suse.de>
4305 PR rtl-optimization/45364
4306 PR rtl-optimization/59890
4307 * var-tracking.c (local_get_addr_clear_given_value): Handle
4308 already cleared slot.
4309 (val_reset): Handle not allocated local_get_addr_cache.
4310 (vt_find_locations): Use post-order on the inverted CFG.
4312 2014-01-28 Richard Biener <rguenther@suse.de>
4314 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4316 2014-01-28 Richard Biener <rguenther@suse.de>
4318 PR rtl-optimization/45364
4319 PR rtl-optimization/59890
4320 * var-tracking.c (local_get_addr_clear_given_value): Handle
4321 already cleared slot.
4322 (val_reset): Handle not allocated local_get_addr_cache.
4323 (vt_find_locations): Use post-order on the inverted CFG.
4325 2014-01-28 Alan Modra <amodra@gmail.com>
4327 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4328 * configure.ac <recursive call for build != host>: Define
4329 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4330 and LD_FOR_BUILD too.
4331 * configure: Regenerate.
4333 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4335 * config/i386/i386.c (get_builtin_code_for_version): Separate
4336 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4337 Broadwell from Haswell.
4339 2014-01-27 Steve Ellcey <sellcey@mips.com>
4341 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4342 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4343 * config/mips/mips.c (mips_option_override): Change setting
4345 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4346 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4347 Change from Mask to Var.
4349 2014-01-27 Jeff Law <law@redhat.com>
4351 * ipa-inline.c (inline_small_functions): Fix typo.
4353 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4355 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4356 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4357 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4358 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4359 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4360 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4361 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4362 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4363 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4364 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4365 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4366 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4367 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4368 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4369 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4370 (_mm512_storeu_epi64): Ditto.
4371 (_mm512_cmpge_epi32_mask): Ditto.
4372 (_mm512_cmpge_epu32_mask): Ditto.
4373 (_mm512_cmpge_epi64_mask): Ditto.
4374 (_mm512_cmpge_epu64_mask): Ditto.
4375 (_mm512_cmple_epi32_mask): Ditto.
4376 (_mm512_cmple_epu32_mask): Ditto.
4377 (_mm512_cmple_epi64_mask): Ditto.
4378 (_mm512_cmple_epu64_mask): Ditto.
4379 (_mm512_cmplt_epi32_mask): Ditto.
4380 (_mm512_cmplt_epu32_mask): Ditto.
4381 (_mm512_cmplt_epi64_mask): Ditto.
4382 (_mm512_cmplt_epu64_mask): Ditto.
4383 (_mm512_cmpneq_epi32_mask): Ditto.
4384 (_mm512_cmpneq_epu32_mask): Ditto.
4385 (_mm512_cmpneq_epi64_mask): Ditto.
4386 (_mm512_cmpneq_epu64_mask): Ditto.
4387 (_mm512_expand_pd): Ditto.
4388 (_mm512_expand_ps): Ditto.
4389 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4390 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4391 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4392 * config/i386/i386.c (ix86_builtins): Add
4393 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4394 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4395 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4396 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4397 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4398 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4399 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4400 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4401 IX86_BUILTIN_PMOVUSQW512_MEM.
4402 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4403 __builtin_ia32_pmovsqd512mem_mask,
4404 __builtin_ia32_pmovqd512mem_mask,
4405 __builtin_ia32_pmovusqw512mem_mask,
4406 __builtin_ia32_pmovsqw512mem_mask,
4407 __builtin_ia32_pmovqw512mem_mask,
4408 __builtin_ia32_pmovusdw512mem_mask,
4409 __builtin_ia32_pmovsdw512mem_mask,
4410 __builtin_ia32_pmovdw512mem_mask,
4411 __builtin_ia32_pmovqb512mem_mask,
4412 __builtin_ia32_pmovusqb512mem_mask,
4413 __builtin_ia32_pmovsqb512mem_mask,
4414 __builtin_ia32_pmovusdb512mem_mask,
4415 __builtin_ia32_pmovsdb512mem_mask,
4416 __builtin_ia32_pmovdb512mem_mask.
4417 (bdesc_args): Add __builtin_ia32_expanddf512,
4418 __builtin_ia32_expandsf512.
4419 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4420 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4421 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4422 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4423 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4424 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4425 (avx512f_<code>v8div16qi2_mask_store): This.
4426 (avx512f_expand<mode>): New.
4428 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4430 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4432 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4433 (_mm512_prefetch_i32scatter_pd): Ditto.
4434 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4435 (_mm512_prefetch_i64scatter_pd): Ditto.
4436 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4437 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4438 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4439 (_mm512_prefetch_i32scatter_ps): Ditto.
4440 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4441 (_mm512_prefetch_i64scatter_ps): Ditto.
4442 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4443 * config/i386/i386-builtin-types.def: Define
4444 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4445 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4446 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4447 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4448 IX86_BUILTIN_SCATTERPFQPD.
4449 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4450 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4451 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4452 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4453 __builtin_ia32_scatterpfqps.
4454 (ix86_expand_builtin): Expand new built-ins.
4455 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4456 fix memory access data type.
4457 (*avx512pf_gatherpf<mode>_mask): Ditto.
4458 (*avx512pf_gatherpf<mode>): Ditto.
4459 (avx512pf_scatterpf<mode>): Ditto.
4460 (*avx512pf_scatterpf<mode>_mask): Ditto.
4461 (*avx512pf_scatterpf<mode>): Ditto.
4462 (GATHER_SCATTER_SF_MEM_MODE): New.
4463 (avx512pf_gatherpf<mode>df): Ditto.
4464 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4465 (*avx512pf_scatterpf<mode>df): Ditto.
4467 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4470 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4471 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4474 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4476 * common/config/arm/arm-common.c
4477 (arm_rewrite_mcpu): Handle multiple names.
4479 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4481 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4483 * gimple-builder.h (create_gimple_tmp): Delete.
4485 2014-01-27 Christian Bruel <christian.bruel@st.com>
4487 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4490 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4492 * config/pa/pa.md (call): Generate indirect long calls to non-local
4493 functions when outputing 32-bit code.
4494 (call_value): Likewise except for special call to buggy powf function.
4496 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4497 portable runtime and PIC indirect calls.
4498 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4499 and PIC call sequences. Use ldo instead of blr to set return register
4500 in PIC call sequence.
4502 2014-01-25 Walter Lee <walt@tilera.com>
4504 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4505 avoid clobbering a live register.
4507 2014-01-25 Walter Lee <walt@tilera.com>
4509 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4510 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4511 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4512 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4514 2014-01-25 Walter Lee <walt@tilera.com>
4516 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4517 arguments on even registers.
