1 2014-04-14 Richard Biener <rguenther@suse.de>
4 * fold-const.c (negate_expr_p): Don't negate directional rounding
6 (fold_negate_expr): Likewise.
8 2014-04-14 Richard Biener <rguenther@suse.de>
10 PR tree-optimization/59817
11 PR tree-optimization/60453
12 * graphite-scop-detection.c (graphite_can_represent_scev): Complete
13 recursion to catch all CHRECs in the scalar evolution and restrict
14 the predicate for the remains appropriately.
16 2014-04-12 Catherine Moore <clm@codesourcery.com>
18 * config/mips/constraints.md: Add new register constraint "kb".
19 * config/mips/mips.md (*mov<mode>_internal): Use constraint "kb".
20 (*movhi_internal): Likewise.
21 (*movqi_internal): Likewise.
22 * config/mips/mips.h (M16_STORE_REGS): New register class.
23 (REG_CLASS_NAMES): Add M16_STORE_REGS.
24 (REG_CLASS_CONTENTS): Likewise.
25 * config/mips/mips.c (mips_regno_to_class): Add M16_STORE_REGS.
27 2014-04-11 Tobias Burnus <burnus@net-b.de>
30 * doc/invoke.texi (-Wformat-signedness): Document it.
31 (Wformat=2): Mention that this enables -Wformat-signedness.
33 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
35 * common/config/epiphany/epiphany-common.c
36 (epiphany_option_optimization_table): Enable section anchors by
37 default at -O1 or higher.
38 * config/epiphany/epiphany.c (TARGET_MAX_ANCHOR_OFFSET): Define.
39 (TARGET_MIN_ANCHOR_OFFSET): Likewise.
40 (epiphany_rtx_costs) <SET>: For binary operators, the set as such
41 carries no extra cost.
42 (epiphany_legitimate_address_p): For BLKmode, apply SImode check.
43 * config/epiphany/epiphany.h (ASM_OUTPUT_DEF): Define.
44 * config/epiphany/predicates.md (memclob_operand): New predicate.
45 * config/epiphany/epiphany.md (stack_adjust_add, stack_adjust_str):
46 Use memclob_operand predicate and X constraint for operand 3.
48 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
50 * config/epiphany/epiphany.c (epiphany_rtx_cost): Compare
51 with CC_N_NE / CC_C_LTU / CC_C_GTU carries no extra cost for
54 2014-04-11 Joern Rennecke <joern.rennecke@embecosm.com>
56 PR rtl-optimization/60651
57 * mode-switching.c (optimize_mode_switching): Make sure to emit
58 sets of a lower numbered entity before sets of a higher numbered
59 entity to a mode of the same or lower priority.
60 When creating a seginfo for a basic block that starts with a code
61 label, move the insertion point past the code label.
62 (new_seginfo): Document and enforce requirement that
63 NOTE_INSN_BASIC_BLOCK only appears for empty blocks.
64 * doc/tm.texi.in: Document ordering constraint for emitted mode sets.
65 * doc/tm.texi: Regenerate.
67 2014-01-11 Joern Rennecke <joern.rennecke@embecosm.com>
70 * config/arc/arc.c (arc_save_restore): Fix assert typo.
72 2013-04-11 Jakub Jelinek <jakub@redhat.com>
74 * BASE-VER: Set to 4.10.0.
76 2014-04-11 Tobias Burnus <burnus@net-b.de>
79 * doc/bugreport.texi (Bugs): Remove nodes pointing to the
81 * doc/gcc.texi (Service): Update description in the @menu
82 * doc/invoke.texi (Option Summary): Remove misplaced and
85 2014-04-11 Steve Ellcey <sellcey@mips.com>
86 Jakub Jelinek <jakub@redhat.com>
89 * expr.c (convert_move): Use emit_store_flag_force instead of
90 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
93 2014-04-11 Richard Biener <rguenther@suse.de>
96 * varasm.c (assemble_alias): Avoid endless error reporting
97 recursion by setting TREE_ASM_WRITTEN.
99 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
101 * config/s390/s390.md: Add a splitter for NOT rtx.
103 2014-04-11 Jakub Jelinek <jakub@redhat.com>
105 PR rtl-optimization/60663
106 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
109 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
110 Jakub Jelinek <jakub@redhat.com>
113 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
116 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
119 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
120 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
121 ameliorating the cases where it can be.
123 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
126 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
128 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
129 (loadsync_<mode>): Change mode.
130 (load_quadpti, store_quadpti): New.
131 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
132 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
133 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
135 2014-04-09 Cong Hou <congh@google.com>
138 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
141 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
143 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
144 instead of vnor to exploit possible fusion opportunity in the
146 (altivec_expand_vec_perm_const_le): Likewise.
148 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
150 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
151 (loadsync_<mode>): Change mode.
152 (load_quadpti, store_quadpti): New.
153 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
154 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
156 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
159 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
160 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
161 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
163 2014-04-08 Richard Biener <rguenther@suse.de>
166 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
167 a 64bit widest int print double-int similar to on HWI64 hosts.
169 2014-04-08 Richard Biener <rguenther@suse.de>
171 PR tree-optimization/60785
172 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
173 default defs properly.
175 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
177 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
180 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
182 * ipa-devirt.c (maybe_record_node): When node is not recorded,
183 set completep to false rather than true.
185 2014-04-07 Douglas B Rupp <rupp@adacore.com>
188 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
189 ARM_TARGET2_DWARF_FORMAT.
191 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
194 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
195 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
198 2014-04-07 Richard Biener <rguenther@suse.de>
200 PR tree-optimization/60766
201 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
202 (may_eliminate_iv): Convert cand_value_at result to desired type.
204 2014-04-07 Jason Merrill <jason@redhat.com>
207 * common.opt (-fno-gnu-unique): Add.
208 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
210 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
212 * haifa-sched.c: Fix outdated function reference and minor
213 grammar errors in introductory comment.
215 2014-04-07 Richard Biener <rguenther@suse.de>
218 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
220 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
222 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
225 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
226 size accounting for thunks.
227 (pa_asm_output_mi_thunk): Use final_start_function() and
228 final_end_function() to output function start and end directives.
230 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
232 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
233 specific ISA/ feature information. Remove short_sp and errata_skip ds.
234 Add avr_device_specific_features enum to have device specific info.
235 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
236 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
237 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
238 updated device specific info.
239 * config/avr/avr-mcus.def: Merge device specific details to
241 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
243 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
244 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
245 assembler if RMW isa supported by current device.
246 * config/avr/genmultilib.awk: Update as device info structure changed.
247 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
249 2014-04-04 Cong Hou <congh@google.com>
251 PR tree-optimization/60656
252 * tree-vect-stmts.c (supportable_widening_operation):
253 Fix a bug that elements in a vector with vect_used_by_reduction
254 property are incorrectly reordered when the operation on it is not
255 consistant with the one in reduction operation.
257 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
259 PR rtl-optimization/60155
260 * gcse.c (record_set_data): New function.
261 (single_set_gcse): New function.
262 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
263 (hoist_code): Likewise.
264 (get_pressure_class_and_nregs): Likewise.
266 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
268 * explow.c (probe_stack_range): Emit a final optimization blockage.
270 2014-04-04 Anthony Green <green@moxielogic.com>
272 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
275 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
278 * lto-cgraph.c (input_overwrite_node): Check that partitioning
279 flags are set only during streaming.
280 * ipa.c (process_references, walk_polymorphic_call_targets,
281 symtab_remove_unreachable_nodes): Drop bodies of always inline
282 after early inlining.
283 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
285 2014-04-04 Jakub Jelinek <jakub@redhat.com>
286 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
289 * dwarf2out.c (const_ok_for_output_1): Reject expressions
292 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
295 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
297 (cortex_a53_fdivd): Likewise.
299 2014-04-04 Martin Jambor <mjambor@suse.cz>
302 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
304 * cgraph.c (clone_of_p): Also return true if thunks match.
305 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
306 cgraph_function_or_thunk_node and an obsolete comment.
307 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
309 (build_function_decl_skip_args): Likewise.
310 (set_new_clone_decl_and_node_flags): New function.
311 (duplicate_thunk_for_node): Likewise.
312 (redirect_edge_duplicating_thunks): Likewise.
313 (cgraph_clone_node): New parameter args_to_skip, pass it to
314 redirect_edge_duplicating_thunks which is called instead of
315 cgraph_redirect_edge_callee.
316 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
317 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
319 2014-04-04 Jeff Law <law@redhat.com>
322 * config/arm/predicates.md (const_int_I_operand): New predicate.
323 (const_int_M_operand): Similarly.
324 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
326 (insv_t2, extv_reg, extzv_t2): Likewise.
327 (load_multiple_with_writeback): Similarly for const_int_I_operand.
328 (pop_multiple_with_writeback_and_return): Likewise.
329 (vfp_pop_multiple_with_writeback): Likewise
331 2014-04-04 Richard Biener <rguenther@suse.de>
334 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
335 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
337 * gimplify.h (gimple_add_tmp_var_fn): Declare.
338 * gimplify.c (gimple_add_tmp_var_fn): New function.
339 * gimple-expr.h (create_tmp_reg_fn): Declare.
340 * gimple-expr.c (create_tmp_reg_fn): New function.
341 * gimple-low.c (record_vars_into): Don't change cfun.
342 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
343 code generation without cfun.
345 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
348 * Makefile.in (install-driver): Fix shell scripting.
350 2014-04-03 Cong Hou <congh@google.com>
352 PR tree-optimization/60505
353 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
354 threshold of number of iterations below which no vectorization will be
356 * tree-vect-loop.c (new_loop_vec_info):
357 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
358 * tree-vect-loop.c (vect_analyze_loop_operations):
359 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
360 * tree-vect-loop.c (vect_transform_loop):
361 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
362 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
363 of iterations of the loop and see if we should build the epilogue.
365 2014-04-03 Richard Biener <rguenther@suse.de>
367 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
369 (streamer_tree_cache_create): Adjust.
370 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
371 to allow optional nodes array.
372 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
373 (streamer_tree_cache_append): Likewise.
374 (streamer_tree_cache_create): Create nodes array optionally
375 as specified by parameter.
376 * lto-streamer-out.c (create_output_block): Avoid maintaining
377 the node array in the writer cache.
378 (DFS_write_tree): Remove assertion.
379 (produce_asm_for_decls): Free the out decl state hash table
381 * lto-streamer-in.c (lto_data_in_create): Adjust for
382 streamer_tree_cache_create prototype change.
384 2014-04-03 Richard Biener <rguenther@suse.de>
386 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
387 set TREE_CHAIN to NULL_TREE.
389 2014-04-03 Richard Biener <rguenther@suse.de>
391 PR tree-optimization/60740
392 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
393 over all GIMPLE_COND operands.
395 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
397 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
398 (Weffc++): Remove Scott's numbering, merge lists and reference
404 * class.c (accessible_nvdtor_p): New.
405 (check_bases): Don't check base destructor here ...
406 (check_bases_and_members): ... check them here. Trigger on
407 Wnon-virtual-dtor flag.
408 (finish_struct_1): Use accessible_nvdtor_p.
411 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
412 * g++.dg/warn/Wnvdtor-2.C: New.
413 * g++.dg/warn/Wnvdtor-3.C: New.
414 * g++.dg/warn/Wnvdtor-4.C: New.
415 * g++.dg/warn/Weff1.C: Delete.
416 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
417 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
419 2014-04-03 Nick Clifton <nickc@redhat.com>
421 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
424 2014-04-03 Martin Jambor <mjambor@suse.cz>
426 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
427 mention gcc_unreachable before failing.
428 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
431 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
434 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
435 code and instead mark the context inconsistent.
436 (possible_polymorphic_call_targets): For inconsistent contexts
437 return empty complete list.
439 2014-04-02 Anthony Green <green@moxielogic.com>
441 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
442 (extendqisi2, extendhisi2): Define.
443 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
444 (WCHAR_TYPE): Change to unsigned int.
446 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
448 PR tree-optimization/60733
449 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
450 insertion point for PHI candidates to be the end of the feeding
451 block for the PHI argument.
453 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
455 PR rtl-optimization/60650
456 * lra-constraints.c (process_alt_operands): Decrease reject for
457 earlyclobber matching.
459 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
461 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
463 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
465 * config/spu/spu.c (pad_bb): Do not crash when the last
466 insn is CODE_FOR_blockage.
468 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
470 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
471 lies outside the target mode.
473 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
476 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
477 software floating point or no floating point registers, do not
478 allow any type in the FPRs. Eliminate a test for SPE SIMD types
479 in GPRs that occurs after we tested for GPRs that would never be
482 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
483 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
484 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
485 specifically allow DDmode, since that does not use the SPE SIMD
488 2014-04-02 Richard Biener <rguenther@suse.de>
491 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
492 MODE_INTs. Properly use negv_optab.
493 (expand_abs): Likewise.
495 2014-04-02 Richard Biener <rguenther@suse.de>
498 * Makefile.in (install-driver): Guard extra installs with special
501 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
503 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
506 2014-04-01 Richard Henderson <rth@redhat.com>
509 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
510 alternative enabled before register allocation.
512 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
514 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
515 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
517 (nios2_large_got_address): Remove unneeded 'sym' parameter.
518 (nios2_got_address): Update nios2_large_got_address call site.
519 (nios2_delegitimize_address): New function.
520 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
521 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
522 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
524 2014-04-01 Martin Husemann <martin@duskware.de>
526 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
529 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
531 PR rtl-optimization/60604
532 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
533 check from register_operand.
534 (register_operand): Redefine in terms of general_operand.
535 (nonmemory_operand): Use register_operand for the non-constant cases.
537 2014-04-01 Richard Biener <rguenther@suse.de>
539 * gimple.h (struct gimple_statement_base): Align subcode to
542 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
544 * doc/invoke.texi (mapp-regs): Clarify.
546 2014-03-31 Ulrich Drepper <drepper@gmail.com>
548 * config/i386/avx512fintrin.h (__v32hi): Define type.
550 (_mm512_set1_epi8): Define.
551 (_mm512_set1_epi16): Define.
552 (_mm512_set4_epi32): Define.
553 (_mm512_set4_epi64): Define.
554 (_mm512_set4_pd): Define.
555 (_mm512_set4_ps): Define.
556 (_mm512_setr4_epi64): Define.
557 (_mm512_setr4_epi32): Define.
558 (_mm512_setr4_pd): Define.
559 (_mm512_setr4_ps): Define.
560 (_mm512_setzero_epi32): Define.
562 2014-03-31 Martin Jambor <mjambor@suse.cz>
565 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
566 callsite_arguments_match_p. Updated all callers. Also check types of
567 corresponding formal parameters and actual arguments.
568 (not_all_callers_have_enough_arguments_p) Renamed to
569 some_callers_have_mismatched_arguments_p.
571 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
573 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
575 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
578 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
581 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
583 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
584 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
586 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
587 Use FMAMODE_NOVF512 mode iterator.
588 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
589 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
590 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
591 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
593 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
594 Use VF_128_256 mode iterator.
595 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
598 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
600 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
601 static chain if needed.
603 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
606 * lra-constraints.c (index_part_to_reg): New.
607 (process_address): Use it.
609 2014-03-27 Jeff Law <law@redhat.com>
610 Jakub Jelinek <jakub@redhat.com>
613 * expr.c (do_tablejump): Use simplify_gen_binary rather than
614 gen_rtx_{PLUS,MULT} to build up the address expression.
616 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
617 creating non-canonical RTL.
619 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
622 * ipa-inline.c (want_inline_small_function_p): Short circuit large
623 functions; reorganize to make cheap checks first.
624 (inline_small_functions): Do not estimate growth when dumping;
626 * ipa-inline.h (inline_summary): Add min_size.
627 (growth_likely_positive): New function.
628 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
629 (set_cond_stmt_execution_predicate): Cleanup.
630 (estimate_edge_size_and_time): Compute min_size.
631 (estimate_calls_size_and_time): Likewise.
632 (estimate_node_size_and_time): Likewise.
633 (inline_update_overall_summary): Update min_size.
634 (do_estimate_edge_time): Likewise.
635 (do_estimate_edge_size): Update.
636 (do_estimate_edge_hints): Update.
637 (growth_likely_positive): New function.
639 2014-03-28 Jakub Jelinek <jakub@redhat.com>
642 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
643 also if addr has VOIDmode.
645 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
647 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
648 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
650 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
651 instructions as well as AdvancedSIMD loads.
653 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
655 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
656 Use crypto_aese type.
657 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
658 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
659 crypto_aese, crypto_aesmc. Move to types.md.
660 * config/arm/types.md (crypto_aes): Split into crypto_aese,
662 * config/arm/iterators.md (crypto_type): Likewise.
664 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
666 * cgraph.c: Include expr.h and tree-dfa.h.
667 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
670 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
673 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
674 regs from checking multi-reg pseudos.
676 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
678 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
680 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
682 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
683 if it would clobber the stack pointer, even temporarily.
685 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
687 * mode-switching.c: Make small adjustments to the top comment.
689 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
691 * config/rs6000/constraints.md (wD constraint): New constraint to
692 match the constant integer to get the top DImode/DFmode out of a
693 vector in a VSX register.
695 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
696 match the constant integer to get the top DImode/DFmode out of a
697 vector in a VSX register.
699 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
702 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
705 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
706 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
708 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
709 Optimize vec_extract of 64-bit values, where the value being
710 extracted is in the top word, where we can use scalar
711 instructions. Add direct move and store support. Combine the big
712 endian/little endian vector select load support into a single insn.
713 (vsx_extract_<mode>_internal1): Likewise.
714 (vsx_extract_<mode>_internal2): Likewise.
715 (vsx_extract_<mode>_load): Likewise.
716 (vsx_extract_<mode>_store): Likewise.
717 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
718 combined into vsx_extract_<mode>_load.
719 (vsx_extract_<mode>_one_le): Likewise.
721 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
722 define the top 64-bit vector element.
724 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
727 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
728 Document vec_vbpermq builtin.
731 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
732 enable use of xxsldwi and xxpermdi builtin functions.
733 (vec_xxpermdi): Likewise.
735 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
736 Document use of vec_xxsldwi and vec_xxpermdi builtins.
738 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
740 PR rtl-optimization/60650
741 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
743 (find_spills_for): New.
744 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
745 Spill all pseudos on the second iteration.
747 2014-03-27 Marek Polacek <polacek@redhat.com>
750 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
753 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
755 * config/s390/s390.c (s390_can_use_return_insn): Check for
756 call-saved FPRs on 31 bit.
758 2014-03-27 Jakub Jelinek <jakub@redhat.com>
761 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
762 if they need regimplification, just drop them instead of
763 calling gimple_regimplify_operands on them.
765 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
768 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
769 (aarch64_frame_pointer_required): Adjust logic.
770 (aarch64_can_eliminate): Adjust logic.
771 (aarch64_override_options_after_change): Adjust logic.
773 2014-03-27 Dehao Chen <dehao@google.com>
775 * ipa-inline.c (early_inliner): Update node's inline info.
777 2014-03-26 Dehao Chen <dehao@google.com>
779 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
780 compiler inserted conditional jumps for NAN float check.
782 2014-03-26 Jakub Jelinek <jakub@redhat.com>
784 * ubsan.h (ubsan_create_data): Change second argument's type
785 to const location_t *.
786 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
788 (ubsan_create_data): Change second argument to const location_t *PLOC.
789 Create Loc field whenever PLOC is non-NULL.
790 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
791 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
795 * real.c (real_to_integer2): Change type of low to UHWI.
797 2014-03-26 Tobias Burnus <burnus@net-b.de>
799 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
800 (CILK_SELF_SPECS): New define.
801 (driver_self_specs): Use it.
803 2014-03-26 Richard Biener <rguenther@suse.de>
805 * tree-pretty-print.c (percent_K_format): Implement special
806 case for LTO and its stripped down BLOCK tree.
808 2014-03-26 Jakub Jelinek <jakub@redhat.com>
811 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
813 * tree-vrp.c (simplify_internal_call_using_ranges): If only
814 one range is range_int_cst_p, but not both, at least optimize
815 addition/subtraction of 0 and multiplication by 0 or 1.
