1 2014-04-08 Pat Haugen <pthaugen@us.ibm.com>
3 * config/rs6000/sync.md (AINT mode_iterator): Move definition.
4 (loadsync_<mode>): Change mode.
5 (load_quadpti, store_quadpti): New.
6 (atomic_load<mode>, atomic_store<mode>): Add support for TI mode.
7 * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_LSQ.
9 2014-04-08 Richard Sandiford <rdsandiford@googlemail.com>
12 * config/rs6000/vsx.md (vsx_xscvdpspn_scalar): Change input to DImode.
13 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Update accordingly.
14 Use gen_rtx_REG rather than simplify_gen_subreg for op0_di.
16 2014-04-08 Richard Biener <rguenther@suse.de>
19 * tree-pretty-print.c (pp_double_int): For HWI32 hosts with
20 a 64bit widest int print double-int similar to on HWI64 hosts.
22 2014-04-08 Richard Biener <rguenther@suse.de>
24 PR tree-optimization/60785
25 * graphite-sese-to-poly.c (rewrite_phi_out_of_ssa): Treat
26 default defs properly.
28 2014-04-08 Nathan Sidwell <nathan@codesourcery.com>
30 * doc/invoke (Wnon-virtual-dtor): Update to match implementation.
33 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
35 * ipa-devirt.c (maybe_record_node): When node is not recorded,
36 set completep to false rather than true.
38 2014-04-07 Douglas B Rupp <rupp@adacore.com>
41 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
42 ARM_TARGET2_DWARF_FORMAT.
44 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
47 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
48 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
51 2014-04-07 Richard Biener <rguenther@suse.de>
53 PR tree-optimization/60766
54 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
55 (may_eliminate_iv): Convert cand_value_at result to desired type.
57 2014-04-07 Jason Merrill <jason@redhat.com>
60 * common.opt (-fno-gnu-unique): Add.
61 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
63 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
65 * haifa-sched.c: Fix outdated function reference and minor
66 grammar errors in introductory comment.
68 2014-04-07 Richard Biener <rguenther@suse.de>
71 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
73 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
75 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
78 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
79 size accounting for thunks.
80 (pa_asm_output_mi_thunk): Use final_start_function() and
81 final_end_function() to output function start and end directives.
83 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
85 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
86 specific ISA/ feature information. Remove short_sp and errata_skip ds.
87 Add avr_device_specific_features enum to have device specific info.
88 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
89 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
90 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
91 updated device specific info.
92 * config/avr/avr-mcus.def: Merge device specific details to
94 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
96 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
97 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
98 assembler if RMW isa supported by current device.
99 * config/avr/genmultilib.awk: Update as device info structure changed.
100 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
102 2014-04-04 Cong Hou <congh@google.com>
104 PR tree-optimization/60656
105 * tree-vect-stmts.c (supportable_widening_operation):
106 Fix a bug that elements in a vector with vect_used_by_reduction
107 property are incorrectly reordered when the operation on it is not
108 consistant with the one in reduction operation.
110 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
112 PR rtl-optimization/60155
113 * gcse.c (record_set_data): New function.
114 (single_set_gcse): New function.
115 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
116 (hoist_code): Likewise.
117 (get_pressure_class_and_nregs): Likewise.
119 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
121 * explow.c (probe_stack_range): Emit a final optimization blockage.
123 2014-04-04 Anthony Green <green@moxielogic.com>
125 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
128 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
131 * lto-cgraph.c (input_overwrite_node): Check that partitioning
132 flags are set only during streaming.
133 * ipa.c (process_references, walk_polymorphic_call_targets,
134 symtab_remove_unreachable_nodes): Drop bodies of always inline
135 after early inlining.
136 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
138 2014-04-04 Jakub Jelinek <jakub@redhat.com>
139 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
142 * dwarf2out.c (const_ok_for_output_1): Reject expressions
145 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
148 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
150 (cortex_a53_fdivd): Likewise.
