1 2014-04-11 Steve Ellcey <sellcey@mips.com>
2 Jakub Jelinek <jakub@redhat.com>
5 * expr.c (convert_move): Use emit_store_flag_force instead of
6 emit_store_flag. Pass lowpart_mode instead of VOIDmode as 5th
9 2014-04-11 Richard Biener <rguenther@suse.de>
12 * varasm.c (assemble_alias): Avoid endless error reporting
13 recursion by setting TREE_ASM_WRITTEN.
15 2014-04-11 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
17 * config/s390/s390.md: Add a splitter for NOT rtx.
19 2014-04-11 Jakub Jelinek <jakub@redhat.com>
21 PR rtl-optimization/60663
22 * cse.c (cse_insn): Set src_volatile on ASM_OPERANDS in
25 2014-04-10 Jan Hubicka <hubicka@ucw.cz>
26 Jakub Jelinek <jakub@redhat.com>
29 * ipa.c (function_and_variable_visibility): Copy forced_by_abi flag from
32 2014-04-10 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
35 * config/arm/arm.c (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Define
36 (arm_const_not_ok_for_debug_p): Reject MINUS with SYM_REF's
37 ameliorating the cases where it can be.
39 2014-04-09 David Edelsohn <dje.gcc@gmail.com>
42 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
44 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
45 (loadsync_<mode>): Change mode.
46 (load_quadpti, store_quadpti): New.
47 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
48 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
49 * config/rs6000/predicates.md (quad_memory_operand): !TARGET_SYNC_TI.
51 2014-04-09 Cong Hou <congh@google.com>
54 * doc/sourcebuild.texi (vect_widen_mult_si_to_di_pattern): Add
57 2014-04-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
59 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Use vnand
60 instead of vnor to exploit possible fusion opportunity in the
62 (altivec_expand_vec_perm_const_le): Likewise.
64 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
66 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
67 (loadsync_<mode>): Change mode.
68 (load_quadpti, store_quadpti): New.
69 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
70 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
72 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
75 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
76 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
77 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
79 2014-04-08 Richard Biener <rguenther@suse.de>
82 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
83 a 64bit widest int print double-int similar to on HWI64 hosts.
85 2014-04-08 Richard Biener <rguenther@suse.de>
87 PR tree-optimization/60785
88 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
89 default defs properly.
91 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
93 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
96 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
98 * ipa-devirt.c (maybe_record_node): When node is not recorded,
99 set completep to false rather than true.
101 2014-04-07 Douglas B Rupp <rupp@adacore.com>
104 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
105 ARM_TARGET2_DWARF_FORMAT.
107 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
110 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
111 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
114 2014-04-07 Richard Biener <rguenther@suse.de>
116 PR tree-optimization/60766
117 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
118 (may_eliminate_iv): Convert cand_value_at result to desired type.
120 2014-04-07 Jason Merrill <jason@redhat.com>
123 * common.opt (-fno-gnu-unique): Add.
124 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
126 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
128 * haifa-sched.c: Fix outdated function reference and minor
129 grammar errors in introductory comment.
131 2014-04-07 Richard Biener <rguenther@suse.de>
134 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
136 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
138 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
141 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
142 size accounting for thunks.
143 (pa_asm_output_mi_thunk): Use final_start_function() and
144 final_end_function() to output function start and end directives.
146 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
148 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
149 specific ISA/ feature information. Remove short_sp and errata_skip ds.
150 Add avr_device_specific_features enum to have device specific info.
151 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
152 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
153 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
154 updated device specific info.
155 * config/avr/avr-mcus.def: Merge device specific details to
157 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
159 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
160 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
161 assembler if RMW isa supported by current device.
162 * config/avr/genmultilib.awk: Update as device info structure changed.
163 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
165 2014-04-04 Cong Hou <congh@google.com>
167 PR tree-optimization/60656
168 * tree-vect-stmts.c (supportable_widening_operation):
169 Fix a bug that elements in a vector with vect_used_by_reduction
170 property are incorrectly reordered when the operation on it is not
171 consistant with the one in reduction operation.
173 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
175 PR rtl-optimization/60155
176 * gcse.c (record_set_data): New function.
177 (single_set_gcse): New function.
178 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
179 (hoist_code): Likewise.
180 (get_pressure_class_and_nregs): Likewise.
182 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
184 * explow.c (probe_stack_range): Emit a final optimization blockage.
186 2014-04-04 Anthony Green <green@moxielogic.com>
188 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
191 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
194 * lto-cgraph.c (input_overwrite_node): Check that partitioning
195 flags are set only during streaming.
196 * ipa.c (process_references, walk_polymorphic_call_targets,
197 symtab_remove_unreachable_nodes): Drop bodies of always inline
198 after early inlining.
199 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
201 2014-04-04 Jakub Jelinek <jakub@redhat.com>
202 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
205 * dwarf2out.c (const_ok_for_output_1): Reject expressions
208 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
211 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
213 (cortex_a53_fdivd): Likewise.
215 2014-04-04 Martin Jambor <mjambor@suse.cz>
218 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
220 * cgraph.c (clone_of_p): Also return true if thunks match.
221 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
222 cgraph_function_or_thunk_node and an obsolete comment.
223 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
225 (build_function_decl_skip_args): Likewise.
226 (set_new_clone_decl_and_node_flags): New function.
227 (duplicate_thunk_for_node): Likewise.
228 (redirect_edge_duplicating_thunks): Likewise.
229 (cgraph_clone_node): New parameter args_to_skip, pass it to
230 redirect_edge_duplicating_thunks which is called instead of
231 cgraph_redirect_edge_callee.
232 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
233 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
235 2014-04-04 Jeff Law <law@redhat.com>
238 * config/arm/predicates.md (const_int_I_operand): New predicate.
239 (const_int_M_operand): Similarly.
240 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
242 (insv_t2, extv_reg, extzv_t2): Likewise.
243 (load_multiple_with_writeback): Similarly for const_int_I_operand.
244 (pop_multiple_with_writeback_and_return): Likewise.
245 (vfp_pop_multiple_with_writeback): Likewise
247 2014-04-04 Richard Biener <rguenther@suse.de>
250 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
251 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
253 * gimplify.h (gimple_add_tmp_var_fn): Declare.
254 * gimplify.c (gimple_add_tmp_var_fn): New function.
255 * gimple-expr.h (create_tmp_reg_fn): Declare.
256 * gimple-expr.c (create_tmp_reg_fn): New function.
257 * gimple-low.c (record_vars_into): Don't change cfun.
258 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
259 code generation without cfun.
261 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
264 * Makefile.in (install-driver): Fix shell scripting.
266 2014-04-03 Cong Hou <congh@google.com>
268 PR tree-optimization/60505
269 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
270 threshold of number of iterations below which no vectorization will be
272 * tree-vect-loop.c (new_loop_vec_info):
273 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
274 * tree-vect-loop.c (vect_analyze_loop_operations):
275 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
276 * tree-vect-loop.c (vect_transform_loop):
277 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
278 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
279 of iterations of the loop and see if we should build the epilogue.
281 2014-04-03 Richard Biener <rguenther@suse.de>
283 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
285 (streamer_tree_cache_create): Adjust.
286 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
287 to allow optional nodes array.
288 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
289 (streamer_tree_cache_append): Likewise.
290 (streamer_tree_cache_create): Create nodes array optionally
291 as specified by parameter.
292 * lto-streamer-out.c (create_output_block): Avoid maintaining
293 the node array in the writer cache.
294 (DFS_write_tree): Remove assertion.
295 (produce_asm_for_decls): Free the out decl state hash table
297 * lto-streamer-in.c (lto_data_in_create): Adjust for
298 streamer_tree_cache_create prototype change.
300 2014-04-03 Richard Biener <rguenther@suse.de>
302 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
303 set TREE_CHAIN to NULL_TREE.
305 2014-04-03 Richard Biener <rguenther@suse.de>
307 PR tree-optimization/60740
308 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
309 over all GIMPLE_COND operands.
311 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
313 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
314 (Weffc++): Remove Scott's numbering, merge lists and reference
320 * class.c (accessible_nvdtor_p): New.
321 (check_bases): Don't check base destructor here ...
322 (check_bases_and_members): ... check them here. Trigger on
323 Wnon-virtual-dtor flag.
324 (finish_struct_1): Use accessible_nvdtor_p.
327 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
328 * g++.dg/warn/Wnvdtor-2.C: New.
329 * g++.dg/warn/Wnvdtor-3.C: New.
330 * g++.dg/warn/Wnvdtor-4.C: New.
331 * g++.dg/warn/Weff1.C: Delete.
332 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
333 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
335 2014-04-03 Nick Clifton <nickc@redhat.com>
337 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
340 2014-04-03 Martin Jambor <mjambor@suse.cz>
342 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
343 mention gcc_unreachable before failing.
344 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
347 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
350 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
351 code and instead mark the context inconsistent.
352 (possible_polymorphic_call_targets): For inconsistent contexts
353 return empty complete list.
355 2014-04-02 Anthony Green <green@moxielogic.com>
357 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
358 (extendqisi2, extendhisi2): Define.
359 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
360 (WCHAR_TYPE): Change to unsigned int.
362 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
364 PR tree-optimization/60733
365 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
366 insertion point for PHI candidates to be the end of the feeding
367 block for the PHI argument.
369 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
371 PR rtl-optimization/60650
372 * lra-constraints.c (process_alt_operands): Decrease reject for
373 earlyclobber matching.
375 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
377 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
379 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
381 * config/spu/spu.c (pad_bb): Do not crash when the last
382 insn is CODE_FOR_blockage.
384 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
386 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
387 lies outside the target mode.
389 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
392 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
393 software floating point or no floating point registers, do not
394 allow any type in the FPRs. Eliminate a test for SPE SIMD types
395 in GPRs that occurs after we tested for GPRs that would never be
398 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
399 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
400 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
401 specifically allow DDmode, since that does not use the SPE SIMD
404 2014-04-02 Richard Biener <rguenther@suse.de>
407 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
408 MODE_INTs. Properly use negv_optab.
409 (expand_abs): Likewise.
411 2014-04-02 Richard Biener <rguenther@suse.de>
414 * Makefile.in (install-driver): Guard extra installs with special
417 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
419 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
422 2014-04-01 Richard Henderson <rth@redhat.com>
425 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
426 alternative enabled before register allocation.
428 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
430 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
431 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
433 (nios2_large_got_address): Remove unneeded 'sym' parameter.
434 (nios2_got_address): Update nios2_large_got_address call site.
435 (nios2_delegitimize_address): New function.
436 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
437 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
438 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
440 2014-04-01 Martin Husemann <martin@duskware.de>
442 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
445 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
447 PR rtl-optimization/60604
448 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
449 check from register_operand.
450 (register_operand): Redefine in terms of general_operand.
451 (nonmemory_operand): Use register_operand for the non-constant cases.
453 2014-04-01 Richard Biener <rguenther@suse.de>
455 * gimple.h (struct gimple_statement_base): Align subcode to
458 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
460 * doc/invoke.texi (mapp-regs): Clarify.
462 2014-03-31 Ulrich Drepper <drepper@gmail.com>
464 * config/i386/avx512fintrin.h (__v32hi): Define type.
466 (_mm512_set1_epi8): Define.
467 (_mm512_set1_epi16): Define.
468 (_mm512_set4_epi32): Define.
469 (_mm512_set4_epi64): Define.
470 (_mm512_set4_pd): Define.
471 (_mm512_set4_ps): Define.
472 (_mm512_setr4_epi64): Define.
473 (_mm512_setr4_epi32): Define.
474 (_mm512_setr4_pd): Define.
475 (_mm512_setr4_ps): Define.
476 (_mm512_setzero_epi32): Define.
478 2014-03-31 Martin Jambor <mjambor@suse.cz>
481 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
482 callsite_arguments_match_p. Updated all callers. Also check types of
483 corresponding formal parameters and actual arguments.
484 (not_all_callers_have_enough_arguments_p) Renamed to
485 some_callers_have_mismatched_arguments_p.
487 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
489 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
491 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
494 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
497 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
499 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
500 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
502 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
503 Use FMAMODE_NOVF512 mode iterator.
504 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
505 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
506 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
507 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
509 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
510 Use VF_128_256 mode iterator.
511 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
514 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
516 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
517 static chain if needed.
519 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
522 * lra-constraints.c (index_part_to_reg): New.
523 (process_address): Use it.
525 2014-03-27 Jeff Law <law@redhat.com>
526 Jakub Jelinek <jakub@redhat.com>
529 * expr.c (do_tablejump): Use simplify_gen_binary rather than
530 gen_rtx_{PLUS,MULT} to build up the address expression.
532 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
533 creating non-canonical RTL.
535 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
538 * ipa-inline.c (want_inline_small_function_p): Short circuit large
539 functions; reorganize to make cheap checks first.
540 (inline_small_functions): Do not estimate growth when dumping;
542 * ipa-inline.h (inline_summary): Add min_size.
543 (growth_likely_positive): New function.
544 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
545 (set_cond_stmt_execution_predicate): Cleanup.
546 (estimate_edge_size_and_time): Compute min_size.
547 (estimate_calls_size_and_time): Likewise.
548 (estimate_node_size_and_time): Likewise.
549 (inline_update_overall_summary): Update min_size.
550 (do_estimate_edge_time): Likewise.
551 (do_estimate_edge_size): Update.
552 (do_estimate_edge_hints): Update.
553 (growth_likely_positive): New function.
555 2014-03-28 Jakub Jelinek <jakub@redhat.com>
558 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
559 also if addr has VOIDmode.
561 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
563 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
564 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
566 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
567 instructions as well as AdvancedSIMD loads.
569 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
571 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
572 Use crypto_aese type.
573 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
574 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
575 crypto_aese, crypto_aesmc. Move to types.md.
576 * config/arm/types.md (crypto_aes): Split into crypto_aese,
578 * config/arm/iterators.md (crypto_type): Likewise.
580 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
582 * cgraph.c: Include expr.h and tree-dfa.h.
583 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
586 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
589 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
590 regs from checking multi-reg pseudos.
592 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
594 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
596 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
598 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
599 if it would clobber the stack pointer, even temporarily.
601 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
603 * mode-switching.c: Make small adjustments to the top comment.
605 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
607 * config/rs6000/constraints.md (wD constraint): New constraint to
608 match the constant integer to get the top DImode/DFmode out of a
609 vector in a VSX register.
611 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
612 match the constant integer to get the top DImode/DFmode out of a
613 vector in a VSX register.
615 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
618 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
621 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
622 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
624 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
625 Optimize vec_extract of 64-bit values, where the value being
626 extracted is in the top word, where we can use scalar
627 instructions. Add direct move and store support. Combine the big
628 endian/little endian vector select load support into a single insn.
629 (vsx_extract_<mode>_internal1): Likewise.
630 (vsx_extract_<mode>_internal2): Likewise.
631 (vsx_extract_<mode>_load): Likewise.
632 (vsx_extract_<mode>_store): Likewise.
633 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
634 combined into vsx_extract_<mode>_load.
635 (vsx_extract_<mode>_one_le): Likewise.
637 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
638 define the top 64-bit vector element.
640 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
643 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
644 Document vec_vbpermq builtin.
647 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
648 enable use of xxsldwi and xxpermdi builtin functions.
649 (vec_xxpermdi): Likewise.
651 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
652 Document use of vec_xxsldwi and vec_xxpermdi builtins.
654 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
656 PR rtl-optimization/60650
657 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
659 (find_spills_for): New.
660 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
661 Spill all pseudos on the second iteration.
663 2014-03-27 Marek Polacek <polacek@redhat.com>
666 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
669 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
671 * config/s390/s390.c (s390_can_use_return_insn): Check for
672 call-saved FPRs on 31 bit.
674 2014-03-27 Jakub Jelinek <jakub@redhat.com>
677 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
678 if they need regimplification, just drop them instead of
679 calling gimple_regimplify_operands on them.
681 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
684 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
685 (aarch64_frame_pointer_required): Adjust logic.
686 (aarch64_can_eliminate): Adjust logic.
687 (aarch64_override_options_after_change): Adjust logic.
689 2014-03-27 Dehao Chen <dehao@google.com>
691 * ipa-inline.c (early_inliner): Update node's inline info.
693 2014-03-26 Dehao Chen <dehao@google.com>
695 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
696 compiler inserted conditional jumps for NAN float check.
698 2014-03-26 Jakub Jelinek <jakub@redhat.com>
700 * ubsan.h (ubsan_create_data): Change second argument's type
701 to const location_t *.
702 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
704 (ubsan_create_data): Change second argument to const location_t *PLOC.
705 Create Loc field whenever PLOC is non-NULL.
706 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
707 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
711 * real.c (real_to_integer2): Change type of low to UHWI.
713 2014-03-26 Tobias Burnus <burnus@net-b.de>
715 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
716 (CILK_SELF_SPECS): New define.
717 (driver_self_specs): Use it.
719 2014-03-26 Richard Biener <rguenther@suse.de>
721 * tree-pretty-print.c (percent_K_format): Implement special
722 case for LTO and its stripped down BLOCK tree.
724 2014-03-26 Jakub Jelinek <jakub@redhat.com>
727 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
729 * tree-vrp.c (simplify_internal_call_using_ranges): If only
730 one range is range_int_cst_p, but not both, at least optimize
731 addition/subtraction of 0 and multiplication by 0 or 1.
732 * gimple-fold.c (gimple_fold_call): Fold
733 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
734 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
735 INTEGER_CSTs, try to fold at least x * 0 and y - y.
737 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
739 PR rtl-optimization/60452
740 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
741 <case REG>: Return 1 for invalid offsets from the frame pointer.
743 2014-03-26 Marek Polacek <polacek@redhat.com>
746 * doc/extend.texi (C Extensions): Mention variable-length arrays in
749 2014-03-26 Marek Polacek <polacek@redhat.com>
752 * doc/extend.texi (Designated Inits): Describe what happens to omitted
755 2014-03-26 Marek Polacek <polacek@redhat.com>
758 * ira-color.c (update_conflict_hard_regno_costs): Perform the
759 multiplication in unsigned type.
761 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
763 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
765 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
767 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
769 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
772 * cif-code.def (UNREACHABLE) New code.
773 * ipa-inline.c (inline_small_functions): Skip edges to
774 __builtlin_unreachable.
775 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
776 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
777 predicate to __bulitin_unreachable.
778 (set_cond_stmt_execution_predicate): Fix issue when
779 invert_tree_comparison returns ERROR_MARK.
780 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
781 propagate to inline clones.
782 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
784 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
785 * cgraphclones.c (cgraph_clone_node): If call destination is already
786 ureachable, do not redirect it back.
787 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
790 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
792 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
793 Do not modify inline clones.
795 2014-03-25 Jakub Jelinek <jakub@redhat.com>
797 * config/i386/i386.md (general_sext_operand): New mode attr.
798 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
799 don't generate (sign_extend (const_int)).
800 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
801 operands[2]. Use We constraint instead of <i> and
802 <general_sext_operand> predicate instead of <general_operand>.
803 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
804 * config/i386/constraints.md (We): New constraint.
805 * config/i386/predicates.md (x86_64_sext_operand,
806 sext_operand): New predicates.
808 2014-03-25 Martin Jambor <mjambor@suse.cz>
811 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
812 inconsistent devirtualizations to __builtin_unreachable.
814 2014-03-25 Marek Polacek <polacek@redhat.com>
817 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
819 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
821 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
822 order of elements for big-endian.