816 * gimple-fold.c (gimple_fold_call): Fold
817 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
818 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
819 INTEGER_CSTs, try to fold at least x * 0 and y - y.
821 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
823 PR rtl-optimization/60452
824 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
825 <case REG>: Return 1 for invalid offsets from the frame pointer.
827 2014-03-26 Marek Polacek <polacek@redhat.com>
830 * doc/extend.texi (C Extensions): Mention variable-length arrays in
833 2014-03-26 Marek Polacek <polacek@redhat.com>
836 * doc/extend.texi (Designated Inits): Describe what happens to omitted
839 2014-03-26 Marek Polacek <polacek@redhat.com>
842 * ira-color.c (update_conflict_hard_regno_costs): Perform the
843 multiplication in unsigned type.
845 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
847 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
849 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
851 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
853 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
856 * cif-code.def (UNREACHABLE) New code.
857 * ipa-inline.c (inline_small_functions): Skip edges to
858 __builtlin_unreachable.
859 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
860 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
861 predicate to __bulitin_unreachable.
862 (set_cond_stmt_execution_predicate): Fix issue when
863 invert_tree_comparison returns ERROR_MARK.
864 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
865 propagate to inline clones.
866 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
868 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
869 * cgraphclones.c (cgraph_clone_node): If call destination is already
870 ureachable, do not redirect it back.
871 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
874 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
876 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
877 Do not modify inline clones.
879 2014-03-25 Jakub Jelinek <jakub@redhat.com>
881 * config/i386/i386.md (general_sext_operand): New mode attr.
882 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
883 don't generate (sign_extend (const_int)).
884 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
885 operands[2]. Use We constraint instead of <i> and
886 <general_sext_operand> predicate instead of <general_operand>.
887 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
888 * config/i386/constraints.md (We): New constraint.
889 * config/i386/predicates.md (x86_64_sext_operand,
890 sext_operand): New predicates.
892 2014-03-25 Martin Jambor <mjambor@suse.cz>
895 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
896 inconsistent devirtualizations to __builtin_unreachable.
898 2014-03-25 Marek Polacek <polacek@redhat.com>
901 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
903 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
905 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
906 order of elements for big-endian.
908 2014-03-25 Richard Biener <rguenther@suse.de>
911 * gimplify-me.c (gimple_regimplify_operands): Update the
914 2014-03-25 Martin Jambor <mjambor@suse.cz>
917 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
918 (lto_output_varpool_node): Likewise.
919 (input_overwrite_node): Likewise.
920 (input_varpool_node): Likewise.
922 2014-03-25 Richard Biener <rguenther@suse.de>
924 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
927 2014-03-25 Jakub Jelinek <jakub@redhat.com>
929 * combine.c (simplify_compare_const): Add MODE argument.
930 Handle mode_width 0 as very large mode_width.
931 (try_combine, simplify_comparison): Adjust callers.
933 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
934 type to avoid signed integer overflow.
935 * explow.c (plus_constant): Likewise.
937 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
939 * doc/generic.texi: Correct typos.
941 2014-03-24 Tobias Burnus <burnus@net-b.de>
943 * doc/invoke.texi (-flto): Expand section about
944 using static libraries with LTO.
946 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
948 PR rtl-optimization/60501
949 * optabs.def (addptr3_optab): New optab.
950 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
951 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
952 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
954 * lra.c (emit_add3_insn): Use the addptr pattern if available.
956 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
958 2014-03-24 Ulrich Drepper <drepper@gmail.com>
960 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
963 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
964 (_mm256_undefined_ps): Define.
965 (_mm256_undefined_pd): Define.
966 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
967 (_mm_undefined_pd): Define.
968 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
969 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
970 (_mm512_undefined_ps): Define.
971 (_mm512_undefined_pd): Define.
972 Use _mm*_undefined_*.
973 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
975 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
977 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
978 (lshr_simd): DI mode added.
979 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
980 (aarch64_ushr_simddi): Likewise.
981 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
982 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
983 (vshrd_n_u64): Likewise.
985 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
987 * Makefile.in (s-macro_list): Depend on cc1.
989 2014-03-23 Teresa Johnson <tejohnson@google.com>
991 * ipa-utils.c (ipa_print_order): Use specified dump file.
993 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
995 PR rtl-optimization/60601
996 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
998 * gcc.c (eval_spec_function): Initialize save_growing_value.
1000 2014-03-22 Jakub Jelinek <jakub@redhat.com>
1003 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
1004 code == MINUS_EXPR, never swap op0 with op1.
1006 * toplev.c (init_local_tick): Avoid signed integer multiplication
1008 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
1009 shift by first operand's bitsize.
1011 2014-03-21 Jakub Jelinek <jakub@redhat.com>
1014 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
1016 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
1017 TARGET_ISA_64BIT_P(x).
1019 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1021 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
1022 pattern for vector nor instead of subtract from splat(-1).
1023 (altivec_expand_vec_perm_const_le): Likewise.
1025 2014-03-21 Richard Henderson <rth@twiddle.net>
1028 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
1029 related insns after epilogue_completed.
1031 2014-03-21 Martin Jambor <mjambor@suse.cz>
1034 * cgraph.h (symtab_node): New flag body_removed.
1035 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
1036 when removing bodies.
1037 * symtab.c (dump_symtab_base): Dump body_removed flag.
1038 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
1039 had their bodies removed.
1041 2014-03-21 Martin Jambor <mjambor@suse.cz>
1044 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
1047 2014-03-21 Richard Biener <rguenther@suse.de>
1049 PR tree-optimization/60577
1050 * tree-core.h (struct tree_base): Document nothrow_flag use
1051 in VAR_DECL_NONALIASED.
1052 * tree.h (VAR_DECL_NONALIASED): New.
1053 (may_be_aliased): Adjust.
1054 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
1056 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1058 * expr.c (expand_expr_real_1): Remove outdated comment.
1060 2014-03-20 Jakub Jelinek <jakub@redhat.com>
1063 * ira.c (adjust_cleared_regs): Call copy_rtx on
1064 *reg_equiv[REGNO (loc)].src_p before passing it to
1065 simplify_replace_fn_rtx.
1068 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
1069 into CONST, put pic register as first operand of PLUS. Use
1070 gen_const_mem for both 32-bit and 64-bit PIC got loads.
1072 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1074 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
1076 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1078 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
1079 around for store forwarding issue in the FPU on the UT699.
1080 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
1081 loads and operations if -mfix-ut699 is specified.
1082 (divtf3_hq): Tweak attribute.
1083 (sqrttf2_hq): Likewise.
1085 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1087 * calls.c (store_one_arg): Remove incorrect const qualification on the
1088 type of the temporary.
1089 * cfgexpand.c (expand_return): Likewise.
1090 * expr.c (expand_constructor): Likewise.
1091 (expand_expr_real_1): Likewise.
1093 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1095 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1098 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1101 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1103 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1105 * config/arm/aarch-common-protos.h
1106 (alu_cost_table): Fix spelling of "extend".
1107 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1109 2014-03-19 Richard Biener <rguenther@suse.de>
1112 * tree-core.h (tree_type_common): Re-order pointer members
1113 to reduce recursion depth during GC walks.
1115 2014-03-19 Marek Polacek <polacek@redhat.com>
1118 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1119 before accessing it.
1121 2014-03-19 Richard Biener <rguenther@suse.de>
1124 * lto-streamer-in.c (input_function): In WPA stage do not drop
1127 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1129 PR tree-optimization/60559
1130 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1131 with build_zero_cst assignment.
1133 2014-03-18 Kai Tietz <ktietz@redhat.com>
1135 PR rtl-optimization/56356
1136 * sdbout.c (sdbout_parms): Verify that parms'
1137 incoming argument is valid.
1138 (sdbout_reg_parms): Likewise.
1140 2014-03-18 Richard Henderson <rth@redhat.com>
1143 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1144 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1145 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1147 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1149 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1150 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1151 Italicize plugin event names in description. Explain that
1152 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1153 Remind that no GCC functions should be called after PLUGIN_FINISH.
1154 Explain what pragmas with expansion are.
1156 2014-03-18 Martin Liska <mliska@suse.cz>
1158 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1159 gimple call statement is update.
1160 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1161 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1163 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1166 * ubsan.c (ubsan_instrument_unreachable): Call
1167 initialize_sanitizer_builtins.
1168 (ubsan_pass): Likewise.
1171 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1172 varpool_finalize_decl instead of rest_of_decl_compilation.
1174 2014-03-18 Richard Biener <rguenther@suse.de>
1176 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1177 by using bitmap_and_compl instead of bitmap_and_compl_into.
1178 (df_rd_transfer_function): Likewise.
1180 2014-03-18 Richard Biener <rguenther@suse.de>
1182 * doc/lto.texi (fresolution): Fix typo.
1184 2014-03-18 Richard Biener <rguenther@suse.de>
1186 * doc/invoke.texi (flto): Update for changes in 4.9.
1188 2014-03-18 Richard Biener <rguenther@suse.de>
1190 * doc/loop.texi: Remove section on the removed lambda framework.
1191 Update loop docs with recent changes in preserving loop structure.
1193 2014-03-18 Richard Biener <rguenther@suse.de>
1195 * doc/lto.texi (-fresolution): Document.
1197 2014-03-18 Richard Biener <rguenther@suse.de>
1199 * doc/contrib.texi: Adjust my name.
1201 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1204 * internal-fn.c: Include diagnostic-core.h.
1205 (expand_BUILTIN_EXPECT): New function.
1206 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1207 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1208 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1209 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1211 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1212 Revert 3 argument __builtin_expect code.
1213 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1214 * gimple-fold.c (gimple_fold_call): Likewise.
1215 * tree.h (fold_builtin_expect): New prototype.
1216 * builtins.c (build_builtin_expect_predicate): Add predictor
1217 argument, if non-NULL, create 3 argument __builtin_expect.
1218 (fold_builtin_expect): No longer static. Add ARG2 argument,
1219 pass it through to build_builtin_expect_predicate.
1220 (fold_builtin_2): Adjust caller.
1221 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1222 * internal-fn.def (BUILTIN_EXPECT): New.
1224 2014-03-18 Tobias Burnus <burnus@net-b.de>
1227 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1228 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1229 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1231 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1234 * predict.c (combine_predictions_for_bb): Fix up formatting.
1235 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1236 fill what it points to if non-NULL.
1237 (tree_predict_by_opcode): Adjust caller, use the predictor.
1238 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1240 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1242 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1243 proper constant for the store mode.
1245 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1247 * symtab.c (change_decl_assembler_name): Fix transparent alias
1250 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1252 * config/aarch64/aarch64.c: Correct the comments about the
1253 aarch64 stack layout.
1255 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1257 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1258 check for GF_OMP_FOR_KIND_FOR.
1260 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1262 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1263 ymm and zmm register names.
1265 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1268 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1269 note creation for the 2010-08-31 changes.
1271 2014-03-17 Marek Polacek <polacek@redhat.com>
1274 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1275 as -fno-tree-loop-vectorize.
1276 (expand_omp_simd): Likewise.
1278 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1280 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1281 (eligible_for_call_delay): New prototype.
1282 * config/sparc/sparc.c (tls_call_delay): Rename into...
1283 (eligible_for_call_delay): ...this. Return false if the instruction
1284 cannot be put in the delay slot of a branch.
1285 (eligible_for_restore_insn): Simplify.
1286 (eligible_for_return_delay): Return false if the instruction cannot be
1287 put in the delay slot of a branch and simplify.
1288 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1289 put in the delay slot of a branch.
1290 * config/sparc/sparc.md (fix_ut699): New attribute.
1291 (tls_call_delay): Delete.
1292 (in_call_delay): Reimplement.
1293 (eligible_for_sibcall_delay): Rename into...
1294 (in_sibcall_delay): ...this.
1295 (eligible_for_return_delay): Rename into...
1296 (in_return_delay): ...this.
1297 (in_branch_delay): Reimplement.
1298 (in_uncond_branch_delay): Delete.
1299 (in_annul_branch_delay): Delete.
1301 2014-03-14 Richard Henderson <rth@redhat.com>
1304 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1305 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1306 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1307 (floathi splitters): Remove.
1308 (float<SWI48x>xf2): New pattern.
1309 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1310 code that tried to handle DImode for 32-bit, but which was excluded
1311 by the pattern's condition. Drop allocation of stack temporary.
1312 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1313 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1314 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1315 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1316 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1317 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1318 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1319 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1320 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1321 (*float<SWI48x><X87MODEF>2_i387): Remove.
1322 (all float _with_temp splitters): Remove.
1323 (*float<SWI48x><MODEF>2_i387): New pattern.
1324 (*float<SWI48><MODEF>2_sse): New pattern.
1325 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1326 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1328 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1329 Marek Polacek <polacek@redhat.com>
1332 * common.opt (dump_base_name_prefixed): New Variable.
1333 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1334 if x_dump_base_name_prefixed is already set, set it at the end.
1336 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1338 PR rtl-optimization/60508
1339 * lra-constraints.c (get_reload_reg): Add new parameter
1341 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1342 Pass the new parameter values.
1344 2014-03-14 Richard Biener <rguenther@suse.de>
1346 * common.opt: Revert unintented changes from r205065.
1349 2014-03-14 Richard Biener <rguenther@suse.de>
1352 * cfghooks.c (split_block): Properly adjust all loops the
1353 block was a latch of.
1355 2014-03-14 Martin Jambor <mjambor@suse.cz>
1358 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1361 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1364 * config/avr/avr.c (avr_set_current_function): Pass function name
1365 through default_strip_name_encoding before sanity checking instead
1366 of skipping the first char of the assembler name.
1368 2014-03-13 Richard Henderson <rth@redhat.com>
1371 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1372 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1373 * config/i386/i386-protos.h: Likewise.
1374 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1375 in the expander instead of a splitter.
1376 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1377 any possibility of requiring a memory.
1378 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1379 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1380 (fp branch splitters): Update for ix86_split_fp_branch.
1381 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1382 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1383 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1384 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1385 (*fop_<MODEF>_3_i387): Likewise.
1386 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1387 (splitters for the fop_* register patterns): Remove.
1388 (fscalexf4_i387): Rename from *fscalexf4_i387.
1389 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1391 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1393 PR tree-optimization/59779
1394 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1395 type for bitsize and maxsize instead of HOST_WIDE_INT.
1397 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1399 PR rtl-optimization/57320
1400 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1401 the CFG after thread_prologue_and_epilogue_insns.
1403 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1405 PR rtl-optimization/57189
1406 * lra-constraints.c (process_alt_operands): Disfavor spilling
1409 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1411 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1413 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1415 PR tree-optimization/59025
1417 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1418 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1420 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1423 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1424 calls of avr_out_plus_1.
1426 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1428 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1429 BB's single pred and update the father loop's latch info later.
1431 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1433 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1437 (VEC_base): Likewise.
1438 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1439 registers, we need to swap double words in little endian mode.
1441 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1442 to be a container mode for 128-bit integer operations added in ISA
1443 2.07. Unlike TImode and PTImode, the preferred register set is
1444 the Altivec/VMX registers for the 128-bit operations.
1446 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1448 (rs6000_split_128bit_ok_p): Likewise.
1450 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1451 macros for creating ISA 2.07 normal and overloaded builtin
1452 functions with 3 arguments.
1453 (BU_P8V_OVERLOAD_3): Likewise.
1454 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1455 for use as overloaded functions.
1456 (VPERM_1TI_UNS): Likewise.
1457 (VSEL_1TI): Likewise.
1458 (VSEL_1TI_UNS): Likewise.
1459 (ST_INTERNAL_1ti): Likewise.
1460 (LD_INTERNAL_1ti): Likewise.
1461 (XXSEL_1TI): Likewise.
1462 (XXSEL_1TI_UNS): Likewise.
1463 (VPERM_1TI): Likewise.
1464 (VPERM_1TI_UNS): Likewise.
1465 (XXPERMDI_1TI): Likewise.
1466 (SET_1TI): Likewise.
1467 (LXVD2X_V1TI): Likewise.
1468 (STXVD2X_V1TI): Likewise.
1469 (VEC_INIT_V1TI): Likewise.
1470 (VEC_SET_V1TI): Likewise.
1471 (VEC_EXT_V1TI): Likewise.
1472 (EQV_V1TI): Likewise.
1473 (NAND_V1TI): Likewise.
1474 (ORC_V1TI): Likewise.
1475 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1476 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1478 (VADDUQM): Likewise.
1479 (VSUBCUQ): Likewise.
1480 (VADDEUQM): Likewise.
1481 (VADDECUQ): Likewise.
1482 (VSUBEUQM): Likewise.
1483 (VSUBECUQ): Likewise.
1485 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1486 __int128_t and __uint128_t types.
1487 (__uint128_type): Likewise.
1488 (altivec_categorize_keyword): Add support for vector __int128_t,
1489 vector __uint128_t, vector __int128, and vector unsigned __int128
1490 as a container type for TImode operations that need to be done in
1491 VSX/Altivec registers.
1492 (rs6000_macro_to_expand): Likewise.
1493 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1494 to support 128-bit integer instructions vaddcuq, vadduqm,
1495 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1496 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1498 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1499 for V1TImode, and set up preferences to use VSX/Altivec registers.
1500 Setup VSX reload handlers.
1501 (rs6000_debug_reg_global): Likewise.
1502 (rs6000_init_hard_regno_mode_ok): Likewise.
1503 (rs6000_preferred_simd_mode): Likewise.
1504 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1505 (easy_altivec_constant): Likewise.
1506 (output_vec_const_move): Likewise.
1507 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1509 (rs6000_expand_vector_extract): Likewise.
1510 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1512 (rs6000_const_vec): Add support for V1TImode.
1513 (rs6000_emit_le_vsx_load): Swap double words when loading or
1514 storing TImode/V1TImode.
1515 (rs6000_emit_le_vsx_store): Likewise.
1516 (rs6000_emit_le_vsx_move): Likewise.
1517 (rs6000_emit_move): Add support for V1TImode.
1518 (altivec_expand_ld_builtin): Likewise.
1519 (altivec_expand_st_builtin): Likewise.
1520 (altivec_expand_vec_init_builtin): Likewise.
1521 (altivec_expand_builtin): Likewise.
1522 (rs6000_init_builtins): Add support for V1TImode type. Add
1523 support for ISA 2.07 128-bit integer builtins. Define type names
1524 for the VSX/Altivec vector types.
1525 (altivec_init_builtins): Add support for overloaded vector
1526 functions with V1TImode type.
1527 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1528 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1530 (rs6000_split_128bit_ok_p): Likewise.
1531 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1532 __int128_t and vector __uint128_t.
1534 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1535 and mode attributes.
1542 (VS_scalar): Likewise.
1543 (VS_double): Likewise.
1544 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1546 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1547 we support the ISA 2.07 128-bit integer arithmetic instructions.
1548 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1549 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1550 and TImode types for use with the builtin functions.
1551 (V1TI_type_node): Likewise.
1552 (unsigned_V1TI_type_node): Likewise.
1553 (intTI_type_internal_node): Likewise.
1554 (uintTI_type_internal_node): Likewise.
1556 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1557 128-bit builtin functions.
1558 (UNSPEC_VADDEUQM): Likewise.
1559 (UNSPEC_VADDECUQ): Likewise.
1560 (UNSPEC_VSUBCUQ): Likewise.
1561 (UNSPEC_VSUBEUQM): Likewise.
1562 (UNSPEC_VSUBECUQ): Likewise.
1563 (VM): Add V1TImode to vector mode iterators.
1565 (VI_unit): Likewise.
1566 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1567 (altivec_vaddcuq): Likewise.
1568 (altivec_vsubuqm): Likewise.
1569 (altivec_vsubcuq): Likewise.
1570 (altivec_vaddeuqm): Likewise.
1571 (altivec_vaddecuq): Likewise.
1572 (altivec_vsubeuqm): Likewise.
1573 (altivec_vsubecuq): Likewise.