152 2014-04-04 Martin Jambor <mjambor@suse.cz>
155 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
157 * cgraph.c (clone_of_p): Also return true if thunks match.
158 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
159 cgraph_function_or_thunk_node and an obsolete comment.
160 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
162 (build_function_decl_skip_args): Likewise.
163 (set_new_clone_decl_and_node_flags): New function.
164 (duplicate_thunk_for_node): Likewise.
165 (redirect_edge_duplicating_thunks): Likewise.
166 (cgraph_clone_node): New parameter args_to_skip, pass it to
167 redirect_edge_duplicating_thunks which is called instead of
168 cgraph_redirect_edge_callee.
169 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
170 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
172 2014-04-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
175 * fortran/trans-types.c (gfc_get_function_type): In case of recursion
176 build a variadic function type with empty argument list instead of a
177 stdarg-like function type with incomplete argument list.
179 2014-04-04 Jeff Law <law@redhat.com>
182 * config/arm/predicates.md (const_int_I_operand): New predicate.
183 (const_int_M_operand): Similarly.
184 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
186 (insv_t2, extv_reg, extzv_t2): Likewise.
187 (load_multiple_with_writeback): Similarly for const_int_I_operand.
188 (pop_multiple_with_writeback_and_return): Likewise.
189 (vfp_pop_multiple_with_writeback): Likewise
191 2014-04-04 Richard Biener <rguenther@suse.de>
194 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
195 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
197 * gimplify.h (gimple_add_tmp_var_fn): Declare.
198 * gimplify.c (gimple_add_tmp_var_fn): New function.
199 * gimple-expr.h (create_tmp_reg_fn): Declare.
200 * gimple-expr.c (create_tmp_reg_fn): New function.
201 * gimple-low.c (record_vars_into): Don't change cfun.
202 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
203 code generation without cfun.
205 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
208 * Makefile.in (install-driver): Fix shell scripting.
210 2014-04-03 Cong Hou <congh@google.com>
212 PR tree-optimization/60505
213 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
214 threshold of number of iterations below which no vectorization will be
216 * tree-vect-loop.c (new_loop_vec_info):
217 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
218 * tree-vect-loop.c (vect_analyze_loop_operations):
219 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
220 * tree-vect-loop.c (vect_transform_loop):
221 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
222 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
223 of iterations of the loop and see if we should build the epilogue.
225 2014-04-03 Richard Biener <rguenther@suse.de>
227 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
229 (streamer_tree_cache_create): Adjust.
230 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
231 to allow optional nodes array.
232 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
233 (streamer_tree_cache_append): Likewise.
234 (streamer_tree_cache_create): Create nodes array optionally
235 as specified by parameter.
236 * lto-streamer-out.c (create_output_block): Avoid maintaining
237 the node array in the writer cache.
238 (DFS_write_tree): Remove assertion.
239 (produce_asm_for_decls): Free the out decl state hash table
241 * lto-streamer-in.c (lto_data_in_create): Adjust for
242 streamer_tree_cache_create prototype change.
244 2014-04-03 Richard Biener <rguenther@suse.de>
246 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
247 set TREE_CHAIN to NULL_TREE.
249 2014-04-03 Richard Biener <rguenther@suse.de>
251 PR tree-optimization/60740
252 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
253 over all GIMPLE_COND operands.
255 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
257 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
258 (Weffc++): Remove Scott's numbering, merge lists and reference
264 * class.c (accessible_nvdtor_p): New.
265 (check_bases): Don't check base destructor here ...
266 (check_bases_and_members): ... check them here. Trigger on
267 Wnon-virtual-dtor flag.
268 (finish_struct_1): Use accessible_nvdtor_p.
271 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
272 * g++.dg/warn/Wnvdtor-2.C: New.
273 * g++.dg/warn/Wnvdtor-3.C: New.
274 * g++.dg/warn/Wnvdtor-4.C: New.