824 2014-03-25 Richard Biener <rguenther@suse.de>
827 * gimplify-me.c (gimple_regimplify_operands): Update the
830 2014-03-25 Martin Jambor <mjambor@suse.cz>
833 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
834 (lto_output_varpool_node): Likewise.
835 (input_overwrite_node): Likewise.
836 (input_varpool_node): Likewise.
838 2014-03-25 Richard Biener <rguenther@suse.de>
840 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
843 2014-03-25 Jakub Jelinek <jakub@redhat.com>
845 * combine.c (simplify_compare_const): Add MODE argument.
846 Handle mode_width 0 as very large mode_width.
847 (try_combine, simplify_comparison): Adjust callers.
849 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
850 type to avoid signed integer overflow.
851 * explow.c (plus_constant): Likewise.
853 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
855 * doc/generic.texi: Correct typos.
857 2014-03-24 Tobias Burnus <burnus@net-b.de>
859 * doc/invoke.texi (-flto): Expand section about
860 using static libraries with LTO.
862 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
864 PR rtl-optimization/60501
865 * optabs.def (addptr3_optab): New optab.
866 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
867 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
868 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
870 * lra.c (emit_add3_insn): Use the addptr pattern if available.
872 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
874 2014-03-24 Ulrich Drepper <drepper@gmail.com>
876 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
879 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
880 (_mm256_undefined_ps): Define.
881 (_mm256_undefined_pd): Define.
882 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
883 (_mm_undefined_pd): Define.
884 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
885 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
886 (_mm512_undefined_ps): Define.
887 (_mm512_undefined_pd): Define.
888 Use _mm*_undefined_*.
889 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
891 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
893 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
894 (lshr_simd): DI mode added.
895 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
896 (aarch64_ushr_simddi): Likewise.
897 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
898 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
899 (vshrd_n_u64): Likewise.
901 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
903 * Makefile.in (s-macro_list): Depend on cc1.
905 2014-03-23 Teresa Johnson <tejohnson@google.com>
907 * ipa-utils.c (ipa_print_order): Use specified dump file.
909 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
911 PR rtl-optimization/60601
912 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
914 * gcc.c (eval_spec_function): Initialize save_growing_value.
916 2014-03-22 Jakub Jelinek <jakub@redhat.com>
919 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
920 code == MINUS_EXPR, never swap op0 with op1.
922 * toplev.c (init_local_tick): Avoid signed integer multiplication
924 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
925 shift by first operand's bitsize.
927 2014-03-21 Jakub Jelinek <jakub@redhat.com>
930 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
932 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
933 TARGET_ISA_64BIT_P(x).
935 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
937 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
938 pattern for vector nor instead of subtract from splat(-1).
939 (altivec_expand_vec_perm_const_le): Likewise.
941 2014-03-21 Richard Henderson <rth@twiddle.net>
944 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
945 related insns after epilogue_completed.
947 2014-03-21 Martin Jambor <mjambor@suse.cz>
950 * cgraph.h (symtab_node): New flag body_removed.
951 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
952 when removing bodies.
953 * symtab.c (dump_symtab_base): Dump body_removed flag.
954 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
955 had their bodies removed.
957 2014-03-21 Martin Jambor <mjambor@suse.cz>
960 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
963 2014-03-21 Richard Biener <rguenther@suse.de>
965 PR tree-optimization/60577
966 * tree-core.h (struct tree_base): Document nothrow_flag use
967 in VAR_DECL_NONALIASED.
968 * tree.h (VAR_DECL_NONALIASED): New.
969 (may_be_aliased): Adjust.
970 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
972 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
974 * expr.c (expand_expr_real_1): Remove outdated comment.
976 2014-03-20 Jakub Jelinek <jakub@redhat.com>
979 * ira.c (adjust_cleared_regs): Call copy_rtx on
980 *reg_equiv[REGNO (loc)].src_p before passing it to
981 simplify_replace_fn_rtx.
984 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
985 into CONST, put pic register as first operand of PLUS. Use
986 gen_const_mem for both 32-bit and 64-bit PIC got loads.
988 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
990 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
992 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
994 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
995 around for store forwarding issue in the FPU on the UT699.
996 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
997 loads and operations if -mfix-ut699 is specified.
998 (divtf3_hq): Tweak attribute.
999 (sqrttf2_hq): Likewise.
1001 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
1003 * calls.c (store_one_arg): Remove incorrect const qualification on the
1004 type of the temporary.
1005 * cfgexpand.c (expand_return): Likewise.
1006 * expr.c (expand_constructor): Likewise.
1007 (expand_expr_real_1): Likewise.
1009 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
1011 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
1014 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
1017 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
1019 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
1021 * config/arm/aarch-common-protos.h
1022 (alu_cost_table): Fix spelling of "extend".
1023 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
1025 2014-03-19 Richard Biener <rguenther@suse.de>
1028 * tree-core.h (tree_type_common): Re-order pointer members
1029 to reduce recursion depth during GC walks.
1031 2014-03-19 Marek Polacek <polacek@redhat.com>
1034 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
1035 before accessing it.
1037 2014-03-19 Richard Biener <rguenther@suse.de>
1040 * lto-streamer-in.c (input_function): In WPA stage do not drop
1043 2014-03-19 Jakub Jelinek <jakub@redhat.com>
1045 PR tree-optimization/60559
1046 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
1047 with build_zero_cst assignment.
1049 2014-03-18 Kai Tietz <ktietz@redhat.com>
1051 PR rtl-optimization/56356
1052 * sdbout.c (sdbout_parms): Verify that parms'
1053 incoming argument is valid.
1054 (sdbout_reg_parms): Likewise.
1056 2014-03-18 Richard Henderson <rth@redhat.com>
1059 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1060 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1061 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1063 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1065 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1066 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1067 Italicize plugin event names in description. Explain that
1068 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1069 Remind that no GCC functions should be called after PLUGIN_FINISH.
1070 Explain what pragmas with expansion are.
1072 2014-03-18 Martin Liska <mliska@suse.cz>
1074 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1075 gimple call statement is update.
1076 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1077 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1079 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1082 * ubsan.c (ubsan_instrument_unreachable): Call
1083 initialize_sanitizer_builtins.
1084 (ubsan_pass): Likewise.
1087 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1088 varpool_finalize_decl instead of rest_of_decl_compilation.
1090 2014-03-18 Richard Biener <rguenther@suse.de>
1092 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1093 by using bitmap_and_compl instead of bitmap_and_compl_into.
1094 (df_rd_transfer_function): Likewise.
1096 2014-03-18 Richard Biener <rguenther@suse.de>
1098 * doc/lto.texi (fresolution): Fix typo.
1100 2014-03-18 Richard Biener <rguenther@suse.de>
1102 * doc/invoke.texi (flto): Update for changes in 4.9.
1104 2014-03-18 Richard Biener <rguenther@suse.de>
1106 * doc/loop.texi: Remove section on the removed lambda framework.
1107 Update loop docs with recent changes in preserving loop structure.
1109 2014-03-18 Richard Biener <rguenther@suse.de>
1111 * doc/lto.texi (-fresolution): Document.
1113 2014-03-18 Richard Biener <rguenther@suse.de>
1115 * doc/contrib.texi: Adjust my name.
1117 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1120 * internal-fn.c: Include diagnostic-core.h.
1121 (expand_BUILTIN_EXPECT): New function.
1122 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1123 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1124 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1125 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1127 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1128 Revert 3 argument __builtin_expect code.
1129 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1130 * gimple-fold.c (gimple_fold_call): Likewise.
1131 * tree.h (fold_builtin_expect): New prototype.
1132 * builtins.c (build_builtin_expect_predicate): Add predictor
1133 argument, if non-NULL, create 3 argument __builtin_expect.
1134 (fold_builtin_expect): No longer static. Add ARG2 argument,
1135 pass it through to build_builtin_expect_predicate.
1136 (fold_builtin_2): Adjust caller.
1137 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1138 * internal-fn.def (BUILTIN_EXPECT): New.
1140 2014-03-18 Tobias Burnus <burnus@net-b.de>
1143 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1144 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1145 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1147 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1150 * predict.c (combine_predictions_for_bb): Fix up formatting.
1151 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1152 fill what it points to if non-NULL.
1153 (tree_predict_by_opcode): Adjust caller, use the predictor.
1154 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1156 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1158 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1159 proper constant for the store mode.
1161 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1163 * symtab.c (change_decl_assembler_name): Fix transparent alias
1166 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1168 * config/aarch64/aarch64.c: Correct the comments about the
1169 aarch64 stack layout.
1171 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1173 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1174 check for GF_OMP_FOR_KIND_FOR.
1176 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1178 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1179 ymm and zmm register names.
1181 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1184 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1185 note creation for the 2010-08-31 changes.
1187 2014-03-17 Marek Polacek <polacek@redhat.com>
1190 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1191 as -fno-tree-loop-vectorize.
1192 (expand_omp_simd): Likewise.
1194 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1196 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1197 (eligible_for_call_delay): New prototype.
1198 * config/sparc/sparc.c (tls_call_delay): Rename into...
1199 (eligible_for_call_delay): ...this. Return false if the instruction
1200 cannot be put in the delay slot of a branch.
1201 (eligible_for_restore_insn): Simplify.
1202 (eligible_for_return_delay): Return false if the instruction cannot be
1203 put in the delay slot of a branch and simplify.
1204 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1205 put in the delay slot of a branch.
1206 * config/sparc/sparc.md (fix_ut699): New attribute.
1207 (tls_call_delay): Delete.
1208 (in_call_delay): Reimplement.
1209 (eligible_for_sibcall_delay): Rename into...
1210 (in_sibcall_delay): ...this.
1211 (eligible_for_return_delay): Rename into...
1212 (in_return_delay): ...this.
1213 (in_branch_delay): Reimplement.
1214 (in_uncond_branch_delay): Delete.
1215 (in_annul_branch_delay): Delete.
1217 2014-03-14 Richard Henderson <rth@redhat.com>
1220 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1221 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1222 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1223 (floathi splitters): Remove.
1224 (float<SWI48x>xf2): New pattern.
1225 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1226 code that tried to handle DImode for 32-bit, but which was excluded
1227 by the pattern's condition. Drop allocation of stack temporary.
1228 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1229 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1230 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1231 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1232 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1233 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1234 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1235 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1236 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1237 (*float<SWI48x><X87MODEF>2_i387): Remove.
1238 (all float _with_temp splitters): Remove.
1239 (*float<SWI48x><MODEF>2_i387): New pattern.
1240 (*float<SWI48><MODEF>2_sse): New pattern.
1241 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1242 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1244 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1245 Marek Polacek <polacek@redhat.com>
1248 * common.opt (dump_base_name_prefixed): New Variable.
1249 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1250 if x_dump_base_name_prefixed is already set, set it at the end.
1252 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1254 PR rtl-optimization/60508
1255 * lra-constraints.c (get_reload_reg): Add new parameter
1257 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1258 Pass the new parameter values.
1260 2014-03-14 Richard Biener <rguenther@suse.de>
1262 * common.opt: Revert unintented changes from r205065.
1265 2014-03-14 Richard Biener <rguenther@suse.de>
1268 * cfghooks.c (split_block): Properly adjust all loops the
1269 block was a latch of.
1271 2014-03-14 Martin Jambor <mjambor@suse.cz>
1274 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1277 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1280 * config/avr/avr.c (avr_set_current_function): Pass function name
1281 through default_strip_name_encoding before sanity checking instead
1282 of skipping the first char of the assembler name.
1284 2014-03-13 Richard Henderson <rth@redhat.com>
1287 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1288 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1289 * config/i386/i386-protos.h: Likewise.
1290 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1291 in the expander instead of a splitter.
1292 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1293 any possibility of requiring a memory.
1294 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1295 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1296 (fp branch splitters): Update for ix86_split_fp_branch.
1297 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1298 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1299 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1300 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1301 (*fop_<MODEF>_3_i387): Likewise.
1302 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1303 (splitters for the fop_* register patterns): Remove.
1304 (fscalexf4_i387): Rename from *fscalexf4_i387.
1305 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1307 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1309 PR tree-optimization/59779
1310 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1311 type for bitsize and maxsize instead of HOST_WIDE_INT.
1313 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1315 PR rtl-optimization/57320
1316 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1317 the CFG after thread_prologue_and_epilogue_insns.
1319 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1321 PR rtl-optimization/57189
1322 * lra-constraints.c (process_alt_operands): Disfavor spilling
1325 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1327 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1329 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1331 PR tree-optimization/59025
1333 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1334 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1336 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1339 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1340 calls of avr_out_plus_1.
1342 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1344 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1345 BB's single pred and update the father loop's latch info later.
1347 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1349 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1353 (VEC_base): Likewise.
1354 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1355 registers, we need to swap double words in little endian mode.
1357 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1358 to be a container mode for 128-bit integer operations added in ISA
1359 2.07. Unlike TImode and PTImode, the preferred register set is
1360 the Altivec/VMX registers for the 128-bit operations.
1362 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1364 (rs6000_split_128bit_ok_p): Likewise.
1366 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1367 macros for creating ISA 2.07 normal and overloaded builtin
1368 functions with 3 arguments.
1369 (BU_P8V_OVERLOAD_3): Likewise.
1370 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1371 for use as overloaded functions.
1372 (VPERM_1TI_UNS): Likewise.
1373 (VSEL_1TI): Likewise.
1374 (VSEL_1TI_UNS): Likewise.
1375 (ST_INTERNAL_1ti): Likewise.
1376 (LD_INTERNAL_1ti): Likewise.
1377 (XXSEL_1TI): Likewise.
1378 (XXSEL_1TI_UNS): Likewise.
1379 (VPERM_1TI): Likewise.
1380 (VPERM_1TI_UNS): Likewise.
1381 (XXPERMDI_1TI): Likewise.
1382 (SET_1TI): Likewise.
1383 (LXVD2X_V1TI): Likewise.
1384 (STXVD2X_V1TI): Likewise.
1385 (VEC_INIT_V1TI): Likewise.
1386 (VEC_SET_V1TI): Likewise.
1387 (VEC_EXT_V1TI): Likewise.
1388 (EQV_V1TI): Likewise.
1389 (NAND_V1TI): Likewise.
1390 (ORC_V1TI): Likewise.
1391 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1392 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1394 (VADDUQM): Likewise.
1395 (VSUBCUQ): Likewise.
1396 (VADDEUQM): Likewise.
1397 (VADDECUQ): Likewise.
1398 (VSUBEUQM): Likewise.
1399 (VSUBECUQ): Likewise.
1401 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1402 __int128_t and __uint128_t types.
1403 (__uint128_type): Likewise.
1404 (altivec_categorize_keyword): Add support for vector __int128_t,
1405 vector __uint128_t, vector __int128, and vector unsigned __int128
1406 as a container type for TImode operations that need to be done in
1407 VSX/Altivec registers.
1408 (rs6000_macro_to_expand): Likewise.
1409 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1410 to support 128-bit integer instructions vaddcuq, vadduqm,
1411 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1412 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1414 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1415 for V1TImode, and set up preferences to use VSX/Altivec registers.
1416 Setup VSX reload handlers.
1417 (rs6000_debug_reg_global): Likewise.
1418 (rs6000_init_hard_regno_mode_ok): Likewise.
1419 (rs6000_preferred_simd_mode): Likewise.
1420 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1421 (easy_altivec_constant): Likewise.
1422 (output_vec_const_move): Likewise.
1423 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1425 (rs6000_expand_vector_extract): Likewise.
1426 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1428 (rs6000_const_vec): Add support for V1TImode.
1429 (rs6000_emit_le_vsx_load): Swap double words when loading or
1430 storing TImode/V1TImode.
1431 (rs6000_emit_le_vsx_store): Likewise.
1432 (rs6000_emit_le_vsx_move): Likewise.
1433 (rs6000_emit_move): Add support for V1TImode.
1434 (altivec_expand_ld_builtin): Likewise.
1435 (altivec_expand_st_builtin): Likewise.
1436 (altivec_expand_vec_init_builtin): Likewise.
1437 (altivec_expand_builtin): Likewise.
1438 (rs6000_init_builtins): Add support for V1TImode type. Add
1439 support for ISA 2.07 128-bit integer builtins. Define type names
1440 for the VSX/Altivec vector types.
1441 (altivec_init_builtins): Add support for overloaded vector
1442 functions with V1TImode type.
1443 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1444 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1446 (rs6000_split_128bit_ok_p): Likewise.
1447 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1448 __int128_t and vector __uint128_t.
1450 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1451 and mode attributes.
1458 (VS_scalar): Likewise.
1459 (VS_double): Likewise.
1460 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1462 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1463 we support the ISA 2.07 128-bit integer arithmetic instructions.
1464 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1465 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1466 and TImode types for use with the builtin functions.
1467 (V1TI_type_node): Likewise.
1468 (unsigned_V1TI_type_node): Likewise.
1469 (intTI_type_internal_node): Likewise.
1470 (uintTI_type_internal_node): Likewise.
1472 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1473 128-bit builtin functions.
1474 (UNSPEC_VADDEUQM): Likewise.
1475 (UNSPEC_VADDECUQ): Likewise.
1476 (UNSPEC_VSUBCUQ): Likewise.
1477 (UNSPEC_VSUBEUQM): Likewise.
1478 (UNSPEC_VSUBECUQ): Likewise.
1479 (VM): Add V1TImode to vector mode iterators.
1481 (VI_unit): Likewise.
1482 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1483 (altivec_vaddcuq): Likewise.
1484 (altivec_vsubuqm): Likewise.
1485 (altivec_vsubcuq): Likewise.
1486 (altivec_vaddeuqm): Likewise.
1487 (altivec_vaddecuq): Likewise.
1488 (altivec_vsubeuqm): Likewise.
1489 (altivec_vsubecuq): Likewise.
1491 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1493 (BOOL_128): Likewise.
1494 (BOOL_REGS_OUTPUT): Likewise.
1495 (BOOL_REGS_OP1): Likewise.
1496 (BOOL_REGS_OP2): Likewise.
1497 (BOOL_REGS_UNARY): Likewise.
1498 (BOOL_REGS_AND_CR0): Likewise.
1500 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1501 128-bit integer builtin support.
1502 (vec_vadduqm): Likewise.
1503 (vec_vaddecuq): Likewise.
1504 (vec_vaddeuqm): Likewise.
1505 (vec_vsubecuq): Likewise.
1506 (vec_vsubeuqm): Likewise.
1507 (vec_vsubcuq): Likewise.
1508 (vec_vsubuqm): Likewise.
1510 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1511 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1512 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1513 128-bit integer add/subtract to ISA 2.07.
1515 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1517 * config/arc/arc.c (arc_predicate_delay_insns):
1518 Fix third argument passed to conditionalize_nonjump.
1520 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1522 * config/aarch64/aarch64-builtins.c
1523 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1524 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1525 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1526 instead of __builtin_lfloor.
1527 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1529 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1531 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1532 (tree_ssa_ifcombine_bb_1): New function.
1533 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1534 is an empty forwarder block to then_bb or vice versa and then_bb
1535 and else_bb are effectively swapped.
1537 2014-03-12 Christian Bruel <christian.bruel@st.com>
1540 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1541 REG_CFA_DEF_CFA note.
1542 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1543 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1545 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1547 PR tree-optimization/60454
1548 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1550 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1552 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1553 Do not define target_cpu_default2 to generic.
1554 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1555 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1556 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1558 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1559 Marc Glisse <marc.glisse@inria.fr>
1561 PR tree-optimization/60502
1562 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1563 instead of build_low_bits_mask.
1565 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1568 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1569 if there are multiple uses, but op doesn't live on E edge.
1570 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1571 clobber stmts before __builtin_unreachable.