1575 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1577 (BOOL_128): Likewise.
1578 (BOOL_REGS_OUTPUT): Likewise.
1579 (BOOL_REGS_OP1): Likewise.
1580 (BOOL_REGS_OP2): Likewise.
1581 (BOOL_REGS_UNARY): Likewise.
1582 (BOOL_REGS_AND_CR0): Likewise.
1584 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1585 128-bit integer builtin support.
1586 (vec_vadduqm): Likewise.
1587 (vec_vaddecuq): Likewise.
1588 (vec_vaddeuqm): Likewise.
1589 (vec_vsubecuq): Likewise.
1590 (vec_vsubeuqm): Likewise.
1591 (vec_vsubcuq): Likewise.
1592 (vec_vsubuqm): Likewise.
1594 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1595 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1596 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1597 128-bit integer add/subtract to ISA 2.07.
1599 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1601 * config/arc/arc.c (arc_predicate_delay_insns):
1602 Fix third argument passed to conditionalize_nonjump.
1604 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1606 * config/aarch64/aarch64-builtins.c
1607 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1608 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1609 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1610 instead of __builtin_lfloor.
1611 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1613 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1615 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1616 (tree_ssa_ifcombine_bb_1): New function.
1617 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1618 is an empty forwarder block to then_bb or vice versa and then_bb
1619 and else_bb are effectively swapped.
1621 2014-03-12 Christian Bruel <christian.bruel@st.com>
1624 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1625 REG_CFA_DEF_CFA note.
1626 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1627 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1629 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1631 PR tree-optimization/60454
1632 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1634 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1636 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1637 Do not define target_cpu_default2 to generic.
1638 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1639 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1640 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1642 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1643 Marc Glisse <marc.glisse@inria.fr>
1645 PR tree-optimization/60502
1646 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1647 instead of build_low_bits_mask.
1649 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1652 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1653 if there are multiple uses, but op doesn't live on E edge.
1654 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1655 clobber stmts before __builtin_unreachable.
1657 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1659 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1660 hard_frame_pointer_rtx.
1661 * cse.c (cse_insn): Remove volatile check.
1662 * cselib.c (cselib_process_insn): Likewise.
1663 * dse.c (scan_insn): Likewise.
1665 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1667 * config/arc/arc.c (conditionalize_nonjump): New function,
1669 (arc_ifcvt): ... this.
1670 (arc_predicate_delay_insns): Use it.
1672 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1674 * config/arc/predicates.md (extend_operand): During/after reload,
1675 allow const_int_operand.
1676 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1677 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1678 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1680 (umulsi3_highpart_i): Likewise.
1682 2014-03-11 Richard Biener <rguenther@suse.de>
1684 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1685 Add asserts to guard possible wrong-code bugs.
1687 2014-03-11 Richard Biener <rguenther@suse.de>
1689 PR tree-optimization/60429
1690 PR tree-optimization/60485
1691 * tree-ssa-structalias.c (set_union_with_increment): Properly
1692 take into account all fields that overlap the shifted vars.
1693 (do_sd_constraint): Likewise.
1694 (do_ds_constraint): Likewise.
1695 (get_constraint_for_ptr_offset): Likewise.
1697 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1699 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1700 (nios2_compute_frame_layout):
1701 Add calculation of cfun->machine->fp_save_offset.
1702 (nios2_expand_prologue): Correct setting of frame pointer register
1704 (nios2_expand_epilogue): Update recovery of stack pointer from
1705 frame pointer accordingly.
1706 (nios2_initial_elimination_offset): Update calculation of offset
1707 for eliminating to HARD_FRAME_POINTER_REGNUM.
1709 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1712 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1713 cgraph_get_create_node on VAR_DECLs.
1715 2014-03-10 Richard Biener <rguenther@suse.de>
1718 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1720 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1722 * config/vms/vms.opt (vms_float_format): New variable.
1724 2014-03-08 Tobias Burnus <burnus@net-b.de>
1726 * doc/invoke.texi (-fcilkplus): Update implementation status.
1728 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1729 Richard Biener <rguenther@suse.de>
1731 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1732 consistently accross all TUs.
1733 (run_gcc): Enable -fshort-double automatically at link at link-time
1734 and disallow override.
1736 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1739 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1740 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1741 if they can't be used.
1743 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1745 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1746 for Solaris 11/x86 ld.
1747 * configure: Regenerate.
1749 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1751 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1752 (LIB_TLS_SPEC): Save as ld_tls_libs.
1753 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1754 (HAVE_AS_IX86_TLSLDM): New test.
1755 * configure, config.in: Regenerate.
1756 * config/i386/i386.c (legitimize_tls_address): Fall back to
1757 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1758 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1759 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1760 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1762 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1764 * common.opt (fira-loop-pressure): Mark as optimization.
1766 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1768 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1769 an OpenMP mappable type.
1771 2014-03-06 Matthias Klose <doko@ubuntu.com>
1773 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1774 MULTILIB_OSDIRNAMES is not defined.
1776 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1777 Meador Inge <meadori@codesourcery.com>
1780 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1781 (arm_legitimize_address): Call legitimize_tls_address for any
1782 arm_tls_referenced_p expression, handle constant addend. Call it
1783 before testing for !TARGET_ARM.
1784 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1786 2014-03-06 Richard Biener <rguenther@suse.de>
1792 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1794 * tree-streamer.c (record_common_node): Assert we don't record
1795 nodes with type double.
1796 (preload_common_node): Skip type double, complex double and double
1797 pointer since it is now frontend dependent due to fshort-double option.
1799 2014-03-06 Richard Biener <rguenther@suse.de>
1801 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1802 or -fno-lto is specified and the linker has full plugin support.
1803 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1804 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1805 * lto-wrapper.c (merge_and_complain): Merge compile-time
1806 optimization levels.
1807 (run_gcc): And pass it through to the link options.
1809 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1813 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1815 * cselib.c (remove_useless_values): Skip to avoid quadratic
1816 behavior if the condition moved from...
1817 (cselib_process_insn): ... here holds.
1819 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1822 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1823 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1826 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1827 (TM_H): Add x86-tune.def.
1829 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1831 * config/aarch64/aarch64.c (generic_tunings):
1832 Use cortexa57_extra_costs.
1834 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1837 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1838 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1839 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1842 2014-03-04 Heiher <r@hev.cc>
1844 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1845 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1847 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1849 * config/i386/predicates.md (const2356_operand): Change to ...
1850 (const2367_operand): ... this.
1851 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1853 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1854 (*avx512pf_scatterpf<mode>sf): Ditto.
1855 (avx512pf_scatterpf<mode>df): Ditto.
1856 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1857 (*avx512pf_scatterpf<mode>df): Ditto.
1858 * config/i386/i386.c (ix86_expand_builtin): Update
1859 incorrect hint operand error message.
1861 2014-03-04 Richard Biener <rguenther@suse.de>
1863 * lto-section-in.c (lto_get_section_data): Fix const cast.
1865 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1867 * tree-streamer.c (record_common_node): Assert we don't record
1868 nodes with type double.
1869 (preload_common_node): Skip type double, complex double and double
1870 pointer since it is now frontend dependent due to fshort-double option.
1872 2014-03-04 Richard Biener <rguenther@suse.de>
1875 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1876 (lto_input_toplevel_asms): Likewise.
1877 * lto-section-in.c (lto_get_section_data): Instead do it here
1880 2014-03-04 Richard Biener <rguenther@suse.de>
1882 PR tree-optimization/60382
1883 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1884 dead PHIs a reduction.
1886 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1888 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1890 (_mm_prefetch): Move out of GCC target("sse") pragma.
1891 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1892 GCC target("prfchw") pragma.
1893 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1895 * config/i386/i386.c (ix86_option_override_internal): Enable
1896 -mprfchw with -mprefetchwt1.
1898 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1900 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1903 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1905 * opts.h (CL_PCH_IGNORE): Define.
1906 * targhooks.c (option_affects_pch_p):
1907 Return false for options that have CL_PCH_IGNORE set.
1908 * opt-functions.awk: Process PchIgnore.
1909 * doc/options.texi: Document PchIgnore.
1911 * config/arc/arc.opt (misize): Add PchIgnore property.
1913 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1915 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1916 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1917 constraint on constants to permit them being loaded into
1918 GENERAL_REGS or BASE_REGS.
1920 2014-03-03 Nick Clifton <nickc@redhat.com>
1922 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1923 anti-cacnonical alternatives.
1924 (negandhi3_real): New pattern.
1925 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1927 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1929 * config/avr/avr-mcus.def: Remove atxmega16x1.
1930 * config/avr/avr-tables.opt: Regenerate.
1931 * config/avr/t-multilib: Regenerate.
1932 * doc/avr-mmcu.texi: Regenerate.
1934 2014-03-03 Tobias Grosser <tobias@grosser.es>
1935 Mircea Namolaru <mircea.namolaru@inria.fr>
1937 PR tree-optimization/58028
1938 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1941 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1943 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1944 not handled by recognizers.
1946 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1949 * function.c (expand_function_end): Don't emit
1950 clobber_return_register sequence if clobber_after is a BARRIER.
1951 * cfgexpand.c (construct_exit_block): Append instructions before
1952 return_label to prev_bb.
1954 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1956 * config/rs6000/constraints.md: Document reserved use of "wc".
1958 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1961 * ipa.c (function_and_variable_visibility): When dissolving comdat
1962 group, also set all symbols to local.
1964 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1969 2013-12-14 Jan Hubicka <jh@suse.cz>
1971 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1973 2014-03-02 Jon Beniston <jon@beniston.com>
1979 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1980 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1981 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1982 (simple_return, *simple_return): New patterns
1983 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1984 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1986 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1988 * dwarf2out.c (gen_subprogram_die): Tidy.
1990 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1993 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1994 (*mov_t_msb_neg_negc): ... this new insn.
1996 2014-02-28 Jason Merrill <jason@redhat.com>
1999 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
2002 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
2005 * dwarf2out.c (decltype_auto_die): New static.
2006 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
2007 (gen_type_die_with_usage): Handle 'decltype(auto)'.
2008 (is_cxx_auto): Likewise.
2010 2014-02-28 Ian Bolton <ian.bolton@arm.com>
2012 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
2013 we are not using general regs only.
2015 2014-02-28 Richard Biener <rguenther@suse.de>
2018 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
2019 previous fix and only allow to remove trivial pre-headers
2020 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
2021 (remove_forwarder_block): Properly update the latch of a loop.
2023 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2026 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
2027 (cselib_preserved_hash_table): New.
2028 (preserve_constants_and_equivs): Move preserved vals to it.
2029 (cselib_find_slot): Look it up first.
2030 (cselib_init): Initialize it.
2031 (cselib_finish): Release it.
2032 (dump_cselib_table): Dump it.
2034 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2037 * cselib.c (remove_useless_values): Skip to avoid quadratic
2038 behavior if the condition moved from...
2039 (cselib_process_insn): ... here holds.
2041 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
2044 * var-tracking.c (vt_initialize): Apply the same condition to
2045 preserve the CFA base value.
2047 2014-02-28 Joey Ye <joey.ye@arm.com>
2050 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
2051 if reload in progress or completed.
2053 2014-02-28 Tobias Burnus <burnus@net-b.de>
2056 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
2059 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
2061 * doc/tm.texi.in (Condition Code Status): Update documention for
2062 relative locations of cc0-setter and cc0-user.
2064 2014-02-27 Jeff Law <law@redhat.com>
2066 PR rtl-optimization/52714
2067 * combine.c (try_combine): When splitting an unrecognized PARALLEL
2068 into two independent simple sets, if I3 is a jump, ensure the
2069 pattern we place into I3 is a (set (pc) ...).
2071 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
2072 Jeff Law <law@redhat.com>
2074 PR rtl-optimization/49847
2075 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
2076 are in different blocks.
2077 * doc/tm.texi (Condition Code Status): Update documention for
2078 relative locations of cc0-setter and cc0-user.
2080 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
2083 * lra.c (lra_emit_add): Check SUBREG too.
2085 2014-02-27 Andreas Schwab <schwab@suse.de>
2087 * config/m68k/m68k.c (m68k_option_override): Disable
2088 -flive-range-shrinkage for classic m68k.
2089 (m68k_override_options_after_change): Likewise.
2091 2014-02-27 Marek Polacek <polacek@redhat.com>
2094 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2095 -Wmaybe-uninitialized.
2097 2014-02-27 Alan Modra <amodra@gmail.com>
2100 * reload1.c (emit_input_reload_insns): When reload_override_in,
2101 set old to rl->in_reg when rl->in_reg is a subreg.
2103 2014-02-26 Richard Biener <rguenther@suse.de>
2106 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2108 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2110 * common/config/i386/predicates.md (const1256_operand): Remove.
2111 (const2356_operand): New.
2112 (const_1_to_2_operand): Remove.
2113 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2114 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2115 (*avx512pf_gatherpf<mode>sf): Ditto.
2116 (avx512pf_gatherpf<mode>df): Ditto.
2117 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2118 (*avx512pf_gatherpf<mode>df): Ditto.
2119 (avx512pf_scatterpf<mode>sf): Ditto.
2120 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2121 (*avx512pf_scatterpf<mode>sf): Ditto.
2122 (avx512pf_scatterpf<mode>df): Ditto.
2123 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2124 (*avx512pf_scatterpf<mode>df): Ditto.
2125 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2127 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2129 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2130 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2131 (_mm512_mask_testn_epi64_mask): Move to ...
2132 * config/i386/avx512cdintrin.h: Here.
2133 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2134 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2135 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2136 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2137 TARGET_AVX512F from TARGET_AVX512CD.
2139 2014-02-26 Richard Biener <rguenther@suse.de>
2142 * ipa.c (walk_polymorphic_call_targets): Properly guard
2143 call to inline_update_overall_summary.
2145 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2148 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2149 and latches only if requested. Fix latch if it is removed.
2150 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2151 LOOPS_HAVE_PREHEADERS.
2153 2014-02-25 Andrew Pinski <apinski@cavium.com>
2155 * builtins.c (expand_builtin_thread_pointer): Create a new target
2156 when the target is NULL.
2158 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2160 PR rtl-optimization/60317
2161 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2162 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2163 * lra-assigns.c: Include params.h.
2164 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2165 other reload pseudos considerations.
2167 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2169 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2170 to use canonical form for nor<mode>3.
2172 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2175 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2178 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2180 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2181 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2182 (ix86_handle_option): Handle OPT_mprefetchwt1.
2183 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2184 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2186 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2187 OPTION_MASK_ISA_PREFETCHWT1.
2188 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2189 (PTA_PREFETCHWT1): New.
2190 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2191 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2192 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2193 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2194 (*prefetch_avx512pf_<mode>_: Change into ...
2195 (*prefetch_prefetchwt1_<mode>: This.
2196 * config/i386/i386.opt (mprefetchwt1): New.
2197 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2198 (_mm_prefetch): Handle intent to write.
2199 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2201 2014-02-25 Richard Biener <rguenther@suse.de>
2204 * emit-rtl.c (mem_attrs_htab): Remove.
2205 (mem_attrs_htab_hash): Likewise.
2206 (mem_attrs_htab_eq): Likewise.
2207 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2208 (init_emit_once): Do not allocate mem_attrs_htab.
2210 2014-02-25 Richard Biener <rguenther@suse.de>
2213 * lto-opts.c (lto_write_options): Output non-explicit conservative
2214 -fwrapv, -fno-trapv and -fno-strict-overflow.
2215 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2216 (run_gcc): And pass them through.
2218 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2220 * sel-sched.c (calculate_new_fences): New parameter ptime.
2221 Calculate it as a maximum over all fence cycles.
2222 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2223 Print the final schedule timing when sched_verbose.
2225 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2227 PR rtl-optimization/60292
2228 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2229 bit fot the fence instruction.
2231 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2233 * calls.h: Fix typo in comment.
2235 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2237 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2238 adjusting offsetable addresses.
2240 2014-02-24 Guozhi Wei <carrot@google.com>
2242 * sparseset.h (sparseset_pop): Fix the wrong index.
2244 2014-02-24 Walter Lee <walt@tilera.com>
2246 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2247 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2249 * common/config/tilegx/tilegx-common.c
2250 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2251 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2253 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2254 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2255 (tilegx_gimplify_va_arg_expr): Handle big endian.
2256 (tilegx_expand_unaligned_load): Ditto.
2257 (tilegx_expand_unaligned_store): Ditto.
2258 (TARGET_RETURN_IN_MSB): New.
2259 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2260 (TARGET_ENDIAN_DEFAULT): New.
2261 (TARGET_BIG_ENDIAN): Handle big endian.
2262 (BYTES_BIG_ENDIAN): Ditto.
2263 (WORDS_BIG_ENDIAN): Ditto.
2264 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2267 * config/tilegx/tilegx.md (extv): Handle big endian.
2269 (insn_st<n>): Ditto.
2270 (insn_st<n>_add<bitsuffix>): Ditto.
2271 (insn_stnt<n>): Ditto.
2272 (insn_stnt<n>_add<bitsuffix>):Ditto.
2273 (vec_interleave_highv8qi): Handle big endian.
2274 (vec_interleave_highv8qi_be): New.
2275 (vec_interleave_highv8qi_le): New.
2276 (insn_v1int_h): Handle big endian.
2277 (vec_interleave_lowv8qi): Handle big endian.
2278 (vec_interleave_lowv8qi_be): New.
2279 (vec_interleave_lowv8qi_le): New.
2280 (insn_v1int_l): Handle big endian.
2281 (vec_interleave_highv4hi): Handle big endian.
2282 (vec_interleave_highv4hi_be): New.
2283 (vec_interleave_highv4hi_le): New.
2284 (insn_v2int_h): Handle big endian.
2285 (vec_interleave_lowv4hi): Handle big endian.
2286 (vec_interleave_lowv4hi_be): New.
2287 (vec_interleave_lowv4hi_le): New.
2288 (insn_v2int_l): Handle big endian.
2289 (vec_interleave_highv2si): Handle big endian.
2290 (vec_interleave_highv2si_be): New.
2291 (vec_interleave_highv2si_le): New.
2292 (insn_v4int_h): Handle big endian.
2293 (vec_interleave_lowv2si): Handle big endian.
2294 (vec_interleave_lowv2si_be): New.
2295 (vec_interleave_lowv2si_le): New.
2296 (insn_v4int_l): Handle big endian.
2297 * config/tilegx/tilegx.opt (mbig-endian): New option.
2298 (mlittle-endian): New option.
2299 * doc/install.texi: Document tilegxbe-linux.
2300 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2302 2014-02-24 Martin Jambor <mjambor@suse.cz>
2305 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2306 there are no parameter descriptors.
2308 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2310 PR rtl-optimization/60268
2311 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2312 initialization to ...
2313 (sched_rgn_init): ... here.
2314 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2316 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2318 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2321 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2323 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2326 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2328 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2329 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2331 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2333 * config/microblaze/predicates.md: Add cmp_op predicate.
2334 * config/microblaze/microblaze.md: Add branch_compare instruction
2335 which uses cmp_op predicate and emits cmp insn before branch.
2336 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2337 to microblaze_expand_conditional_branch and consolidate logic.
2338 (microblaze_expand_conditional_branch): emit branch_compare
2339 insn instead of handling cmp op separate from branch insn.
2341 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2343 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2346 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2348 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2349 define_insn with define_expand and new define_insn
2350 *altivec_lve<VI_char>x_internal.
2351 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2352 and new define_insn *altivec_stve<VI_char>x_internal.
2353 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2355 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2357 (altivec_expand_stvex_be): New function.
2359 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2361 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2362 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2363 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2364 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2366 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2369 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2370 instead of emit_move_insn.
2372 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2374 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2376 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2377 gen_altivec_vsumsws.
2379 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2381 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2382 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2383 (altivec_lvxl_<mode>): New define_expand incorporating
2384 -maltivec=be semantics where needed.
2385 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2386 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2387 semantics where needed.
2388 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2389 (altivec_stvx_<mode>): New define_expand incorporating
2390 -maltivec=be semantics where needed.