275 * g++.dg/warn/Weff1.C: Delete.
276 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
277 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
279 2014-04-03 Nick Clifton <nickc@redhat.com>
281 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
284 2014-04-03 Martin Jambor <mjambor@suse.cz>
286 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
287 mention gcc_unreachable before failing.
288 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
291 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
294 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
295 code and instead mark the context inconsistent.
296 (possible_polymorphic_call_targets): For inconsistent contexts
297 return empty complete list.
299 2014-04-02 Anthony Green <green@moxielogic.com>
301 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
302 (extendqisi2, extendhisi2): Define.
303 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
304 (WCHAR_TYPE): Change to unsigned int.
306 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
308 PR tree-optimization/60733
309 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
310 insertion point for PHI candidates to be the end of the feeding
311 block for the PHI argument.
313 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
315 PR rtl-optimization/60650
316 * lra-constraints.c (process_alt_operands): Decrease reject for
317 earlyclobber matching.
319 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
321 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
323 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
325 * config/spu/spu.c (pad_bb): Do not crash when the last
326 insn is CODE_FOR_blockage.
328 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
330 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
331 lies outside the target mode.
333 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
336 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
337 software floating point or no floating point registers, do not
338 allow any type in the FPRs. Eliminate a test for SPE SIMD types
339 in GPRs that occurs after we tested for GPRs that would never be
342 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
343 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
344 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
345 specifically allow DDmode, since that does not use the SPE SIMD
348 2014-04-02 Richard Biener <rguenther@suse.de>
351 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
352 MODE_INTs. Properly use negv_optab.
353 (expand_abs): Likewise.
355 2014-04-02 Richard Biener <rguenther@suse.de>
358 * Makefile.in (install-driver): Guard extra installs with special
361 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
363 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
366 2014-04-01 Richard Henderson <rth@redhat.com>
369 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
370 alternative enabled before register allocation.
372 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
374 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
375 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
377 (nios2_large_got_address): Remove unneeded 'sym' parameter.
378 (nios2_got_address): Update nios2_large_got_address call site.
379 (nios2_delegitimize_address): New function.
380 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
381 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
382 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
384 2014-04-01 Martin Husemann <martin@duskware.de>
386 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
389 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
391 PR rtl-optimization/60604
392 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
393 check from register_operand.
394 (register_operand): Redefine in terms of general_operand.
395 (nonmemory_operand): Use register_operand for the non-constant cases.
397 2014-04-01 Richard Biener <rguenther@suse.de>
399 * gimple.h (struct gimple_statement_base): Align subcode to
402 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
404 * doc/invoke.texi (mapp-regs): Clarify.
406 2014-03-31 Ulrich Drepper <drepper@gmail.com>
408 * config/i386/avx512fintrin.h (__v32hi): Define type.
410 (_mm512_set1_epi8): Define.
411 (_mm512_set1_epi16): Define.
412 (_mm512_set4_epi32): Define.
413 (_mm512_set4_epi64): Define.
414 (_mm512_set4_pd): Define.
415 (_mm512_set4_ps): Define.
416 (_mm512_setr4_epi64): Define.
417 (_mm512_setr4_epi32): Define.
418 (_mm512_setr4_pd): Define.
419 (_mm512_setr4_ps): Define.
420 (_mm512_setzero_epi32): Define.
422 2014-03-31 Martin Jambor <mjambor@suse.cz>
425 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
426 callsite_arguments_match_p. Updated all callers. Also check types of
427 corresponding formal parameters and actual arguments.
428 (not_all_callers_have_enough_arguments_p) Renamed to
429 some_callers_have_mismatched_arguments_p.
431 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
433 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
435 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
438 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
441 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
443 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
444 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
446 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
447 Use FMAMODE_NOVF512 mode iterator.
448 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
449 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
450 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
451 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
453 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
454 Use VF_128_256 mode iterator.
455 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
458 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
460 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
461 static chain if needed.