4518 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4520 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4521 (BIGGEST_ALIGNMENT): Ditto.
4522 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4524 2014-01-25 Walter Lee <walt@tilera.com>
4526 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4527 insns before bundling.
4528 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4530 2014-01-25 Walter Lee <walt@tilera.com>
4532 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4533 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4534 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4536 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4538 * config/mips/constraints.md (kl): Delete.
4539 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4540 define expands, using...
4541 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4542 instructions for MIPS16.
4543 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4544 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4546 2014-01-25 Walter Lee <walt@tilera.com>
4548 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4552 2014-01-25 Walter Lee <walt@tilera.com>
4554 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4555 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4557 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4559 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4562 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4564 * print-rtl.c (in_call_function_usage): New var.
4565 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4566 EXPR_LIST mode as mode and not as reg note name.
4569 * cfgloopmanip.c (copy_loop_info): If
4570 loop->warned_aggressive_loop_optimizations, make sure
4571 the flag is set in target loop too.
4573 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4575 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4577 * builtins.def: Likewise.
4578 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4579 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4580 * ira.c (ira_setup_eliminable_regset): Likewise.
4581 * omp-low.c (gate_expand_omp): Likewise.
4582 (execute_lower_omp): Likewise.
4583 (diagnose_sb_0): Likewise.
4584 (gate_diagnose_omp_blocks): Likewise.
4585 (simd_clone_clauses_extract): Likewise.
4588 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4590 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4591 correction for little endian...
4592 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4595 2014-01-24 Jeff Law <law@redhat.com>
4597 PR tree-optimization/59919
4598 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4599 for non-returning calls.
4601 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4603 * common/config/aarch64/aarch64-common.c
4604 (aarch64_rewrite_mcpu): Handle multiple names.
4605 * config/aarch64/aarch64.h
4606 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4608 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4610 * input.c (add_file_to_cache_tab): Handle the case where fopen
4613 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4616 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4617 from push operand if code of push isn't PRE_DEC.
4619 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4622 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4623 -mquad-memory-atomic. Update -mquad-memory documentation to say
4624 it is only used for non-atomic loads/stores.
4626 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4627 -mquad-memory or -mquad-memory-atomic switches.
4629 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4630 -mquad-memory-atomic to ISA 2.07 support.
4632 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4633 to separate support of normal quad word memory operations (ldq, stq)
4634 from the atomic quad word memory operations.
4636 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4637 support to separate non-atomic quad word operations from atomic
4638 quad word operations. Disable non-atomic quad word operations in
4639 little endian mode so that we don't have to swap words after the
4640 load and before the store.
4641 (quad_load_store_p): Add comment about atomic quad word support.
4642 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4643 options printed with -mdebug=reg.
4645 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4646 -mquad-memory-atomic as the test for whether we have quad word
4647 atomic instructions.
4648 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4649 or -mp8-vector are used, allow byte/half-word atomic operations.
4651 * config/rs6000/sync.md (load_lockedti): Insure that the address
4652 is a proper indexed or indirect address for the lqarx instruction.
4653 On little endian systems, swap the hi/lo registers after the lqarx
4655 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4656 insure the address is valid for the lqarx instruction.
4657 (store_conditionalti): Insure that the address is a proper indexed
4658 or indirect address for the stqcrx. instruction. On little endian
4659 systems, swap the hi/lo registers before doing the stqcrx.
4661 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4662 insure the address is valid for the stqcrx. instruction.
4664 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4665 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4666 type of quad memory support is available.
4668 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4671 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4672 there is a danger of looping.
4674 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4676 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4677 force flag_ira_loop_pressure if set via command line.
4679 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4681 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4682 (ashr_simd): New builtin handling DI mode.
4683 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4684 (aarch64_sshr_simddi): New match pattern.
4685 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4686 (vshrd_n_s64): Likewise.
4687 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4689 2014-01-23 Nick Clifton <nickc@redhat.com>
4691 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4692 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4693 favour of mcu specific scripts.
4694 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4697 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4698 Alex Velenko <Alex.Velenko@arm.com>
4700 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4701 (vaddv_s16): Likewise.
4702 (vaddv_s32): Likewise.
4703 (vaddv_u8): Likewise.
4704 (vaddv_u16): Likewise.
4705 (vaddv_u32): Likewise.
4706 (vaddvq_s8): Likewise.
4707 (vaddvq_s16): Likewise.
4708 (vaddvq_s32): Likewise.
4709 (vaddvq_s64): Likewise.
4710 (vaddvq_u8): Likewise.
4711 (vaddvq_u16): Likewise.
4712 (vaddvq_u32): Likewise.
4713 (vaddvq_u64): Likewise.
4714 (vaddv_f32): Likewise.
4715 (vaddvq_f32): Likewise.
4716 (vaddvq_f64): Likewise.
4717 (vmaxv_f32): Likewise.
4718 (vmaxv_s8): Likewise.
4719 (vmaxv_s16): Likewise.
4720 (vmaxv_s32): Likewise.
4721 (vmaxv_u8): Likewise.
4722 (vmaxv_u16): Likewise.
4723 (vmaxv_u32): Likewise.
4724 (vmaxvq_f32): Likewise.
4725 (vmaxvq_f64): Likewise.
4726 (vmaxvq_s8): Likewise.
4727 (vmaxvq_s16): Likewise.
4728 (vmaxvq_s32): Likewise.
4729 (vmaxvq_u8): Likewise.
4730 (vmaxvq_u16): Likewise.
4731 (vmaxvq_u32): Likewise.
4732 (vmaxnmv_f32): Likewise.
4733 (vmaxnmvq_f32): Likewise.
4734 (vmaxnmvq_f64): Likewise.
4735 (vminv_f32): Likewise.
4736 (vminv_s8): Likewise.
4737 (vminv_s16): Likewise.
4738 (vminv_s32): Likewise.
4739 (vminv_u8): Likewise.
4740 (vminv_u16): Likewise.
4741 (vminv_u32): Likewise.
4742 (vminvq_f32): Likewise.
4743 (vminvq_f64): Likewise.
4744 (vminvq_s8): Likewise.
4745 (vminvq_s16): Likewise.
4746 (vminvq_s32): Likewise.
4747 (vminvq_u8): Likewise.
4748 (vminvq_u16): Likewise.
4749 (vminvq_u32): Likewise.
4750 (vminnmv_f32): Likewise.
4751 (vminnmvq_f32): Likewise.
4752 (vminnmvq_f64): Likewise.
4754 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4756 * config/aarch64/aarch64-simd.md
4757 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4758 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4759 (*aarch64_mul3_elt<mode>): Likewise.