2391 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2392 VM2 iterator instead of V4SI.
2393 (altivec_stvxl_<mode>): New define_expand incorporating
2394 -maltivec=be semantics where needed.
2395 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2396 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2397 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2398 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2399 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2400 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2401 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2402 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2403 ALTIVEC_BUILTIN_STVXL.
2404 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2405 (altivec_expand_stvx_be): Likewise.
2406 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2407 (altivec_expand_lvx_be): Likewise.
2408 (altivec_expand_stvx_be): Likewise.
2409 (altivec_expand_builtin): Add cases for
2410 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2411 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2412 (altivec_init_builtins): Add definitions for
2413 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2414 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2416 2014-02-21 Catherine Moore <clm@codesourcery.com>
2418 * doc/invoke.texi (mvirt, mno-virt): Document.
2419 * config/mips/mips.opt (mvirt): New option.
2420 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2422 2014-02-21 Richard Biener <rguenther@suse.de>
2424 PR tree-optimization/60276
2425 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2426 (STMT_VINFO_MIN_NEG_DIST): New macro.
2427 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2428 STMT_VINFO_MIN_NEG_DIST.
2429 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2430 made for negative dependence distances still hold.
2432 2014-02-21 Richard Biener <rguenther@suse.de>
2435 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2436 DECL_INITIAL for globals not in the current function context.
2438 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2440 PR tree-optimization/56490
2441 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2442 * tree-ssa-uninit.c: Include params.h.
2443 (compute_control_dep_chain): Add num_calls argument, return false
2444 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2445 num_calls to recursive call.
2446 (find_predicates): Change dep_chain into normal array,
2447 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2448 variable and adjust compute_control_dep_chain caller.
2449 (find_def_preds): Likewise.
2451 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2453 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2454 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2456 2014-02-21 Nick Clifton <nickc@redhat.com>
2458 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2459 (pushhi1): Likewise.
2460 (popqi1): Add mode to pre_dec.
2463 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2465 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2466 mode for mask of V8SFmode permutation.
2468 2014-02-20 Richard Henderson <rth@redhat.com>
2471 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2472 a new pseudo for OLDVAL.
2474 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2477 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2478 gen_reg_rtx if d->testing_p.
2479 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2480 if d->testing_p and we will certainly return true.
2481 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2484 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2486 * emit-rtl.c (gen_reg_rtx): Assert that
2487 crtl->emit.regno_pointer_align_length is non-zero.
2489 2014-02-20 Richard Henderson <rth@redhat.com>
2492 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2493 on failure the store back into EXPECT.
2495 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2496 Sandra Loosemore <sandra@codesourcery.com>
2498 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2499 * config/nios2/nios2.c (nios2_function_profiler): Add
2500 -fPIC (flag_pic == 2) support.
2501 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2502 (nios2_large_offset_p): New function.
2503 (nios2_unspec_reloc_p): Move up position, update to use
2504 nios2_large_offset_p.
2505 (nios2_unspec_address): Remove function.
2506 (nios2_unspec_offset): New function.
2507 (nios2_large_got_address): New function.
2508 (nios2_got_address): Add large offset support.
2509 (nios2_legitimize_tls_address): Update usage of removed and new
2511 (nios2_symbol_binds_local_p): New function.
2512 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2513 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2514 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2515 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2516 processing for (const (unspec ...)).
2517 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2519 2014-02-20 Richard Biener <rguenther@suse.de>
2521 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2522 doing the substitution.
2523 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2525 2014-02-20 Martin Jambor <mjambor@suse.cz>
2528 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2529 info when checking whether lattices are bottom.
2531 2014-02-20 Richard Biener <rguenther@suse.de>
2534 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2537 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2540 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2541 parameter specifying the scaling.
2542 (inline_call): Update.
2543 (want_inline_recursively): Guard division by zero.
2544 (recursive_inlining): Update.
2545 * ipa-inline.h (clone_inlined_nodes): Update.
2547 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2550 * config/i386/i386.c (classify_argument): Pass structures of size
2551 64 bytes or less in register.
2553 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2554 Kirill Yukhin <kirill.yukhin@intel.com>
2556 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2557 (_mm_rcp28_round_ss): Ditto.
2558 (_mm_rsqrt28_round_sd): Ditto.
2559 (_mm_rsqrt28_round_ss): Ditto.
2560 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2561 (_mm_rcp14_round_ss): Ditto.
2562 (_mm_rsqrt14_round_sd): Ditto.
2563 (_mm_rsqrt14_round_ss): Ditto.
2564 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2565 the first input operand, get rid of match_dup.
2566 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2568 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2570 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2571 operand as the first input operand, set type attribute.
2572 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2574 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2575 operand as the first input operand, set type attribute.
2577 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2579 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2582 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2585 * config/i386/i386.c (construct_container): Remove TFmode check
2586 for X86_64_INTEGER_CLASS.
2588 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2591 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2592 only when -Wpsabi is enabled.
2594 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2597 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2598 passing arrays in registers are the same as for structs, so remove the
2599 special case for them.
2601 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2603 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2604 destination type, extract only the valid bits if the source type is not
2605 integral and has a different mode.
2607 2014-02-19 Richard Biener <rguenther@suse.de>
2610 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2613 2014-02-19 Richard Biener <rguenther@suse.de>
2616 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2617 (ipa_modify_call_arguments): Emit an argument load explicitely and
2618 preserve virtual SSA form there and for the replacement call.
2619 Do not update SSA form nor free dominance info.
2621 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2623 * ipa.c (function_and_variable_visibility): Also clear WEAK
2624 flag when disolving COMDAT_GROUP.
2626 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2628 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2629 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2631 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2632 do more sanity checks.
2633 (detect_type_change): Return true when giving up early.
2634 (compute_complex_assign_jump_func): Fix type parameter of
2635 ipa_set_ancestor_jf.
2636 (compute_complex_ancestor_jump_func): Likewise.
2637 (update_jump_functions_after_inlining): Fix updating of
2639 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2641 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2643 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2644 inline clones when edge disappears.
2646 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2649 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2650 Split 64-bit moves into 2 patterns. Do not allow the use of
2651 direct move for TDmode in little endian, since the decimal value
2652 has little endian bytes within a word, but the 64-bit pieces are
2653 ordered in a big endian fashion, and normal subreg's of TDmode are
2655 (mov<mode>_64bit_dm): Likewise.
2656 (movtd_64bit_nodm): Likewise.
2658 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2660 PR tree-optimization/60174
2661 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2662 statement of an SSA_NAME that occurs in an abnormal PHI node.
2664 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2667 * final.c (SEEN_BB): Remove.
2668 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2669 (final_scan_insn): Don't force_source_line on second
2670 NOTE_INSN_BASIC_BLOCK.
2672 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2675 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2676 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2677 (type_natural_mode): Warn ABI change when %zmm register is not
2678 available for AVX512F vector value passing.
2680 2014-02-18 Kai Tietz <ktietz@redhat.com>
2683 * config/i386/i386.c (ix86_expand_prologue): Use value in
2684 rax register as displacement when restoring %r10 or %rax.
2685 Fix wrong offset when restoring both registers.
2687 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2689 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2690 assertion with conditional return.
2692 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2693 Uros Bizjak <ubizjak@gmail.com>
2696 * config/i386/driver-i386.c (host_detect_local_cpu): If
2697 YMM state is not saved by the OS, also clear has_f16c. Move
2698 CPUID 0x80000001 handling before YMM state saving checking.
2700 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2702 PR rtl-optimization/58960
2703 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2704 factored out from ...
2705 (sched_init): ... here.
2706 (free_global_sched_pressure_data): New, factored out from ...
2707 (sched_finish): ... here.
2708 * sched-int.h (free_global_sched_pressure_data): Declare.
2709 * sched-rgn.c (nr_regions_initial): New static global.
2710 (haifa_find_rgns): Initialize it.
2711 (schedule_region): Disable sched-pressure for the newly
2714 2014-02-17 Richard Biener <rguenther@suse.de>
2716 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2717 release SSA defs of pattern stmts.
2719 2014-02-17 Richard Biener <rguenther@suse.de>
2721 * tree-inline.c (expand_call_inline): Release the virtual
2722 operand defined by the call we are about to inline.
2724 2014-02-17 Richard Biener <rguenther@suse.de>
2726 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2728 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2729 Ilya Tocar <ilya.tocar@intel.com>
2731 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2732 arguments order in builtin.
2733 (_mm512_permutexvar_epi64): Ditto.
2734 (_mm512_mask_permutexvar_epi64): Ditto
2735 (_mm512_maskz_permutexvar_epi32): Ditto
2736 (_mm512_permutexvar_epi32): Ditto
2737 (_mm512_mask_permutexvar_epi32): Ditto
2739 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2741 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2742 (p8_vmrgow): Likewise.
2744 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2746 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2749 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2752 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2753 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2754 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2755 using direct move instructions on ISA 2.07. Also adjust
2756 instruction length for 64-bit.
2757 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2758 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2760 2014-02-15 Alan Modra <amodra@gmail.com>
2764 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2765 find_replacement on parts of insn rtl that might be reloaded.
2767 2014-02-15 Richard Biener <rguenther@suse.de>
2769 PR tree-optimization/60183
2770 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2771 (tree_ssa_phiprop): Calculate and free post-dominators.
2773 2014-02-14 Jeff Law <law@redhat.com>
2775 PR rtl-optimization/60131
2776 * ree.c (get_extended_src_reg): New function.
2777 (combine_reaching_defs): Use it rather than assuming location of REG.
2778 (find_and_remove_re): Verify first operand of extension is
2779 a REG before adding the insns to the copy list.
2781 2014-02-14 Roland McGrath <mcgrathr@google.com>
2783 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2784 * configure: Regenerated.
2785 * config.in: Regenerated.
2786 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2787 instead of ASM_SHORT.
2789 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2790 Richard Earnshaw <rearnsha@arm.com>
2792 PR rtl-optimization/59535
2793 * lra-constraints.c (process_alt_operands): Encourage alternative
2794 when unassigned pseudo class is superset of the alternative class.
2795 (inherit_reload_reg): Don't inherit when optimizing for code size.
2796 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2797 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2798 modes not less than 4 for Thumb1.
2800 2014-02-14 Kyle McMartin <kyle@redhat.com>
2803 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2805 2014-02-14 Richard Biener <rguenther@suse.de>
2807 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2808 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2809 Do not strip INDIRECT_REFs.
2811 2014-02-14 Richard Biener <rguenther@suse.de>
2814 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2815 DECL_FUNCTION_SPECIFIC_TARGET.
2816 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2817 * tree-streamer-out.c (pack_ts_target_option): Remove.
2818 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2819 (write_ts_function_decl_tree_pointers): Do not stream
2820 DECL_FUNCTION_SPECIFIC_TARGET.
2821 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2822 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2823 (lto_input_ts_function_decl_tree_pointers): Do not stream
2824 DECL_FUNCTION_SPECIFIC_TARGET.
2826 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2828 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2829 (get_initial_def_for_induction, vectorizable_induction): Ignore
2830 debug stmts when looking for exit_phi.
2831 (vectorizable_live_operation): Fix up condition.
2833 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2835 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2836 nreverse() because it changes the content of original tree list.
2838 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2840 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2841 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2843 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2845 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2846 GNU coding standards.
2848 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2851 * dwarf2out.c (gen_subprogram_die): Don't call
2852 add_calling_convention_attribute if subr_die is old_die.
2854 2014-02-13 Sharad Singhai <singhai@google.com>
2856 * doc/optinfo.texi: Fix order of nodes.
2858 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2860 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2861 operands[2], not operands[3].
2863 2014-02-13 Richard Biener <rguenther@suse.de>
2866 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2867 mention the possibility of an in-tree build.
2868 (CLooG): Update recommended version to 0.18.1, mention the
2869 possibility of an in-tree build and clarify that the ISL
2870 bundled with CLooG does not work.
2872 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2875 * expr.c (compress_float_constant): If x is a hard register,
2876 extend into a pseudo and then move to x.
2878 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2880 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2881 caused by bad second argument to warning_at() with -mhotpatch and
2882 nested functions (e.g. with gfortran).
2884 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2886 * opts.c (option_name): Remove "enabled by default" rider.
2888 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2890 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2892 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2893 Uros Bizjak <ubizjak@gmail.com>
2896 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2897 * configure: Regenerated.
2899 2014-02-12 Richard Biener <rguenther@suse.de>
2901 * vec.c (vec_prefix::calculate_allocation): Move as
2902 inline variant to vec.h.
2903 (vec_prefix::calculate_allocation_1): New out-of-line version.
2904 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2905 (vec_prefix::m_has_auto_buf): Rename to ...
2906 (vec_prefix::m_using_auto_storage): ... this.
2907 (vec_prefix::calculate_allocation): Inline the easy cases
2908 and dispatch to calculate_allocation_1 which doesn't need the
2910 (va_heap::reserve): Use gcc_checking_assert.
2911 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2912 m_using_auto_storage.
2913 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2915 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2916 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2917 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2919 2014-02-12 Richard Biener <rguenther@suse.de>
2921 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2922 when we found a dependence.
2924 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2926 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2928 (maybe_fold_stmt): ... into this new function.
2929 * omp-low.c (lower_omp): Update comment.
2931 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2934 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2937 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2939 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2940 identifiers in comments.
2941 (cortexa53_extra_costs): Likewise.
2942 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2943 (cortexa7_extra_costs): Likewise.
2944 (cortexa12_extra_costs): Likewise.
2945 (cortexa15_extra_costs): Likewise.
2946 (v7m_extra_costs): Likewise.
2948 2014-02-12 Richard Biener <rguenther@suse.de>
2951 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2952 of posix_memalign being successful.
2953 (lower_stmt): Restrict lowering of posix_memalign to when
2954 -ftree-bit-ccp is enabled.
2956 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2958 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2960 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2962 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2964 PR rtl-optimization/60116
2965 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2966 other_insn once the combination has been validated.
2968 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2971 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2973 * ipa-devirt.c: Include demangle.h
2974 (odr_violation_reported): New static variable.
2975 (add_type_duplicate): Update odr_violations.
2976 (maybe_record_node): Add completep parameter; update it.
2977 (record_target_from_binfo): Add COMPLETEP parameter;
2978 update it as needed.
2979 (possible_polymorphic_call_targets_1): Likewise.
2980 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2981 rename FINAL to COMPLETE.
2982 (record_targets_from_bases): Sanity check we found the binfo;
2983 fix COMPLETEP updating.
2984 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2985 parameter, fix computing of COMPLETEP.
2986 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2987 at LTO time do demangling.
2988 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2989 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2991 (gimple_get_virt_method_for_binfo): Likewise.
2992 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2993 gimple_get_virt_method_for_vtable): Update prototypes.
2995 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2998 * genautomata.c (add_presence_absence): Fix typo with
2999 {final_}presence_list.
3001 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
3004 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
3005 for VSX/Altivec vectors that land in GPR registers.
3007 2014-02-11 Richard Henderson <rth@redhat.com>
3008 Jakub Jelinek <jakub@redhat.com>
3011 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
3012 around drhs if type conversion to lacc->type is not useless.
3014 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3016 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
3018 (cortex-a57.cortex-a53): Likewise.
3019 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
3021 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3023 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
3026 2014-02-11 Renlin Li <Renlin.Li@arm.com>
3028 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
3029 add_options_for_arm_vfp3.
3031 2014-02-11 Jeff Law <law@redhat.com>
3034 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
3035 object with an undesirable mode.
3037 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3040 * config/i386/sol2-9.h: New file.
3041 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
3042 *-*-solaris2.9*): Use it.
3044 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3046 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
3047 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
3049 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
3051 * config/microblaze/microblaze.c: Extend mcpu version format
3053 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
3055 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
3057 2014-02-10 Richard Henderson <rth@redhat.com>
3060 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
3061 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
3062 ms-abi vs -mno-accumulate-outgoing-args.
3063 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
3064 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
3067 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3070 * cfgexpand.c (expand_asm_operands): Attach source location to
3071 ASM_INPUT rtx objects.
3072 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
3074 2014-02-10 Nick Clifton <nickc@redhat.com>
3076 * config/mn10300/mn10300.c (popcount): New function.
3077 (mn10300_expand_prologue): Include saved registers in stack usage
3080 2014-02-10 Jeff Law <law@redhat.com>
3083 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3084 when changing the SET_DEST of a prior insn to avoid an input reload.
3086 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3088 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3089 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3090 -mcall-openbsd, or -mcall-linux.
3091 (CC1_ENDIAN_BIG_SPEC): Remove.
3092 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3093 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3094 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3095 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3096 and %cc1_endian_default.
3097 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3099 2014-02-10 Richard Biener <rguenther@suse.de>
3101 PR tree-optimization/60115
3102 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3103 MEM_REF handling. Properly verify that the accesses are not
3104 out of the objects bound.
3106 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3108 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3111 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3113 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3114 proper constants and fix formatting.
3115 (possible_polymorphic_call_targets): Fix formatting.
3117 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3118 Ilya Tocar <ilya.tocar@intel.com>
3120 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3121 (_mm512_loadu_epi32): Renamed into...
3122 (_mm512_loadu_si512): This.
3123 (_mm512_storeu_epi32): Renamed into...
3124 (_mm512_storeu_si512): This.
3125 (_mm512_maskz_ceil_ps): Removed.
3126 (_mm512_maskz_ceil_pd): Ditto.
3127 (_mm512_maskz_floor_ps): Ditto.
3128 (_mm512_maskz_floor_pd): Ditto.
3129 (_mm512_floor_round_ps): Ditto.
3130 (_mm512_floor_round_pd): Ditto.
3131 (_mm512_ceil_round_ps): Ditto.
3132 (_mm512_ceil_round_pd): Ditto.
3133 (_mm512_mask_floor_round_ps): Ditto.
3134 (_mm512_mask_floor_round_pd): Ditto.
3135 (_mm512_mask_ceil_round_ps): Ditto.
3136 (_mm512_mask_ceil_round_pd): Ditto.
3137 (_mm512_maskz_floor_round_ps): Ditto.
3138 (_mm512_maskz_floor_round_pd): Ditto.
3139 (_mm512_maskz_ceil_round_ps): Ditto.
3140 (_mm512_maskz_ceil_round_pd): Ditto.
3141 (_mm512_expand_pd): Ditto.
3142 (_mm512_expand_ps): Ditto.
3143 * config/i386/i386.c (ix86_builtins): Remove
3144 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3145 (bdesc_args): Ditto.
3146 * config/i386/predicates.md (const1256_operand): New.
3147 (const_1_to_2_operand): Ditto.
3148 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3149 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3150 (*avx512pf_gatherpf<mode>sf): Ditto.
3151 (avx512pf_gatherpf<mode>df): Ditto.
3152 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3153 (*avx512pf_gatherpf<mode>df): Ditto.
3154 (avx512pf_scatterpf<mode>sf): Ditto.
3155 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3156 (*avx512pf_scatterpf<mode>sf): Ditto.
3157 (avx512pf_scatterpf<mode>df): Ditto.
3158 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3159 (*avx512pf_scatterpf<mode>df): Ditto.
3160 (avx512f_expand<mode>): Removed.
3161 (<shift_insn><mode>3<mask_name>): Change predicate type.
3163 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3165 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3166 not at the end of datarefs vector use ordered_remove to avoid
3167 reordering datarefs vector.
3170 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3171 mark local addressable non-static vars as GOVD_PRIVATE
3172 instead of GOVD_LOCAL.
3173 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3174 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3178 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3179 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3180 assume_aligned or alloc_align attributes.
3181 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3182 arguments. Handle also assume_aligned and alloc_align attributes.
3183 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3184 calls to functions with assume_aligned or alloc_align attributes.
3185 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3187 2014-02-08 Terry Guo <terry.guo@arm.com>
3189 * doc/invoke.texi: Document ARM -march=armv7e-m.
3191 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3193 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3194 flag on __cilkrts_rethrow builtin.