463 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
466 * lra-constraints.c (index_part_to_reg): New.
467 (process_address): Use it.
469 2014-03-27 Jeff Law <law@redhat.com>
470 Jakub Jelinek <jakub@redhat.com>
473 * expr.c (do_tablejump): Use simplify_gen_binary rather than
474 gen_rtx_{PLUS,MULT} to build up the address expression.
476 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
477 creating non-canonical RTL.
479 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
482 * ipa-inline.c (want_inline_small_function_p): Short circuit large
483 functions; reorganize to make cheap checks first.
484 (inline_small_functions): Do not estimate growth when dumping;
486 * ipa-inline.h (inline_summary): Add min_size.
487 (growth_likely_positive): New function.
488 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
489 (set_cond_stmt_execution_predicate): Cleanup.
490 (estimate_edge_size_and_time): Compute min_size.
491 (estimate_calls_size_and_time): Likewise.
492 (estimate_node_size_and_time): Likewise.
493 (inline_update_overall_summary): Update min_size.
494 (do_estimate_edge_time): Likewise.
495 (do_estimate_edge_size): Update.
496 (do_estimate_edge_hints): Update.
497 (growth_likely_positive): New function.
499 2014-03-28 Jakub Jelinek <jakub@redhat.com>
502 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
503 also if addr has VOIDmode.
505 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
507 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
508 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
510 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
511 instructions as well as AdvancedSIMD loads.
513 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
515 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
516 Use crypto_aese type.
517 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
518 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
519 crypto_aese, crypto_aesmc. Move to types.md.
520 * config/arm/types.md (crypto_aes): Split into crypto_aese,
522 * config/arm/iterators.md (crypto_type): Likewise.
524 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
526 * cgraph.c: Include expr.h and tree-dfa.h.
527 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
530 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
533 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
534 regs from checking multi-reg pseudos.
536 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
538 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
540 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
542 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
543 if it would clobber the stack pointer, even temporarily.
545 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
547 * mode-switching.c: Make small adjustments to the top comment.
549 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
551 * config/rs6000/constraints.md (wD constraint): New constraint to
552 match the constant integer to get the top DImode/DFmode out of a
553 vector in a VSX register.
555 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
556 match the constant integer to get the top DImode/DFmode out of a
557 vector in a VSX register.
559 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
562 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
565 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
566 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
568 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
569 Optimize vec_extract of 64-bit values, where the value being
570 extracted is in the top word, where we can use scalar
571 instructions. Add direct move and store support. Combine the big
572 endian/little endian vector select load support into a single insn.
573 (vsx_extract_<mode>_internal1): Likewise.
574 (vsx_extract_<mode>_internal2): Likewise.
575 (vsx_extract_<mode>_load): Likewise.
576 (vsx_extract_<mode>_store): Likewise.
577 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
578 combined into vsx_extract_<mode>_load.
579 (vsx_extract_<mode>_one_le): Likewise.
581 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
582 define the top 64-bit vector element.
584 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
587 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
588 Document vec_vbpermq builtin.
591 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
592 enable use of xxsldwi and xxpermdi builtin functions.
593 (vec_xxpermdi): Likewise.
595 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
596 Document use of vec_xxsldwi and vec_xxpermdi builtins.
598 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
600 PR rtl-optimization/60650
601 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
603 (find_spills_for): New.
604 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
605 Spill all pseudos on the second iteration.
607 2014-03-27 Marek Polacek <polacek@redhat.com>
610 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
613 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
615 * config/s390/s390.c (s390_can_use_return_insn): Check for
616 call-saved FPRs on 31 bit.
618 2014-03-27 Jakub Jelinek <jakub@redhat.com>
621 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
622 if they need regimplification, just drop them instead of
623 calling gimple_regimplify_operands on them.
625 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
628 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
629 (aarch64_frame_pointer_required): Adjust logic.
630 (aarch64_can_eliminate): Adjust logic.