4760 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4761 (*aarch64_mul3_elt_to_64v2df): Likewise.
4762 (*aarch64_mla_elt<mode>): Likewise.
4763 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4764 (*aarch64_mls_elt<mode>): Likewise.
4765 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4766 (*aarch64_fma4_elt<mode>): Likewise.
4767 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4768 (*aarch64_fma4_elt_to_64v2df): Likewise.
4769 (*aarch64_fnma4_elt<mode>): Likewise.
4770 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4771 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4772 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4773 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4774 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4775 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4776 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4777 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4778 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4780 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4782 * config/aarch64/aarch64-simd.md
4783 (aarch64_be_checked_get_lane<mode>): New define_expand.
4784 * config/aarch64/aarch64-simd-builtins.def
4785 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4786 New builtin definition.
4787 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4788 Use new safe be builtin.
4790 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4792 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4794 (aarch64_be_st1<mode>): Likewise.
4795 (aarch_ld1<VALL:mode>): Define_expand modified.
4796 (aarch_st1<VALL:mode>): Likewise.
4797 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4798 (UNSPEC_ST1): Likewise.
4800 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4802 * config/microblaze/microblaze.md: Add trap insn and attribute
4804 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4806 PR preprocessor/58580
4807 * input.h (location_get_source_line): Take an additional line_size
4809 (void diagnostics_file_cache_fini): Declare new function.
4810 * input.c (struct fcache): New type.
4811 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4812 New static constants.
4813 (diagnostic_file_cache_init, total_lines_num)
4814 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4815 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4816 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4817 (get_next_line, read_next_line, goto_next_line, read_line_num):
4818 New static function definitions.
4819 (diagnostic_file_cache_fini): New function.
4820 (location_get_source_line): Take an additional output line_len
4821 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4823 * diagnostic.c (diagnostic_finish): Call
4824 diagnostic_file_cache_fini.
4825 (adjust_line): Take an additional input parameter for the length
4826 of the line, rather than calculating it with strlen.
4827 (diagnostic_show_locus): Adjust the use of
4828 location_get_source_line and adjust_line with respect to their new
4829 signature. While displaying a line now, do not stop at the first
4830 null byte. Rather, display the zero byte as a space and keep
4831 going until we reach the size of the line.
4832 * Makefile.in: Add vec.o to OBJS-libcommon
4834 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4835 Ilya Tocar <ilya.tocar@intel.com>
4837 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4838 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4839 (__builtin_ia32_kmov16): Ditto.
4840 * config/i386/i386.md (UNSPEC_KMOV): New.
4843 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4845 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4846 (_mm512_storeu_si512): Ditto.
4848 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4851 * rtl.h (get_referenced_operands): Declare.
4852 * recog.c (get_referenced_operands): New function.
4853 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4854 operands have been referenced when recording LO_SUM references.
4856 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4858 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4860 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4862 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4863 Enable for generic and recent AMD targets.
4865 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4867 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4868 ARG_SIZE note when adjustment was eliminated.
4870 2014-01-22 Jeff Law <law@redhat.com>
4872 PR tree-optimization/59597
4873 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4874 in file. Accept new argument REGISTERING and use it to modify
4875 dump output appropriately.
4876 (register_jump_thread): Corresponding changes.
4877 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4878 thread paths involving joiner blocks. Add code to dump cancelled
4879 jump threading paths.
4881 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4883 PR rtl-optimization/59477
4884 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4885 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4887 2014-01-22 Tom Tromey <tromey@redhat.com>
4889 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4891 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4893 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4895 PR rtl-optimization/59896
4896 * lra-constraints.c (process_alt_operands): Check unused note for
4897 matched operands of insn with no output reloads.
4899 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4901 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4902 (mips_move_from_gpr_cost): Likewise.
4904 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4906 PR rtl-optimization/59858
4907 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4908 ira_class_hard_regs_num.
4909 (process_alt_operands): Increase reject for dying matched operand.
4911 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4914 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4915 smaller than size, perform several stores or loads and stores
4916 at dst + count - size to store or copy all of size bytes, rather
4917 than just last modesize bytes.
4919 2014-01-20 DJ Delorie <dj@redhat.com>
4921 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4922 that CLOBBERs are REGs before propogating their values.
4924 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4927 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4928 (cgraph_inline_failed_type): New function.
4929 * cgraph.h (DEFCIFCODE): Add type.
4930 (cgraph_inline_failed_type_t): New enum.
4931 (cgraph_inline_failed_type): New prototype.
4932 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4933 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4934 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4935 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4936 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4937 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4938 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4939 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4940 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4941 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4942 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4943 OPTIMIZATION_MISMATCH.
4944 * tree-inline.c (expand_call_inline): Emit errors during
4945 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4947 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4950 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4951 mode attribute in insn output.
4953 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4955 * output.h (output_constant): Delete.
4956 * varasm.c (output_constant): Make private.
4958 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4960 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4962 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4965 * tree.h (fold_builtin_strcat): New prototype.
4966 * builtins.c (fold_builtin_strcat): No longer static. Add len
4967 argument, if non-NULL, don't call c_strlen. Optimize
4968 directly into __builtin_memcpy instead of __builtin_strcpy.
4969 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4970 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4972 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4974 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4975 for SImode_address_operand operands, having only a REG argument.
4977 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4979 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4980 loader name using mbig-endian.
4981 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4983 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4985 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4989 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4991 * config/aarch64/aarch64-protos.h
4992 (aarch64_cannot_change_mode_class_ptr): Declare.
4993 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4994 aarch64_cannot_change_mode_class_ptr): New.
4995 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4996 backend hook aarch64_cannot_change_mode_class.
4998 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5000 * common/config/aarch64/aarch64-common.c
5001 (aarch64_handle_option): Don't handle any option order logic here.
5002 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5003 selected_cpu, warn on architecture version mismatch.
5004 (aarch64_override_options): Fix parsing order for option strings.
5006 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5007 Iain Sandoe <iain@codesourcery.com>
5010 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5011 warning. Amend comment to reflect current functionality.
5013 2014-01-20 Richard Biener <rguenther@suse.de>
5016 * builtins.c (fold_builtin_strcat): Remove case better handled
5017 by tree-ssa-strlen.c.
5019 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5021 * config/aarch64/aarch64.opt
5022 (mcpu, march, mtune): Make case-insensitive.
5024 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5027 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5028 if operands[1] is a REG or ZERO_EXTEND of a REG.
5030 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5032 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5034 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5036 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5037 long non-pic millicode calls.