1573 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1575 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1576 hard_frame_pointer_rtx.
1577 * cse.c (cse_insn): Remove volatile check.
1578 * cselib.c (cselib_process_insn): Likewise.
1579 * dse.c (scan_insn): Likewise.
1581 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1583 * config/arc/arc.c (conditionalize_nonjump): New function,
1585 (arc_ifcvt): ... this.
1586 (arc_predicate_delay_insns): Use it.
1588 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1590 * config/arc/predicates.md (extend_operand): During/after reload,
1591 allow const_int_operand.
1592 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1593 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1594 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1596 (umulsi3_highpart_i): Likewise.
1598 2014-03-11 Richard Biener <rguenther@suse.de>
1600 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1601 Add asserts to guard possible wrong-code bugs.
1603 2014-03-11 Richard Biener <rguenther@suse.de>
1605 PR tree-optimization/60429
1606 PR tree-optimization/60485
1607 * tree-ssa-structalias.c (set_union_with_increment): Properly
1608 take into account all fields that overlap the shifted vars.
1609 (do_sd_constraint): Likewise.
1610 (do_ds_constraint): Likewise.
1611 (get_constraint_for_ptr_offset): Likewise.
1613 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1615 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1616 (nios2_compute_frame_layout):
1617 Add calculation of cfun->machine->fp_save_offset.
1618 (nios2_expand_prologue): Correct setting of frame pointer register
1620 (nios2_expand_epilogue): Update recovery of stack pointer from
1621 frame pointer accordingly.
1622 (nios2_initial_elimination_offset): Update calculation of offset
1623 for eliminating to HARD_FRAME_POINTER_REGNUM.
1625 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1628 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1629 cgraph_get_create_node on VAR_DECLs.
1631 2014-03-10 Richard Biener <rguenther@suse.de>
1634 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1636 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1638 * config/vms/vms.opt (vms_float_format): New variable.
1640 2014-03-08 Tobias Burnus <burnus@net-b.de>
1642 * doc/invoke.texi (-fcilkplus): Update implementation status.
1644 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1645 Richard Biener <rguenther@suse.de>
1647 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1648 consistently accross all TUs.
1649 (run_gcc): Enable -fshort-double automatically at link at link-time
1650 and disallow override.
1652 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1655 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1656 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1657 if they can't be used.
1659 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1661 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1662 for Solaris 11/x86 ld.
1663 * configure: Regenerate.
1665 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1667 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1668 (LIB_TLS_SPEC): Save as ld_tls_libs.
1669 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1670 (HAVE_AS_IX86_TLSLDM): New test.
1671 * configure, config.in: Regenerate.
1672 * config/i386/i386.c (legitimize_tls_address): Fall back to
1673 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1674 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1675 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1676 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1678 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1680 * common.opt (fira-loop-pressure): Mark as optimization.
1682 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1684 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1685 an OpenMP mappable type.
1687 2014-03-06 Matthias Klose <doko@ubuntu.com>
1689 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1690 MULTILIB_OSDIRNAMES is not defined.
1692 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1693 Meador Inge <meadori@codesourcery.com>
1696 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1697 (arm_legitimize_address): Call legitimize_tls_address for any
1698 arm_tls_referenced_p expression, handle constant addend. Call it
1699 before testing for !TARGET_ARM.
1700 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1702 2014-03-06 Richard Biener <rguenther@suse.de>
1708 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1710 * tree-streamer.c (record_common_node): Assert we don't record
1711 nodes with type double.
1712 (preload_common_node): Skip type double, complex double and double
1713 pointer since it is now frontend dependent due to fshort-double option.
1715 2014-03-06 Richard Biener <rguenther@suse.de>
1717 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1718 or -fno-lto is specified and the linker has full plugin support.
1719 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1720 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1721 * lto-wrapper.c (merge_and_complain): Merge compile-time
1722 optimization levels.
1723 (run_gcc): And pass it through to the link options.
1725 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1729 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1731 * cselib.c (remove_useless_values): Skip to avoid quadratic
1732 behavior if the condition moved from...
1733 (cselib_process_insn): ... here holds.
1735 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1738 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1739 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1742 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1743 (TM_H): Add x86-tune.def.
1745 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1747 * config/aarch64/aarch64.c (generic_tunings):
1748 Use cortexa57_extra_costs.
1750 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1753 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1754 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1755 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1758 2014-03-04 Heiher <r@hev.cc>
1760 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1761 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1763 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1765 * config/i386/predicates.md (const2356_operand): Change to ...
1766 (const2367_operand): ... this.
1767 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1769 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1770 (*avx512pf_scatterpf<mode>sf): Ditto.
1771 (avx512pf_scatterpf<mode>df): Ditto.
1772 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1773 (*avx512pf_scatterpf<mode>df): Ditto.
1774 * config/i386/i386.c (ix86_expand_builtin): Update
1775 incorrect hint operand error message.
1777 2014-03-04 Richard Biener <rguenther@suse.de>
1779 * lto-section-in.c (lto_get_section_data): Fix const cast.
1781 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1783 * tree-streamer.c (record_common_node): Assert we don't record
1784 nodes with type double.
1785 (preload_common_node): Skip type double, complex double and double
1786 pointer since it is now frontend dependent due to fshort-double option.
1788 2014-03-04 Richard Biener <rguenther@suse.de>
1791 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1792 (lto_input_toplevel_asms): Likewise.
1793 * lto-section-in.c (lto_get_section_data): Instead do it here
1796 2014-03-04 Richard Biener <rguenther@suse.de>
1798 PR tree-optimization/60382
1799 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1800 dead PHIs a reduction.
1802 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1804 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1806 (_mm_prefetch): Move out of GCC target("sse") pragma.
1807 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1808 GCC target("prfchw") pragma.
1809 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1811 * config/i386/i386.c (ix86_option_override_internal): Enable
1812 -mprfchw with -mprefetchwt1.
1814 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1816 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1819 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1821 * opts.h (CL_PCH_IGNORE): Define.
1822 * targhooks.c (option_affects_pch_p):
1823 Return false for options that have CL_PCH_IGNORE set.
1824 * opt-functions.awk: Process PchIgnore.
1825 * doc/options.texi: Document PchIgnore.
1827 * config/arc/arc.opt (misize): Add PchIgnore property.
1829 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1831 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1832 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1833 constraint on constants to permit them being loaded into
1834 GENERAL_REGS or BASE_REGS.
1836 2014-03-03 Nick Clifton <nickc@redhat.com>
1838 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1839 anti-cacnonical alternatives.
1840 (negandhi3_real): New pattern.
1841 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1843 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1845 * config/avr/avr-mcus.def: Remove atxmega16x1.
1846 * config/avr/avr-tables.opt: Regenerate.
1847 * config/avr/t-multilib: Regenerate.
1848 * doc/avr-mmcu.texi: Regenerate.
1850 2014-03-03 Tobias Grosser <tobias@grosser.es>
1851 Mircea Namolaru <mircea.namolaru@inria.fr>
1853 PR tree-optimization/58028
1854 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1857 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1859 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1860 not handled by recognizers.
1862 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1865 * function.c (expand_function_end): Don't emit
1866 clobber_return_register sequence if clobber_after is a BARRIER.
1867 * cfgexpand.c (construct_exit_block): Append instructions before
1868 return_label to prev_bb.
1870 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1872 * config/rs6000/constraints.md: Document reserved use of "wc".
1874 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1877 * ipa.c (function_and_variable_visibility): When dissolving comdat
1878 group, also set all symbols to local.
1880 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1885 2013-12-14 Jan Hubicka <jh@suse.cz>
1887 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1889 2014-03-02 Jon Beniston <jon@beniston.com>
1895 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1896 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1897 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1898 (simple_return, *simple_return): New patterns
1899 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1900 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1902 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1904 * dwarf2out.c (gen_subprogram_die): Tidy.
1906 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1909 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1910 (*mov_t_msb_neg_negc): ... this new insn.
1912 2014-02-28 Jason Merrill <jason@redhat.com>
1915 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1918 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1921 * dwarf2out.c (decltype_auto_die): New static.
1922 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1923 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1924 (is_cxx_auto): Likewise.
1926 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1928 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1929 we are not using general regs only.
1931 2014-02-28 Richard Biener <rguenther@suse.de>
1934 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1935 previous fix and only allow to remove trivial pre-headers
1936 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1937 (remove_forwarder_block): Properly update the latch of a loop.
1939 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1942 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1943 (cselib_preserved_hash_table): New.
1944 (preserve_constants_and_equivs): Move preserved vals to it.
1945 (cselib_find_slot): Look it up first.
1946 (cselib_init): Initialize it.
1947 (cselib_finish): Release it.
1948 (dump_cselib_table): Dump it.
1950 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1953 * cselib.c (remove_useless_values): Skip to avoid quadratic
1954 behavior if the condition moved from...
1955 (cselib_process_insn): ... here holds.
1957 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1960 * var-tracking.c (vt_initialize): Apply the same condition to
1961 preserve the CFA base value.
1963 2014-02-28 Joey Ye <joey.ye@arm.com>
1966 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1967 if reload in progress or completed.
1969 2014-02-28 Tobias Burnus <burnus@net-b.de>
1972 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1975 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1977 * doc/tm.texi.in (Condition Code Status): Update documention for
1978 relative locations of cc0-setter and cc0-user.
1980 2014-02-27 Jeff Law <law@redhat.com>
1982 PR rtl-optimization/52714
1983 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1984 into two independent simple sets, if I3 is a jump, ensure the
1985 pattern we place into I3 is a (set (pc) ...).
1987 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1988 Jeff Law <law@redhat.com>
1990 PR rtl-optimization/49847
1991 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1992 are in different blocks.
1993 * doc/tm.texi (Condition Code Status): Update documention for
1994 relative locations of cc0-setter and cc0-user.
1996 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1999 * lra.c (lra_emit_add): Check SUBREG too.
2001 2014-02-27 Andreas Schwab <schwab@suse.de>
2003 * config/m68k/m68k.c (m68k_option_override): Disable
2004 -flive-range-shrinkage for classic m68k.
2005 (m68k_override_options_after_change): Likewise.
2007 2014-02-27 Marek Polacek <polacek@redhat.com>
2010 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
2011 -Wmaybe-uninitialized.
2013 2014-02-27 Alan Modra <amodra@gmail.com>
2016 * reload1.c (emit_input_reload_insns): When reload_override_in,
2017 set old to rl->in_reg when rl->in_reg is a subreg.
2019 2014-02-26 Richard Biener <rguenther@suse.de>
2022 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
2024 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2026 * common/config/i386/predicates.md (const1256_operand): Remove.
2027 (const2356_operand): New.
2028 (const_1_to_2_operand): Remove.
2029 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2030 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2031 (*avx512pf_gatherpf<mode>sf): Ditto.
2032 (avx512pf_gatherpf<mode>df): Ditto.
2033 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2034 (*avx512pf_gatherpf<mode>df): Ditto.
2035 (avx512pf_scatterpf<mode>sf): Ditto.
2036 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2037 (*avx512pf_scatterpf<mode>sf): Ditto.
2038 (avx512pf_scatterpf<mode>df): Ditto.
2039 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2040 (*avx512pf_scatterpf<mode>df): Ditto.
2041 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
2043 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
2045 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
2046 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
2047 (_mm512_mask_testn_epi64_mask): Move to ...
2048 * config/i386/avx512cdintrin.h: Here.
2049 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
2050 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
2051 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
2052 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
2053 TARGET_AVX512F from TARGET_AVX512CD.
2055 2014-02-26 Richard Biener <rguenther@suse.de>
2058 * ipa.c (walk_polymorphic_call_targets): Properly guard
2059 call to inline_update_overall_summary.
2061 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2064 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2065 and latches only if requested. Fix latch if it is removed.
2066 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2067 LOOPS_HAVE_PREHEADERS.
2069 2014-02-25 Andrew Pinski <apinski@cavium.com>
2071 * builtins.c (expand_builtin_thread_pointer): Create a new target
2072 when the target is NULL.
2074 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2076 PR rtl-optimization/60317
2077 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2078 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2079 * lra-assigns.c: Include params.h.
2080 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2081 other reload pseudos considerations.
2083 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2085 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2086 to use canonical form for nor<mode>3.
2088 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2091 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2094 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2096 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2097 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2098 (ix86_handle_option): Handle OPT_mprefetchwt1.
2099 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2100 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2102 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2103 OPTION_MASK_ISA_PREFETCHWT1.
2104 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2105 (PTA_PREFETCHWT1): New.
2106 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2107 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2108 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2109 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2110 (*prefetch_avx512pf_<mode>_: Change into ...
2111 (*prefetch_prefetchwt1_<mode>: This.
2112 * config/i386/i386.opt (mprefetchwt1): New.
2113 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2114 (_mm_prefetch): Handle intent to write.
2115 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2117 2014-02-25 Richard Biener <rguenther@suse.de>
2120 * emit-rtl.c (mem_attrs_htab): Remove.
2121 (mem_attrs_htab_hash): Likewise.
2122 (mem_attrs_htab_eq): Likewise.
2123 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2124 (init_emit_once): Do not allocate mem_attrs_htab.
2126 2014-02-25 Richard Biener <rguenther@suse.de>
2129 * lto-opts.c (lto_write_options): Output non-explicit conservative
2130 -fwrapv, -fno-trapv and -fno-strict-overflow.
2131 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2132 (run_gcc): And pass them through.
2134 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2136 * sel-sched.c (calculate_new_fences): New parameter ptime.
2137 Calculate it as a maximum over all fence cycles.
2138 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2139 Print the final schedule timing when sched_verbose.
2141 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2143 PR rtl-optimization/60292
2144 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2145 bit fot the fence instruction.
2147 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2149 * calls.h: Fix typo in comment.
2151 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2153 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2154 adjusting offsetable addresses.
2156 2014-02-24 Guozhi Wei <carrot@google.com>
2158 * sparseset.h (sparseset_pop): Fix the wrong index.
2160 2014-02-24 Walter Lee <walt@tilera.com>
2162 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2163 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2165 * common/config/tilegx/tilegx-common.c
2166 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2167 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2169 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2170 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2171 (tilegx_gimplify_va_arg_expr): Handle big endian.
2172 (tilegx_expand_unaligned_load): Ditto.
2173 (tilegx_expand_unaligned_store): Ditto.
2174 (TARGET_RETURN_IN_MSB): New.
2175 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2176 (TARGET_ENDIAN_DEFAULT): New.
2177 (TARGET_BIG_ENDIAN): Handle big endian.
2178 (BYTES_BIG_ENDIAN): Ditto.
2179 (WORDS_BIG_ENDIAN): Ditto.
2180 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2183 * config/tilegx/tilegx.md (extv): Handle big endian.
2185 (insn_st<n>): Ditto.
2186 (insn_st<n>_add<bitsuffix>): Ditto.
2187 (insn_stnt<n>): Ditto.
2188 (insn_stnt<n>_add<bitsuffix>):Ditto.
2189 (vec_interleave_highv8qi): Handle big endian.
2190 (vec_interleave_highv8qi_be): New.
2191 (vec_interleave_highv8qi_le): New.
2192 (insn_v1int_h): Handle big endian.
2193 (vec_interleave_lowv8qi): Handle big endian.
2194 (vec_interleave_lowv8qi_be): New.
2195 (vec_interleave_lowv8qi_le): New.
2196 (insn_v1int_l): Handle big endian.
2197 (vec_interleave_highv4hi): Handle big endian.
2198 (vec_interleave_highv4hi_be): New.
2199 (vec_interleave_highv4hi_le): New.
2200 (insn_v2int_h): Handle big endian.
2201 (vec_interleave_lowv4hi): Handle big endian.
2202 (vec_interleave_lowv4hi_be): New.
2203 (vec_interleave_lowv4hi_le): New.
2204 (insn_v2int_l): Handle big endian.
2205 (vec_interleave_highv2si): Handle big endian.
2206 (vec_interleave_highv2si_be): New.
2207 (vec_interleave_highv2si_le): New.
2208 (insn_v4int_h): Handle big endian.
2209 (vec_interleave_lowv2si): Handle big endian.
2210 (vec_interleave_lowv2si_be): New.
2211 (vec_interleave_lowv2si_le): New.
2212 (insn_v4int_l): Handle big endian.
2213 * config/tilegx/tilegx.opt (mbig-endian): New option.
2214 (mlittle-endian): New option.
2215 * doc/install.texi: Document tilegxbe-linux.
2216 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2218 2014-02-24 Martin Jambor <mjambor@suse.cz>
2221 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2222 there are no parameter descriptors.
2224 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2226 PR rtl-optimization/60268
2227 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2228 initialization to ...
2229 (sched_rgn_init): ... here.
2230 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2232 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2234 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2237 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2239 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2242 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2244 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2245 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2247 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2249 * config/microblaze/predicates.md: Add cmp_op predicate.
2250 * config/microblaze/microblaze.md: Add branch_compare instruction
2251 which uses cmp_op predicate and emits cmp insn before branch.
2252 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2253 to microblaze_expand_conditional_branch and consolidate logic.
2254 (microblaze_expand_conditional_branch): emit branch_compare
2255 insn instead of handling cmp op separate from branch insn.
2257 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2259 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2262 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2264 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2265 define_insn with define_expand and new define_insn
2266 *altivec_lve<VI_char>x_internal.
2267 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2268 and new define_insn *altivec_stve<VI_char>x_internal.
2269 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2271 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2273 (altivec_expand_stvex_be): New function.
2275 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2277 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2278 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2279 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2280 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2282 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2285 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2286 instead of emit_move_insn.
2288 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2290 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2292 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2293 gen_altivec_vsumsws.
2295 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2297 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2298 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2299 (altivec_lvxl_<mode>): New define_expand incorporating
2300 -maltivec=be semantics where needed.
2301 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2302 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2303 semantics where needed.
2304 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2305 (altivec_stvx_<mode>): New define_expand incorporating
2306 -maltivec=be semantics where needed.
2307 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2308 VM2 iterator instead of V4SI.
2309 (altivec_stvxl_<mode>): New define_expand incorporating
2310 -maltivec=be semantics where needed.
2311 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2312 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2313 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2314 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2315 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2316 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2317 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2318 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2319 ALTIVEC_BUILTIN_STVXL.
2320 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2321 (altivec_expand_stvx_be): Likewise.
2322 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2323 (altivec_expand_lvx_be): Likewise.
2324 (altivec_expand_stvx_be): Likewise.
2325 (altivec_expand_builtin): Add cases for
2326 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2327 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2328 (altivec_init_builtins): Add definitions for
2329 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2330 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2332 2014-02-21 Catherine Moore <clm@codesourcery.com>
2334 * doc/invoke.texi (mvirt, mno-virt): Document.
2335 * config/mips/mips.opt (mvirt): New option.
2336 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2338 2014-02-21 Richard Biener <rguenther@suse.de>
2340 PR tree-optimization/60276
2341 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2342 (STMT_VINFO_MIN_NEG_DIST): New macro.
2343 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2344 STMT_VINFO_MIN_NEG_DIST.
2345 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2346 made for negative dependence distances still hold.
2348 2014-02-21 Richard Biener <rguenther@suse.de>
2351 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2352 DECL_INITIAL for globals not in the current function context.
2354 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2356 PR tree-optimization/56490
2357 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2358 * tree-ssa-uninit.c: Include params.h.
2359 (compute_control_dep_chain): Add num_calls argument, return false
2360 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2361 num_calls to recursive call.
2362 (find_predicates): Change dep_chain into normal array,
2363 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2364 variable and adjust compute_control_dep_chain caller.