631 (aarch64_override_options_after_change): Adjust logic.
633 2014-03-27 Dehao Chen <dehao@google.com>
635 * ipa-inline.c (early_inliner): Update node's inline info.
637 2014-03-26 Dehao Chen <dehao@google.com>
639 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
640 compiler inserted conditional jumps for NAN float check.
642 2014-03-26 Jakub Jelinek <jakub@redhat.com>
644 * ubsan.h (ubsan_create_data): Change second argument's type
645 to const location_t *.
646 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
648 (ubsan_create_data): Change second argument to const location_t *PLOC.
649 Create Loc field whenever PLOC is non-NULL.
650 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
651 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
655 * real.c (real_to_integer2): Change type of low to UHWI.
657 2014-03-26 Tobias Burnus <burnus@net-b.de>
659 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
660 (CILK_SELF_SPECS): New define.
661 (driver_self_specs): Use it.
663 2014-03-26 Richard Biener <rguenther@suse.de>
665 * tree-pretty-print.c (percent_K_format): Implement special
666 case for LTO and its stripped down BLOCK tree.
668 2014-03-26 Jakub Jelinek <jakub@redhat.com>
671 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
673 * tree-vrp.c (simplify_internal_call_using_ranges): If only
674 one range is range_int_cst_p, but not both, at least optimize
675 addition/subtraction of 0 and multiplication by 0 or 1.
676 * gimple-fold.c (gimple_fold_call): Fold
677 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
678 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
679 INTEGER_CSTs, try to fold at least x * 0 and y - y.
681 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
683 PR rtl-optimization/60452
684 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
685 <case REG>: Return 1 for invalid offsets from the frame pointer.
687 2014-03-26 Marek Polacek <polacek@redhat.com>
690 * doc/extend.texi (C Extensions): Mention variable-length arrays in
693 2014-03-26 Marek Polacek <polacek@redhat.com>
696 * doc/extend.texi (Designated Inits): Describe what happens to omitted
699 2014-03-26 Marek Polacek <polacek@redhat.com>
702 * ira-color.c (update_conflict_hard_regno_costs): Perform the
703 multiplication in unsigned type.
705 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
707 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
709 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
711 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
713 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
716 * cif-code.def (UNREACHABLE) New code.
717 * ipa-inline.c (inline_small_functions): Skip edges to
718 __builtlin_unreachable.
719 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
720 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
721 predicate to __bulitin_unreachable.
722 (set_cond_stmt_execution_predicate): Fix issue when
723 invert_tree_comparison returns ERROR_MARK.
724 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
725 propagate to inline clones.
726 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
728 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
729 * cgraphclones.c (cgraph_clone_node): If call destination is already
730 ureachable, do not redirect it back.
731 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
734 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
736 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
737 Do not modify inline clones.
739 2014-03-25 Jakub Jelinek <jakub@redhat.com>
741 * config/i386/i386.md (general_sext_operand): New mode attr.
742 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
743 don't generate (sign_extend (const_int)).
744 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
745 operands[2]. Use We constraint instead of <i> and
746 <general_sext_operand> predicate instead of <general_operand>.
747 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
748 * config/i386/constraints.md (We): New constraint.
749 * config/i386/predicates.md (x86_64_sext_operand,
750 sext_operand): New predicates.
752 2014-03-25 Martin Jambor <mjambor@suse.cz>
755 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
756 inconsistent devirtualizations to __builtin_unreachable.
758 2014-03-25 Marek Polacek <polacek@redhat.com>
761 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
763 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
765 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
766 order of elements for big-endian.
768 2014-03-25 Richard Biener <rguenther@suse.de>
771 * gimplify-me.c (gimple_regimplify_operands): Update the
774 2014-03-25 Martin Jambor <mjambor@suse.cz>
777 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
778 (lto_output_varpool_node): Likewise.
779 (input_overwrite_node): Likewise.