5039 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5041 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5043 2014-01-19 Kito Cheng <kito@0xlab.org>
5045 * builtins.c (expand_movstr): Check movstr expand done or fail.
5047 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5048 H.J. Lu <hongjiu.lu@intel.com>
5051 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5052 to DImode for zero-extended addresses.
5054 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5056 PR rtl-optimization/57763
5057 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5058 on the new indirect jump_insn and increment LABEL_NUSES (label).
5060 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5064 * config.gcc (x86_archs): New variable.
5065 (x86_64_archs): Likewise.
5066 (x86_cpus): Likewise.
5067 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5068 --with-arch/--with-cpu= options.
5069 Support --with-arch=/--with-cpu={nehalem,westmere,
5070 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5072 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5074 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5075 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5077 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5079 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5081 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5084 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5085 clear cpp_get_options (parse_in)->warn_unused_macros for
5086 ix86_target_macros_internal with cpp_define.
5088 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5090 * jump.c (delete_related_insns): Keep (use (insn))s.
5091 * reorg.c (redundant_insn): Check for barriers too.
5093 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5095 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5097 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5099 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5100 call to $$dyncall when TARGET_LONG_CALLS is true.
5102 2014-01-17 Jeff Law <law@redhat.com>
5104 * ree.c (combine_set_extension): Temporarily disable test for
5105 changing number of hard registers.
5107 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5110 * ipa-inline-analysis.c (inline_free_summary):
5111 Do not free summary of aliases.
5113 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5116 * gimplify.c (gimplify_expr): Use create_tmp_var
5117 instead of create_tmp_var_raw. If cond doesn't have
5118 integral type, don't add the IFN_ANNOTATE builtin at all.
5120 2014-01-17 Martin Jambor <mjambor@suse.cz>
5123 * ipa-cp.c (prev_edge_clone): New variable.
5124 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5125 Also resize prev_edge_clone vector.
5126 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5127 (ipcp_edge_removal_hook): New function.
5128 (ipcp_driver): Register ipcp_edge_removal_hook.
5130 2014-01-17 Andrew Pinski <apinski@cavium.com>
5131 Steve Ellcey <sellcey@mips.com>
5134 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5137 2014-01-17 Jeff Law <law@redhat.com>
5140 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5141 so that pass_ccp runs first.
5143 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5145 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5146 (ix86_adjust_cost): Use !TARGET_XXX.
5147 (do_reorder_for_imul): Likewise.
5148 (swap_top_of_ready_list): Likewise.
5149 (ix86_sched_reorder): Likewise.
5151 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5153 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5154 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5155 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5156 (intel_memset): New. Duplicate slm_memset.
5157 (intel_cost): New. Duplicate slm_cost.
5158 (m_INTEL): New macro.
5159 (processor_target_table): Add "intel".
5160 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5161 with PROCESSOR_INTEL for "intel".
5162 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5163 PROCESSOR_SILVERMONT.
5164 (ix86_issue_rate): Likewise.
5165 (ix86_adjust_cost): Likewise.
5166 (ia32_multipass_dfa_lookahead): Likewise.
5167 (swap_top_of_ready_list): Likewise.
5168 (ix86_sched_reorder): Likewise.
5169 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5170 instead of TARGET_OPT_AGU.
5171 * config/i386/i386.h (TARGET_INTEL): New.
5172 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5173 (processor_type): Add PROCESSOR_INTEL.
5174 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5175 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5177 2014-01-17 Marek Polacek <polacek@redhat.com>
5180 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5183 2014-01-17 Richard Biener <rguenther@suse.de>
5185 PR tree-optimization/46590
5186 * opts.c (default_options_table): Add entries for
5187 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5188 all enabled at -O1 but not for -Og.
5189 * common.opt (fbranch-count-reg): Remove Init(1).
5190 (fmove-loop-invariants): Likewise.
5191 (ftree-pta): Likewise.
5193 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5195 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5196 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5197 decls to at least the GCC 4.8 used alignments.
5200 * tree-nested.c (convert_nonlocal_reference_stmt,
5201 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5202 of GIMPLE_BIND stmts, adjust associated decls.
5204 2014-01-17 Richard Biener <rguenther@suse.de>
5206 PR tree-optimization/46590
5207 * vec.h (vec<>::bseach): New member function implementing
5208 binary search according to C89 bsearch.
5209 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5210 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5211 bitmap pointer again. Make accesses_in_loop a flat array.
5212 (mem_ref_obstack): New global.
5213 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5214 (mark_ref_stored): Likewise.
5215 (ref_indep_loop_p_2): Likewise.
5216 (set_ref_stored_in_loop): New helper function.
5217 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5218 (memref_free): Adjust.
5219 (record_mem_ref_loc): Simplify.
5220 (gather_mem_refs_stmt): Adjust.
5221 (sort_locs_in_loop_postorder_cmp): New function.
5222 (analyze_memory_references): Sort accesses_in_loop after
5223 loop postorder number.
5224 (find_ref_loc_in_loop_cmp): New function.
5225 (for_all_locs_in_loop): Find relevant cluster of locs in
5226 accesses_in_loop and iterate without recursion.
5227 (execute_sm): Avoid uninit warning.
5228 (struct ref_always_accessed): Simplify.
5229 (ref_always_accessed::operator ()): Likewise.
5230 (ref_always_accessed_p): Likewise.
5231 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5232 loop postorder numbers here.
5233 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5236 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5239 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5240 on decls for which assemble_alias has been called.
5242 2014-01-17 Nick Clifton <nickc@redhat.com>
5244 * config/msp430/msp430.opt: (mcpu): New option.
5245 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5246 (msp430_option_override): Parse target_cpu. If the MCU name
5247 matches a generic string, clear target_mcu.
5248 (msp430_attr): Allow numeric interrupt values up to 63.
5249 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5250 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5252 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5254 * config/msp430/msp430.md (popm): Use %J rather than %I.
5255 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5256 (addhi_cy_i): Use immediate_operand for operand 2.
5257 * doc/invoke.texi: Document -mcpu option.
5259 2014-01-17 Richard Biener <rguenther@suse.de>
5261 PR rtl-optimization/38518
5262 * df.h (df_analyze_loop): Declare.
5263 * df-core.c: Include cfgloop.h.
5264 (df_analyze_1): Split out main part of df_analyze.
5265 (df_analyze): Adjust.
5266 (loop_inverted_post_order_compute): New function.
5267 (loop_post_order_compute): Likewise.
5268 (df_analyze_loop): New function avoiding whole-function
5270 * loop-invariant.c (find_defs): Use df_analyze_loop.
5271 (find_invariants): Adjust.
5272 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5274 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5276 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5277 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5279 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5281 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5282 traversal when removing references.