2365 (find_def_preds): Likewise.
2367 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2369 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2370 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2372 2014-02-21 Nick Clifton <nickc@redhat.com>
2374 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2375 (pushhi1): Likewise.
2376 (popqi1): Add mode to pre_dec.
2379 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2381 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2382 mode for mask of V8SFmode permutation.
2384 2014-02-20 Richard Henderson <rth@redhat.com>
2387 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2388 a new pseudo for OLDVAL.
2390 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2393 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2394 gen_reg_rtx if d->testing_p.
2395 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2396 if d->testing_p and we will certainly return true.
2397 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2400 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2402 * emit-rtl.c (gen_reg_rtx): Assert that
2403 crtl->emit.regno_pointer_align_length is non-zero.
2405 2014-02-20 Richard Henderson <rth@redhat.com>
2408 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2409 on failure the store back into EXPECT.
2411 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2412 Sandra Loosemore <sandra@codesourcery.com>
2414 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2415 * config/nios2/nios2.c (nios2_function_profiler): Add
2416 -fPIC (flag_pic == 2) support.
2417 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2418 (nios2_large_offset_p): New function.
2419 (nios2_unspec_reloc_p): Move up position, update to use
2420 nios2_large_offset_p.
2421 (nios2_unspec_address): Remove function.
2422 (nios2_unspec_offset): New function.
2423 (nios2_large_got_address): New function.
2424 (nios2_got_address): Add large offset support.
2425 (nios2_legitimize_tls_address): Update usage of removed and new
2427 (nios2_symbol_binds_local_p): New function.
2428 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2429 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2430 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2431 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2432 processing for (const (unspec ...)).
2433 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2435 2014-02-20 Richard Biener <rguenther@suse.de>
2437 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2438 doing the substitution.
2439 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2441 2014-02-20 Martin Jambor <mjambor@suse.cz>
2444 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2445 info when checking whether lattices are bottom.
2447 2014-02-20 Richard Biener <rguenther@suse.de>
2450 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2453 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2456 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2457 parameter specifying the scaling.
2458 (inline_call): Update.
2459 (want_inline_recursively): Guard division by zero.
2460 (recursive_inlining): Update.
2461 * ipa-inline.h (clone_inlined_nodes): Update.
2463 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2466 * config/i386/i386.c (classify_argument): Pass structures of size
2467 64 bytes or less in register.
2469 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2470 Kirill Yukhin <kirill.yukhin@intel.com>
2472 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2473 (_mm_rcp28_round_ss): Ditto.
2474 (_mm_rsqrt28_round_sd): Ditto.
2475 (_mm_rsqrt28_round_ss): Ditto.
2476 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2477 (_mm_rcp14_round_ss): Ditto.
2478 (_mm_rsqrt14_round_sd): Ditto.
2479 (_mm_rsqrt14_round_ss): Ditto.
2480 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2481 the first input operand, get rid of match_dup.
2482 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2484 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2486 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2487 operand as the first input operand, set type attribute.
2488 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2490 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2491 operand as the first input operand, set type attribute.
2493 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2495 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2498 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2501 * config/i386/i386.c (construct_container): Remove TFmode check
2502 for X86_64_INTEGER_CLASS.
2504 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2507 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2508 only when -Wpsabi is enabled.
2510 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2513 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2514 passing arrays in registers are the same as for structs, so remove the
2515 special case for them.
2517 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2519 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2520 destination type, extract only the valid bits if the source type is not
2521 integral and has a different mode.
2523 2014-02-19 Richard Biener <rguenther@suse.de>
2526 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2529 2014-02-19 Richard Biener <rguenther@suse.de>
2532 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2533 (ipa_modify_call_arguments): Emit an argument load explicitely and
2534 preserve virtual SSA form there and for the replacement call.
2535 Do not update SSA form nor free dominance info.
2537 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2539 * ipa.c (function_and_variable_visibility): Also clear WEAK
2540 flag when disolving COMDAT_GROUP.
2542 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2544 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2545 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2547 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2548 do more sanity checks.
2549 (detect_type_change): Return true when giving up early.
2550 (compute_complex_assign_jump_func): Fix type parameter of
2551 ipa_set_ancestor_jf.
2552 (compute_complex_ancestor_jump_func): Likewise.
2553 (update_jump_functions_after_inlining): Fix updating of
2555 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2557 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2559 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2560 inline clones when edge disappears.
2562 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2565 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2566 Split 64-bit moves into 2 patterns. Do not allow the use of
2567 direct move for TDmode in little endian, since the decimal value
2568 has little endian bytes within a word, but the 64-bit pieces are
2569 ordered in a big endian fashion, and normal subreg's of TDmode are
2571 (mov<mode>_64bit_dm): Likewise.
2572 (movtd_64bit_nodm): Likewise.
2574 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2576 PR tree-optimization/60174
2577 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2578 statement of an SSA_NAME that occurs in an abnormal PHI node.
2580 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2583 * final.c (SEEN_BB): Remove.
2584 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2585 (final_scan_insn): Don't force_source_line on second
2586 NOTE_INSN_BASIC_BLOCK.
2588 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2591 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2592 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2593 (type_natural_mode): Warn ABI change when %zmm register is not
2594 available for AVX512F vector value passing.
2596 2014-02-18 Kai Tietz <ktietz@redhat.com>
2599 * config/i386/i386.c (ix86_expand_prologue): Use value in
2600 rax register as displacement when restoring %r10 or %rax.
2601 Fix wrong offset when restoring both registers.
2603 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2605 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2606 assertion with conditional return.
2608 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2609 Uros Bizjak <ubizjak@gmail.com>
2612 * config/i386/driver-i386.c (host_detect_local_cpu): If
2613 YMM state is not saved by the OS, also clear has_f16c. Move
2614 CPUID 0x80000001 handling before YMM state saving checking.
2616 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2618 PR rtl-optimization/58960
2619 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2620 factored out from ...
2621 (sched_init): ... here.
2622 (free_global_sched_pressure_data): New, factored out from ...
2623 (sched_finish): ... here.
2624 * sched-int.h (free_global_sched_pressure_data): Declare.
2625 * sched-rgn.c (nr_regions_initial): New static global.
2626 (haifa_find_rgns): Initialize it.
2627 (schedule_region): Disable sched-pressure for the newly
2630 2014-02-17 Richard Biener <rguenther@suse.de>
2632 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2633 release SSA defs of pattern stmts.
2635 2014-02-17 Richard Biener <rguenther@suse.de>
2637 * tree-inline.c (expand_call_inline): Release the virtual
2638 operand defined by the call we are about to inline.
2640 2014-02-17 Richard Biener <rguenther@suse.de>
2642 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2644 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2645 Ilya Tocar <ilya.tocar@intel.com>
2647 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2648 arguments order in builtin.
2649 (_mm512_permutexvar_epi64): Ditto.
2650 (_mm512_mask_permutexvar_epi64): Ditto
2651 (_mm512_maskz_permutexvar_epi32): Ditto
2652 (_mm512_permutexvar_epi32): Ditto
2653 (_mm512_mask_permutexvar_epi32): Ditto
2655 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2657 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2658 (p8_vmrgow): Likewise.
2660 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2662 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2665 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2668 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2669 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2670 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2671 using direct move instructions on ISA 2.07. Also adjust
2672 instruction length for 64-bit.
2673 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2674 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2676 2014-02-15 Alan Modra <amodra@gmail.com>
2680 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2681 find_replacement on parts of insn rtl that might be reloaded.
2683 2014-02-15 Richard Biener <rguenther@suse.de>
2685 PR tree-optimization/60183
2686 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2687 (tree_ssa_phiprop): Calculate and free post-dominators.
2689 2014-02-14 Jeff Law <law@redhat.com>
2691 PR rtl-optimization/60131
2692 * ree.c (get_extended_src_reg): New function.
2693 (combine_reaching_defs): Use it rather than assuming location of REG.
2694 (find_and_remove_re): Verify first operand of extension is
2695 a REG before adding the insns to the copy list.
2697 2014-02-14 Roland McGrath <mcgrathr@google.com>
2699 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2700 * configure: Regenerated.
2701 * config.in: Regenerated.
2702 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2703 instead of ASM_SHORT.
2705 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2706 Richard Earnshaw <rearnsha@arm.com>
2708 PR rtl-optimization/59535
2709 * lra-constraints.c (process_alt_operands): Encourage alternative
2710 when unassigned pseudo class is superset of the alternative class.
2711 (inherit_reload_reg): Don't inherit when optimizing for code size.
2712 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2713 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2714 modes not less than 4 for Thumb1.
2716 2014-02-14 Kyle McMartin <kyle@redhat.com>
2719 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2721 2014-02-14 Richard Biener <rguenther@suse.de>
2723 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2724 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2725 Do not strip INDIRECT_REFs.
2727 2014-02-14 Richard Biener <rguenther@suse.de>
2730 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2731 DECL_FUNCTION_SPECIFIC_TARGET.
2732 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2733 * tree-streamer-out.c (pack_ts_target_option): Remove.
2734 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2735 (write_ts_function_decl_tree_pointers): Do not stream
2736 DECL_FUNCTION_SPECIFIC_TARGET.
2737 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2738 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2739 (lto_input_ts_function_decl_tree_pointers): Do not stream
2740 DECL_FUNCTION_SPECIFIC_TARGET.
2742 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2744 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2745 (get_initial_def_for_induction, vectorizable_induction): Ignore
2746 debug stmts when looking for exit_phi.
2747 (vectorizable_live_operation): Fix up condition.
2749 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2751 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2752 nreverse() because it changes the content of original tree list.
2754 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2756 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2757 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2759 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2761 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2762 GNU coding standards.
2764 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2767 * dwarf2out.c (gen_subprogram_die): Don't call
2768 add_calling_convention_attribute if subr_die is old_die.
2770 2014-02-13 Sharad Singhai <singhai@google.com>
2772 * doc/optinfo.texi: Fix order of nodes.
2774 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2776 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2777 operands[2], not operands[3].
2779 2014-02-13 Richard Biener <rguenther@suse.de>
2782 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2783 mention the possibility of an in-tree build.
2784 (CLooG): Update recommended version to 0.18.1, mention the
2785 possibility of an in-tree build and clarify that the ISL
2786 bundled with CLooG does not work.
2788 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2791 * expr.c (compress_float_constant): If x is a hard register,
2792 extend into a pseudo and then move to x.
2794 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2796 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2797 caused by bad second argument to warning_at() with -mhotpatch and
2798 nested functions (e.g. with gfortran).
2800 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2802 * opts.c (option_name): Remove "enabled by default" rider.
2804 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2806 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2808 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2809 Uros Bizjak <ubizjak@gmail.com>
2812 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2813 * configure: Regenerated.
2815 2014-02-12 Richard Biener <rguenther@suse.de>
2817 * vec.c (vec_prefix::calculate_allocation): Move as
2818 inline variant to vec.h.
2819 (vec_prefix::calculate_allocation_1): New out-of-line version.
2820 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2821 (vec_prefix::m_has_auto_buf): Rename to ...
2822 (vec_prefix::m_using_auto_storage): ... this.
2823 (vec_prefix::calculate_allocation): Inline the easy cases
2824 and dispatch to calculate_allocation_1 which doesn't need the
2826 (va_heap::reserve): Use gcc_checking_assert.
2827 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2828 m_using_auto_storage.
2829 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2831 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2832 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2833 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2835 2014-02-12 Richard Biener <rguenther@suse.de>
2837 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2838 when we found a dependence.
2840 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2842 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2844 (maybe_fold_stmt): ... into this new function.
2845 * omp-low.c (lower_omp): Update comment.
2847 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2850 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2853 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2855 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2856 identifiers in comments.
2857 (cortexa53_extra_costs): Likewise.
2858 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2859 (cortexa7_extra_costs): Likewise.
2860 (cortexa12_extra_costs): Likewise.
2861 (cortexa15_extra_costs): Likewise.
2862 (v7m_extra_costs): Likewise.
2864 2014-02-12 Richard Biener <rguenther@suse.de>
2867 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2868 of posix_memalign being successful.
2869 (lower_stmt): Restrict lowering of posix_memalign to when
2870 -ftree-bit-ccp is enabled.
2872 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2874 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2876 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2878 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2880 PR rtl-optimization/60116
2881 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2882 other_insn once the combination has been validated.
2884 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2887 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2889 * ipa-devirt.c: Include demangle.h
2890 (odr_violation_reported): New static variable.
2891 (add_type_duplicate): Update odr_violations.
2892 (maybe_record_node): Add completep parameter; update it.
2893 (record_target_from_binfo): Add COMPLETEP parameter;
2894 update it as needed.
2895 (possible_polymorphic_call_targets_1): Likewise.
2896 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2897 rename FINAL to COMPLETE.
2898 (record_targets_from_bases): Sanity check we found the binfo;
2899 fix COMPLETEP updating.
2900 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2901 parameter, fix computing of COMPLETEP.
2902 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2903 at LTO time do demangling.
2904 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2905 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2907 (gimple_get_virt_method_for_binfo): Likewise.
2908 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2909 gimple_get_virt_method_for_vtable): Update prototypes.
2911 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2914 * genautomata.c (add_presence_absence): Fix typo with
2915 {final_}presence_list.
2917 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2920 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2921 for VSX/Altivec vectors that land in GPR registers.
2923 2014-02-11 Richard Henderson <rth@redhat.com>
2924 Jakub Jelinek <jakub@redhat.com>
2927 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2928 around drhs if type conversion to lacc->type is not useless.
2930 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2932 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2934 (cortex-a57.cortex-a53): Likewise.
2935 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2937 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2939 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2942 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2944 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2945 add_options_for_arm_vfp3.
2947 2014-02-11 Jeff Law <law@redhat.com>
2950 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2951 object with an undesirable mode.
2953 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2956 * config/i386/sol2-9.h: New file.
2957 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2958 *-*-solaris2.9*): Use it.
2960 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2962 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2963 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2965 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2967 * config/microblaze/microblaze.c: Extend mcpu version format
2969 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2971 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2973 2014-02-10 Richard Henderson <rth@redhat.com>
2976 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2977 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2978 ms-abi vs -mno-accumulate-outgoing-args.
2979 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2980 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2983 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2986 * cfgexpand.c (expand_asm_operands): Attach source location to
2987 ASM_INPUT rtx objects.
2988 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2990 2014-02-10 Nick Clifton <nickc@redhat.com>
2992 * config/mn10300/mn10300.c (popcount): New function.
2993 (mn10300_expand_prologue): Include saved registers in stack usage
2996 2014-02-10 Jeff Law <law@redhat.com>
2999 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
3000 when changing the SET_DEST of a prior insn to avoid an input reload.
3002 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
3004 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
3005 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
3006 -mcall-openbsd, or -mcall-linux.
3007 (CC1_ENDIAN_BIG_SPEC): Remove.
3008 (CC1_ENDIAN_LITTLE_SPEC): Remove.
3009 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3010 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
3011 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
3012 and %cc1_endian_default.
3013 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
3015 2014-02-10 Richard Biener <rguenther@suse.de>
3017 PR tree-optimization/60115
3018 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
3019 MEM_REF handling. Properly verify that the accesses are not
3020 out of the objects bound.
3022 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3024 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
3027 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3029 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
3030 proper constants and fix formatting.
3031 (possible_polymorphic_call_targets): Fix formatting.
3033 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
3034 Ilya Tocar <ilya.tocar@intel.com>
3036 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
3037 (_mm512_loadu_epi32): Renamed into...
3038 (_mm512_loadu_si512): This.
3039 (_mm512_storeu_epi32): Renamed into...
3040 (_mm512_storeu_si512): This.
3041 (_mm512_maskz_ceil_ps): Removed.
3042 (_mm512_maskz_ceil_pd): Ditto.
3043 (_mm512_maskz_floor_ps): Ditto.
3044 (_mm512_maskz_floor_pd): Ditto.
3045 (_mm512_floor_round_ps): Ditto.
3046 (_mm512_floor_round_pd): Ditto.
3047 (_mm512_ceil_round_ps): Ditto.
3048 (_mm512_ceil_round_pd): Ditto.
3049 (_mm512_mask_floor_round_ps): Ditto.
3050 (_mm512_mask_floor_round_pd): Ditto.
3051 (_mm512_mask_ceil_round_ps): Ditto.
3052 (_mm512_mask_ceil_round_pd): Ditto.
3053 (_mm512_maskz_floor_round_ps): Ditto.
3054 (_mm512_maskz_floor_round_pd): Ditto.
3055 (_mm512_maskz_ceil_round_ps): Ditto.
3056 (_mm512_maskz_ceil_round_pd): Ditto.
3057 (_mm512_expand_pd): Ditto.
3058 (_mm512_expand_ps): Ditto.
3059 * config/i386/i386.c (ix86_builtins): Remove
3060 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3061 (bdesc_args): Ditto.
3062 * config/i386/predicates.md (const1256_operand): New.
3063 (const_1_to_2_operand): Ditto.
3064 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3065 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3066 (*avx512pf_gatherpf<mode>sf): Ditto.
3067 (avx512pf_gatherpf<mode>df): Ditto.
3068 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3069 (*avx512pf_gatherpf<mode>df): Ditto.
3070 (avx512pf_scatterpf<mode>sf): Ditto.
3071 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3072 (*avx512pf_scatterpf<mode>sf): Ditto.
3073 (avx512pf_scatterpf<mode>df): Ditto.
3074 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3075 (*avx512pf_scatterpf<mode>df): Ditto.
3076 (avx512f_expand<mode>): Removed.
3077 (<shift_insn><mode>3<mask_name>): Change predicate type.
3079 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3081 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3082 not at the end of datarefs vector use ordered_remove to avoid
3083 reordering datarefs vector.
3086 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3087 mark local addressable non-static vars as GOVD_PRIVATE
3088 instead of GOVD_LOCAL.
3089 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3090 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3094 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3095 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3096 assume_aligned or alloc_align attributes.
3097 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3098 arguments. Handle also assume_aligned and alloc_align attributes.
3099 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3100 calls to functions with assume_aligned or alloc_align attributes.
3101 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3103 2014-02-08 Terry Guo <terry.guo@arm.com>
3105 * doc/invoke.texi: Document ARM -march=armv7e-m.
3107 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3109 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3110 flag on __cilkrts_rethrow builtin.
3113 * ipa-cp.c (determine_versionability): Fail at -O0
3114 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3115 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3118 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3121 * tree-inline.c (copy_forbidden): Fail for
3122 __attribute__((optimize (0))) functions.
3124 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3126 * varpool.c: Include pointer-set.h.
3127 (varpool_remove_unreferenced_decls): Variables in other partitions
3128 will not be output; be however careful to not lose information
3131 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3133 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3134 lookup in the vtable constructor.
3136 2014-02-07 Jeff Law <law@redhat.com>
3139 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3140 define_insn_and_split.
3142 * ipa-inline.c (inline_small_functions): Fix typos.
3144 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3146 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3147 (s390_can_use_return_insn): Declare.
3148 * config/s390/s390.h (EPILOGUE_USES): Define.
3149 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3151 (s390_chunkify_start): Handle return JUMP_LABELs.
3152 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3153 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3154 (s390_can_use_return_insn): New functions.
3155 (s390_fix_long_loop_prediction): Handle conditional returns.
3156 (TARGET_SET_UP_BY_PROLOGUE): Define.
3157 * config/s390/s390.md (ANY_RETURN): New code iterator.
3158 (*creturn, *csimple_return, return, simple_return): New patterns.
3160 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3162 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3163 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3164 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3165 REG_CFA_RESTORE list when deciding not to restore a register.