3197 * ipa-cp.c (determine_versionability): Fail at -O0
3198 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3199 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3202 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3205 * tree-inline.c (copy_forbidden): Fail for
3206 __attribute__((optimize (0))) functions.
3208 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3210 * varpool.c: Include pointer-set.h.
3211 (varpool_remove_unreferenced_decls): Variables in other partitions
3212 will not be output; be however careful to not lose information
3215 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3217 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3218 lookup in the vtable constructor.
3220 2014-02-07 Jeff Law <law@redhat.com>
3223 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3224 define_insn_and_split.
3226 * ipa-inline.c (inline_small_functions): Fix typos.
3228 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3230 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3231 (s390_can_use_return_insn): Declare.
3232 * config/s390/s390.h (EPILOGUE_USES): Define.
3233 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3235 (s390_chunkify_start): Handle return JUMP_LABELs.
3236 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3237 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3238 (s390_can_use_return_insn): New functions.
3239 (s390_fix_long_loop_prediction): Handle conditional returns.
3240 (TARGET_SET_UP_BY_PROLOGUE): Define.
3241 * config/s390/s390.md (ANY_RETURN): New code iterator.
3242 (*creturn, *csimple_return, return, simple_return): New patterns.
3244 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3246 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3247 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3248 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3249 REG_CFA_RESTORE list when deciding not to restore a register.
3251 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3253 * config/s390/s390.c: Include tree-pass.h and context.h.
3254 (s390_early_mach): New function, split out from...
3255 (s390_emit_prologue): ...here.
3256 (pass_data_s390_early_mach): New pass structure.
3257 (pass_s390_early_mach): New class.
3258 (s390_option_override): Create and register early_mach pass.
3259 Move to end of file.
3261 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3263 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3264 to match for the exit block.
3266 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3268 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3269 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3270 Reject misaligned operands.
3272 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3274 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3276 2014-02-07 Richard Biener <rguenther@suse.de>
3279 * gimple-low.c (lower_builtin_posix_memalign): New function.
3280 (lower_stmt): Call it to lower posix_memalign in a way
3281 to make alignment info accessible.
3283 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3286 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3287 __builtin_setjmp_receiver.
3289 2014-02-07 Richard Biener <rguenther@suse.de>
3292 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3293 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3294 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3295 Handle BUILT_IN_POSIX_MEMALIGN.
3296 (find_func_clobbers): Likewise.
3297 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3298 (call_may_clobber_ref_p_1): Likewise.
3300 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3303 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3306 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3309 * lto-cgraph.c (lto_output_node): Use
3310 symtab_get_symbol_partitioning_class.
3311 (lto_output_varpool_node): likewise.
3312 (symtab_get_symbol_partitioning_class): Move here from
3314 * cgraph.h (symbol_partitioning_class): Likewise.
3315 (symtab_get_symbol_partitioning_class): Declare.
3317 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3319 * ggc.h (ggc_internal_cleared_alloc): New macro.
3320 * vec.h (vec_safe_copy): Handle memory stats.
3321 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3322 * target-globals.c (save_target_globals): Likewise.
3324 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3327 * expr.c (emit_move_resolve_push): Export; be bit more selective
3328 on when to clear alias set.
3329 * expr.h (emit_move_resolve_push): Declare.
3330 * function.h (struct function): Add tail_call_marked.
3331 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3332 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3333 * config/i386/i386.md (TImode move expander): De not call
3335 (FP push expanders): Preserve memory attributes.
3336 * config/i386/sse.md (push<mode>1): Remove.
3337 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3338 (ix86_expand_push): Remove.
3339 * config/i386/mmx.md (push<mode>1): Remove.
3341 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3343 PR rtl-optimization/60030
3344 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3345 lopart with paradoxical subreg before shifting it up by hprec.
3347 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3349 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3350 Remove extra newline at end of file.
3351 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3352 (arm_issue_rate): Handle cortexa57.
3353 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3354 (cortex-a57.cortex-a53): Likewise.
3356 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3359 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3360 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3362 (arm_expand_prologue): Adjust all callers.
3363 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3364 info, registers also at the lowest numbered registers side. Use
3365 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3369 * var-tracking.c (adjust_mems): Before adding a SET to
3370 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3372 2014-02-06 Alan Modra <amodra@gmail.com>
3375 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3376 change SDmode to DDmode when lra_in_progress.
3378 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3381 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3382 free_data_ref on the dr first, and before goto again also set dr
3383 to the next dr. For simd_lane_access, free old datarefs[i] before
3384 overwriting it. For get_vectype_for_scalar_type failure, don't
3385 free_data_ref if simd_lane_access.
3387 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3390 * tree.h (opts_for_fn): New inline function.
3391 (opt_for_fn): Define.
3392 * config/i386/i386.c (ix86_function_regparm): Use
3393 opt_for_fn (decl, optimize) instead of optimize.
3395 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3397 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3398 for SYMBOL_REF in large memory model.
3400 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3402 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3404 (cortex-a57): Likewise.
3405 (cortex-a57.cortex-a53): Likewise.
3407 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3408 Kugan Vivekanandarajah <kuganv@linaro.org>
3410 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3412 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3414 2014-02-06 Richard Biener <rguenther@suse.de>
3416 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3417 set_loop_copy and initialize_original_copy_tables.
3419 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3421 * config/aarch64/aarch64-simd.md
3422 (aarch64_ashr_simddi): Change QI to SI.
3424 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3425 Jakub Jelinek <jakub@redhat.com>
3428 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3431 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3433 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3434 CODE_FOR_altivec_vpku[hw]um to
3435 CODE_FOR_altivec_vpku[hw]um_direct.
3436 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3437 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3438 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3439 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3441 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3443 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3444 generation for -maltivec=be.
3445 (altivec_vsumsws): Simplify redundant test.
3447 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3449 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3450 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3451 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3452 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3453 gen_altivec_vpkuwum.
3454 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3456 (altivec_vpks<VI_char>ss): Likewise.
3457 (altivec_vpks<VI_char>us): Likewise.
3458 (altivec_vpku<VI_char>us): Likewise.
3459 (altivec_vpku<VI_char>um): Likewise.
3460 (altivec_vpku<VI_char>um_direct): New (copy of
3461 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3463 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3464 target is little endian and -maltivec=be is not specified.
3465 (*altivec_vupkhs<VU_char>_direct): New (copy of
3466 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3467 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3468 target is little endian and -maltivec=be is not specified.
3469 (*altivec_vupkls<VU_char>_direct): New (copy of
3470 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3471 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3472 little endian and -maltivec=be is not specified.
3473 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3474 little endian and -maltivec=be is not specified.
3476 2014-02-05 Richard Henderson <rth@redhat.com>
3479 * combine-stack-adj.c: Revert r206943.
3480 * sched-int.h (struct deps_desc): Add last_args_size.
3481 * sched-deps.c (init_deps): Initialize it.
3482 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3483 contain REG_ARGS_SIZE notes.
3485 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3487 * lto-cgraph.c (asm_nodes_output): Make global.
3488 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3489 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3490 (driver_handle_option): Handle OPT_fwpa.
3492 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3495 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3496 a comment typo and formatting issue. If odr_hash hasn't been
3497 created, return vNULL and set *completep to false.
3500 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3501 bb with no successors.
3503 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3506 * doc/invoke.texi (-march): Clarify documentation for ARM.
3510 2014-02-05 Richard Biener <rguenther@suse.de>
3512 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3513 when not vectorizing because of too many alias checks.
3514 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3515 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3517 2014-02-05 Nick Clifton <nickc@redhat.com>
3519 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3520 accept extended registers in any mode when compiling for the MN10300.
3522 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3524 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3525 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3526 sanitization attributes.
3527 (can_inline_edge_p): Likewise.
3528 (sanitize_attrs_match_for_inline_p): New function.
3530 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3532 * ipa-prop.c (detect_type_change): Shor circuit testing of
3533 type changes on THIS pointer.
3535 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3538 * config/pa/pa.c (legitimize_tls_address): Return original address
3539 if not passed a SYMBOL_REF rtx.
3540 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3542 (pa_emit_move_sequence): Simplify TLS source operands.
3543 (pa_legitimate_constant_p): Reject all TLS constants.
3544 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3545 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3547 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3549 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3550 groups when we know they are controlled by LTO.
3551 * varasm.c (default_binds_local_p_1): If object is in other partition,
3552 it will be resolved locally.
3554 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3556 * config/host-linux.c (linux_gt_pch_use_address): Don't
3557 use SSIZE_MAX because it is not always defined.
3559 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3562 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3563 threshold for pseudo splitting.
3564 (update_ebb_live_info): Process call argument hard registers and
3565 hard registers from insn definition too.
3566 (max_small_class_regs_num): New constant.
3567 (inherit_in_ebb): Update live hard regs through EBBs. Update
3568 reloads_num only for small register classes. Don't split for
3571 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3574 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3577 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3579 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3582 2014-02-04 Marek Polacek <polacek@redhat.com>
3584 * gdbinit.in (pel): Define.
3586 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3588 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3591 2014-02-04 Richard Biener <rguenther@suse.de>
3594 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3595 in function context local.
3596 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3597 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3598 similar to LTO_imported_decl_ref.
3600 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3602 PR tree-optimization/60002
3603 * cgraphclones.c (build_function_decl_skip_args): Clear
3606 PR tree-optimization/60023
3607 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3608 false to gsi_replace.
3609 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3610 has been in some EH region and vec_stmt could throw, add
3611 vec_stmt into the same EH region.
3612 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3613 has no lhs, ignore it.
3614 * internal-fn.c (expand_MASK_LOAD): Likewise.
3617 * tree-inline.c (copy_forbidden): Fail for
3618 __attribute__((optimize (0))) functions.
3621 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3622 is set, copy one less argument.
3623 (expand_simd_clones): Don't subtract clone_info->inbranch
3624 from simd_clone_struct_alloc argument.
3626 PR rtl-optimization/57915
3627 * recog.c (simplify_while_replacing): If all unary/binary/relational
3628 operation arguments are constant, attempt to simplify those.
3631 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3632 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3634 2014-02-04 Richard Biener <rguenther@suse.de>
3636 PR tree-optimization/60012
3637 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3638 TBAA disambiguation to all DDRs.
3640 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3643 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3644 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3646 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3649 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3651 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3653 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3654 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3656 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3659 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3660 to figure out targets of polymorphic calls with known decl.
3661 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3662 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3663 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3664 (get_polymorphic_call_info): ... here.
3665 (get_polymorphic_call_info_from_invariant): New function.
3667 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3669 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3670 lookup via vtable pointer; check for type consistency
3671 and turn inconsitent facts into UNREACHABLE.
3672 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3673 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3674 type inconsistent querries; return UNREACHABLE instead.
3676 2014-02-03 Richard Henderson <rth@twiddle.net>
3679 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3680 already processed this node.
3681 (normalize_one_pred_1): Pass along mark_set.
3682 (normalize_one_pred): Create and destroy a pointer_set_t.
3683 (normalize_one_pred_chain): Likewise.
3685 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3687 PR gcov-profile/58602
3688 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3690 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3693 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3694 -fno-devirtualize; try to devirtualize by the knowledge of
3695 virtual table pointer given by aggregate propagation.
3696 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3697 (ipa_print_node_jump_functions): Dump also offset that
3698 is relevant for polymorphic calls.
3699 (determine_known_aggregate_parts): Add arg_type parameter; use it
3700 instead of determining the type from pointer type.
3701 (ipa_compute_jump_functions_for_edge): Update call of
3702 determine_known_aggregate_parts.
3703 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3704 (gimple_get_virt_method_for_binfo): ... here; simplify using
3705 vtable_pointer_value_to_vtable.
3706 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3707 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3708 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3709 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3711 (vtable_pointer_value_to_binfo): ... here.
3712 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3714 2014-02-03 Teresa Johnson <tejohnson@google.com>
3716 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3717 redef of outer loop index variable.
3719 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3723 * doc/extend.texi (Function Attributes): Typo.
3725 2014-02-03 Cong Hou <congh@google.com>
3727 PR tree-optimization/60000
3728 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3729 if the vectorized statement is a store. A store statement can only
3730 appear at the end of pattern statements.
3732 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3734 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3735 (ix86_option_override_internal): Default long double to 64-bit for
3736 32-bit Bionic and to 128-bit for 64-bit Bionic.
3738 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3739 TARGET_LONG_DOUBLE_128 is true.
3740 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3742 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3743 (mlong-double-64): Negate -mlong-double-128.
3744 (mlong-double-128): New option.
3746 * config/i386/i386-c.c (ix86_target_macros): Define
3747 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3749 * doc/invoke.texi: Document -mlong-double-128.
3751 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3753 PR rtl-optimization/60024
3754 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3756 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3758 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3760 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3762 PR rtl-optimization/57662
3763 * sel-sched.c (code_motion_path_driver): Do not mark already not
3764 existing blocks in the visiting bitmap.
3766 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3768 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3769 on the insn being emitted.
3771 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3772 Will Deacon <will.deacon@arm.com>
3774 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3776 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3778 * config/arm/arm-tables.opt: Regenerate.
3780 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3782 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3783 for vector types other than V16QImode.
3784 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3785 define_expand, and call altivec_expand_vec_perm_le when producing
3786 code with little endian element order.
3787 (*altivec_vperm_<mode>_internal): New insn having previous
3788 behavior of altivec_vperm_<mode>.
3789 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3790 altivec_expand_vec_perm_le when producing code with little endian
3792 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3793 behavior of altivec_vperm_<mode>_uns.
3795 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3797 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3798 (altivec_vsumsws): Add handling for -maltivec=be with a little
3800 (altivec_vsumsws_direct): New.
3801 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3802 gen_altivec_vsumsws.
3804 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3806 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3807 vtable_pointer_value_to_binfo): New functions.
3808 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3809 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3811 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3813 * config/nios2/nios2.md (load_got_register): Initialize GOT
3814 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3815 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3817 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3819 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3820 preserverd by passthrough, do not propagate the type.
3822 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3824 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3825 (mips_atomic_assign_expand_fenv): New function.
3826 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3828 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3830 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3831 (__builtin_mips_set_fcsr): Likewise.
3832 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3833 MIPS_USI_FTYPE_VOID.
3834 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3835 (mips16_expand_set_fcsr): Likewise.
3836 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3837 (mips16_set_fcsr_stub): Likewise.
3838 (mips16_get_fcsr_one_only_stub): New class.
3839 (mips16_set_fcsr_one_only_stub): Likewise.
3840 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3841 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3842 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3843 (hard_float): New availability predicate.
3844 (mips_builtins): Add get_fcsr and set_fcsr.
3845 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3846 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3847 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3848 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3849 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3852 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3854 * config/mips/mips.c (mips_one_only_stub): New class.
3855 (mips_need_mips16_rdhwr_p): Replace with...
3856 (mips16_rdhwr_stub): ...this new variable.
3857 (mips16_stub_call_address): New function.
3858 (mips16_rdhwr_one_only_stub): New class.
3859 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3860 (mips_output_mips16_rdhwr): Delete.
3861 (mips_finish_stub): New function.
3862 (mips_code_end): Use it to handle rdhwr stubs.
3864 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3867 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3868 when calculating size of integer atomic types.
3870 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3872 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3874 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3876 PR tree-optimization/60003
3877 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3878 * profile.c (branch_prob): Use gimple_call_builtin_p
3879 to check for BUILT_IN_SETJMP_RECEIVER.
3880 * tree-inline.c (copy_bb): Call notice_special_calls.
3882 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3885 * lra-constraints.c (process_alt_operands): Update reload_sum only
3888 2014-01-31 Richard Henderson <rth@redhat.com>
3891 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3892 until after else_eh is processed.
3894 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3896 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3897 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3898 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3899 in smmintrin.h, remove them.
3900 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3901 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3902 * config/i386/i386.md (ROUND_SAE): Fix value.
3903 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3904 (const48_operand): New.
3905 * config/i386/subst.md (round), (round_expand): Use
3906 const_4_or_8_to_11_operand.
3907 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3909 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3911 * config/i386/constraints.md (Yk): Swap meaning with k.
3912 * config/i386/i386.md (movhi_internal): Change Yk to k.
3913 (movqi_internal): Ditto.
3914 (*k<logic><mode>): Ditto.
3917 (kandn<mode>): Ditto.
3918 (*<code>hi_1): Ditto.
3919 (*<code>qi_1): Ditto.
3920 (kxnor<mode>): Ditto.
3921 (kortestzhi): Ditto.
3922 (kortestchi): Ditto.
3924 (*one_cmplhi2_1): Ditto.
3925 (*one_cmplqi2_1): Ditto.
3926 * config/i386/sse.md (): Change k to Yk.
3927 (avx512f_load<mode>_mask): Ditto.
3928 (avx512f_blendm<mode>): Ditto.
3929 (avx512f_store<mode>_mask): Ditto.
3930 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3931 (avx512f_storedqu<mode>_mask): Ditto.
3932 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3934 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3935 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3936 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3937 (avx512f_maskcmp<mode>3): Ditto.
3938 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3939 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3940 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3941 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3942 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3943 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3944 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3945 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3946 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3947 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3948 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3949 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3950 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3951 (vec_extract_lo_<mode>_maskm): Ditto.
3952 (vec_extract_hi_<mode>_maskm): Ditto.
3953 (avx512f_vternlog<mode>_mask): Ditto.
3954 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3955 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3956 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3957 (avx512f_<code>v8div16qi2_mask): Ditto.
3958 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3959 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3960 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3961 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3962 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3963 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3964 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3965 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3966 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3967 (avx512cd_maskb_vec_dupv8di): Ditto.
3968 (avx512cd_maskw_vec_dupv16si): Ditto.
3969 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3970 (avx512f_vpermi2var<mode>3_mask): Ditto.
3971 (avx512f_vpermi2var<mode>3_mask): Ditto.
3972 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3973 (*avx512f_gathersi<mode>): Ditto.
3974 (*avx512f_gathersi<mode>_2): Ditto.
3975 (*avx512f_gatherdi<mode>): Ditto.
3976 (*avx512f_gatherdi<mode>_2): Ditto.
3977 (*avx512f_scattersi<mode>): Ditto.
3978 (*avx512f_scatterdi<mode>): Ditto.
3979 (avx512f_compress<mode>_mask): Ditto.
3980 (avx512f_compressstore<mode>_mask): Ditto.
3981 (avx512f_expand<mode>_mask): Ditto.
3982 * config/i386/subst.md (mask): Change k to Yk.
3983 (mask_scalar_merge): Ditto.
3986 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3988 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3990 2014-01-31 Richard Biener <rguenther@suse.de>
3993 * builtins.c (fold_builtin_memory_op): Make sure to not
3994 use a floating-point mode or a boolean or enumeral type for
3997 2014-01-30 DJ Delorie <dj@redhat.com>
3999 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
4000 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
4001 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
4002 whenever main() has an epilogue.
4004 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4006 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
4007 unused variable "field".
4008 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
4009 (vsx_mergeh_<mode>): Likewise.
4010 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
4011 (altivec_vmrghh): Likewise.
4012 (altivec_vmrghw): Likewise.
4013 (altivec_vmrglb): Likewise.
4014 (altivec_vmrglh): Likewise.
4015 (altivec_vmrglw): Likewise.
4016 (altivec_vspltb): Add missing uses.
4017 (altivec_vsplth): Likewise.
4018 (altivec_vspltw): Likewise.
4019 (altivec_vspltsf): Likewise.
4021 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4024 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
4025 frame related instructions.
4027 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
4029 PR rtl-optimization/59959
4030 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
4031 any reload of register whose subreg is invalid.
4033 2014-01-30 Jakub Jelinek <jakub@redhat.com>
4035 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
4036 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
4037 Add missing return type - void.
4039 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4041 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
4042 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
4043 remove element index adjustment for endian (now handled in vsx.md
4045 (altivec_expand_vec_perm_const): Use
4046 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
4047 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
4048 (vsx_xxspltw_<mode>): Adjust element index for little endian.
4049 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
4050 define_expand and a new define_insn *altivec_vspltb_internal;
4051 adjust for -maltivec=be on a little endian target.