5284 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5287 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5289 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5292 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5293 pushing a reload for an autoinc when we had previously reloaded an
5294 inner part of the address.
5296 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5298 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5300 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5301 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5302 when not giving up or versioning for alias only because of
5304 (vect_analyze_data_ref_dependences): Set to true.
5305 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5307 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5308 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5312 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5315 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5316 operand 0 predicate for gathers, use a new pseudo as subtarget.
5318 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5321 * lra-constraints.c (process_alt_operands): Add printing debug info.
5322 Check absence of input/output reloads for matched operands too.
5324 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5326 PR rtl-optimization/59835
5327 * ira.c (ira_init_register_move_cost): Increase cost for
5330 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5332 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5334 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5337 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5338 non-register objects. Use gen_(high/low)part more consistently.
5341 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5344 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5345 endian support, remove tests for WORDS_BIG_ENDIAN.
5346 (p8_mfvsrd_3_<mode>): Likewise.
5347 (reload_gpr_from_vsx<mode>): Likewise.
5348 (reload_gpr_from_vsxsf): Likewise.
5349 (p8_mfvsrd_4_disf): Likewise.
5351 2014-01-16 Richard Biener <rguenther@suse.de>
5353 PR rtl-optimization/46590
5354 * lcm.c (compute_antinout_edge): Use postorder iteration.
5355 (compute_laterin): Use inverted postorder iteration.
5357 2014-01-16 Nick Clifton <nickc@redhat.com>
5360 * varasm.c (output_constant): Return the number of bytes actually
5362 (output_constructor_array_range): Update the field size with the
5363 number of bytes emitted by output_constant.
5364 (output_constructor_regular_field): Likewise. Also do not
5365 complain if the total number of bytes emitted is now greater
5366 than the expected fieldpos.
5367 * output.h (output_constant): Update prototype and descriptive comment.
5369 2014-01-16 Marek Polacek <polacek@redhat.com>
5372 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5373 it is error_mark_node.
5375 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5377 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5378 VALID_AVX256_REG_OR_OI_MODE.
5380 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5382 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5383 current procedure should be profiled.
5385 2014-01-15 Andrew Pinski <apinski@cavium.com>
5387 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5388 of moving from/to the STACK_REG register class.
5390 2014-01-15 Richard Henderson <rth@redhat.com>
5393 * reginfo.c (global_regs_decl): Globalize.
5394 * rtl.h (global_regs_decl): Declare.
5395 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5396 reserved via global_regs.
5398 2014-01-15 Teresa Johnson <tejohnson@google.com>
5400 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5402 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5404 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5405 and vmulosh rather than call gen_vec_widen_smult_*.
5406 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5407 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5408 (vec_widen_smult_even_v16qi): Likewise.
5409 (vec_widen_umult_even_v8hi): Likewise.
5410 (vec_widen_smult_even_v8hi): Likewise.
5411 (vec_widen_umult_odd_v16qi): Likewise.
5412 (vec_widen_smult_odd_v16qi): Likewise.
5413 (vec_widen_umult_odd_v8hi): Likewise.
5414 (vec_widen_smult_odd_v8hi): Likewise.
5415 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5416 vmuloub rather than call gen_vec_widen_umult_*.
5417 (vec_widen_umult_lo_v16qi): Likewise.
5418 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5419 vmulosb rather than call gen_vec_widen_smult_*.
5420 (vec_widen_smult_lo_v16qi): Likewise.
5421 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5422 rather than call gen_vec_widen_umult_*.
5423 (vec_widen_umult_lo_v8hi): Likewise.
5424 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5425 rather than call gen_vec_widen_smult_*.
5426 (vec_widen_smult_lo_v8hi): Likewise.
5428 2014-01-15 Jeff Law <law@redhat.com>
5430 PR tree-optimization/59747
5431 * ree.c (find_and_remove_re): Properly handle case where a second
5432 eliminated extension requires widening a copy created for elimination
5433 of a prior extension.
5434 (combine_set_extension): Ensure that the number of hard regs needed
5435 for a destination register does not change when we widen it.
5437 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5439 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5440 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5441 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5442 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5443 (avr-*-rtems*): Likewise.
5444 (bfin*-rtems*): Likewise.
5445 (moxie-*-rtems*): Likewise.
5446 (h8300-*-rtems*): Likewise.
5447 (i[34567]86-*-rtems*): Likewise.
5448 (lm32-*-rtems*): Likewise.
5449 (m32r-*-rtems*): Likewise.
5450 (m68k-*-rtems*): Likewise.
5451 (microblaze*-*-rtems*): Likewise.
5452 (mips*-*-rtems*): Likewise.
5453 (powerpc-*-rtems*): Likewise.
5454 (sh-*-rtems*): Likewise.
5455 (sparc-*-rtems*): Likewise.
5456 (sparc64-*-rtems*): Likewise.
5457 (v850-*-rtems*): Likewise.
5458 (m32c-*-rtems*): Likewise.
5460 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5462 PR rtl-optimization/59511
5463 * ira.c (ira_init_register_move_cost): Use memory costs for some
5464 cases of register move cost calculations.
5465 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5466 instead of BB frequency.
5467 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5468 * lra-assigns.c (find_hard_regno_for): Ditto.
5470 2014-01-15 Richard Biener <rguenther@suse.de>
5472 PR tree-optimization/59822
5473 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5474 (vectorizable_load): Use it to hoist defs of uses of invariant
5475 loads out of the loop.
5477 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5478 Kugan Vivekanandarajah <kuganv@linaro.org>
5481 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5484 2014-01-15 Richard Biener <rguenther@suse.de>
5486 PR rtl-optimization/59802
5487 * lcm.c (compute_available): Use inverted postorder to seed
5488 the initial worklist.
5490 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5493 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5494 ADDR_REGS for invalid symrefs in non-PIC code.
5496 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5499 * builtins.c (determine_block_size): Initialize *probable_max_size
5500 even if len_rtx is CONST_INT.
5502 2014-01-14 Andrew Pinski <apinski@cavium.com>
5504 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5505 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5506 (cortexa53_tunings): Likewise.
5507 (aarch64_sched_issue_rate): New function.
5508 (TARGET_SCHED_ISSUE_RATE): Define.
5510 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5512 * ira-costs.c (find_costs_and_classes): Add missed
5513 ira_init_register_move_cost_if_necessary.
5515 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5518 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5520 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5523 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5524 to indicate if type is used for function return value. Warn ABI
5525 change if the vector mode isn't available for function return value.
5526 (ix86_function_arg_advance): Pass false to type_natural_mode.
5527 (ix86_function_arg): Likewise.