3167 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3169 * config/s390/s390.c: Include tree-pass.h and context.h.
3170 (s390_early_mach): New function, split out from...
3171 (s390_emit_prologue): ...here.
3172 (pass_data_s390_early_mach): New pass structure.
3173 (pass_s390_early_mach): New class.
3174 (s390_option_override): Create and register early_mach pass.
3175 Move to end of file.
3177 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3179 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3180 to match for the exit block.
3182 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3184 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3185 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3186 Reject misaligned operands.
3188 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3190 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3192 2014-02-07 Richard Biener <rguenther@suse.de>
3195 * gimple-low.c (lower_builtin_posix_memalign): New function.
3196 (lower_stmt): Call it to lower posix_memalign in a way
3197 to make alignment info accessible.
3199 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3202 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3203 __builtin_setjmp_receiver.
3205 2014-02-07 Richard Biener <rguenther@suse.de>
3208 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3209 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3210 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3211 Handle BUILT_IN_POSIX_MEMALIGN.
3212 (find_func_clobbers): Likewise.
3213 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3214 (call_may_clobber_ref_p_1): Likewise.
3216 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3219 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3222 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3225 * lto-cgraph.c (lto_output_node): Use
3226 symtab_get_symbol_partitioning_class.
3227 (lto_output_varpool_node): likewise.
3228 (symtab_get_symbol_partitioning_class): Move here from
3230 * cgraph.h (symbol_partitioning_class): Likewise.
3231 (symtab_get_symbol_partitioning_class): Declare.
3233 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3235 * ggc.h (ggc_internal_cleared_alloc): New macro.
3236 * vec.h (vec_safe_copy): Handle memory stats.
3237 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3238 * target-globals.c (save_target_globals): Likewise.
3240 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3243 * expr.c (emit_move_resolve_push): Export; be bit more selective
3244 on when to clear alias set.
3245 * expr.h (emit_move_resolve_push): Declare.
3246 * function.h (struct function): Add tail_call_marked.
3247 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3248 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3249 * config/i386/i386.md (TImode move expander): De not call
3251 (FP push expanders): Preserve memory attributes.
3252 * config/i386/sse.md (push<mode>1): Remove.
3253 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3254 (ix86_expand_push): Remove.
3255 * config/i386/mmx.md (push<mode>1): Remove.
3257 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3259 PR rtl-optimization/60030
3260 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3261 lopart with paradoxical subreg before shifting it up by hprec.
3263 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3265 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3266 Remove extra newline at end of file.
3267 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3268 (arm_issue_rate): Handle cortexa57.
3269 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3270 (cortex-a57.cortex-a53): Likewise.
3272 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3275 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3276 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3278 (arm_expand_prologue): Adjust all callers.
3279 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3280 info, registers also at the lowest numbered registers side. Use
3281 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3285 * var-tracking.c (adjust_mems): Before adding a SET to
3286 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3288 2014-02-06 Alan Modra <amodra@gmail.com>
3291 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3292 change SDmode to DDmode when lra_in_progress.
3294 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3297 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3298 free_data_ref on the dr first, and before goto again also set dr
3299 to the next dr. For simd_lane_access, free old datarefs[i] before
3300 overwriting it. For get_vectype_for_scalar_type failure, don't
3301 free_data_ref if simd_lane_access.
3303 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3306 * tree.h (opts_for_fn): New inline function.
3307 (opt_for_fn): Define.
3308 * config/i386/i386.c (ix86_function_regparm): Use
3309 opt_for_fn (decl, optimize) instead of optimize.
3311 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3313 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3314 for SYMBOL_REF in large memory model.
3316 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3318 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3320 (cortex-a57): Likewise.
3321 (cortex-a57.cortex-a53): Likewise.
3323 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3324 Kugan Vivekanandarajah <kuganv@linaro.org>
3326 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3328 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3330 2014-02-06 Richard Biener <rguenther@suse.de>
3332 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3333 set_loop_copy and initialize_original_copy_tables.
3335 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3337 * config/aarch64/aarch64-simd.md
3338 (aarch64_ashr_simddi): Change QI to SI.
3340 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3341 Jakub Jelinek <jakub@redhat.com>
3344 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3347 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3349 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3350 CODE_FOR_altivec_vpku[hw]um to
3351 CODE_FOR_altivec_vpku[hw]um_direct.
3352 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3353 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3354 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3355 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3357 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3359 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3360 generation for -maltivec=be.
3361 (altivec_vsumsws): Simplify redundant test.
3363 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3365 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3366 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3367 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3368 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3369 gen_altivec_vpkuwum.
3370 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3372 (altivec_vpks<VI_char>ss): Likewise.
3373 (altivec_vpks<VI_char>us): Likewise.
3374 (altivec_vpku<VI_char>us): Likewise.
3375 (altivec_vpku<VI_char>um): Likewise.
3376 (altivec_vpku<VI_char>um_direct): New (copy of
3377 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3379 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3380 target is little endian and -maltivec=be is not specified.
3381 (*altivec_vupkhs<VU_char>_direct): New (copy of
3382 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3383 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3384 target is little endian and -maltivec=be is not specified.
3385 (*altivec_vupkls<VU_char>_direct): New (copy of
3386 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3387 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3388 little endian and -maltivec=be is not specified.
3389 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3390 little endian and -maltivec=be is not specified.
3392 2014-02-05 Richard Henderson <rth@redhat.com>
3395 * combine-stack-adj.c: Revert r206943.
3396 * sched-int.h (struct deps_desc): Add last_args_size.
3397 * sched-deps.c (init_deps): Initialize it.
3398 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3399 contain REG_ARGS_SIZE notes.
3401 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3403 * lto-cgraph.c (asm_nodes_output): Make global.
3404 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3405 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3406 (driver_handle_option): Handle OPT_fwpa.
3408 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3411 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3412 a comment typo and formatting issue. If odr_hash hasn't been
3413 created, return vNULL and set *completep to false.
3416 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3417 bb with no successors.
3419 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3422 * doc/invoke.texi (-march): Clarify documentation for ARM.
3426 2014-02-05 Richard Biener <rguenther@suse.de>
3428 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3429 when not vectorizing because of too many alias checks.
3430 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3431 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3433 2014-02-05 Nick Clifton <nickc@redhat.com>
3435 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3436 accept extended registers in any mode when compiling for the MN10300.
3438 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3440 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3441 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3442 sanitization attributes.
3443 (can_inline_edge_p): Likewise.
3444 (sanitize_attrs_match_for_inline_p): New function.
3446 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3448 * ipa-prop.c (detect_type_change): Shor circuit testing of
3449 type changes on THIS pointer.
3451 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3454 * config/pa/pa.c (legitimize_tls_address): Return original address
3455 if not passed a SYMBOL_REF rtx.
3456 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3458 (pa_emit_move_sequence): Simplify TLS source operands.
3459 (pa_legitimate_constant_p): Reject all TLS constants.
3460 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3461 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3463 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3465 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3466 groups when we know they are controlled by LTO.
3467 * varasm.c (default_binds_local_p_1): If object is in other partition,
3468 it will be resolved locally.
3470 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3472 * config/host-linux.c (linux_gt_pch_use_address): Don't
3473 use SSIZE_MAX because it is not always defined.
3475 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3478 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3479 threshold for pseudo splitting.
3480 (update_ebb_live_info): Process call argument hard registers and
3481 hard registers from insn definition too.
3482 (max_small_class_regs_num): New constant.
3483 (inherit_in_ebb): Update live hard regs through EBBs. Update
3484 reloads_num only for small register classes. Don't split for
3487 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3490 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3493 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3495 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3498 2014-02-04 Marek Polacek <polacek@redhat.com>
3500 * gdbinit.in (pel): Define.
3502 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3504 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3507 2014-02-04 Richard Biener <rguenther@suse.de>
3510 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3511 in function context local.
3512 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3513 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3514 similar to LTO_imported_decl_ref.
3516 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3518 PR tree-optimization/60002
3519 * cgraphclones.c (build_function_decl_skip_args): Clear
3522 PR tree-optimization/60023
3523 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3524 false to gsi_replace.
3525 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3526 has been in some EH region and vec_stmt could throw, add
3527 vec_stmt into the same EH region.
3528 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3529 has no lhs, ignore it.
3530 * internal-fn.c (expand_MASK_LOAD): Likewise.
3533 * tree-inline.c (copy_forbidden): Fail for
3534 __attribute__((optimize (0))) functions.
3537 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3538 is set, copy one less argument.
3539 (expand_simd_clones): Don't subtract clone_info->inbranch
3540 from simd_clone_struct_alloc argument.
3542 PR rtl-optimization/57915
3543 * recog.c (simplify_while_replacing): If all unary/binary/relational
3544 operation arguments are constant, attempt to simplify those.
3547 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3548 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3550 2014-02-04 Richard Biener <rguenther@suse.de>
3552 PR tree-optimization/60012
3553 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3554 TBAA disambiguation to all DDRs.
3556 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3559 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3560 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3562 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3565 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3567 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3569 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3570 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3572 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3575 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3576 to figure out targets of polymorphic calls with known decl.
3577 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3578 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3579 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3580 (get_polymorphic_call_info): ... here.
3581 (get_polymorphic_call_info_from_invariant): New function.
3583 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3585 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3586 lookup via vtable pointer; check for type consistency
3587 and turn inconsitent facts into UNREACHABLE.
3588 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3589 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3590 type inconsistent querries; return UNREACHABLE instead.
3592 2014-02-03 Richard Henderson <rth@twiddle.net>
3595 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3596 already processed this node.
3597 (normalize_one_pred_1): Pass along mark_set.
3598 (normalize_one_pred): Create and destroy a pointer_set_t.
3599 (normalize_one_pred_chain): Likewise.
3601 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3603 PR gcov-profile/58602
3604 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3606 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3609 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3610 -fno-devirtualize; try to devirtualize by the knowledge of
3611 virtual table pointer given by aggregate propagation.
3612 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3613 (ipa_print_node_jump_functions): Dump also offset that
3614 is relevant for polymorphic calls.
3615 (determine_known_aggregate_parts): Add arg_type parameter; use it
3616 instead of determining the type from pointer type.
3617 (ipa_compute_jump_functions_for_edge): Update call of
3618 determine_known_aggregate_parts.
3619 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3620 (gimple_get_virt_method_for_binfo): ... here; simplify using
3621 vtable_pointer_value_to_vtable.
3622 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3623 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3624 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3625 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3627 (vtable_pointer_value_to_binfo): ... here.
3628 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3630 2014-02-03 Teresa Johnson <tejohnson@google.com>
3632 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3633 redef of outer loop index variable.
3635 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3639 * doc/extend.texi (Function Attributes): Typo.
3641 2014-02-03 Cong Hou <congh@google.com>
3643 PR tree-optimization/60000
3644 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3645 if the vectorized statement is a store. A store statement can only
3646 appear at the end of pattern statements.
3648 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3650 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3651 (ix86_option_override_internal): Default long double to 64-bit for
3652 32-bit Bionic and to 128-bit for 64-bit Bionic.
3654 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3655 TARGET_LONG_DOUBLE_128 is true.
3656 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3658 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3659 (mlong-double-64): Negate -mlong-double-128.
3660 (mlong-double-128): New option.
3662 * config/i386/i386-c.c (ix86_target_macros): Define
3663 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3665 * doc/invoke.texi: Document -mlong-double-128.
3667 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3669 PR rtl-optimization/60024
3670 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3672 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3674 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3676 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3678 PR rtl-optimization/57662
3679 * sel-sched.c (code_motion_path_driver): Do not mark already not
3680 existing blocks in the visiting bitmap.
3682 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3684 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3685 on the insn being emitted.
3687 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3688 Will Deacon <will.deacon@arm.com>
3690 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3692 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3694 * config/arm/arm-tables.opt: Regenerate.
3696 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3698 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3699 for vector types other than V16QImode.
3700 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3701 define_expand, and call altivec_expand_vec_perm_le when producing
3702 code with little endian element order.
3703 (*altivec_vperm_<mode>_internal): New insn having previous
3704 behavior of altivec_vperm_<mode>.
3705 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3706 altivec_expand_vec_perm_le when producing code with little endian
3708 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3709 behavior of altivec_vperm_<mode>_uns.
3711 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3713 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3714 (altivec_vsumsws): Add handling for -maltivec=be with a little
3716 (altivec_vsumsws_direct): New.
3717 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3718 gen_altivec_vsumsws.
3720 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3722 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3723 vtable_pointer_value_to_binfo): New functions.
3724 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3725 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3727 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3729 * config/nios2/nios2.md (load_got_register): Initialize GOT
3730 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3731 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3733 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3735 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3736 preserverd by passthrough, do not propagate the type.
3738 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3740 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3741 (mips_atomic_assign_expand_fenv): New function.
3742 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3744 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3746 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3747 (__builtin_mips_set_fcsr): Likewise.
3748 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3749 MIPS_USI_FTYPE_VOID.
3750 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3751 (mips16_expand_set_fcsr): Likewise.
3752 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3753 (mips16_set_fcsr_stub): Likewise.
3754 (mips16_get_fcsr_one_only_stub): New class.
3755 (mips16_set_fcsr_one_only_stub): Likewise.
3756 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3757 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3758 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3759 (hard_float): New availability predicate.
3760 (mips_builtins): Add get_fcsr and set_fcsr.
3761 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3762 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3763 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3764 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3765 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3768 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3770 * config/mips/mips.c (mips_one_only_stub): New class.
3771 (mips_need_mips16_rdhwr_p): Replace with...
3772 (mips16_rdhwr_stub): ...this new variable.
3773 (mips16_stub_call_address): New function.
3774 (mips16_rdhwr_one_only_stub): New class.
3775 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3776 (mips_output_mips16_rdhwr): Delete.
3777 (mips_finish_stub): New function.
3778 (mips_code_end): Use it to handle rdhwr stubs.
3780 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3783 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3784 when calculating size of integer atomic types.
3786 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3788 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3790 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3792 PR tree-optimization/60003
3793 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3794 * profile.c (branch_prob): Use gimple_call_builtin_p
3795 to check for BUILT_IN_SETJMP_RECEIVER.
3796 * tree-inline.c (copy_bb): Call notice_special_calls.
3798 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3801 * lra-constraints.c (process_alt_operands): Update reload_sum only
3804 2014-01-31 Richard Henderson <rth@redhat.com>
3807 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3808 until after else_eh is processed.
3810 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3812 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3813 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3814 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3815 in smmintrin.h, remove them.
3816 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3817 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3818 * config/i386/i386.md (ROUND_SAE): Fix value.
3819 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3820 (const48_operand): New.
3821 * config/i386/subst.md (round), (round_expand): Use
3822 const_4_or_8_to_11_operand.
3823 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3825 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3827 * config/i386/constraints.md (Yk): Swap meaning with k.
3828 * config/i386/i386.md (movhi_internal): Change Yk to k.
3829 (movqi_internal): Ditto.
3830 (*k<logic><mode>): Ditto.
3833 (kandn<mode>): Ditto.
3834 (*<code>hi_1): Ditto.
3835 (*<code>qi_1): Ditto.
3836 (kxnor<mode>): Ditto.
3837 (kortestzhi): Ditto.
3838 (kortestchi): Ditto.
3840 (*one_cmplhi2_1): Ditto.
3841 (*one_cmplqi2_1): Ditto.
3842 * config/i386/sse.md (): Change k to Yk.
3843 (avx512f_load<mode>_mask): Ditto.
3844 (avx512f_blendm<mode>): Ditto.
3845 (avx512f_store<mode>_mask): Ditto.
3846 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3847 (avx512f_storedqu<mode>_mask): Ditto.
3848 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3850 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3851 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3852 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3853 (avx512f_maskcmp<mode>3): Ditto.
3854 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3855 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3856 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3857 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3858 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3859 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3860 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3861 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3862 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3863 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3864 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3865 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3866 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3867 (vec_extract_lo_<mode>_maskm): Ditto.
3868 (vec_extract_hi_<mode>_maskm): Ditto.
3869 (avx512f_vternlog<mode>_mask): Ditto.
3870 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3871 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3872 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3873 (avx512f_<code>v8div16qi2_mask): Ditto.
3874 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3875 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3876 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3877 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3878 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3879 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3880 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3881 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3882 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3883 (avx512cd_maskb_vec_dupv8di): Ditto.
3884 (avx512cd_maskw_vec_dupv16si): Ditto.
3885 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3886 (avx512f_vpermi2var<mode>3_mask): Ditto.
3887 (avx512f_vpermi2var<mode>3_mask): Ditto.
3888 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3889 (*avx512f_gathersi<mode>): Ditto.
3890 (*avx512f_gathersi<mode>_2): Ditto.
3891 (*avx512f_gatherdi<mode>): Ditto.
3892 (*avx512f_gatherdi<mode>_2): Ditto.
3893 (*avx512f_scattersi<mode>): Ditto.
3894 (*avx512f_scatterdi<mode>): Ditto.
3895 (avx512f_compress<mode>_mask): Ditto.
3896 (avx512f_compressstore<mode>_mask): Ditto.
3897 (avx512f_expand<mode>_mask): Ditto.
3898 * config/i386/subst.md (mask): Change k to Yk.
3899 (mask_scalar_merge): Ditto.
3902 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3904 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3906 2014-01-31 Richard Biener <rguenther@suse.de>
3909 * builtins.c (fold_builtin_memory_op): Make sure to not
3910 use a floating-point mode or a boolean or enumeral type for
3913 2014-01-30 DJ Delorie <dj@redhat.com>
3915 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3916 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3917 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3918 whenever main() has an epilogue.
3920 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3922 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3923 unused variable "field".
3924 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3925 (vsx_mergeh_<mode>): Likewise.
3926 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3927 (altivec_vmrghh): Likewise.
3928 (altivec_vmrghw): Likewise.
3929 (altivec_vmrglb): Likewise.
3930 (altivec_vmrglh): Likewise.
3931 (altivec_vmrglw): Likewise.
3932 (altivec_vspltb): Add missing uses.
3933 (altivec_vsplth): Likewise.
3934 (altivec_vspltw): Likewise.
3935 (altivec_vspltsf): Likewise.
3937 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3940 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3941 frame related instructions.
3943 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3945 PR rtl-optimization/59959
3946 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3947 any reload of register whose subreg is invalid.
3949 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3951 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3952 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3953 Add missing return type - void.
3955 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3957 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3958 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3959 remove element index adjustment for endian (now handled in vsx.md
3961 (altivec_expand_vec_perm_const): Use
3962 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3963 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3964 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3965 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3966 define_expand and a new define_insn *altivec_vspltb_internal;
3967 adjust for -maltivec=be on a little endian target.
3968 (altivec_vspltb_direct): New.
3969 (altivec_vsplth): Divide into a define_expand and a new
3970 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3971 little endian target.
3972 (altivec_vsplth_direct): New.
3973 (altivec_vspltw): Divide into a define_expand and a new
3974 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3975 little endian target.
3976 (altivec_vspltw_direct): New.
3977 (altivec_vspltsf): Divide into a define_expand and a new
3978 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3979 a little endian target.
3981 2014-01-30 Richard Biener <rguenther@suse.de>
3983 PR tree-optimization/59993
3984 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3985 can propagate form the earlier stmt and avoid the transform
3986 when the intermediate result is needed.
3988 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3990 * README.Portability: Fix typo.
3992 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3994 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3995 comparison_operator with ordered_comparison_operator.
3997 2014-01-30 Nick Clifton <nickc@redhat.com>
3999 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
4000 Rename to mn10300_store_multiple_regs.
4001 * config/mn10300/mn10300.c: Likewise.