4052 (altivec_vspltb_direct): New.
4053 (altivec_vsplth): Divide into a define_expand and a new
4054 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
4055 little endian target.
4056 (altivec_vsplth_direct): New.
4057 (altivec_vspltw): Divide into a define_expand and a new
4058 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
4059 little endian target.
4060 (altivec_vspltw_direct): New.
4061 (altivec_vspltsf): Divide into a define_expand and a new
4062 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
4063 a little endian target.
4065 2014-01-30 Richard Biener <rguenther@suse.de>
4067 PR tree-optimization/59993
4068 * tree-ssa-forwprop.c (associate_pointerplus): Check we
4069 can propagate form the earlier stmt and avoid the transform
4070 when the intermediate result is needed.
4072 2014-01-30 Alangi Derick <alangiderick@gmail.com>
4074 * README.Portability: Fix typo.
4076 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
4078 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
4079 comparison_operator with ordered_comparison_operator.
4081 2014-01-30 Nick Clifton <nickc@redhat.com>
4083 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4084 Rename to mn10300_store_multiple_regs.
4085 * config/mn10300/mn10300.c: Likewise.
4086 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4087 store_multiple_regs.
4088 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4089 Call mn10300_store_multiple_regs.
4091 2014-01-30 Nick Clifton <nickc@redhat.com>
4092 DJ Delorie <dj@redhat.com>
4094 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4095 %fp 2 to keep registers after it properly word-aligned.
4096 (rl78_alloc_physical_registers_umul): Handle the case where both
4097 input operands are the same.
4099 2014-01-30 Richard Biener <rguenther@suse.de>
4101 PR tree-optimization/59903
4102 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4105 2014-01-30 Jason Merrill <jason@redhat.com>
4108 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4111 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4113 2014-01-30 Richard Biener <rguenther@suse.de>
4115 PR tree-optimization/59951
4116 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4118 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4121 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4122 SFmode to DFmode case.
4124 2014-01-29 DJ Delorie <dj@redhat.com>
4126 * config/msp430/msp430.opt (-minrt): New.
4127 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4129 (ENDFILE_SPEC): Likewise.
4131 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4133 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4134 (estimate_function_body_sizes): Use it.
4136 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4139 * dwarf2out.c (is_cxx_auto): New.
4140 (is_base_type): Use it.
4141 (gen_type_die_with_usage): Likewise.
4143 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4145 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4146 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4147 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4148 -maltivec=be with LE targets.
4149 (vsx_mergeh_<mode>): Likewise.
4150 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4151 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4152 (altivec_vmrghb): Replace with define_expand and new
4153 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4154 (altivec_vmrghb_direct): New define_insn.
4155 (altivec_vmrghh): Replace with define_expand and new
4156 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4157 (altivec_vmrghh_direct): New define_insn.
4158 (altivec_vmrghw): Replace with define_expand and new
4159 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4160 (altivec_vmrghw_direct): New define_insn.
4161 (*altivec_vmrghsf): Adjust for endianness.
4162 (altivec_vmrglb): Replace with define_expand and new
4163 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4164 (altivec_vmrglb_direct): New define_insn.
4165 (altivec_vmrglh): Replace with define_expand and new
4166 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4167 (altivec_vmrglh_direct): New define_insn.
4168 (altivec_vmrglw): Replace with define_expand and new
4169 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4170 (altivec_vmrglw_direct): New define_insn.
4171 (*altivec_vmrglsf): Adjust for endianness.
4172 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4173 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4174 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4175 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4176 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4177 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4178 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4179 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4181 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4183 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4184 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4187 2014-01-29 Richard Biener <rguenther@suse.de>
4189 PR tree-optimization/58742
4190 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4191 associate_pointerplus_align.
4192 (associate_pointerplus_diff): New function.
4193 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4194 and associate_pointerplus_diff.
4196 2014-01-29 Richard Biener <rguenther@suse.de>
4198 * lto-streamer.h (LTO_major_version): Bump to 3.
4199 (LTO_minor_version): Reset to 0.
4201 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4203 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4204 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4205 (arm_file_start): Generate correct asm header for armv7ve.
4206 * config/arm/bpabi.h: Add multilib support for armv7ve.
4207 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4208 and cortex-a15 to armv7ve.
4209 * config/arm/t-aprofile: Add multilib support for armv7ve.
4210 * doc/invoke.texi: Document -march=armv7ve.
4212 2014-01-29 Richard Biener <rguenther@suse.de>
4214 PR tree-optimization/58742
4215 * tree-ssa-forwprop.c (associate_plusminus): Return true
4216 if we changed sth, defer EH cleanup to ...
4217 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4218 (simplify_mult): New function.
4220 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4223 PR tree-optimization/59920
4224 * tree.c (build_common_builtin_nodes): Remove
4225 __builtin_setjmp_dispatcher initialization.
4226 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4227 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4228 instead of gsi_after_labels + manually skipping debug stmts.
4229 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4230 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4231 * tree-inline.c (copy_edges_for_bb): Remove
4232 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4233 argument. Ignore computed_goto_p stmts. Don't call
4234 make_abnormal_goto_edges. If a call might need abnormal edges
4235 for non-local gotos, see if it already has an edge to
4236 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4237 with true argument, don't do anything then, otherwise add
4238 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4239 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4241 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4242 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4243 (lower_stmt): Don't set data->calls_builtin_setjmp.
4244 (lower_builtin_setjmp): Adjust comment.
4245 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4246 * tree-cfg.c (found_computed_goto): Remove.
4247 (factor_computed_gotos): Remove.
4248 (make_goto_expr_edges): Return bool, true for computed gotos.
4249 Don't call make_abnormal_goto_edges.
4250 (build_gimple_cfg): Don't set found_computed_goto, don't call
4251 factor_computed_gotos.
4252 (computed_goto_p): No longer static.
4253 (make_blocks): Don't set found_computed_goto.
4254 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4255 (make_edges): If make_goto_expr_edges returns true, push bb
4256 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4257 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4258 vector. Record mapping between bbs and OpenMP regions if there
4259 are any, adjust make_gimple_omp_edges caller. Call
4260 handle_abnormal_edges.
4261 (make_abnormal_goto_edges): Remove.
4262 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4263 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4264 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4265 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4266 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4267 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4268 filling *region also set *region_idx to (*region)->entry->index.
4271 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4272 For REGs set ORIGINAL_REGNO.
4274 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4276 * doc/md.texi: Mention that a target shouldn't implement
4277 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4280 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4282 PR tree-optimization/59594
4283 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4284 a copy of the datarefs vector rather than the vector itself.
4286 2014-01-28 Jason Merrill <jason@redhat.com>
4289 * dwarf2out.c (auto_die): New static.
4290 (gen_type_die_with_usage): Handle C++1y 'auto'.
4291 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4294 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4297 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4298 (SPEC_X32): Likewise.
4299 (SPEC_64): Likewise.
4300 * config/i386/i386.c (ix86_option_override_internal): Turn off
4301 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4303 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4304 * config/i386/i386.h (TARGET_16BIT): New macro.
4305 (TARGET_16BIT_P): Likewise.
4306 * config/i386/i386.opt: Add m16.
4307 * doc/invoke.texi: Document -m16.
4309 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4311 PR preprocessor/59935
4312 * input.c (location_get_source_line): Bail out on when line number
4313 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4315 2014-01-28 Richard Biener <rguenther@suse.de>
4317 PR tree-optimization/58742
4318 * tree-ssa-forwprop.c (associate_plusminus): Handle
4319 pointer subtraction of the form (T)(P + A) - (T)P.
4321 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4323 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4326 2014-01-28 Richard Biener <rguenther@suse.de>
4329 2014-01-28 Richard Biener <rguenther@suse.de>
4331 PR rtl-optimization/45364
4332 PR rtl-optimization/59890
4333 * var-tracking.c (local_get_addr_clear_given_value): Handle
4334 already cleared slot.
4335 (val_reset): Handle not allocated local_get_addr_cache.
4336 (vt_find_locations): Use post-order on the inverted CFG.
4338 2014-01-28 Richard Biener <rguenther@suse.de>
4340 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4342 2014-01-28 Richard Biener <rguenther@suse.de>
4344 PR rtl-optimization/45364
4345 PR rtl-optimization/59890
4346 * var-tracking.c (local_get_addr_clear_given_value): Handle
4347 already cleared slot.
4348 (val_reset): Handle not allocated local_get_addr_cache.
4349 (vt_find_locations): Use post-order on the inverted CFG.
4351 2014-01-28 Alan Modra <amodra@gmail.com>
4353 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4354 * configure.ac <recursive call for build != host>: Define
4355 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4356 and LD_FOR_BUILD too.
4357 * configure: Regenerate.
4359 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4361 * config/i386/i386.c (get_builtin_code_for_version): Separate
4362 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4363 Broadwell from Haswell.
4365 2014-01-27 Steve Ellcey <sellcey@mips.com>
4367 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4368 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4369 * config/mips/mips.c (mips_option_override): Change setting
4371 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4372 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4373 Change from Mask to Var.
4375 2014-01-27 Jeff Law <law@redhat.com>
4377 * ipa-inline.c (inline_small_functions): Fix typo.
4379 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4381 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4382 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4383 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4384 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4385 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4386 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4387 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4388 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4389 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4390 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4391 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4392 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4393 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4394 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4395 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4396 (_mm512_storeu_epi64): Ditto.
4397 (_mm512_cmpge_epi32_mask): Ditto.
4398 (_mm512_cmpge_epu32_mask): Ditto.
4399 (_mm512_cmpge_epi64_mask): Ditto.
4400 (_mm512_cmpge_epu64_mask): Ditto.
4401 (_mm512_cmple_epi32_mask): Ditto.
4402 (_mm512_cmple_epu32_mask): Ditto.
4403 (_mm512_cmple_epi64_mask): Ditto.
4404 (_mm512_cmple_epu64_mask): Ditto.
4405 (_mm512_cmplt_epi32_mask): Ditto.
4406 (_mm512_cmplt_epu32_mask): Ditto.
4407 (_mm512_cmplt_epi64_mask): Ditto.
4408 (_mm512_cmplt_epu64_mask): Ditto.
4409 (_mm512_cmpneq_epi32_mask): Ditto.
4410 (_mm512_cmpneq_epu32_mask): Ditto.
4411 (_mm512_cmpneq_epi64_mask): Ditto.
4412 (_mm512_cmpneq_epu64_mask): Ditto.
4413 (_mm512_expand_pd): Ditto.
4414 (_mm512_expand_ps): Ditto.
4415 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4416 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4417 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4418 * config/i386/i386.c (ix86_builtins): Add
4419 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4420 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4421 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4422 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4423 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4424 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4425 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4426 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4427 IX86_BUILTIN_PMOVUSQW512_MEM.
4428 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4429 __builtin_ia32_pmovsqd512mem_mask,
4430 __builtin_ia32_pmovqd512mem_mask,
4431 __builtin_ia32_pmovusqw512mem_mask,
4432 __builtin_ia32_pmovsqw512mem_mask,
4433 __builtin_ia32_pmovqw512mem_mask,
4434 __builtin_ia32_pmovusdw512mem_mask,
4435 __builtin_ia32_pmovsdw512mem_mask,
4436 __builtin_ia32_pmovdw512mem_mask,
4437 __builtin_ia32_pmovqb512mem_mask,
4438 __builtin_ia32_pmovusqb512mem_mask,
4439 __builtin_ia32_pmovsqb512mem_mask,
4440 __builtin_ia32_pmovusdb512mem_mask,
4441 __builtin_ia32_pmovsdb512mem_mask,
4442 __builtin_ia32_pmovdb512mem_mask.
4443 (bdesc_args): Add __builtin_ia32_expanddf512,
4444 __builtin_ia32_expandsf512.
4445 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4446 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4447 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4448 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4449 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4450 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4451 (avx512f_<code>v8div16qi2_mask_store): This.
4452 (avx512f_expand<mode>): New.
4454 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4456 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4458 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4459 (_mm512_prefetch_i32scatter_pd): Ditto.
4460 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4461 (_mm512_prefetch_i64scatter_pd): Ditto.
4462 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4463 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4464 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4465 (_mm512_prefetch_i32scatter_ps): Ditto.
4466 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4467 (_mm512_prefetch_i64scatter_ps): Ditto.
4468 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4469 * config/i386/i386-builtin-types.def: Define
4470 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4471 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4472 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4473 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4474 IX86_BUILTIN_SCATTERPFQPD.
4475 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4476 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4477 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4478 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4479 __builtin_ia32_scatterpfqps.
4480 (ix86_expand_builtin): Expand new built-ins.
4481 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4482 fix memory access data type.
4483 (*avx512pf_gatherpf<mode>_mask): Ditto.
4484 (*avx512pf_gatherpf<mode>): Ditto.
4485 (avx512pf_scatterpf<mode>): Ditto.
4486 (*avx512pf_scatterpf<mode>_mask): Ditto.
4487 (*avx512pf_scatterpf<mode>): Ditto.
4488 (GATHER_SCATTER_SF_MEM_MODE): New.
4489 (avx512pf_gatherpf<mode>df): Ditto.
4490 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4491 (*avx512pf_scatterpf<mode>df): Ditto.
4493 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4496 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4497 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4500 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4502 * common/config/arm/arm-common.c
4503 (arm_rewrite_mcpu): Handle multiple names.
4505 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4507 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4509 * gimple-builder.h (create_gimple_tmp): Delete.
4511 2014-01-27 Christian Bruel <christian.bruel@st.com>
4513 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4516 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4518 * config/pa/pa.md (call): Generate indirect long calls to non-local
4519 functions when outputing 32-bit code.
4520 (call_value): Likewise except for special call to buggy powf function.
4522 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4523 portable runtime and PIC indirect calls.
4524 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4525 and PIC call sequences. Use ldo instead of blr to set return register
4526 in PIC call sequence.
4528 2014-01-25 Walter Lee <walt@tilera.com>
4530 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4531 avoid clobbering a live register.
4533 2014-01-25 Walter Lee <walt@tilera.com>
4535 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4536 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4537 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4538 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4540 2014-01-25 Walter Lee <walt@tilera.com>
4542 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4543 arguments on even registers.
4544 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4546 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4547 (BIGGEST_ALIGNMENT): Ditto.
4548 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4550 2014-01-25 Walter Lee <walt@tilera.com>
4552 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4553 insns before bundling.
4554 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4556 2014-01-25 Walter Lee <walt@tilera.com>
4558 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4559 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4560 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4562 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4564 * config/mips/constraints.md (kl): Delete.
4565 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4566 define expands, using...
4567 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4568 instructions for MIPS16.
4569 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4570 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4572 2014-01-25 Walter Lee <walt@tilera.com>
4574 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4578 2014-01-25 Walter Lee <walt@tilera.com>
4580 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4581 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4583 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4585 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4588 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4590 * print-rtl.c (in_call_function_usage): New var.
4591 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4592 EXPR_LIST mode as mode and not as reg note name.
4595 * cfgloopmanip.c (copy_loop_info): If
4596 loop->warned_aggressive_loop_optimizations, make sure
4597 the flag is set in target loop too.
4599 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4601 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4603 * builtins.def: Likewise.
4604 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4605 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4606 * ira.c (ira_setup_eliminable_regset): Likewise.
4607 * omp-low.c (gate_expand_omp): Likewise.
4608 (execute_lower_omp): Likewise.
4609 (diagnose_sb_0): Likewise.
4610 (gate_diagnose_omp_blocks): Likewise.
4611 (simd_clone_clauses_extract): Likewise.
4614 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4616 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4617 correction for little endian...
4618 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4621 2014-01-24 Jeff Law <law@redhat.com>
4623 PR tree-optimization/59919
4624 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4625 for non-returning calls.
4627 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4629 * common/config/aarch64/aarch64-common.c
4630 (aarch64_rewrite_mcpu): Handle multiple names.
4631 * config/aarch64/aarch64.h
4632 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4634 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4636 * input.c (add_file_to_cache_tab): Handle the case where fopen
4639 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4642 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4643 from push operand if code of push isn't PRE_DEC.
4645 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4648 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4649 -mquad-memory-atomic. Update -mquad-memory documentation to say
4650 it is only used for non-atomic loads/stores.
4652 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4653 -mquad-memory or -mquad-memory-atomic switches.
4655 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4656 -mquad-memory-atomic to ISA 2.07 support.
4658 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4659 to separate support of normal quad word memory operations (ldq, stq)
4660 from the atomic quad word memory operations.
4662 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4663 support to separate non-atomic quad word operations from atomic
4664 quad word operations. Disable non-atomic quad word operations in
4665 little endian mode so that we don't have to swap words after the
4666 load and before the store.
4667 (quad_load_store_p): Add comment about atomic quad word support.
4668 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4669 options printed with -mdebug=reg.
4671 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4672 -mquad-memory-atomic as the test for whether we have quad word
4673 atomic instructions.
4674 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4675 or -mp8-vector are used, allow byte/half-word atomic operations.
4677 * config/rs6000/sync.md (load_lockedti): Insure that the address
4678 is a proper indexed or indirect address for the lqarx instruction.
4679 On little endian systems, swap the hi/lo registers after the lqarx
4681 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4682 insure the address is valid for the lqarx instruction.
4683 (store_conditionalti): Insure that the address is a proper indexed
4684 or indirect address for the stqcrx. instruction. On little endian
4685 systems, swap the hi/lo registers before doing the stqcrx.
4687 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4688 insure the address is valid for the stqcrx. instruction.
4690 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4691 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4692 type of quad memory support is available.
4694 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4697 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4698 there is a danger of looping.
4700 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4703 force flag_ira_loop_pressure if set via command line.
4705 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4707 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4708 (ashr_simd): New builtin handling DI mode.
4709 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4710 (aarch64_sshr_simddi): New match pattern.
4711 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4712 (vshrd_n_s64): Likewise.
4713 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4715 2014-01-23 Nick Clifton <nickc@redhat.com>
4717 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4718 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4719 favour of mcu specific scripts.
4720 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4723 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4724 Alex Velenko <Alex.Velenko@arm.com>
4726 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4727 (vaddv_s16): Likewise.
4728 (vaddv_s32): Likewise.
4729 (vaddv_u8): Likewise.
4730 (vaddv_u16): Likewise.
4731 (vaddv_u32): Likewise.
4732 (vaddvq_s8): Likewise.
4733 (vaddvq_s16): Likewise.
4734 (vaddvq_s32): Likewise.
4735 (vaddvq_s64): Likewise.
4736 (vaddvq_u8): Likewise.
4737 (vaddvq_u16): Likewise.
4738 (vaddvq_u32): Likewise.
4739 (vaddvq_u64): Likewise.
4740 (vaddv_f32): Likewise.
4741 (vaddvq_f32): Likewise.
4742 (vaddvq_f64): Likewise.
4743 (vmaxv_f32): Likewise.
4744 (vmaxv_s8): Likewise.
4745 (vmaxv_s16): Likewise.
4746 (vmaxv_s32): Likewise.
4747 (vmaxv_u8): Likewise.
4748 (vmaxv_u16): Likewise.
4749 (vmaxv_u32): Likewise.
4750 (vmaxvq_f32): Likewise.
4751 (vmaxvq_f64): Likewise.
4752 (vmaxvq_s8): Likewise.
4753 (vmaxvq_s16): Likewise.
4754 (vmaxvq_s32): Likewise.
4755 (vmaxvq_u8): Likewise.
4756 (vmaxvq_u16): Likewise.
4757 (vmaxvq_u32): Likewise.
4758 (vmaxnmv_f32): Likewise.
4759 (vmaxnmvq_f32): Likewise.
4760 (vmaxnmvq_f64): Likewise.
4761 (vminv_f32): Likewise.
4762 (vminv_s8): Likewise.
4763 (vminv_s16): Likewise.
4764 (vminv_s32): Likewise.
4765 (vminv_u8): Likewise.
4766 (vminv_u16): Likewise.
4767 (vminv_u32): Likewise.
4768 (vminvq_f32): Likewise.