5528 (ix86_gimplify_va_arg): Likewise.
5529 (function_arg_32): Don't warn ABI change.
5530 (ix86_function_value): Pass true to type_natural_mode.
5531 (ix86_return_in_memory): Likewise.
5532 (ix86_struct_value_rtx): Removed.
5533 (TARGET_STRUCT_VALUE_RTX): Likewise.
5535 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5537 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5538 converting a conditional jump into a conditional return.
5540 2014-01-14 Richard Biener <rguenther@suse.de>
5542 PR tree-optimization/58921
5543 PR tree-optimization/59006
5544 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5545 hoisting invariant stmts.
5546 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5547 invariant loads on the preheader edge if possible.
5549 2014-01-14 Joey Ye <joey.ye@arm.com>
5551 * doc/plugin.texi (Building GCC plugins): Update to C++.
5553 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5555 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5556 (_mm_rcp28_round_ss): Ditto.
5557 (_mm_rsqrt28_round_sd): Ditto.
5558 (_mm_rsqrt28_round_ss): Ditto.
5559 (_mm_rcp28_sd): Ditto.
5560 (_mm_rcp28_ss): Ditto.
5561 (_mm_rsqrt28_sd): Ditto.
5562 (_mm_rsqrt28_ss): Ditto.
5563 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5564 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5565 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5566 (IX86_BUILTIN_RCP28SD): Ditto.
5567 (IX86_BUILTIN_RCP28SS): Ditto.
5568 (IX86_BUILTIN_RSQRT28SD): Ditto.
5569 (IX86_BUILTIN_RSQRT28SS): Ditto.
5570 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5571 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5572 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5573 (ix86_expand_special_args_builtin): Expand new FTYPE.
5574 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5575 (srcp14<mode>): Make insn unary.
5576 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5577 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5578 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5579 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5580 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5581 Fix rounding: make it SAE only.
5582 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5584 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5586 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5587 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5588 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5589 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5590 (round_saeonly_mask_scalar_operand4): Ditto.
5591 (round_saeonly_mask_scalar_op3): Ditto.
5592 (round_saeonly_mask_scalar_op4): Ditto.
5594 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5596 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5597 Implement -maltivec=be for vec_insert and vec_extract.
5599 2014-01-10 DJ Delorie <dj@redhat.com>
5601 * config/msp430/msp430.md (call_internal): Don't allow memory
5602 references with SP as the base register.
5603 (call_value_internal): Likewise.
5604 * config/msp430/constraints.md (Yc): New. For memory references
5605 that don't use SP as a base register.
5607 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5608 "an integer without a # prefix"
5609 * config/msp430/msp430.md (epilogue_helper): Use it.
5611 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5614 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5615 AVX512F gather builtins.
5616 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5617 on gather decls with INTEGER_TYPE masktype.
5618 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5619 directly into the builtin rather than hoisting it before loop.
5621 PR tree-optimization/59387
5622 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5623 (scev_const_prop): If folded_casts and type has undefined overflow,
5624 use force_gimple_operand instead of force_gimple_operand_gsi and
5625 for each added stmt if it is assign with
5626 arith_code_with_undefined_signed_overflow, call
5627 rewrite_to_defined_overflow.
5628 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5629 gimple-fold.h instead.
5630 (arith_code_with_undefined_signed_overflow,
5631 rewrite_to_defined_overflow): Moved to ...
5632 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5633 rewrite_to_defined_overflow): ... here. No longer static.
5634 Include gimplify-me.h.
5635 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5636 rewrite_to_defined_overflow): New prototypes.
5638 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5640 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5642 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5644 * builtins.c (get_object_alignment_2): Minor tweak.
5645 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5647 2014-01-13 Christian Bruel <christian.bruel@st.com>
5649 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5650 optimized non constant lengths.
5652 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5655 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5656 load as __atomic_load_N if possible.
5658 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5660 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5662 (rs6000_expand_builtin): Adjust call.
5664 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5667 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5668 * config/rs6000/rs6000.c: Include target-globals.h.
5669 (rs6000_set_current_function): Instead of doing target_reinit
5670 unconditionally, use save_target_globals_default_opts and
5671 restore_target_globals.
5673 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5675 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5676 (rs6000_expand_builtin): Handle mffs and mtfsf.
5677 (rs6000_init_builtins): Define mffs and mtfsf.
5678 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5679 (rs6000_mffs): New pattern.
5680 (rs6000_mtfsf): New pattern.
5682 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5684 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5685 Start narrowing with START. Apply candidate-use pair
5686 and check overall cost in narrowing.
5687 (iv_ca_prune): Pass new argument.
5689 2014-01-10 Jeff Law <law@redhat.com>
5692 * ree.c (combine_reaching_defs): Ensure the defining statement
5693 occurs before the extension when optimizing extensions with
5694 different source and destination hard registers.
5696 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5699 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5700 vtables into the type inheritance graph.
5702 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5704 PR rtl-optimization/59754
5705 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5706 modes in the REGNO != REGNO case.
5708 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5710 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5712 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5714 PR tree-optimization/59745
5715 * tree-predcom.c (tree_predictive_commoning_loop): Call
5716 free_affine_expand_cache if giving up because components is NULL.
5718 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5719 GC in payload of target_globals struct instead of allocating them on
5720 the heap and the larger structs separately using GC.
5721 * target-globals.h (struct target_globals): Make regs, hard_regs,
5722 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5723 of GTY((skip)) and change type to void *.
5724 (reset_target_globals): Cast loads from those fields to corresponding
5727 2014-01-10 Steve Ellcey <sellcey@mips.com>
5730 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5731 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5732 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5734 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5737 * aarch64-modes.def (CC_Zmode): New flags mode.
5738 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5739 represents an equality.
5740 (aarch64_get_condition_code): Handle CC_Zmode.
5741 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5743 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5745 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5746 extraction in good case.
5748 2014-01-10 Richard Biener <rguenther@suse.de>
5750 PR tree-optimization/59374
5751 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5752 checking after SLP discovery. Mark stmts not participating
5753 in any SLP instance properly.
5755 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5757 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5758 when handling a SET rtx.
5760 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5762 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5763 (cortex-a57): Likewise.
5764 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5766 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5768 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5769 non-iwmmxt builtins.
5771 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5775 * ipa-devirt.c record_target_from_binfo): Take as argument
5776 stack of binfos and lookup matching one for virtual inheritance.
5777 (possible_polymorphic_call_targets_1): Update.
5779 2014-01-10 Huacai Chen <chenhc@lemote.com>
5781 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5782 kernel strings for Loongson-2E/2F/3A.