4002 * config/mn10300/mn10300.md (store_movm): Fix typo: call
4003 store_multiple_regs.
4004 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
4005 Call mn10300_store_multiple_regs.
4007 2014-01-30 Nick Clifton <nickc@redhat.com>
4008 DJ Delorie <dj@redhat.com>
4010 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
4011 %fp 2 to keep registers after it properly word-aligned.
4012 (rl78_alloc_physical_registers_umul): Handle the case where both
4013 input operands are the same.
4015 2014-01-30 Richard Biener <rguenther@suse.de>
4017 PR tree-optimization/59903
4018 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
4021 2014-01-30 Jason Merrill <jason@redhat.com>
4024 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
4027 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
4029 2014-01-30 Richard Biener <rguenther@suse.de>
4031 PR tree-optimization/59951
4032 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
4034 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
4037 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
4038 SFmode to DFmode case.
4040 2014-01-29 DJ Delorie <dj@redhat.com>
4042 * config/msp430/msp430.opt (-minrt): New.
4043 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
4045 (ENDFILE_SPEC): Likewise.
4047 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
4049 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
4050 (estimate_function_body_sizes): Use it.
4052 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
4055 * dwarf2out.c (is_cxx_auto): New.
4056 (is_base_type): Use it.
4057 (gen_type_die_with_usage): Likewise.
4059 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4061 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4062 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4063 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4064 -maltivec=be with LE targets.
4065 (vsx_mergeh_<mode>): Likewise.
4066 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4067 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4068 (altivec_vmrghb): Replace with define_expand and new
4069 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4070 (altivec_vmrghb_direct): New define_insn.
4071 (altivec_vmrghh): Replace with define_expand and new
4072 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4073 (altivec_vmrghh_direct): New define_insn.
4074 (altivec_vmrghw): Replace with define_expand and new
4075 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4076 (altivec_vmrghw_direct): New define_insn.
4077 (*altivec_vmrghsf): Adjust for endianness.
4078 (altivec_vmrglb): Replace with define_expand and new
4079 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4080 (altivec_vmrglb_direct): New define_insn.
4081 (altivec_vmrglh): Replace with define_expand and new
4082 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4083 (altivec_vmrglh_direct): New define_insn.
4084 (altivec_vmrglw): Replace with define_expand and new
4085 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4086 (altivec_vmrglw_direct): New define_insn.
4087 (*altivec_vmrglsf): Adjust for endianness.
4088 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4089 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4090 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4091 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4092 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4093 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4094 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4095 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4097 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4099 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4100 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4103 2014-01-29 Richard Biener <rguenther@suse.de>
4105 PR tree-optimization/58742
4106 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4107 associate_pointerplus_align.
4108 (associate_pointerplus_diff): New function.
4109 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4110 and associate_pointerplus_diff.
4112 2014-01-29 Richard Biener <rguenther@suse.de>
4114 * lto-streamer.h (LTO_major_version): Bump to 3.
4115 (LTO_minor_version): Reset to 0.
4117 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4119 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4120 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4121 (arm_file_start): Generate correct asm header for armv7ve.
4122 * config/arm/bpabi.h: Add multilib support for armv7ve.
4123 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4124 and cortex-a15 to armv7ve.
4125 * config/arm/t-aprofile: Add multilib support for armv7ve.
4126 * doc/invoke.texi: Document -march=armv7ve.
4128 2014-01-29 Richard Biener <rguenther@suse.de>
4130 PR tree-optimization/58742
4131 * tree-ssa-forwprop.c (associate_plusminus): Return true
4132 if we changed sth, defer EH cleanup to ...
4133 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4134 (simplify_mult): New function.
4136 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4139 PR tree-optimization/59920
4140 * tree.c (build_common_builtin_nodes): Remove
4141 __builtin_setjmp_dispatcher initialization.
4142 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4143 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4144 instead of gsi_after_labels + manually skipping debug stmts.
4145 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4146 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4147 * tree-inline.c (copy_edges_for_bb): Remove
4148 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4149 argument. Ignore computed_goto_p stmts. Don't call
4150 make_abnormal_goto_edges. If a call might need abnormal edges
4151 for non-local gotos, see if it already has an edge to
4152 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4153 with true argument, don't do anything then, otherwise add
4154 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4155 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4157 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4158 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4159 (lower_stmt): Don't set data->calls_builtin_setjmp.
4160 (lower_builtin_setjmp): Adjust comment.
4161 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4162 * tree-cfg.c (found_computed_goto): Remove.
4163 (factor_computed_gotos): Remove.
4164 (make_goto_expr_edges): Return bool, true for computed gotos.
4165 Don't call make_abnormal_goto_edges.
4166 (build_gimple_cfg): Don't set found_computed_goto, don't call
4167 factor_computed_gotos.
4168 (computed_goto_p): No longer static.
4169 (make_blocks): Don't set found_computed_goto.
4170 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4171 (make_edges): If make_goto_expr_edges returns true, push bb
4172 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4173 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4174 vector. Record mapping between bbs and OpenMP regions if there
4175 are any, adjust make_gimple_omp_edges caller. Call
4176 handle_abnormal_edges.
4177 (make_abnormal_goto_edges): Remove.
4178 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4179 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4180 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4181 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4182 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4183 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4184 filling *region also set *region_idx to (*region)->entry->index.
4187 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4188 For REGs set ORIGINAL_REGNO.
4190 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4192 * doc/md.texi: Mention that a target shouldn't implement
4193 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4196 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4198 PR tree-optimization/59594
4199 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4200 a copy of the datarefs vector rather than the vector itself.
4202 2014-01-28 Jason Merrill <jason@redhat.com>
4205 * dwarf2out.c (auto_die): New static.
4206 (gen_type_die_with_usage): Handle C++1y 'auto'.
4207 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4210 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4213 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4214 (SPEC_X32): Likewise.
4215 (SPEC_64): Likewise.
4216 * config/i386/i386.c (ix86_option_override_internal): Turn off
4217 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4219 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4220 * config/i386/i386.h (TARGET_16BIT): New macro.
4221 (TARGET_16BIT_P): Likewise.
4222 * config/i386/i386.opt: Add m16.
4223 * doc/invoke.texi: Document -m16.
4225 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4227 PR preprocessor/59935
4228 * input.c (location_get_source_line): Bail out on when line number
4229 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4231 2014-01-28 Richard Biener <rguenther@suse.de>
4233 PR tree-optimization/58742
4234 * tree-ssa-forwprop.c (associate_plusminus): Handle
4235 pointer subtraction of the form (T)(P + A) - (T)P.
4237 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4239 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4242 2014-01-28 Richard Biener <rguenther@suse.de>
4245 2014-01-28 Richard Biener <rguenther@suse.de>
4247 PR rtl-optimization/45364
4248 PR rtl-optimization/59890
4249 * var-tracking.c (local_get_addr_clear_given_value): Handle
4250 already cleared slot.
4251 (val_reset): Handle not allocated local_get_addr_cache.
4252 (vt_find_locations): Use post-order on the inverted CFG.
4254 2014-01-28 Richard Biener <rguenther@suse.de>
4256 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4258 2014-01-28 Richard Biener <rguenther@suse.de>
4260 PR rtl-optimization/45364
4261 PR rtl-optimization/59890
4262 * var-tracking.c (local_get_addr_clear_given_value): Handle
4263 already cleared slot.
4264 (val_reset): Handle not allocated local_get_addr_cache.
4265 (vt_find_locations): Use post-order on the inverted CFG.
4267 2014-01-28 Alan Modra <amodra@gmail.com>
4269 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4270 * configure.ac <recursive call for build != host>: Define
4271 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4272 and LD_FOR_BUILD too.
4273 * configure: Regenerate.
4275 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4277 * config/i386/i386.c (get_builtin_code_for_version): Separate
4278 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4279 Broadwell from Haswell.
4281 2014-01-27 Steve Ellcey <sellcey@mips.com>
4283 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4284 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4285 * config/mips/mips.c (mips_option_override): Change setting
4287 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4288 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4289 Change from Mask to Var.
4291 2014-01-27 Jeff Law <law@redhat.com>
4293 * ipa-inline.c (inline_small_functions): Fix typo.
4295 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4297 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4298 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4299 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4300 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4301 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4302 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4303 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4304 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4305 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4306 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4307 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4308 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4309 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4310 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4311 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4312 (_mm512_storeu_epi64): Ditto.
4313 (_mm512_cmpge_epi32_mask): Ditto.
4314 (_mm512_cmpge_epu32_mask): Ditto.
4315 (_mm512_cmpge_epi64_mask): Ditto.
4316 (_mm512_cmpge_epu64_mask): Ditto.
4317 (_mm512_cmple_epi32_mask): Ditto.
4318 (_mm512_cmple_epu32_mask): Ditto.
4319 (_mm512_cmple_epi64_mask): Ditto.
4320 (_mm512_cmple_epu64_mask): Ditto.
4321 (_mm512_cmplt_epi32_mask): Ditto.
4322 (_mm512_cmplt_epu32_mask): Ditto.
4323 (_mm512_cmplt_epi64_mask): Ditto.
4324 (_mm512_cmplt_epu64_mask): Ditto.
4325 (_mm512_cmpneq_epi32_mask): Ditto.
4326 (_mm512_cmpneq_epu32_mask): Ditto.
4327 (_mm512_cmpneq_epi64_mask): Ditto.
4328 (_mm512_cmpneq_epu64_mask): Ditto.
4329 (_mm512_expand_pd): Ditto.
4330 (_mm512_expand_ps): Ditto.
4331 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4332 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4333 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4334 * config/i386/i386.c (ix86_builtins): Add
4335 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4336 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4337 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4338 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4339 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4340 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4341 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4342 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4343 IX86_BUILTIN_PMOVUSQW512_MEM.
4344 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4345 __builtin_ia32_pmovsqd512mem_mask,
4346 __builtin_ia32_pmovqd512mem_mask,
4347 __builtin_ia32_pmovusqw512mem_mask,
4348 __builtin_ia32_pmovsqw512mem_mask,
4349 __builtin_ia32_pmovqw512mem_mask,
4350 __builtin_ia32_pmovusdw512mem_mask,
4351 __builtin_ia32_pmovsdw512mem_mask,
4352 __builtin_ia32_pmovdw512mem_mask,
4353 __builtin_ia32_pmovqb512mem_mask,
4354 __builtin_ia32_pmovusqb512mem_mask,
4355 __builtin_ia32_pmovsqb512mem_mask,
4356 __builtin_ia32_pmovusdb512mem_mask,
4357 __builtin_ia32_pmovsdb512mem_mask,
4358 __builtin_ia32_pmovdb512mem_mask.
4359 (bdesc_args): Add __builtin_ia32_expanddf512,
4360 __builtin_ia32_expandsf512.
4361 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4362 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4363 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4364 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4365 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4366 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4367 (avx512f_<code>v8div16qi2_mask_store): This.
4368 (avx512f_expand<mode>): New.
4370 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4372 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4374 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4375 (_mm512_prefetch_i32scatter_pd): Ditto.
4376 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4377 (_mm512_prefetch_i64scatter_pd): Ditto.
4378 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4379 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4380 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4381 (_mm512_prefetch_i32scatter_ps): Ditto.
4382 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4383 (_mm512_prefetch_i64scatter_ps): Ditto.
4384 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4385 * config/i386/i386-builtin-types.def: Define
4386 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4387 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4388 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4389 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4390 IX86_BUILTIN_SCATTERPFQPD.
4391 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4392 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4393 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4394 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4395 __builtin_ia32_scatterpfqps.
4396 (ix86_expand_builtin): Expand new built-ins.
4397 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4398 fix memory access data type.
4399 (*avx512pf_gatherpf<mode>_mask): Ditto.
4400 (*avx512pf_gatherpf<mode>): Ditto.
4401 (avx512pf_scatterpf<mode>): Ditto.
4402 (*avx512pf_scatterpf<mode>_mask): Ditto.
4403 (*avx512pf_scatterpf<mode>): Ditto.
4404 (GATHER_SCATTER_SF_MEM_MODE): New.
4405 (avx512pf_gatherpf<mode>df): Ditto.
4406 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4407 (*avx512pf_scatterpf<mode>df): Ditto.
4409 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4412 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4413 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4416 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4418 * common/config/arm/arm-common.c
4419 (arm_rewrite_mcpu): Handle multiple names.
4421 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4423 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4425 * gimple-builder.h (create_gimple_tmp): Delete.
4427 2014-01-27 Christian Bruel <christian.bruel@st.com>
4429 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4432 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4434 * config/pa/pa.md (call): Generate indirect long calls to non-local
4435 functions when outputing 32-bit code.
4436 (call_value): Likewise except for special call to buggy powf function.
4438 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4439 portable runtime and PIC indirect calls.
4440 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4441 and PIC call sequences. Use ldo instead of blr to set return register
4442 in PIC call sequence.
4444 2014-01-25 Walter Lee <walt@tilera.com>
4446 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4447 avoid clobbering a live register.
4449 2014-01-25 Walter Lee <walt@tilera.com>
4451 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4452 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4453 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4454 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4456 2014-01-25 Walter Lee <walt@tilera.com>
4458 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4459 arguments on even registers.
4460 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4462 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4463 (BIGGEST_ALIGNMENT): Ditto.
4464 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4466 2014-01-25 Walter Lee <walt@tilera.com>
4468 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4469 insns before bundling.
4470 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4472 2014-01-25 Walter Lee <walt@tilera.com>
4474 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4475 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4476 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4478 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4480 * config/mips/constraints.md (kl): Delete.
4481 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4482 define expands, using...
4483 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4484 instructions for MIPS16.
4485 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4486 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4488 2014-01-25 Walter Lee <walt@tilera.com>
4490 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4494 2014-01-25 Walter Lee <walt@tilera.com>
4496 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4497 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4499 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4501 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4504 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4506 * print-rtl.c (in_call_function_usage): New var.
4507 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4508 EXPR_LIST mode as mode and not as reg note name.
4511 * cfgloopmanip.c (copy_loop_info): If
4512 loop->warned_aggressive_loop_optimizations, make sure
4513 the flag is set in target loop too.
4515 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4517 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4519 * builtins.def: Likewise.
4520 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4521 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4522 * ira.c (ira_setup_eliminable_regset): Likewise.
4523 * omp-low.c (gate_expand_omp): Likewise.
4524 (execute_lower_omp): Likewise.
4525 (diagnose_sb_0): Likewise.
4526 (gate_diagnose_omp_blocks): Likewise.
4527 (simd_clone_clauses_extract): Likewise.
4530 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4532 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4533 correction for little endian...
4534 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4537 2014-01-24 Jeff Law <law@redhat.com>
4539 PR tree-optimization/59919
4540 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4541 for non-returning calls.
4543 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4545 * common/config/aarch64/aarch64-common.c
4546 (aarch64_rewrite_mcpu): Handle multiple names.
4547 * config/aarch64/aarch64.h
4548 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4550 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4552 * input.c (add_file_to_cache_tab): Handle the case where fopen
4555 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4558 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4559 from push operand if code of push isn't PRE_DEC.
4561 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4564 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4565 -mquad-memory-atomic. Update -mquad-memory documentation to say
4566 it is only used for non-atomic loads/stores.
4568 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4569 -mquad-memory or -mquad-memory-atomic switches.
4571 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4572 -mquad-memory-atomic to ISA 2.07 support.
4574 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4575 to separate support of normal quad word memory operations (ldq, stq)
4576 from the atomic quad word memory operations.
4578 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4579 support to separate non-atomic quad word operations from atomic
4580 quad word operations. Disable non-atomic quad word operations in
4581 little endian mode so that we don't have to swap words after the
4582 load and before the store.
4583 (quad_load_store_p): Add comment about atomic quad word support.
4584 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4585 options printed with -mdebug=reg.
4587 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4588 -mquad-memory-atomic as the test for whether we have quad word
4589 atomic instructions.
4590 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4591 or -mp8-vector are used, allow byte/half-word atomic operations.
4593 * config/rs6000/sync.md (load_lockedti): Insure that the address
4594 is a proper indexed or indirect address for the lqarx instruction.
4595 On little endian systems, swap the hi/lo registers after the lqarx
4597 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4598 insure the address is valid for the lqarx instruction.
4599 (store_conditionalti): Insure that the address is a proper indexed
4600 or indirect address for the stqcrx. instruction. On little endian
4601 systems, swap the hi/lo registers before doing the stqcrx.
4603 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4604 insure the address is valid for the stqcrx. instruction.
4606 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4607 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4608 type of quad memory support is available.
4610 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4613 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4614 there is a danger of looping.
4616 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4618 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4619 force flag_ira_loop_pressure if set via command line.
4621 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4623 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4624 (ashr_simd): New builtin handling DI mode.
4625 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4626 (aarch64_sshr_simddi): New match pattern.
4627 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4628 (vshrd_n_s64): Likewise.
4629 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4631 2014-01-23 Nick Clifton <nickc@redhat.com>
4633 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4634 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4635 favour of mcu specific scripts.
4636 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4639 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4640 Alex Velenko <Alex.Velenko@arm.com>
4642 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4643 (vaddv_s16): Likewise.
4644 (vaddv_s32): Likewise.
4645 (vaddv_u8): Likewise.
4646 (vaddv_u16): Likewise.
4647 (vaddv_u32): Likewise.
4648 (vaddvq_s8): Likewise.
4649 (vaddvq_s16): Likewise.
4650 (vaddvq_s32): Likewise.
4651 (vaddvq_s64): Likewise.
4652 (vaddvq_u8): Likewise.
4653 (vaddvq_u16): Likewise.
4654 (vaddvq_u32): Likewise.
4655 (vaddvq_u64): Likewise.
4656 (vaddv_f32): Likewise.
4657 (vaddvq_f32): Likewise.
4658 (vaddvq_f64): Likewise.
4659 (vmaxv_f32): Likewise.
4660 (vmaxv_s8): Likewise.
4661 (vmaxv_s16): Likewise.
4662 (vmaxv_s32): Likewise.
4663 (vmaxv_u8): Likewise.
4664 (vmaxv_u16): Likewise.
4665 (vmaxv_u32): Likewise.
4666 (vmaxvq_f32): Likewise.
4667 (vmaxvq_f64): Likewise.
4668 (vmaxvq_s8): Likewise.
4669 (vmaxvq_s16): Likewise.
4670 (vmaxvq_s32): Likewise.
4671 (vmaxvq_u8): Likewise.
4672 (vmaxvq_u16): Likewise.
4673 (vmaxvq_u32): Likewise.
4674 (vmaxnmv_f32): Likewise.
4675 (vmaxnmvq_f32): Likewise.
4676 (vmaxnmvq_f64): Likewise.
4677 (vminv_f32): Likewise.
4678 (vminv_s8): Likewise.
4679 (vminv_s16): Likewise.
4680 (vminv_s32): Likewise.
4681 (vminv_u8): Likewise.
4682 (vminv_u16): Likewise.
4683 (vminv_u32): Likewise.
4684 (vminvq_f32): Likewise.
4685 (vminvq_f64): Likewise.
4686 (vminvq_s8): Likewise.
4687 (vminvq_s16): Likewise.
4688 (vminvq_s32): Likewise.
4689 (vminvq_u8): Likewise.
4690 (vminvq_u16): Likewise.
4691 (vminvq_u32): Likewise.
4692 (vminnmv_f32): Likewise.
4693 (vminnmvq_f32): Likewise.
4694 (vminnmvq_f64): Likewise.
4696 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4698 * config/aarch64/aarch64-simd.md
4699 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4700 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4701 (*aarch64_mul3_elt<mode>): Likewise.