4769 (vminvq_f64): Likewise.
4770 (vminvq_s8): Likewise.
4771 (vminvq_s16): Likewise.
4772 (vminvq_s32): Likewise.
4773 (vminvq_u8): Likewise.
4774 (vminvq_u16): Likewise.
4775 (vminvq_u32): Likewise.
4776 (vminnmv_f32): Likewise.
4777 (vminnmvq_f32): Likewise.
4778 (vminnmvq_f64): Likewise.
4780 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4782 * config/aarch64/aarch64-simd.md
4783 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4784 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4785 (*aarch64_mul3_elt<mode>): Likewise.
4786 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4787 (*aarch64_mul3_elt_to_64v2df): Likewise.
4788 (*aarch64_mla_elt<mode>): Likewise.
4789 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4790 (*aarch64_mls_elt<mode>): Likewise.
4791 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4792 (*aarch64_fma4_elt<mode>): Likewise.
4793 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4794 (*aarch64_fma4_elt_to_64v2df): Likewise.
4795 (*aarch64_fnma4_elt<mode>): Likewise.
4796 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4797 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4798 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4799 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4800 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4801 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4802 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4803 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4804 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4806 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4808 * config/aarch64/aarch64-simd.md
4809 (aarch64_be_checked_get_lane<mode>): New define_expand.
4810 * config/aarch64/aarch64-simd-builtins.def
4811 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4812 New builtin definition.
4813 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4814 Use new safe be builtin.
4816 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4818 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4820 (aarch64_be_st1<mode>): Likewise.
4821 (aarch_ld1<VALL:mode>): Define_expand modified.
4822 (aarch_st1<VALL:mode>): Likewise.
4823 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4824 (UNSPEC_ST1): Likewise.
4826 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4828 * config/microblaze/microblaze.md: Add trap insn and attribute
4830 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4832 PR preprocessor/58580
4833 * input.h (location_get_source_line): Take an additional line_size
4835 (void diagnostics_file_cache_fini): Declare new function.
4836 * input.c (struct fcache): New type.
4837 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4838 New static constants.
4839 (diagnostic_file_cache_init, total_lines_num)
4840 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4841 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4842 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4843 (get_next_line, read_next_line, goto_next_line, read_line_num):
4844 New static function definitions.
4845 (diagnostic_file_cache_fini): New function.
4846 (location_get_source_line): Take an additional output line_len
4847 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4849 * diagnostic.c (diagnostic_finish): Call
4850 diagnostic_file_cache_fini.
4851 (adjust_line): Take an additional input parameter for the length
4852 of the line, rather than calculating it with strlen.
4853 (diagnostic_show_locus): Adjust the use of
4854 location_get_source_line and adjust_line with respect to their new
4855 signature. While displaying a line now, do not stop at the first
4856 null byte. Rather, display the zero byte as a space and keep
4857 going until we reach the size of the line.
4858 * Makefile.in: Add vec.o to OBJS-libcommon
4860 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4861 Ilya Tocar <ilya.tocar@intel.com>
4863 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4864 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4865 (__builtin_ia32_kmov16): Ditto.
4866 * config/i386/i386.md (UNSPEC_KMOV): New.
4869 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4871 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4872 (_mm512_storeu_si512): Ditto.
4874 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4877 * rtl.h (get_referenced_operands): Declare.
4878 * recog.c (get_referenced_operands): New function.
4879 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4880 operands have been referenced when recording LO_SUM references.
4882 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4884 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4886 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4888 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4889 Enable for generic and recent AMD targets.
4891 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4893 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4894 ARG_SIZE note when adjustment was eliminated.
4896 2014-01-22 Jeff Law <law@redhat.com>
4898 PR tree-optimization/59597
4899 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4900 in file. Accept new argument REGISTERING and use it to modify
4901 dump output appropriately.
4902 (register_jump_thread): Corresponding changes.
4903 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4904 thread paths involving joiner blocks. Add code to dump cancelled
4905 jump threading paths.
4907 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4909 PR rtl-optimization/59477
4910 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4911 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4913 2014-01-22 Tom Tromey <tromey@redhat.com>
4915 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4917 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4919 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4921 PR rtl-optimization/59896
4922 * lra-constraints.c (process_alt_operands): Check unused note for
4923 matched operands of insn with no output reloads.
4925 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4927 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4928 (mips_move_from_gpr_cost): Likewise.
4930 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4932 PR rtl-optimization/59858
4933 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4934 ira_class_hard_regs_num.
4935 (process_alt_operands): Increase reject for dying matched operand.
4937 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4940 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4941 smaller than size, perform several stores or loads and stores
4942 at dst + count - size to store or copy all of size bytes, rather
4943 than just last modesize bytes.
4945 2014-01-20 DJ Delorie <dj@redhat.com>
4947 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4948 that CLOBBERs are REGs before propogating their values.
4950 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4953 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4954 (cgraph_inline_failed_type): New function.
4955 * cgraph.h (DEFCIFCODE): Add type.
4956 (cgraph_inline_failed_type_t): New enum.
4957 (cgraph_inline_failed_type): New prototype.
4958 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4959 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4960 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4961 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4962 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4963 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4964 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4965 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4966 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4967 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4968 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4969 OPTIMIZATION_MISMATCH.
4970 * tree-inline.c (expand_call_inline): Emit errors during
4971 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4973 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4976 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4977 mode attribute in insn output.
4979 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4981 * output.h (output_constant): Delete.
4982 * varasm.c (output_constant): Make private.
4984 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4986 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4988 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4991 * tree.h (fold_builtin_strcat): New prototype.
4992 * builtins.c (fold_builtin_strcat): No longer static. Add len
4993 argument, if non-NULL, don't call c_strlen. Optimize
4994 directly into __builtin_memcpy instead of __builtin_strcpy.
4995 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4996 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4998 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
5000 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5001 for SImode_address_operand operands, having only a REG argument.
5003 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
5005 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
5006 loader name using mbig-endian.
5007 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
5009 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5011 * doc/invoke.texi (-march): Clarify documentation for AArch64.
5015 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
5017 * config/aarch64/aarch64-protos.h
5018 (aarch64_cannot_change_mode_class_ptr): Declare.
5019 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
5020 aarch64_cannot_change_mode_class_ptr): New.
5021 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
5022 backend hook aarch64_cannot_change_mode_class.
5024 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
5026 * common/config/aarch64/aarch64-common.c
5027 (aarch64_handle_option): Don't handle any option order logic here.
5028 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
5029 selected_cpu, warn on architecture version mismatch.
5030 (aarch64_override_options): Fix parsing order for option strings.
5032 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5033 Iain Sandoe <iain@codesourcery.com>
5036 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
5037 warning. Amend comment to reflect current functionality.
5039 2014-01-20 Richard Biener <rguenther@suse.de>
5042 * builtins.c (fold_builtin_strcat): Remove case better handled
5043 by tree-ssa-strlen.c.
5045 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
5047 * config/aarch64/aarch64.opt
5048 (mcpu, march, mtune): Make case-insensitive.
5050 2014-01-20 Jakub Jelinek <jakub@redhat.com>
5053 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
5054 if operands[1] is a REG or ZERO_EXTEND of a REG.
5056 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
5058 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
5060 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
5062 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
5063 long non-pic millicode calls.
5065 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
5067 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
5069 2014-01-19 Kito Cheng <kito@0xlab.org>
5071 * builtins.c (expand_movstr): Check movstr expand done or fail.
5073 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5074 H.J. Lu <hongjiu.lu@intel.com>
5077 * config/i386/i386.md (*lea<mode>): Zero-extend return register
5078 to DImode for zero-extended addresses.
5080 2014-01-19 Jakub Jelinek <jakub@redhat.com>
5082 PR rtl-optimization/57763
5083 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5084 on the new indirect jump_insn and increment LABEL_NUSES (label).
5086 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5090 * config.gcc (x86_archs): New variable.
5091 (x86_64_archs): Likewise.
5092 (x86_cpus): Likewise.
5093 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5094 --with-arch/--with-cpu= options.
5095 Support --with-arch=/--with-cpu={nehalem,westmere,
5096 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5098 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5100 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5101 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5103 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5105 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5107 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5110 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5111 clear cpp_get_options (parse_in)->warn_unused_macros for
5112 ix86_target_macros_internal with cpp_define.
5114 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5116 * jump.c (delete_related_insns): Keep (use (insn))s.
5117 * reorg.c (redundant_insn): Check for barriers too.
5119 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5121 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5123 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5125 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5126 call to $$dyncall when TARGET_LONG_CALLS is true.
5128 2014-01-17 Jeff Law <law@redhat.com>
5130 * ree.c (combine_set_extension): Temporarily disable test for
5131 changing number of hard registers.
5133 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5136 * ipa-inline-analysis.c (inline_free_summary):
5137 Do not free summary of aliases.
5139 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5142 * gimplify.c (gimplify_expr): Use create_tmp_var
5143 instead of create_tmp_var_raw. If cond doesn't have
5144 integral type, don't add the IFN_ANNOTATE builtin at all.
5146 2014-01-17 Martin Jambor <mjambor@suse.cz>
5149 * ipa-cp.c (prev_edge_clone): New variable.
5150 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5151 Also resize prev_edge_clone vector.
5152 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5153 (ipcp_edge_removal_hook): New function.
5154 (ipcp_driver): Register ipcp_edge_removal_hook.
5156 2014-01-17 Andrew Pinski <apinski@cavium.com>
5157 Steve Ellcey <sellcey@mips.com>
5160 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5163 2014-01-17 Jeff Law <law@redhat.com>
5166 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5167 so that pass_ccp runs first.
5169 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5171 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5172 (ix86_adjust_cost): Use !TARGET_XXX.
5173 (do_reorder_for_imul): Likewise.
5174 (swap_top_of_ready_list): Likewise.
5175 (ix86_sched_reorder): Likewise.
5177 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5179 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5180 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5181 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5182 (intel_memset): New. Duplicate slm_memset.
5183 (intel_cost): New. Duplicate slm_cost.
5184 (m_INTEL): New macro.
5185 (processor_target_table): Add "intel".
5186 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5187 with PROCESSOR_INTEL for "intel".
5188 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5189 PROCESSOR_SILVERMONT.
5190 (ix86_issue_rate): Likewise.
5191 (ix86_adjust_cost): Likewise.
5192 (ia32_multipass_dfa_lookahead): Likewise.
5193 (swap_top_of_ready_list): Likewise.
5194 (ix86_sched_reorder): Likewise.
5195 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5196 instead of TARGET_OPT_AGU.
5197 * config/i386/i386.h (TARGET_INTEL): New.
5198 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5199 (processor_type): Add PROCESSOR_INTEL.
5200 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5201 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5203 2014-01-17 Marek Polacek <polacek@redhat.com>
5206 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5209 2014-01-17 Richard Biener <rguenther@suse.de>
5211 PR tree-optimization/46590
5212 * opts.c (default_options_table): Add entries for
5213 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5214 all enabled at -O1 but not for -Og.
5215 * common.opt (fbranch-count-reg): Remove Init(1).
5216 (fmove-loop-invariants): Likewise.
5217 (ftree-pta): Likewise.
5219 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5221 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5222 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5223 decls to at least the GCC 4.8 used alignments.
5226 * tree-nested.c (convert_nonlocal_reference_stmt,
5227 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5228 of GIMPLE_BIND stmts, adjust associated decls.
5230 2014-01-17 Richard Biener <rguenther@suse.de>
5232 PR tree-optimization/46590
5233 * vec.h (vec<>::bseach): New member function implementing
5234 binary search according to C89 bsearch.
5235 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5236 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5237 bitmap pointer again. Make accesses_in_loop a flat array.
5238 (mem_ref_obstack): New global.
5239 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5240 (mark_ref_stored): Likewise.
5241 (ref_indep_loop_p_2): Likewise.
5242 (set_ref_stored_in_loop): New helper function.
5243 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5244 (memref_free): Adjust.
5245 (record_mem_ref_loc): Simplify.
5246 (gather_mem_refs_stmt): Adjust.
5247 (sort_locs_in_loop_postorder_cmp): New function.
5248 (analyze_memory_references): Sort accesses_in_loop after
5249 loop postorder number.
5250 (find_ref_loc_in_loop_cmp): New function.
5251 (for_all_locs_in_loop): Find relevant cluster of locs in
5252 accesses_in_loop and iterate without recursion.
5253 (execute_sm): Avoid uninit warning.
5254 (struct ref_always_accessed): Simplify.
5255 (ref_always_accessed::operator ()): Likewise.
5256 (ref_always_accessed_p): Likewise.
5257 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5258 loop postorder numbers here.
5259 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5262 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5265 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5266 on decls for which assemble_alias has been called.
5268 2014-01-17 Nick Clifton <nickc@redhat.com>
5270 * config/msp430/msp430.opt: (mcpu): New option.
5271 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5272 (msp430_option_override): Parse target_cpu. If the MCU name
5273 matches a generic string, clear target_mcu.
5274 (msp430_attr): Allow numeric interrupt values up to 63.
5275 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5276 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5278 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5280 * config/msp430/msp430.md (popm): Use %J rather than %I.
5281 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5282 (addhi_cy_i): Use immediate_operand for operand 2.
5283 * doc/invoke.texi: Document -mcpu option.
5285 2014-01-17 Richard Biener <rguenther@suse.de>
5287 PR rtl-optimization/38518
5288 * df.h (df_analyze_loop): Declare.
5289 * df-core.c: Include cfgloop.h.
5290 (df_analyze_1): Split out main part of df_analyze.
5291 (df_analyze): Adjust.
5292 (loop_inverted_post_order_compute): New function.
5293 (loop_post_order_compute): Likewise.
5294 (df_analyze_loop): New function avoiding whole-function
5296 * loop-invariant.c (find_defs): Use df_analyze_loop.
5297 (find_invariants): Adjust.
5298 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5300 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5302 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5303 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5305 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5307 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5308 traversal when removing references.
5310 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5313 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5315 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5318 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5319 pushing a reload for an autoinc when we had previously reloaded an
5320 inner part of the address.
5322 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5324 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5326 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5327 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5328 when not giving up or versioning for alias only because of
5330 (vect_analyze_data_ref_dependences): Set to true.
5331 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5333 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5334 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5338 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5341 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5342 operand 0 predicate for gathers, use a new pseudo as subtarget.
5344 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5347 * lra-constraints.c (process_alt_operands): Add printing debug info.
5348 Check absence of input/output reloads for matched operands too.
5350 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5352 PR rtl-optimization/59835
5353 * ira.c (ira_init_register_move_cost): Increase cost for
5356 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5358 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5360 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5363 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5364 non-register objects. Use gen_(high/low)part more consistently.
5367 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5370 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5371 endian support, remove tests for WORDS_BIG_ENDIAN.
5372 (p8_mfvsrd_3_<mode>): Likewise.
5373 (reload_gpr_from_vsx<mode>): Likewise.
5374 (reload_gpr_from_vsxsf): Likewise.
5375 (p8_mfvsrd_4_disf): Likewise.
5377 2014-01-16 Richard Biener <rguenther@suse.de>
5379 PR rtl-optimization/46590
5380 * lcm.c (compute_antinout_edge): Use postorder iteration.
5381 (compute_laterin): Use inverted postorder iteration.
5383 2014-01-16 Nick Clifton <nickc@redhat.com>
5386 * varasm.c (output_constant): Return the number of bytes actually
5388 (output_constructor_array_range): Update the field size with the
5389 number of bytes emitted by output_constant.
5390 (output_constructor_regular_field): Likewise. Also do not
5391 complain if the total number of bytes emitted is now greater
5392 than the expected fieldpos.
5393 * output.h (output_constant): Update prototype and descriptive comment.
5395 2014-01-16 Marek Polacek <polacek@redhat.com>
5398 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5399 it is error_mark_node.
5401 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5403 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5404 VALID_AVX256_REG_OR_OI_MODE.
5406 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5408 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5409 current procedure should be profiled.
5411 2014-01-15 Andrew Pinski <apinski@cavium.com>
5413 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5414 of moving from/to the STACK_REG register class.
5416 2014-01-15 Richard Henderson <rth@redhat.com>
5419 * reginfo.c (global_regs_decl): Globalize.
5420 * rtl.h (global_regs_decl): Declare.
5421 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5422 reserved via global_regs.
5424 2014-01-15 Teresa Johnson <tejohnson@google.com>
5426 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5428 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5430 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5431 and vmulosh rather than call gen_vec_widen_smult_*.
5432 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5433 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5434 (vec_widen_smult_even_v16qi): Likewise.
5435 (vec_widen_umult_even_v8hi): Likewise.
5436 (vec_widen_smult_even_v8hi): Likewise.
5437 (vec_widen_umult_odd_v16qi): Likewise.
5438 (vec_widen_smult_odd_v16qi): Likewise.
5439 (vec_widen_umult_odd_v8hi): Likewise.
5440 (vec_widen_smult_odd_v8hi): Likewise.
5441 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5442 vmuloub rather than call gen_vec_widen_umult_*.
5443 (vec_widen_umult_lo_v16qi): Likewise.
5444 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5445 vmulosb rather than call gen_vec_widen_smult_*.
5446 (vec_widen_smult_lo_v16qi): Likewise.
5447 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5448 rather than call gen_vec_widen_umult_*.
5449 (vec_widen_umult_lo_v8hi): Likewise.
5450 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5451 rather than call gen_vec_widen_smult_*.
5452 (vec_widen_smult_lo_v8hi): Likewise.
5454 2014-01-15 Jeff Law <law@redhat.com>
5456 PR tree-optimization/59747
5457 * ree.c (find_and_remove_re): Properly handle case where a second
5458 eliminated extension requires widening a copy created for elimination
5459 of a prior extension.
5460 (combine_set_extension): Ensure that the number of hard regs needed
5461 for a destination register does not change when we widen it.
5463 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5465 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5466 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5467 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5468 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5469 (avr-*-rtems*): Likewise.
5470 (bfin*-rtems*): Likewise.
5471 (moxie-*-rtems*): Likewise.
5472 (h8300-*-rtems*): Likewise.
5473 (i[34567]86-*-rtems*): Likewise.
5474 (lm32-*-rtems*): Likewise.
5475 (m32r-*-rtems*): Likewise.
5476 (m68k-*-rtems*): Likewise.
5477 (microblaze*-*-rtems*): Likewise.
5478 (mips*-*-rtems*): Likewise.
5479 (powerpc-*-rtems*): Likewise.
5480 (sh-*-rtems*): Likewise.
5481 (sparc-*-rtems*): Likewise.
5482 (sparc64-*-rtems*): Likewise.
5483 (v850-*-rtems*): Likewise.
5484 (m32c-*-rtems*): Likewise.
5486 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5488 PR rtl-optimization/59511
5489 * ira.c (ira_init_register_move_cost): Use memory costs for some
5490 cases of register move cost calculations.
5491 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5492 instead of BB frequency.
5493 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5494 * lra-assigns.c (find_hard_regno_for): Ditto.
5496 2014-01-15 Richard Biener <rguenther@suse.de>
5498 PR tree-optimization/59822
5499 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5500 (vectorizable_load): Use it to hoist defs of uses of invariant
5501 loads out of the loop.
5503 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5504 Kugan Vivekanandarajah <kuganv@linaro.org>
5507 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5510 2014-01-15 Richard Biener <rguenther@suse.de>
5512 PR rtl-optimization/59802
5513 * lcm.c (compute_available): Use inverted postorder to seed
5514 the initial worklist.
5516 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5519 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5520 ADDR_REGS for invalid symrefs in non-PIC code.
5522 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5525 * builtins.c (determine_block_size): Initialize *probable_max_size
5526 even if len_rtx is CONST_INT.
5528 2014-01-14 Andrew Pinski <apinski@cavium.com>
5530 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5531 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5532 (cortexa53_tunings): Likewise.
5533 (aarch64_sched_issue_rate): New function.
5534 (TARGET_SCHED_ISSUE_RATE): Define.
5536 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5538 * ira-costs.c (find_costs_and_classes): Add missed
5539 ira_init_register_move_cost_if_necessary.