5784 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5787 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5788 is_gimple_call before calling gimple_call_internal_p.
5790 2014-01-09 Steve Ellcey <sellcey@mips.com>
5792 * Makefile.in (TREE_FLOW_H): Remove.
5793 (TREE_SSA_H): Add file names from tree-flow.h.
5794 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5795 * tree.h: Remove tree-flow.h reference.
5796 * hash-table.h: Remove tree-flow.h reference.
5797 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5798 reference with tree-ssa-loop.h.
5800 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5802 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5803 default element-order behavior for -maltivec.
5804 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5805 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5806 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5807 when targeting big endian, at least for now.
5808 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5810 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5813 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5814 var satisfies use_register_for_decl, just take into account type
5815 alignment, rather than decl alignment.
5817 PR tree-optimization/59622
5818 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5819 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5820 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5821 Don't devirtualize for inplace at all. For targets.length () == 1,
5822 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5824 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5826 * config/i386/i386.md (cpu): Remove the unused btver1.
5828 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5830 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5832 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5835 * tree-core.h (struct target_globals): New forward declaration.
5836 (struct tree_target_option): Add globals field.
5837 * tree.h (TREE_TARGET_GLOBALS): Define.
5838 (prepare_target_option_nodes_for_pch): New prototype.
5839 * target-globals.h (struct target_globals): Define even if
5841 * tree.c (prepare_target_option_node_for_pch,
5842 prepare_target_option_nodes_for_pch): New functions.
5843 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5844 * config/i386/i386.c: Include target-globals.h.
5845 (ix86_set_current_function): Instead of doing target_reinit
5846 unconditionally, use save_target_globals_default_opts and
5847 restore_target_globals.
5849 2014-01-09 Richard Biener <rguenther@suse.de>
5851 PR tree-optimization/59715
5852 * tree-cfg.h (split_critical_edges): Declare.
5853 * tree-cfg.c (split_critical_edges): Export.
5854 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5856 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5858 * cfgexpand.c (expand_stack_vars): Optionally disable
5859 asan stack protection.
5860 (expand_used_vars): Likewise.
5861 (partition_stack_vars): Likewise.
5862 * asan.c (asan_emit_stack_protection): Optionally disable
5863 after return stack usage.
5864 (instrument_derefs): Optionally disable memory access instrumentation.
5865 (instrument_builtin_call): Likewise.
5866 (instrument_strlen_call): Likewise.
5867 (asan_protect_global): Optionally disable global variables protection.
5868 * doc/invoke.texi: Added doc for new options.
5869 * params.def: Added new options.
5870 * params.h: Likewise.
5872 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5874 PR rtl-optimization/59724
5875 * ifcvt.c (cond_exec_process_if_block): Don't call
5876 flow_find_head_matching_sequence with 0 longest_match.
5877 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5878 non-active insns if !stop_after.
5879 (try_head_merge_bb): Revert 2014-01-07 changes.
5881 2014-01-08 Jeff Law <law@redhat.com>
5883 * ree.c (get_sub_rtx): New function, extracted from...
5884 (merge_def_and_ext): Here.
5885 (combine_reaching_defs): Use get_sub_rtx.
5887 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5889 * cgraph.h (varpool_variable_node): Do not choke on null node.
5891 2014-01-08 Catherine Moore <clm@codesourcery.com>
5893 * config/mips/mips.md (simple_return): Attempt to use JRC
5895 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5897 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5899 PR rtl-optimization/59137
5900 * reorg.c (steal_delay_list_from_target): Call update_block for
5902 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5904 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5906 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5907 two duplicate entries.
5909 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5912 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5914 * config/mips/mips.c (mips_truncated_op_cost): New function.
5915 (mips_rtx_costs): Adjust test for BADDU.
5916 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5918 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5920 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5921 (*baddu_si): ...this new pattern.
5923 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5926 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5928 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5931 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5933 (expand_expr, expand_normal): Adjust.
5934 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5935 inner_reference_p. Use inner_reference_p to expand inner references.
5936 (store_expr): Adjust.
5937 * cfgexpand.c (expand_call_stmt): Adjust.
5939 2014-01-08 Rong Xu <xur@google.com>
5941 * gcov-io.c (gcov_var): Move from gcov-io.h.
5942 (gcov_position): Ditto.
5943 (gcov_is_error): Ditto.
5944 (gcov_rewrite): Ditto.
5945 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5946 only part to libgcc/libgcov.h.
5948 2014-01-08 Marek Polacek <polacek@redhat.com>
5951 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5953 2014-01-08 Marek Polacek <polacek@redhat.com>
5956 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5958 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5960 PR rtl-optimization/59649
5961 * stor-layout.c (get_mode_bounds): For BImode return
5962 0 and STORE_FLAG_VALUE.
5964 2014-01-08 Richard Biener <rguenther@suse.de>
5967 * gimple.h (is_gimple_builtin_call): Remove.
5968 (gimple_builtin_call_types_compatible_p): New.
5969 (gimple_call_builtin_p): New overload.
5970 * gimple.c (is_gimple_builtin_call): Remove.
5971 (validate_call): Rename to ...
5972 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5974 (validate_type): New static function.
5975 (gimple_call_builtin_p): New overload and adjust.
5976 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5977 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5978 (gimple_fold_stmt_to_constant_1): Likewise.
5979 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5981 2014-01-08 Richard Biener <rguenther@suse.de>
5984 * gimplify.c (gimplify_expr): Gimplify register-register type
5985 VIEW_CONVERT_EXPRs to separate stmts.
5987 2014-01-07 Jeff Law <law@redhat.com>
5990 * ree.c (combine_set_extension): Handle case where source
5991 and destination registers in an extension insn are different.
5992 (combine_reaching_defs): Allow source and destination registers
5993 in extension to be different under limited circumstances.
5994 (add_removable_extension): Remove restriction that the
5995 source and destination registers in the extension are the same.
5996 (find_and_remove_re): Emit a copy from the extension's
5997 destination to its source after the defining insn if
5998 the source and destination registers are different.
6001 * ifcvt.c (merge_if_block): If we are merging a block with more than
6002 one successor with a block with no successors, remove any BARRIER
6003 after the second block.
6005 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6007 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6009 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6012 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6013 for 14-bit register offsets when INT14_OK_STRICT is false.
6015 2014-01-07 Roland Stigge <stigge@antcom.de>
6016 Michael Meissner <meissner@linux.vnet.ibm.com>
6019 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6020 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6022 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6024 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6027 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6029 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6030 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6031 rtx is const0_rtx or not.
6033 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6036 * target-globals.c (save_target_globals): Remove this_fn_optab
6038 * toplev.c: Include optabs.h.