4702 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4703 (*aarch64_mul3_elt_to_64v2df): Likewise.
4704 (*aarch64_mla_elt<mode>): Likewise.
4705 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4706 (*aarch64_mls_elt<mode>): Likewise.
4707 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4708 (*aarch64_fma4_elt<mode>): Likewise.
4709 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4710 (*aarch64_fma4_elt_to_64v2df): Likewise.
4711 (*aarch64_fnma4_elt<mode>): Likewise.
4712 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4713 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4714 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4715 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4716 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4717 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4718 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4719 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4720 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4722 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4724 * config/aarch64/aarch64-simd.md
4725 (aarch64_be_checked_get_lane<mode>): New define_expand.
4726 * config/aarch64/aarch64-simd-builtins.def
4727 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4728 New builtin definition.
4729 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4730 Use new safe be builtin.
4732 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4734 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4736 (aarch64_be_st1<mode>): Likewise.
4737 (aarch_ld1<VALL:mode>): Define_expand modified.
4738 (aarch_st1<VALL:mode>): Likewise.
4739 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4740 (UNSPEC_ST1): Likewise.
4742 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4744 * config/microblaze/microblaze.md: Add trap insn and attribute
4746 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4748 PR preprocessor/58580
4749 * input.h (location_get_source_line): Take an additional line_size
4751 (void diagnostics_file_cache_fini): Declare new function.
4752 * input.c (struct fcache): New type.
4753 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4754 New static constants.
4755 (diagnostic_file_cache_init, total_lines_num)
4756 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4757 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4758 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4759 (get_next_line, read_next_line, goto_next_line, read_line_num):
4760 New static function definitions.
4761 (diagnostic_file_cache_fini): New function.
4762 (location_get_source_line): Take an additional output line_len
4763 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4765 * diagnostic.c (diagnostic_finish): Call
4766 diagnostic_file_cache_fini.
4767 (adjust_line): Take an additional input parameter for the length
4768 of the line, rather than calculating it with strlen.
4769 (diagnostic_show_locus): Adjust the use of
4770 location_get_source_line and adjust_line with respect to their new
4771 signature. While displaying a line now, do not stop at the first
4772 null byte. Rather, display the zero byte as a space and keep
4773 going until we reach the size of the line.
4774 * Makefile.in: Add vec.o to OBJS-libcommon
4776 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4777 Ilya Tocar <ilya.tocar@intel.com>
4779 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4780 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4781 (__builtin_ia32_kmov16): Ditto.
4782 * config/i386/i386.md (UNSPEC_KMOV): New.
4785 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4787 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4788 (_mm512_storeu_si512): Ditto.
4790 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4793 * rtl.h (get_referenced_operands): Declare.
4794 * recog.c (get_referenced_operands): New function.
4795 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4796 operands have been referenced when recording LO_SUM references.
4798 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4800 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4802 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4804 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4805 Enable for generic and recent AMD targets.
4807 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4809 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4810 ARG_SIZE note when adjustment was eliminated.
4812 2014-01-22 Jeff Law <law@redhat.com>
4814 PR tree-optimization/59597
4815 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4816 in file. Accept new argument REGISTERING and use it to modify
4817 dump output appropriately.
4818 (register_jump_thread): Corresponding changes.
4819 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4820 thread paths involving joiner blocks. Add code to dump cancelled
4821 jump threading paths.
4823 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4825 PR rtl-optimization/59477
4826 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4827 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4829 2014-01-22 Tom Tromey <tromey@redhat.com>
4831 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4833 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4835 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4837 PR rtl-optimization/59896
4838 * lra-constraints.c (process_alt_operands): Check unused note for
4839 matched operands of insn with no output reloads.
4841 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4843 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4844 (mips_move_from_gpr_cost): Likewise.
4846 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4848 PR rtl-optimization/59858
4849 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4850 ira_class_hard_regs_num.
4851 (process_alt_operands): Increase reject for dying matched operand.
4853 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4856 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4857 smaller than size, perform several stores or loads and stores
4858 at dst + count - size to store or copy all of size bytes, rather
4859 than just last modesize bytes.
4861 2014-01-20 DJ Delorie <dj@redhat.com>
4863 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4864 that CLOBBERs are REGs before propogating their values.
4866 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4869 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4870 (cgraph_inline_failed_type): New function.
4871 * cgraph.h (DEFCIFCODE): Add type.
4872 (cgraph_inline_failed_type_t): New enum.
4873 (cgraph_inline_failed_type): New prototype.
4874 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4875 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4876 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4877 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4878 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4879 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4880 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4881 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4882 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4883 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4884 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4885 OPTIMIZATION_MISMATCH.
4886 * tree-inline.c (expand_call_inline): Emit errors during
4887 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4889 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4892 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4893 mode attribute in insn output.
4895 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4897 * output.h (output_constant): Delete.
4898 * varasm.c (output_constant): Make private.
4900 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4902 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4904 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4907 * tree.h (fold_builtin_strcat): New prototype.
4908 * builtins.c (fold_builtin_strcat): No longer static. Add len
4909 argument, if non-NULL, don't call c_strlen. Optimize
4910 directly into __builtin_memcpy instead of __builtin_strcpy.
4911 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4912 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4914 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4916 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4917 for SImode_address_operand operands, having only a REG argument.
4919 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4921 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4922 loader name using mbig-endian.
4923 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4925 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4927 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4931 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4933 * config/aarch64/aarch64-protos.h
4934 (aarch64_cannot_change_mode_class_ptr): Declare.
4935 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4936 aarch64_cannot_change_mode_class_ptr): New.
4937 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4938 backend hook aarch64_cannot_change_mode_class.
4940 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4942 * common/config/aarch64/aarch64-common.c
4943 (aarch64_handle_option): Don't handle any option order logic here.
4944 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4945 selected_cpu, warn on architecture version mismatch.
4946 (aarch64_override_options): Fix parsing order for option strings.
4948 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4949 Iain Sandoe <iain@codesourcery.com>
4952 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4953 warning. Amend comment to reflect current functionality.
4955 2014-01-20 Richard Biener <rguenther@suse.de>
4958 * builtins.c (fold_builtin_strcat): Remove case better handled
4959 by tree-ssa-strlen.c.
4961 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4963 * config/aarch64/aarch64.opt
4964 (mcpu, march, mtune): Make case-insensitive.
4966 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4969 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4970 if operands[1] is a REG or ZERO_EXTEND of a REG.
4972 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4974 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4976 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4978 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4979 long non-pic millicode calls.
4981 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4983 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4985 2014-01-19 Kito Cheng <kito@0xlab.org>
4987 * builtins.c (expand_movstr): Check movstr expand done or fail.
4989 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4990 H.J. Lu <hongjiu.lu@intel.com>
4993 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4994 to DImode for zero-extended addresses.
4996 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4998 PR rtl-optimization/57763
4999 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
5000 on the new indirect jump_insn and increment LABEL_NUSES (label).
5002 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
5006 * config.gcc (x86_archs): New variable.
5007 (x86_64_archs): Likewise.
5008 (x86_cpus): Likewise.
5009 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
5010 --with-arch/--with-cpu= options.
5011 Support --with-arch=/--with-cpu={nehalem,westmere,
5012 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
5014 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5016 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
5017 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
5019 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
5021 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
5023 2014-01-18 Jakub Jelinek <jakub@redhat.com>
5026 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
5027 clear cpp_get_options (parse_in)->warn_unused_macros for
5028 ix86_target_macros_internal with cpp_define.
5030 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
5032 * jump.c (delete_related_insns): Keep (use (insn))s.
5033 * reorg.c (redundant_insn): Check for barriers too.
5035 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5037 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
5039 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
5041 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
5042 call to $$dyncall when TARGET_LONG_CALLS is true.
5044 2014-01-17 Jeff Law <law@redhat.com>
5046 * ree.c (combine_set_extension): Temporarily disable test for
5047 changing number of hard registers.
5049 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5052 * ipa-inline-analysis.c (inline_free_summary):
5053 Do not free summary of aliases.
5055 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5058 * gimplify.c (gimplify_expr): Use create_tmp_var
5059 instead of create_tmp_var_raw. If cond doesn't have
5060 integral type, don't add the IFN_ANNOTATE builtin at all.
5062 2014-01-17 Martin Jambor <mjambor@suse.cz>
5065 * ipa-cp.c (prev_edge_clone): New variable.
5066 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5067 Also resize prev_edge_clone vector.
5068 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5069 (ipcp_edge_removal_hook): New function.
5070 (ipcp_driver): Register ipcp_edge_removal_hook.
5072 2014-01-17 Andrew Pinski <apinski@cavium.com>
5073 Steve Ellcey <sellcey@mips.com>
5076 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5079 2014-01-17 Jeff Law <law@redhat.com>
5082 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5083 so that pass_ccp runs first.
5085 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5087 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5088 (ix86_adjust_cost): Use !TARGET_XXX.
5089 (do_reorder_for_imul): Likewise.
5090 (swap_top_of_ready_list): Likewise.
5091 (ix86_sched_reorder): Likewise.
5093 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5095 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5096 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5097 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5098 (intel_memset): New. Duplicate slm_memset.
5099 (intel_cost): New. Duplicate slm_cost.
5100 (m_INTEL): New macro.
5101 (processor_target_table): Add "intel".
5102 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5103 with PROCESSOR_INTEL for "intel".
5104 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5105 PROCESSOR_SILVERMONT.
5106 (ix86_issue_rate): Likewise.
5107 (ix86_adjust_cost): Likewise.
5108 (ia32_multipass_dfa_lookahead): Likewise.
5109 (swap_top_of_ready_list): Likewise.
5110 (ix86_sched_reorder): Likewise.
5111 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5112 instead of TARGET_OPT_AGU.
5113 * config/i386/i386.h (TARGET_INTEL): New.
5114 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5115 (processor_type): Add PROCESSOR_INTEL.
5116 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5117 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5119 2014-01-17 Marek Polacek <polacek@redhat.com>
5122 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5125 2014-01-17 Richard Biener <rguenther@suse.de>
5127 PR tree-optimization/46590
5128 * opts.c (default_options_table): Add entries for
5129 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5130 all enabled at -O1 but not for -Og.
5131 * common.opt (fbranch-count-reg): Remove Init(1).
5132 (fmove-loop-invariants): Likewise.
5133 (ftree-pta): Likewise.
5135 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5137 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5138 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5139 decls to at least the GCC 4.8 used alignments.
5142 * tree-nested.c (convert_nonlocal_reference_stmt,
5143 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5144 of GIMPLE_BIND stmts, adjust associated decls.
5146 2014-01-17 Richard Biener <rguenther@suse.de>
5148 PR tree-optimization/46590
5149 * vec.h (vec<>::bseach): New member function implementing
5150 binary search according to C89 bsearch.
5151 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5152 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5153 bitmap pointer again. Make accesses_in_loop a flat array.
5154 (mem_ref_obstack): New global.
5155 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5156 (mark_ref_stored): Likewise.
5157 (ref_indep_loop_p_2): Likewise.
5158 (set_ref_stored_in_loop): New helper function.
5159 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5160 (memref_free): Adjust.
5161 (record_mem_ref_loc): Simplify.
5162 (gather_mem_refs_stmt): Adjust.
5163 (sort_locs_in_loop_postorder_cmp): New function.
5164 (analyze_memory_references): Sort accesses_in_loop after
5165 loop postorder number.
5166 (find_ref_loc_in_loop_cmp): New function.
5167 (for_all_locs_in_loop): Find relevant cluster of locs in
5168 accesses_in_loop and iterate without recursion.
5169 (execute_sm): Avoid uninit warning.
5170 (struct ref_always_accessed): Simplify.
5171 (ref_always_accessed::operator ()): Likewise.
5172 (ref_always_accessed_p): Likewise.
5173 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5174 loop postorder numbers here.
5175 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5178 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5181 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5182 on decls for which assemble_alias has been called.
5184 2014-01-17 Nick Clifton <nickc@redhat.com>
5186 * config/msp430/msp430.opt: (mcpu): New option.
5187 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5188 (msp430_option_override): Parse target_cpu. If the MCU name
5189 matches a generic string, clear target_mcu.
5190 (msp430_attr): Allow numeric interrupt values up to 63.
5191 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5192 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5194 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5196 * config/msp430/msp430.md (popm): Use %J rather than %I.
5197 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5198 (addhi_cy_i): Use immediate_operand for operand 2.
5199 * doc/invoke.texi: Document -mcpu option.
5201 2014-01-17 Richard Biener <rguenther@suse.de>
5203 PR rtl-optimization/38518
5204 * df.h (df_analyze_loop): Declare.
5205 * df-core.c: Include cfgloop.h.
5206 (df_analyze_1): Split out main part of df_analyze.
5207 (df_analyze): Adjust.
5208 (loop_inverted_post_order_compute): New function.
5209 (loop_post_order_compute): Likewise.
5210 (df_analyze_loop): New function avoiding whole-function
5212 * loop-invariant.c (find_defs): Use df_analyze_loop.
5213 (find_invariants): Adjust.
5214 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5216 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5218 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5219 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5221 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5223 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5224 traversal when removing references.
5226 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5229 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5231 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5234 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5235 pushing a reload for an autoinc when we had previously reloaded an
5236 inner part of the address.
5238 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5240 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5242 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5243 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5244 when not giving up or versioning for alias only because of
5246 (vect_analyze_data_ref_dependences): Set to true.
5247 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5249 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5250 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5254 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5257 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5258 operand 0 predicate for gathers, use a new pseudo as subtarget.
5260 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5263 * lra-constraints.c (process_alt_operands): Add printing debug info.
5264 Check absence of input/output reloads for matched operands too.
5266 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5268 PR rtl-optimization/59835
5269 * ira.c (ira_init_register_move_cost): Increase cost for
5272 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5274 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5276 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5279 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5280 non-register objects. Use gen_(high/low)part more consistently.
5283 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5286 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5287 endian support, remove tests for WORDS_BIG_ENDIAN.
5288 (p8_mfvsrd_3_<mode>): Likewise.
5289 (reload_gpr_from_vsx<mode>): Likewise.
5290 (reload_gpr_from_vsxsf): Likewise.
5291 (p8_mfvsrd_4_disf): Likewise.
5293 2014-01-16 Richard Biener <rguenther@suse.de>
5295 PR rtl-optimization/46590
5296 * lcm.c (compute_antinout_edge): Use postorder iteration.
5297 (compute_laterin): Use inverted postorder iteration.
5299 2014-01-16 Nick Clifton <nickc@redhat.com>
5302 * varasm.c (output_constant): Return the number of bytes actually
5304 (output_constructor_array_range): Update the field size with the
5305 number of bytes emitted by output_constant.
5306 (output_constructor_regular_field): Likewise. Also do not
5307 complain if the total number of bytes emitted is now greater
5308 than the expected fieldpos.
5309 * output.h (output_constant): Update prototype and descriptive comment.
5311 2014-01-16 Marek Polacek <polacek@redhat.com>
5314 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5315 it is error_mark_node.
5317 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5319 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5320 VALID_AVX256_REG_OR_OI_MODE.
5322 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5324 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5325 current procedure should be profiled.
5327 2014-01-15 Andrew Pinski <apinski@cavium.com>
5329 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5330 of moving from/to the STACK_REG register class.
5332 2014-01-15 Richard Henderson <rth@redhat.com>
5335 * reginfo.c (global_regs_decl): Globalize.
5336 * rtl.h (global_regs_decl): Declare.
5337 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5338 reserved via global_regs.
5340 2014-01-15 Teresa Johnson <tejohnson@google.com>
5342 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5344 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5346 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5347 and vmulosh rather than call gen_vec_widen_smult_*.
5348 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5349 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5350 (vec_widen_smult_even_v16qi): Likewise.
5351 (vec_widen_umult_even_v8hi): Likewise.
5352 (vec_widen_smult_even_v8hi): Likewise.
5353 (vec_widen_umult_odd_v16qi): Likewise.
5354 (vec_widen_smult_odd_v16qi): Likewise.
5355 (vec_widen_umult_odd_v8hi): Likewise.
5356 (vec_widen_smult_odd_v8hi): Likewise.
5357 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5358 vmuloub rather than call gen_vec_widen_umult_*.
5359 (vec_widen_umult_lo_v16qi): Likewise.
5360 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5361 vmulosb rather than call gen_vec_widen_smult_*.
5362 (vec_widen_smult_lo_v16qi): Likewise.
5363 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5364 rather than call gen_vec_widen_umult_*.
5365 (vec_widen_umult_lo_v8hi): Likewise.
5366 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5367 rather than call gen_vec_widen_smult_*.
5368 (vec_widen_smult_lo_v8hi): Likewise.
5370 2014-01-15 Jeff Law <law@redhat.com>
5372 PR tree-optimization/59747
5373 * ree.c (find_and_remove_re): Properly handle case where a second
5374 eliminated extension requires widening a copy created for elimination
5375 of a prior extension.
5376 (combine_set_extension): Ensure that the number of hard regs needed
5377 for a destination register does not change when we widen it.
5379 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5381 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5382 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5383 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5384 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5385 (avr-*-rtems*): Likewise.
5386 (bfin*-rtems*): Likewise.
5387 (moxie-*-rtems*): Likewise.
5388 (h8300-*-rtems*): Likewise.
5389 (i[34567]86-*-rtems*): Likewise.
5390 (lm32-*-rtems*): Likewise.
5391 (m32r-*-rtems*): Likewise.
5392 (m68k-*-rtems*): Likewise.
5393 (microblaze*-*-rtems*): Likewise.
5394 (mips*-*-rtems*): Likewise.
5395 (powerpc-*-rtems*): Likewise.
5396 (sh-*-rtems*): Likewise.
5397 (sparc-*-rtems*): Likewise.
5398 (sparc64-*-rtems*): Likewise.
5399 (v850-*-rtems*): Likewise.
5400 (m32c-*-rtems*): Likewise.
5402 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5404 PR rtl-optimization/59511
5405 * ira.c (ira_init_register_move_cost): Use memory costs for some
5406 cases of register move cost calculations.
5407 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5408 instead of BB frequency.
5409 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5410 * lra-assigns.c (find_hard_regno_for): Ditto.
5412 2014-01-15 Richard Biener <rguenther@suse.de>
5414 PR tree-optimization/59822
5415 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5416 (vectorizable_load): Use it to hoist defs of uses of invariant
5417 loads out of the loop.
5419 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5420 Kugan Vivekanandarajah <kuganv@linaro.org>
5423 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5426 2014-01-15 Richard Biener <rguenther@suse.de>
5428 PR rtl-optimization/59802
5429 * lcm.c (compute_available): Use inverted postorder to seed
5430 the initial worklist.
5432 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5435 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5436 ADDR_REGS for invalid symrefs in non-PIC code.
5438 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5441 * builtins.c (determine_block_size): Initialize *probable_max_size
5442 even if len_rtx is CONST_INT.
5444 2014-01-14 Andrew Pinski <apinski@cavium.com>
5446 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5447 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5448 (cortexa53_tunings): Likewise.
5449 (aarch64_sched_issue_rate): New function.
5450 (TARGET_SCHED_ISSUE_RATE): Define.
5452 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5454 * ira-costs.c (find_costs_and_classes): Add missed
5455 ira_init_register_move_cost_if_necessary.
5457 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5460 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5462 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5465 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5466 to indicate if type is used for function return value. Warn ABI
5467 change if the vector mode isn't available for function return value.
5468 (ix86_function_arg_advance): Pass false to type_natural_mode.
5469 (ix86_function_arg): Likewise.
5470 (ix86_gimplify_va_arg): Likewise.