780 (input_varpool_node): Likewise.
782 2014-03-25 Richard Biener <rguenther@suse.de>
784 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
787 2014-03-25 Jakub Jelinek <jakub@redhat.com>
789 * combine.c (simplify_compare_const): Add MODE argument.
790 Handle mode_width 0 as very large mode_width.
791 (try_combine, simplify_comparison): Adjust callers.
793 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
794 type to avoid signed integer overflow.
795 * explow.c (plus_constant): Likewise.
797 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
799 * doc/generic.texi: Correct typos.
801 2014-03-24 Tobias Burnus <burnus@net-b.de>
803 * doc/invoke.texi (-flto): Expand section about
804 using static libraries with LTO.
806 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
808 PR rtl-optimization/60501
809 * optabs.def (addptr3_optab): New optab.
810 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
811 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
812 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
814 * lra.c (emit_add3_insn): Use the addptr pattern if available.
816 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
818 2014-03-24 Ulrich Drepper <drepper@gmail.com>
820 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
823 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
824 (_mm256_undefined_ps): Define.
825 (_mm256_undefined_pd): Define.
826 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
827 (_mm_undefined_pd): Define.
828 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
829 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
830 (_mm512_undefined_ps): Define.
831 (_mm512_undefined_pd): Define.
832 Use _mm*_undefined_*.
833 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
835 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
837 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
838 (lshr_simd): DI mode added.
839 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
840 (aarch64_ushr_simddi): Likewise.
841 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
842 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
843 (vshrd_n_u64): Likewise.
845 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
847 * Makefile.in (s-macro_list): Depend on cc1.
849 2014-03-23 Teresa Johnson <tejohnson@google.com>
851 * ipa-utils.c (ipa_print_order): Use specified dump file.
853 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
855 PR rtl-optimization/60601
856 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
858 * gcc.c (eval_spec_function): Initialize save_growing_value.
860 2014-03-22 Jakub Jelinek <jakub@redhat.com>
863 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
864 code == MINUS_EXPR, never swap op0 with op1.
866 * toplev.c (init_local_tick): Avoid signed integer multiplication
868 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
869 shift by first operand's bitsize.
871 2014-03-21 Jakub Jelinek <jakub@redhat.com>
874 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
876 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
877 TARGET_ISA_64BIT_P(x).
879 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
881 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
882 pattern for vector nor instead of subtract from splat(-1).
883 (altivec_expand_vec_perm_const_le): Likewise.
885 2014-03-21 Richard Henderson <rth@twiddle.net>
888 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
889 related insns after epilogue_completed.
891 2014-03-21 Martin Jambor <mjambor@suse.cz>
894 * cgraph.h (symtab_node): New flag body_removed.
895 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
896 when removing bodies.
897 * symtab.c (dump_symtab_base): Dump body_removed flag.
898 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
899 had their bodies removed.
901 2014-03-21 Martin Jambor <mjambor@suse.cz>
904 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
907 2014-03-21 Richard Biener <rguenther@suse.de>
909 PR tree-optimization/60577
910 * tree-core.h (struct tree_base): Document nothrow_flag use
911 in VAR_DECL_NONALIASED.
912 * tree.h (VAR_DECL_NONALIASED): New.
913 (may_be_aliased): Adjust.
914 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
916 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
918 * expr.c (expand_expr_real_1): Remove outdated comment.
920 2014-03-20 Jakub Jelinek <jakub@redhat.com>
923 * ira.c (adjust_cleared_regs): Call copy_rtx on
924 *reg_equiv[REGNO (loc)].src_p before passing it to
925 simplify_replace_fn_rtx.
928 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
929 into CONST, put pic register as first operand of PLUS. Use
930 gen_const_mem for both 32-bit and 64-bit PIC got loads.
932 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
934 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
936 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
938 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
939 around for store forwarding issue in the FPU on the UT699.
940 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
941 loads and operations if -mfix-ut699 is specified.