6039 (target_reinit): Temporarily restore the global options if another
6040 set of options are in force.
6042 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6044 PR rtl-optimization/58668
6045 * cfgcleanup.c (flow_find_cross_jump): Don't count
6046 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6047 to determine what is counted.
6048 (flow_find_head_matching_sequence): Use active_insn_p to determine
6050 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6052 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6053 determine what is counted.
6055 PR tree-optimization/59643
6056 * tree-predcom.c (split_data_refs_to_components): If one dr is
6057 read and one write, determine_offset fails and the write isn't
6058 in the bad component, just put the read into the bad component.
6060 2014-01-07 Mike Stump <mikestump@comcast.net>
6061 Jakub Jelinek <jakub@redhat.com>
6064 * tree-core.h (struct tree_optimization_option): Change optabs
6065 type from unsigned char * to void *.
6066 * optabs.c (init_tree_optimization_optabs): Adjust
6067 TREE_OPTIMIZATION_OPTABS initialization.
6069 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6072 * config/i386/i386.h (struct machine_function): Add
6073 no_drap_save_restore field.
6074 * config/i386/i386.c (ix86_save_reg): Use
6075 !cfun->machine->no_drap_save_restore instead of
6076 crtl->stack_realign_needed.
6077 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6078 this function clears frame_pointer_needed. Set
6079 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6080 and DRAP reg is needed.
6082 2014-01-06 Marek Polacek <polacek@redhat.com>
6085 * doc/implement-c.texi: Mention that other integer types are
6086 permitted as bit-field types in strictly conforming mode.
6088 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6090 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6093 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6095 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6097 2014-01-06 Martin Jambor <mjambor@suse.cz>
6100 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6102 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6104 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6108 * var-tracking.c (add_stores): Preserve the value of the source even if
6109 we don't record the store.
6111 2014-01-06 Terry Guo <terry.guo@arm.com>
6113 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6115 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6118 * config/darwin.c (darwin_function_section): Adjust return values to
6119 correspond to optimisation changes made in r206070.
6121 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6123 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6124 from prefetch_block tune setting.
6125 (nocona_cost): Correct size of prefetch block to 64.
6127 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6129 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6130 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6131 used to save the static chain register in the computation of the offset
6132 from which the FP registers need to be restored.
6134 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6136 PR tree-optimization/59519
6137 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6138 ICE if get_current_def (current_new_name) is already non-NULL, as long
6139 as it is a phi result of some other phi in *new_exit_bb that has
6142 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6143 or vmovdqu* for misaligned_operand.
6144 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6145 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6146 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6147 aligned_mem for AVX512F masked aligned load and store builtins and for
6150 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6152 PR tree-optimization/59651
6153 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6154 Address range for negative step should be added by TYPE_SIZE_UNIT.
6156 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6158 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6159 overlapping registers also for registers other than the stack pointer.
6161 2014-01-03 Marek Polacek <polacek@redhat.com>
6164 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6167 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6170 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6173 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6174 (push splitter): Use <P:MODE_SIZE> instead of
6175 GET_MODE_SIZE (<P:MODE>mode).
6176 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6177 (mov -1, reg peephole2): Likewise.
6178 * config/i386/sse.md (*mov<mode>_internal,
6179 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6180 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6181 *<code><mode>3, *andnot<mode>3<mask_name>,
6182 <mask_codefor><code><mode>3<mask_name>): Likewise.
6183 * config/i386/subst.md (mask_mode512bit_condition,
6184 sd_mask_mode512bit_condition): Likewise.
6186 2014-01-02 Xinliang David Li <davidxl@google.com>
6188 PR tree-optimization/59303
6189 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6190 (dump_predicates): Better output format.
6191 (pred_equal_p): New function.
6192 (is_neq_relop_p): Ditto.
6193 (is_neq_zero_form_p): Ditto.
6194 (pred_expr_equal_p): Ditto.
6195 (pred_neg_p): Ditto.
6196 (simplify_pred): Ditto.
6197 (simplify_preds_2): Ditto.
6198 (simplify_preds_3): Ditto.
6199 (simplify_preds_4): Ditto.
6200 (simplify_preds): Ditto.
6202 (push_to_worklist): Ditto.
6203 (get_pred_info_from_cmp): Ditto.
6204 (is_degenerated_phi): Ditto.
6205 (normalize_one_pred_1): Ditto.
6206 (normalize_one_pred): Ditto.
6207 (normalize_one_pred_chain): Ditto.
6208 (normalize_preds): Ditto.
6209 (normalize_cond_1): Remove function.
6210 (normalize_cond): Ditto.
6211 (is_gcond_subset_of): Ditto.
6212 (is_subset_of_any): Ditto.
6213 (is_or_set_subset_of): Ditto.
6214 (is_and_set_subset_of): Ditto.
6215 (is_norm_cond_subset_of): Ditto.
6216 (pred_chain_length_cmp): Ditto.
6217 (convert_control_dep_chain_into_preds): Type change.
6218 (find_predicates): Ditto.
6219 (find_def_preds): Ditto.
6220 (destroy_predicates_vecs): Ditto.
6221 (find_matching_predicates_in_rest_chains): Ditto.
6222 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6223 (is_pred_expr_subset): Ditto.
6224 (is_pred_chain_subset_of): Ditto.
6225 (is_included_in): Ditto.
6226 (is_superset_of): Ditto.
6228 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6230 Update copyright years.
6232 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6234 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6235 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6236 config/arc/arc.md, config/arc/arc.opt,
6237 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6238 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6239 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6240 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6241 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6242 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6243 vtable-verify.c, vtable-verify.h: Use the standard form for the
6246 2014-01-02 Tobias Burnus <burnus@net-b.de>
6248 * gcc.c (process_command): Update copyright notice dates.
6249 * gcov-dump.c: Ditto.
6251 * doc/cpp.texi: Bump @copying's copyright year.
6252 * doc/cppinternals.texi: Ditto.
6253 * doc/gcc.texi: Ditto.
6254 * doc/gccint.texi: Ditto.
6255 * doc/gcov.texi: Ditto.
6256 * doc/install.texi: Ditto.
6257 * doc/invoke.texi: Ditto.
6259 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6261 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6263 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6265 * config/i386/sse.md (*mov<mode>_internal): Guard
6266 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6268 PR rtl-optimization/59647
6269 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6270 new_rtx into UNSIGNED_FLOAT rtxes.
6272 Copyright (C) 2014 Free Software Foundation, Inc.
6274 Copying and distribution of this file, with or without modification,
6275 are permitted in any medium without royalty provided the copyright
6276 notice and this notice are preserved.