5541 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5544 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5546 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5549 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5550 to indicate if type is used for function return value. Warn ABI
5551 change if the vector mode isn't available for function return value.
5552 (ix86_function_arg_advance): Pass false to type_natural_mode.
5553 (ix86_function_arg): Likewise.
5554 (ix86_gimplify_va_arg): Likewise.
5555 (function_arg_32): Don't warn ABI change.
5556 (ix86_function_value): Pass true to type_natural_mode.
5557 (ix86_return_in_memory): Likewise.
5558 (ix86_struct_value_rtx): Removed.
5559 (TARGET_STRUCT_VALUE_RTX): Likewise.
5561 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5563 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5564 converting a conditional jump into a conditional return.
5566 2014-01-14 Richard Biener <rguenther@suse.de>
5568 PR tree-optimization/58921
5569 PR tree-optimization/59006
5570 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5571 hoisting invariant stmts.
5572 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5573 invariant loads on the preheader edge if possible.
5575 2014-01-14 Joey Ye <joey.ye@arm.com>
5577 * doc/plugin.texi (Building GCC plugins): Update to C++.
5579 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5581 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5582 (_mm_rcp28_round_ss): Ditto.
5583 (_mm_rsqrt28_round_sd): Ditto.
5584 (_mm_rsqrt28_round_ss): Ditto.
5585 (_mm_rcp28_sd): Ditto.
5586 (_mm_rcp28_ss): Ditto.
5587 (_mm_rsqrt28_sd): Ditto.
5588 (_mm_rsqrt28_ss): Ditto.
5589 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5590 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5591 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5592 (IX86_BUILTIN_RCP28SD): Ditto.
5593 (IX86_BUILTIN_RCP28SS): Ditto.
5594 (IX86_BUILTIN_RSQRT28SD): Ditto.
5595 (IX86_BUILTIN_RSQRT28SS): Ditto.
5596 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5597 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5598 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5599 (ix86_expand_special_args_builtin): Expand new FTYPE.
5600 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5601 (srcp14<mode>): Make insn unary.
5602 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5603 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5604 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5605 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5606 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5607 Fix rounding: make it SAE only.
5608 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5610 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5612 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5613 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5614 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5615 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5616 (round_saeonly_mask_scalar_operand4): Ditto.
5617 (round_saeonly_mask_scalar_op3): Ditto.
5618 (round_saeonly_mask_scalar_op4): Ditto.
5620 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5622 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5623 Implement -maltivec=be for vec_insert and vec_extract.
5625 2014-01-10 DJ Delorie <dj@redhat.com>
5627 * config/msp430/msp430.md (call_internal): Don't allow memory
5628 references with SP as the base register.
5629 (call_value_internal): Likewise.
5630 * config/msp430/constraints.md (Yc): New. For memory references
5631 that don't use SP as a base register.
5633 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5634 "an integer without a # prefix"
5635 * config/msp430/msp430.md (epilogue_helper): Use it.
5637 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5640 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5641 AVX512F gather builtins.
5642 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5643 on gather decls with INTEGER_TYPE masktype.
5644 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5645 directly into the builtin rather than hoisting it before loop.
5647 PR tree-optimization/59387
5648 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5649 (scev_const_prop): If folded_casts and type has undefined overflow,
5650 use force_gimple_operand instead of force_gimple_operand_gsi and
5651 for each added stmt if it is assign with
5652 arith_code_with_undefined_signed_overflow, call
5653 rewrite_to_defined_overflow.
5654 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5655 gimple-fold.h instead.
5656 (arith_code_with_undefined_signed_overflow,
5657 rewrite_to_defined_overflow): Moved to ...
5658 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5659 rewrite_to_defined_overflow): ... here. No longer static.
5660 Include gimplify-me.h.
5661 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5662 rewrite_to_defined_overflow): New prototypes.
5664 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5666 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5668 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5670 * builtins.c (get_object_alignment_2): Minor tweak.
5671 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5673 2014-01-13 Christian Bruel <christian.bruel@st.com>
5675 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5676 optimized non constant lengths.
5678 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5681 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5682 load as __atomic_load_N if possible.
5684 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5686 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5688 (rs6000_expand_builtin): Adjust call.
5690 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5693 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5694 * config/rs6000/rs6000.c: Include target-globals.h.
5695 (rs6000_set_current_function): Instead of doing target_reinit
5696 unconditionally, use save_target_globals_default_opts and
5697 restore_target_globals.
5699 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5701 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5702 (rs6000_expand_builtin): Handle mffs and mtfsf.
5703 (rs6000_init_builtins): Define mffs and mtfsf.
5704 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5705 (rs6000_mffs): New pattern.
5706 (rs6000_mtfsf): New pattern.
5708 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5710 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5711 Start narrowing with START. Apply candidate-use pair
5712 and check overall cost in narrowing.
5713 (iv_ca_prune): Pass new argument.
5715 2014-01-10 Jeff Law <law@redhat.com>
5718 * ree.c (combine_reaching_defs): Ensure the defining statement
5719 occurs before the extension when optimizing extensions with
5720 different source and destination hard registers.
5722 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5725 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5726 vtables into the type inheritance graph.
5728 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5730 PR rtl-optimization/59754
5731 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5732 modes in the REGNO != REGNO case.
5734 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5736 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5738 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5740 PR tree-optimization/59745
5741 * tree-predcom.c (tree_predictive_commoning_loop): Call
5742 free_affine_expand_cache if giving up because components is NULL.
5744 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5745 GC in payload of target_globals struct instead of allocating them on
5746 the heap and the larger structs separately using GC.
5747 * target-globals.h (struct target_globals): Make regs, hard_regs,
5748 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5749 of GTY((skip)) and change type to void *.
5750 (reset_target_globals): Cast loads from those fields to corresponding
5753 2014-01-10 Steve Ellcey <sellcey@mips.com>
5756 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5757 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5758 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5760 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5763 * aarch64-modes.def (CC_Zmode): New flags mode.
5764 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5765 represents an equality.
5766 (aarch64_get_condition_code): Handle CC_Zmode.
5767 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5769 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5771 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5772 extraction in good case.
5774 2014-01-10 Richard Biener <rguenther@suse.de>
5776 PR tree-optimization/59374
5777 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5778 checking after SLP discovery. Mark stmts not participating
5779 in any SLP instance properly.
5781 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5783 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5784 when handling a SET rtx.
5786 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5788 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5789 (cortex-a57): Likewise.
5790 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5792 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5794 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5795 non-iwmmxt builtins.
5797 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5801 * ipa-devirt.c record_target_from_binfo): Take as argument
5802 stack of binfos and lookup matching one for virtual inheritance.
5803 (possible_polymorphic_call_targets_1): Update.
5805 2014-01-10 Huacai Chen <chenhc@lemote.com>
5807 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5808 kernel strings for Loongson-2E/2F/3A.
5810 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5813 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5814 is_gimple_call before calling gimple_call_internal_p.
5816 2014-01-09 Steve Ellcey <sellcey@mips.com>
5818 * Makefile.in (TREE_FLOW_H): Remove.
5819 (TREE_SSA_H): Add file names from tree-flow.h.
5820 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5821 * tree.h: Remove tree-flow.h reference.
5822 * hash-table.h: Remove tree-flow.h reference.
5823 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5824 reference with tree-ssa-loop.h.
5826 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5828 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5829 default element-order behavior for -maltivec.
5830 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5831 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5832 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5833 when targeting big endian, at least for now.
5834 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5836 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5839 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5840 var satisfies use_register_for_decl, just take into account type
5841 alignment, rather than decl alignment.
5843 PR tree-optimization/59622
5844 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5845 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5846 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5847 Don't devirtualize for inplace at all. For targets.length () == 1,
5848 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5850 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5852 * config/i386/i386.md (cpu): Remove the unused btver1.
5854 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5856 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5858 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5861 * tree-core.h (struct target_globals): New forward declaration.
5862 (struct tree_target_option): Add globals field.
5863 * tree.h (TREE_TARGET_GLOBALS): Define.
5864 (prepare_target_option_nodes_for_pch): New prototype.
5865 * target-globals.h (struct target_globals): Define even if
5867 * tree.c (prepare_target_option_node_for_pch,
5868 prepare_target_option_nodes_for_pch): New functions.
5869 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5870 * config/i386/i386.c: Include target-globals.h.
5871 (ix86_set_current_function): Instead of doing target_reinit
5872 unconditionally, use save_target_globals_default_opts and
5873 restore_target_globals.
5875 2014-01-09 Richard Biener <rguenther@suse.de>
5877 PR tree-optimization/59715
5878 * tree-cfg.h (split_critical_edges): Declare.
5879 * tree-cfg.c (split_critical_edges): Export.
5880 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5882 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5884 * cfgexpand.c (expand_stack_vars): Optionally disable
5885 asan stack protection.
5886 (expand_used_vars): Likewise.
5887 (partition_stack_vars): Likewise.
5888 * asan.c (asan_emit_stack_protection): Optionally disable
5889 after return stack usage.
5890 (instrument_derefs): Optionally disable memory access instrumentation.
5891 (instrument_builtin_call): Likewise.
5892 (instrument_strlen_call): Likewise.
5893 (asan_protect_global): Optionally disable global variables protection.
5894 * doc/invoke.texi: Added doc for new options.
5895 * params.def: Added new options.
5896 * params.h: Likewise.
5898 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5900 PR rtl-optimization/59724
5901 * ifcvt.c (cond_exec_process_if_block): Don't call
5902 flow_find_head_matching_sequence with 0 longest_match.
5903 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5904 non-active insns if !stop_after.
5905 (try_head_merge_bb): Revert 2014-01-07 changes.
5907 2014-01-08 Jeff Law <law@redhat.com>
5909 * ree.c (get_sub_rtx): New function, extracted from...
5910 (merge_def_and_ext): Here.
5911 (combine_reaching_defs): Use get_sub_rtx.
5913 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5915 * cgraph.h (varpool_variable_node): Do not choke on null node.
5917 2014-01-08 Catherine Moore <clm@codesourcery.com>
5919 * config/mips/mips.md (simple_return): Attempt to use JRC
5921 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5923 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5925 PR rtl-optimization/59137
5926 * reorg.c (steal_delay_list_from_target): Call update_block for
5928 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5930 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5932 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5933 two duplicate entries.
5935 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5938 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5940 * config/mips/mips.c (mips_truncated_op_cost): New function.
5941 (mips_rtx_costs): Adjust test for BADDU.
5942 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5944 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5946 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5947 (*baddu_si): ...this new pattern.
5949 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5952 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5954 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5957 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5959 (expand_expr, expand_normal): Adjust.
5960 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5961 inner_reference_p. Use inner_reference_p to expand inner references.
5962 (store_expr): Adjust.
5963 * cfgexpand.c (expand_call_stmt): Adjust.
5965 2014-01-08 Rong Xu <xur@google.com>
5967 * gcov-io.c (gcov_var): Move from gcov-io.h.
5968 (gcov_position): Ditto.
5969 (gcov_is_error): Ditto.
5970 (gcov_rewrite): Ditto.
5971 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5972 only part to libgcc/libgcov.h.
5974 2014-01-08 Marek Polacek <polacek@redhat.com>
5977 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5979 2014-01-08 Marek Polacek <polacek@redhat.com>
5982 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5984 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5986 PR rtl-optimization/59649
5987 * stor-layout.c (get_mode_bounds): For BImode return
5988 0 and STORE_FLAG_VALUE.
5990 2014-01-08 Richard Biener <rguenther@suse.de>
5993 * gimple.h (is_gimple_builtin_call): Remove.
5994 (gimple_builtin_call_types_compatible_p): New.
5995 (gimple_call_builtin_p): New overload.
5996 * gimple.c (is_gimple_builtin_call): Remove.
5997 (validate_call): Rename to ...
5998 (gimple_builtin_call_types_compatible_p): ... this and export. Also
6000 (validate_type): New static function.
6001 (gimple_call_builtin_p): New overload and adjust.
6002 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
6003 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
6004 (gimple_fold_stmt_to_constant_1): Likewise.
6005 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
6007 2014-01-08 Richard Biener <rguenther@suse.de>
6010 * gimplify.c (gimplify_expr): Gimplify register-register type
6011 VIEW_CONVERT_EXPRs to separate stmts.
6013 2014-01-07 Jeff Law <law@redhat.com>
6016 * ree.c (combine_set_extension): Handle case where source
6017 and destination registers in an extension insn are different.
6018 (combine_reaching_defs): Allow source and destination registers
6019 in extension to be different under limited circumstances.
6020 (add_removable_extension): Remove restriction that the
6021 source and destination registers in the extension are the same.
6022 (find_and_remove_re): Emit a copy from the extension's
6023 destination to its source after the defining insn if
6024 the source and destination registers are different.
6027 * ifcvt.c (merge_if_block): If we are merging a block with more than
6028 one successor with a block with no successors, remove any BARRIER
6029 after the second block.
6031 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
6033 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
6035 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
6038 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
6039 for 14-bit register offsets when INT14_OK_STRICT is false.
6041 2014-01-07 Roland Stigge <stigge@antcom.de>
6042 Michael Meissner <meissner@linux.vnet.ibm.com>
6045 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
6046 Only check TFmode for SPE constants. Don't check TImode or TDmode.
6048 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
6050 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
6053 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
6055 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
6056 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
6057 rtx is const0_rtx or not.
6059 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
6062 * target-globals.c (save_target_globals): Remove this_fn_optab
6064 * toplev.c: Include optabs.h.
6065 (target_reinit): Temporarily restore the global options if another
6066 set of options are in force.
6068 2014-01-07 Jakub Jelinek <jakub@redhat.com>
6070 PR rtl-optimization/58668
6071 * cfgcleanup.c (flow_find_cross_jump): Don't count
6072 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
6073 to determine what is counted.
6074 (flow_find_head_matching_sequence): Use active_insn_p to determine
6076 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
6078 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
6079 determine what is counted.
6081 PR tree-optimization/59643
6082 * tree-predcom.c (split_data_refs_to_components): If one dr is
6083 read and one write, determine_offset fails and the write isn't
6084 in the bad component, just put the read into the bad component.
6086 2014-01-07 Mike Stump <mikestump@comcast.net>
6087 Jakub Jelinek <jakub@redhat.com>
6090 * tree-core.h (struct tree_optimization_option): Change optabs
6091 type from unsigned char * to void *.
6092 * optabs.c (init_tree_optimization_optabs): Adjust
6093 TREE_OPTIMIZATION_OPTABS initialization.
6095 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6098 * config/i386/i386.h (struct machine_function): Add
6099 no_drap_save_restore field.
6100 * config/i386/i386.c (ix86_save_reg): Use
6101 !cfun->machine->no_drap_save_restore instead of
6102 crtl->stack_realign_needed.
6103 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6104 this function clears frame_pointer_needed. Set
6105 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6106 and DRAP reg is needed.
6108 2014-01-06 Marek Polacek <polacek@redhat.com>
6111 * doc/implement-c.texi: Mention that other integer types are
6112 permitted as bit-field types in strictly conforming mode.
6114 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6116 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6119 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6121 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6123 2014-01-06 Martin Jambor <mjambor@suse.cz>
6126 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6128 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6130 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6134 * var-tracking.c (add_stores): Preserve the value of the source even if
6135 we don't record the store.
6137 2014-01-06 Terry Guo <terry.guo@arm.com>
6139 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6141 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6144 * config/darwin.c (darwin_function_section): Adjust return values to
6145 correspond to optimisation changes made in r206070.
6147 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6149 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6150 from prefetch_block tune setting.
6151 (nocona_cost): Correct size of prefetch block to 64.
6153 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6155 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6156 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6157 used to save the static chain register in the computation of the offset
6158 from which the FP registers need to be restored.
6160 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6162 PR tree-optimization/59519
6163 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6164 ICE if get_current_def (current_new_name) is already non-NULL, as long
6165 as it is a phi result of some other phi in *new_exit_bb that has
6168 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6169 or vmovdqu* for misaligned_operand.
6170 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6171 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6172 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6173 aligned_mem for AVX512F masked aligned load and store builtins and for
6176 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6178 PR tree-optimization/59651
6179 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6180 Address range for negative step should be added by TYPE_SIZE_UNIT.
6182 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6184 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6185 overlapping registers also for registers other than the stack pointer.
6187 2014-01-03 Marek Polacek <polacek@redhat.com>
6190 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6193 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6196 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6199 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6200 (push splitter): Use <P:MODE_SIZE> instead of
6201 GET_MODE_SIZE (<P:MODE>mode).
6202 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6203 (mov -1, reg peephole2): Likewise.
6204 * config/i386/sse.md (*mov<mode>_internal,
6205 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6206 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6207 *<code><mode>3, *andnot<mode>3<mask_name>,
6208 <mask_codefor><code><mode>3<mask_name>): Likewise.
6209 * config/i386/subst.md (mask_mode512bit_condition,
6210 sd_mask_mode512bit_condition): Likewise.
6212 2014-01-02 Xinliang David Li <davidxl@google.com>
6214 PR tree-optimization/59303
6215 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6216 (dump_predicates): Better output format.
6217 (pred_equal_p): New function.
6218 (is_neq_relop_p): Ditto.
6219 (is_neq_zero_form_p): Ditto.
6220 (pred_expr_equal_p): Ditto.
6221 (pred_neg_p): Ditto.
6222 (simplify_pred): Ditto.
6223 (simplify_preds_2): Ditto.
6224 (simplify_preds_3): Ditto.
6225 (simplify_preds_4): Ditto.
6226 (simplify_preds): Ditto.
6228 (push_to_worklist): Ditto.
6229 (get_pred_info_from_cmp): Ditto.
6230 (is_degenerated_phi): Ditto.
6231 (normalize_one_pred_1): Ditto.
6232 (normalize_one_pred): Ditto.
6233 (normalize_one_pred_chain): Ditto.
6234 (normalize_preds): Ditto.
6235 (normalize_cond_1): Remove function.
6236 (normalize_cond): Ditto.
6237 (is_gcond_subset_of): Ditto.
6238 (is_subset_of_any): Ditto.
6239 (is_or_set_subset_of): Ditto.
6240 (is_and_set_subset_of): Ditto.
6241 (is_norm_cond_subset_of): Ditto.
6242 (pred_chain_length_cmp): Ditto.
6243 (convert_control_dep_chain_into_preds): Type change.
6244 (find_predicates): Ditto.
6245 (find_def_preds): Ditto.
6246 (destroy_predicates_vecs): Ditto.
6247 (find_matching_predicates_in_rest_chains): Ditto.
6248 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6249 (is_pred_expr_subset): Ditto.
6250 (is_pred_chain_subset_of): Ditto.
6251 (is_included_in): Ditto.
6252 (is_superset_of): Ditto.
6254 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6256 Update copyright years.
6258 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6260 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6261 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6262 config/arc/arc.md, config/arc/arc.opt,
6263 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6264 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6265 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6266 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6267 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6268 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6269 vtable-verify.c, vtable-verify.h: Use the standard form for the
6272 2014-01-02 Tobias Burnus <burnus@net-b.de>
6274 * gcc.c (process_command): Update copyright notice dates.
6275 * gcov-dump.c: Ditto.
6277 * doc/cpp.texi: Bump @copying's copyright year.
6278 * doc/cppinternals.texi: Ditto.
6279 * doc/gcc.texi: Ditto.
6280 * doc/gccint.texi: Ditto.
6281 * doc/gcov.texi: Ditto.
6282 * doc/install.texi: Ditto.
6283 * doc/invoke.texi: Ditto.
6285 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6287 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6289 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6291 * config/i386/sse.md (*mov<mode>_internal): Guard
6292 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6294 PR rtl-optimization/59647
6295 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6296 new_rtx into UNSIGNED_FLOAT rtxes.
6298 Copyright (C) 2014 Free Software Foundation, Inc.
6300 Copying and distribution of this file, with or without modification,
6301 are permitted in any medium without royalty provided the copyright
6302 notice and this notice are preserved.