5471 (function_arg_32): Don't warn ABI change.
5472 (ix86_function_value): Pass true to type_natural_mode.
5473 (ix86_return_in_memory): Likewise.
5474 (ix86_struct_value_rtx): Removed.
5475 (TARGET_STRUCT_VALUE_RTX): Likewise.
5477 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5479 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5480 converting a conditional jump into a conditional return.
5482 2014-01-14 Richard Biener <rguenther@suse.de>
5484 PR tree-optimization/58921
5485 PR tree-optimization/59006
5486 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5487 hoisting invariant stmts.
5488 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5489 invariant loads on the preheader edge if possible.
5491 2014-01-14 Joey Ye <joey.ye@arm.com>
5493 * doc/plugin.texi (Building GCC plugins): Update to C++.
5495 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5497 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5498 (_mm_rcp28_round_ss): Ditto.
5499 (_mm_rsqrt28_round_sd): Ditto.
5500 (_mm_rsqrt28_round_ss): Ditto.
5501 (_mm_rcp28_sd): Ditto.
5502 (_mm_rcp28_ss): Ditto.
5503 (_mm_rsqrt28_sd): Ditto.
5504 (_mm_rsqrt28_ss): Ditto.
5505 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5506 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5507 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5508 (IX86_BUILTIN_RCP28SD): Ditto.
5509 (IX86_BUILTIN_RCP28SS): Ditto.
5510 (IX86_BUILTIN_RSQRT28SD): Ditto.
5511 (IX86_BUILTIN_RSQRT28SS): Ditto.
5512 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5513 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5514 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5515 (ix86_expand_special_args_builtin): Expand new FTYPE.
5516 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5517 (srcp14<mode>): Make insn unary.
5518 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5519 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5520 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5521 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5522 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5523 Fix rounding: make it SAE only.
5524 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5526 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5528 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5529 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5530 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5531 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5532 (round_saeonly_mask_scalar_operand4): Ditto.
5533 (round_saeonly_mask_scalar_op3): Ditto.
5534 (round_saeonly_mask_scalar_op4): Ditto.
5536 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5538 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5539 Implement -maltivec=be for vec_insert and vec_extract.
5541 2014-01-10 DJ Delorie <dj@redhat.com>
5543 * config/msp430/msp430.md (call_internal): Don't allow memory
5544 references with SP as the base register.
5545 (call_value_internal): Likewise.
5546 * config/msp430/constraints.md (Yc): New. For memory references
5547 that don't use SP as a base register.
5549 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5550 "an integer without a # prefix"
5551 * config/msp430/msp430.md (epilogue_helper): Use it.
5553 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5556 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5557 AVX512F gather builtins.
5558 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5559 on gather decls with INTEGER_TYPE masktype.
5560 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5561 directly into the builtin rather than hoisting it before loop.
5563 PR tree-optimization/59387
5564 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5565 (scev_const_prop): If folded_casts and type has undefined overflow,
5566 use force_gimple_operand instead of force_gimple_operand_gsi and
5567 for each added stmt if it is assign with
5568 arith_code_with_undefined_signed_overflow, call
5569 rewrite_to_defined_overflow.
5570 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5571 gimple-fold.h instead.
5572 (arith_code_with_undefined_signed_overflow,
5573 rewrite_to_defined_overflow): Moved to ...
5574 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5575 rewrite_to_defined_overflow): ... here. No longer static.
5576 Include gimplify-me.h.
5577 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5578 rewrite_to_defined_overflow): New prototypes.
5580 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5582 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5584 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5586 * builtins.c (get_object_alignment_2): Minor tweak.
5587 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5589 2014-01-13 Christian Bruel <christian.bruel@st.com>
5591 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5592 optimized non constant lengths.
5594 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5597 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5598 load as __atomic_load_N if possible.
5600 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5602 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5604 (rs6000_expand_builtin): Adjust call.
5606 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5609 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5610 * config/rs6000/rs6000.c: Include target-globals.h.
5611 (rs6000_set_current_function): Instead of doing target_reinit
5612 unconditionally, use save_target_globals_default_opts and
5613 restore_target_globals.
5615 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5617 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5618 (rs6000_expand_builtin): Handle mffs and mtfsf.
5619 (rs6000_init_builtins): Define mffs and mtfsf.
5620 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5621 (rs6000_mffs): New pattern.
5622 (rs6000_mtfsf): New pattern.
5624 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5626 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5627 Start narrowing with START. Apply candidate-use pair
5628 and check overall cost in narrowing.
5629 (iv_ca_prune): Pass new argument.
5631 2014-01-10 Jeff Law <law@redhat.com>
5634 * ree.c (combine_reaching_defs): Ensure the defining statement
5635 occurs before the extension when optimizing extensions with
5636 different source and destination hard registers.
5638 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5641 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5642 vtables into the type inheritance graph.
5644 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5646 PR rtl-optimization/59754
5647 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5648 modes in the REGNO != REGNO case.
5650 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5652 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5654 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5656 PR tree-optimization/59745
5657 * tree-predcom.c (tree_predictive_commoning_loop): Call
5658 free_affine_expand_cache if giving up because components is NULL.
5660 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5661 GC in payload of target_globals struct instead of allocating them on
5662 the heap and the larger structs separately using GC.
5663 * target-globals.h (struct target_globals): Make regs, hard_regs,
5664 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5665 of GTY((skip)) and change type to void *.
5666 (reset_target_globals): Cast loads from those fields to corresponding
5669 2014-01-10 Steve Ellcey <sellcey@mips.com>
5672 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5673 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5674 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5676 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5679 * aarch64-modes.def (CC_Zmode): New flags mode.
5680 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5681 represents an equality.
5682 (aarch64_get_condition_code): Handle CC_Zmode.
5683 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5685 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5687 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5688 extraction in good case.
5690 2014-01-10 Richard Biener <rguenther@suse.de>
5692 PR tree-optimization/59374
5693 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5694 checking after SLP discovery. Mark stmts not participating
5695 in any SLP instance properly.
5697 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5699 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5700 when handling a SET rtx.
5702 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5704 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5705 (cortex-a57): Likewise.
5706 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5708 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5710 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5711 non-iwmmxt builtins.
5713 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5717 * ipa-devirt.c record_target_from_binfo): Take as argument
5718 stack of binfos and lookup matching one for virtual inheritance.
5719 (possible_polymorphic_call_targets_1): Update.
5721 2014-01-10 Huacai Chen <chenhc@lemote.com>
5723 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5724 kernel strings for Loongson-2E/2F/3A.
5726 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5729 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5730 is_gimple_call before calling gimple_call_internal_p.
5732 2014-01-09 Steve Ellcey <sellcey@mips.com>
5734 * Makefile.in (TREE_FLOW_H): Remove.
5735 (TREE_SSA_H): Add file names from tree-flow.h.
5736 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5737 * tree.h: Remove tree-flow.h reference.
5738 * hash-table.h: Remove tree-flow.h reference.
5739 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5740 reference with tree-ssa-loop.h.
5742 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5744 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5745 default element-order behavior for -maltivec.
5746 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5747 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5748 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5749 when targeting big endian, at least for now.
5750 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5752 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5755 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5756 var satisfies use_register_for_decl, just take into account type
5757 alignment, rather than decl alignment.
5759 PR tree-optimization/59622
5760 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5761 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5762 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5763 Don't devirtualize for inplace at all. For targets.length () == 1,
5764 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5766 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5768 * config/i386/i386.md (cpu): Remove the unused btver1.
5770 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5772 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5774 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5777 * tree-core.h (struct target_globals): New forward declaration.
5778 (struct tree_target_option): Add globals field.
5779 * tree.h (TREE_TARGET_GLOBALS): Define.
5780 (prepare_target_option_nodes_for_pch): New prototype.
5781 * target-globals.h (struct target_globals): Define even if
5783 * tree.c (prepare_target_option_node_for_pch,
5784 prepare_target_option_nodes_for_pch): New functions.
5785 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5786 * config/i386/i386.c: Include target-globals.h.
5787 (ix86_set_current_function): Instead of doing target_reinit
5788 unconditionally, use save_target_globals_default_opts and
5789 restore_target_globals.
5791 2014-01-09 Richard Biener <rguenther@suse.de>
5793 PR tree-optimization/59715
5794 * tree-cfg.h (split_critical_edges): Declare.
5795 * tree-cfg.c (split_critical_edges): Export.
5796 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5798 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5800 * cfgexpand.c (expand_stack_vars): Optionally disable
5801 asan stack protection.
5802 (expand_used_vars): Likewise.
5803 (partition_stack_vars): Likewise.
5804 * asan.c (asan_emit_stack_protection): Optionally disable
5805 after return stack usage.
5806 (instrument_derefs): Optionally disable memory access instrumentation.
5807 (instrument_builtin_call): Likewise.
5808 (instrument_strlen_call): Likewise.
5809 (asan_protect_global): Optionally disable global variables protection.
5810 * doc/invoke.texi: Added doc for new options.
5811 * params.def: Added new options.
5812 * params.h: Likewise.
5814 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5816 PR rtl-optimization/59724
5817 * ifcvt.c (cond_exec_process_if_block): Don't call
5818 flow_find_head_matching_sequence with 0 longest_match.
5819 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5820 non-active insns if !stop_after.
5821 (try_head_merge_bb): Revert 2014-01-07 changes.
5823 2014-01-08 Jeff Law <law@redhat.com>
5825 * ree.c (get_sub_rtx): New function, extracted from...
5826 (merge_def_and_ext): Here.
5827 (combine_reaching_defs): Use get_sub_rtx.
5829 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5831 * cgraph.h (varpool_variable_node): Do not choke on null node.
5833 2014-01-08 Catherine Moore <clm@codesourcery.com>
5835 * config/mips/mips.md (simple_return): Attempt to use JRC
5837 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5839 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5841 PR rtl-optimization/59137
5842 * reorg.c (steal_delay_list_from_target): Call update_block for
5844 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5846 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5848 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5849 two duplicate entries.
5851 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5854 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5856 * config/mips/mips.c (mips_truncated_op_cost): New function.
5857 (mips_rtx_costs): Adjust test for BADDU.
5858 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5860 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5862 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5863 (*baddu_si): ...this new pattern.
5865 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5868 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5870 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5873 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5875 (expand_expr, expand_normal): Adjust.
5876 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5877 inner_reference_p. Use inner_reference_p to expand inner references.
5878 (store_expr): Adjust.
5879 * cfgexpand.c (expand_call_stmt): Adjust.
5881 2014-01-08 Rong Xu <xur@google.com>
5883 * gcov-io.c (gcov_var): Move from gcov-io.h.
5884 (gcov_position): Ditto.
5885 (gcov_is_error): Ditto.
5886 (gcov_rewrite): Ditto.
5887 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5888 only part to libgcc/libgcov.h.
5890 2014-01-08 Marek Polacek <polacek@redhat.com>
5893 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5895 2014-01-08 Marek Polacek <polacek@redhat.com>
5898 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5900 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5902 PR rtl-optimization/59649
5903 * stor-layout.c (get_mode_bounds): For BImode return
5904 0 and STORE_FLAG_VALUE.
5906 2014-01-08 Richard Biener <rguenther@suse.de>
5909 * gimple.h (is_gimple_builtin_call): Remove.
5910 (gimple_builtin_call_types_compatible_p): New.
5911 (gimple_call_builtin_p): New overload.
5912 * gimple.c (is_gimple_builtin_call): Remove.
5913 (validate_call): Rename to ...
5914 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5916 (validate_type): New static function.
5917 (gimple_call_builtin_p): New overload and adjust.
5918 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5919 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5920 (gimple_fold_stmt_to_constant_1): Likewise.
5921 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5923 2014-01-08 Richard Biener <rguenther@suse.de>
5926 * gimplify.c (gimplify_expr): Gimplify register-register type
5927 VIEW_CONVERT_EXPRs to separate stmts.
5929 2014-01-07 Jeff Law <law@redhat.com>
5932 * ree.c (combine_set_extension): Handle case where source
5933 and destination registers in an extension insn are different.
5934 (combine_reaching_defs): Allow source and destination registers
5935 in extension to be different under limited circumstances.
5936 (add_removable_extension): Remove restriction that the
5937 source and destination registers in the extension are the same.
5938 (find_and_remove_re): Emit a copy from the extension's
5939 destination to its source after the defining insn if
5940 the source and destination registers are different.
5943 * ifcvt.c (merge_if_block): If we are merging a block with more than
5944 one successor with a block with no successors, remove any BARRIER
5945 after the second block.
5947 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5949 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5951 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5954 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5955 for 14-bit register offsets when INT14_OK_STRICT is false.
5957 2014-01-07 Roland Stigge <stigge@antcom.de>
5958 Michael Meissner <meissner@linux.vnet.ibm.com>
5961 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5962 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5964 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5966 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5969 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5971 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5972 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5973 rtx is const0_rtx or not.
5975 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5978 * target-globals.c (save_target_globals): Remove this_fn_optab
5980 * toplev.c: Include optabs.h.
5981 (target_reinit): Temporarily restore the global options if another
5982 set of options are in force.
5984 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5986 PR rtl-optimization/58668
5987 * cfgcleanup.c (flow_find_cross_jump): Don't count
5988 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5989 to determine what is counted.
5990 (flow_find_head_matching_sequence): Use active_insn_p to determine
5992 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5994 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5995 determine what is counted.
5997 PR tree-optimization/59643
5998 * tree-predcom.c (split_data_refs_to_components): If one dr is
5999 read and one write, determine_offset fails and the write isn't
6000 in the bad component, just put the read into the bad component.
6002 2014-01-07 Mike Stump <mikestump@comcast.net>
6003 Jakub Jelinek <jakub@redhat.com>
6006 * tree-core.h (struct tree_optimization_option): Change optabs
6007 type from unsigned char * to void *.
6008 * optabs.c (init_tree_optimization_optabs): Adjust
6009 TREE_OPTIMIZATION_OPTABS initialization.
6011 2014-01-06 Jakub Jelinek <jakub@redhat.com>
6014 * config/i386/i386.h (struct machine_function): Add
6015 no_drap_save_restore field.
6016 * config/i386/i386.c (ix86_save_reg): Use
6017 !cfun->machine->no_drap_save_restore instead of
6018 crtl->stack_realign_needed.
6019 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
6020 this function clears frame_pointer_needed. Set
6021 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
6022 and DRAP reg is needed.
6024 2014-01-06 Marek Polacek <polacek@redhat.com>
6027 * doc/implement-c.texi: Mention that other integer types are
6028 permitted as bit-field types in strictly conforming mode.
6030 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
6032 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
6035 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
6037 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
6039 2014-01-06 Martin Jambor <mjambor@suse.cz>
6042 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
6044 * ipa-prop.c (ipa_print_node_params): Fix indentation.
6046 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
6050 * var-tracking.c (add_stores): Preserve the value of the source even if
6051 we don't record the store.
6053 2014-01-06 Terry Guo <terry.guo@arm.com>
6055 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6057 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6060 * config/darwin.c (darwin_function_section): Adjust return values to
6061 correspond to optimisation changes made in r206070.
6063 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6065 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6066 from prefetch_block tune setting.
6067 (nocona_cost): Correct size of prefetch block to 64.
6069 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6071 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6072 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6073 used to save the static chain register in the computation of the offset
6074 from which the FP registers need to be restored.
6076 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6078 PR tree-optimization/59519
6079 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6080 ICE if get_current_def (current_new_name) is already non-NULL, as long
6081 as it is a phi result of some other phi in *new_exit_bb that has
6084 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6085 or vmovdqu* for misaligned_operand.
6086 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6087 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6088 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6089 aligned_mem for AVX512F masked aligned load and store builtins and for
6092 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6094 PR tree-optimization/59651
6095 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6096 Address range for negative step should be added by TYPE_SIZE_UNIT.
6098 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6100 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6101 overlapping registers also for registers other than the stack pointer.
6103 2014-01-03 Marek Polacek <polacek@redhat.com>
6106 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6109 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6112 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6115 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6116 (push splitter): Use <P:MODE_SIZE> instead of
6117 GET_MODE_SIZE (<P:MODE>mode).
6118 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6119 (mov -1, reg peephole2): Likewise.
6120 * config/i386/sse.md (*mov<mode>_internal,
6121 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6122 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6123 *<code><mode>3, *andnot<mode>3<mask_name>,
6124 <mask_codefor><code><mode>3<mask_name>): Likewise.
6125 * config/i386/subst.md (mask_mode512bit_condition,
6126 sd_mask_mode512bit_condition): Likewise.
6128 2014-01-02 Xinliang David Li <davidxl@google.com>
6130 PR tree-optimization/59303
6131 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6132 (dump_predicates): Better output format.
6133 (pred_equal_p): New function.
6134 (is_neq_relop_p): Ditto.
6135 (is_neq_zero_form_p): Ditto.
6136 (pred_expr_equal_p): Ditto.
6137 (pred_neg_p): Ditto.
6138 (simplify_pred): Ditto.
6139 (simplify_preds_2): Ditto.
6140 (simplify_preds_3): Ditto.
6141 (simplify_preds_4): Ditto.
6142 (simplify_preds): Ditto.
6144 (push_to_worklist): Ditto.
6145 (get_pred_info_from_cmp): Ditto.
6146 (is_degenerated_phi): Ditto.
6147 (normalize_one_pred_1): Ditto.
6148 (normalize_one_pred): Ditto.
6149 (normalize_one_pred_chain): Ditto.
6150 (normalize_preds): Ditto.
6151 (normalize_cond_1): Remove function.
6152 (normalize_cond): Ditto.
6153 (is_gcond_subset_of): Ditto.
6154 (is_subset_of_any): Ditto.
6155 (is_or_set_subset_of): Ditto.
6156 (is_and_set_subset_of): Ditto.
6157 (is_norm_cond_subset_of): Ditto.
6158 (pred_chain_length_cmp): Ditto.
6159 (convert_control_dep_chain_into_preds): Type change.
6160 (find_predicates): Ditto.
6161 (find_def_preds): Ditto.
6162 (destroy_predicates_vecs): Ditto.
6163 (find_matching_predicates_in_rest_chains): Ditto.
6164 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6165 (is_pred_expr_subset): Ditto.
6166 (is_pred_chain_subset_of): Ditto.
6167 (is_included_in): Ditto.
6168 (is_superset_of): Ditto.
6170 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6172 Update copyright years.
6174 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6176 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6177 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6178 config/arc/arc.md, config/arc/arc.opt,
6179 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6180 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6181 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6182 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6183 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6184 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6185 vtable-verify.c, vtable-verify.h: Use the standard form for the
6188 2014-01-02 Tobias Burnus <burnus@net-b.de>
6190 * gcc.c (process_command): Update copyright notice dates.
6191 * gcov-dump.c: Ditto.
6193 * doc/cpp.texi: Bump @copying's copyright year.
6194 * doc/cppinternals.texi: Ditto.
6195 * doc/gcc.texi: Ditto.
6196 * doc/gccint.texi: Ditto.
6197 * doc/gcov.texi: Ditto.
6198 * doc/install.texi: Ditto.
6199 * doc/invoke.texi: Ditto.
6201 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6203 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6205 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6207 * config/i386/sse.md (*mov<mode>_internal): Guard
6208 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6210 PR rtl-optimization/59647
6211 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6212 new_rtx into UNSIGNED_FLOAT rtxes.
6214 Copyright (C) 2014 Free Software Foundation, Inc.
6216 Copying and distribution of this file, with or without modification,
6217 are permitted in any medium without royalty provided the copyright
6218 notice and this notice are preserved.