942 (divtf3_hq): Tweak attribute.
943 (sqrttf2_hq): Likewise.
945 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
947 * calls.c (store_one_arg): Remove incorrect const qualification on the
948 type of the temporary.
949 * cfgexpand.c (expand_return): Likewise.
950 * expr.c (expand_constructor): Likewise.
951 (expand_expr_real_1): Likewise.
953 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
955 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
958 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
961 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
963 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
965 * config/arm/aarch-common-protos.h
966 (alu_cost_table): Fix spelling of "extend".
967 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
969 2014-03-19 Richard Biener <rguenther@suse.de>
972 * tree-core.h (tree_type_common): Re-order pointer members
973 to reduce recursion depth during GC walks.
975 2014-03-19 Marek Polacek <polacek@redhat.com>
978 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
981 2014-03-19 Richard Biener <rguenther@suse.de>
984 * lto-streamer-in.c (input_function): In WPA stage do not drop
987 2014-03-19 Jakub Jelinek <jakub@redhat.com>
989 PR tree-optimization/60559
990 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
991 with build_zero_cst assignment.
993 2014-03-18 Kai Tietz <ktietz@redhat.com>
995 PR rtl-optimization/56356
996 * sdbout.c (sdbout_parms): Verify that parms'
997 incoming argument is valid.
998 (sdbout_reg_parms): Likewise.
1000 2014-03-18 Richard Henderson <rth@redhat.com>
1003 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
1004 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
1005 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
1007 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
1009 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
1010 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
1011 Italicize plugin event names in description. Explain that
1012 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
1013 Remind that no GCC functions should be called after PLUGIN_FINISH.
1014 Explain what pragmas with expansion are.
1016 2014-03-18 Martin Liska <mliska@suse.cz>
1018 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
1019 gimple call statement is update.
1020 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
1021 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
1023 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1026 * ubsan.c (ubsan_instrument_unreachable): Call
1027 initialize_sanitizer_builtins.
1028 (ubsan_pass): Likewise.
1031 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1032 varpool_finalize_decl instead of rest_of_decl_compilation.
1034 2014-03-18 Richard Biener <rguenther@suse.de>
1036 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1037 by using bitmap_and_compl instead of bitmap_and_compl_into.
1038 (df_rd_transfer_function): Likewise.
1040 2014-03-18 Richard Biener <rguenther@suse.de>
1042 * doc/lto.texi (fresolution): Fix typo.
1044 2014-03-18 Richard Biener <rguenther@suse.de>
1046 * doc/invoke.texi (flto): Update for changes in 4.9.
1048 2014-03-18 Richard Biener <rguenther@suse.de>
1050 * doc/loop.texi: Remove section on the removed lambda framework.
1051 Update loop docs with recent changes in preserving loop structure.
1053 2014-03-18 Richard Biener <rguenther@suse.de>
1055 * doc/lto.texi (-fresolution): Document.
1057 2014-03-18 Richard Biener <rguenther@suse.de>
1059 * doc/contrib.texi: Adjust my name.
1061 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1064 * internal-fn.c: Include diagnostic-core.h.
1065 (expand_BUILTIN_EXPECT): New function.
1066 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1067 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1068 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1069 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1071 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1072 Revert 3 argument __builtin_expect code.
1073 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1074 * gimple-fold.c (gimple_fold_call): Likewise.
1075 * tree.h (fold_builtin_expect): New prototype.
1076 * builtins.c (build_builtin_expect_predicate): Add predictor
1077 argument, if non-NULL, create 3 argument __builtin_expect.
1078 (fold_builtin_expect): No longer static. Add ARG2 argument,
1079 pass it through to build_builtin_expect_predicate.
1080 (fold_builtin_2): Adjust caller.
1081 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1082 * internal-fn.def (BUILTIN_EXPECT): New.
1084 2014-03-18 Tobias Burnus <burnus@net-b.de>
1087 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1088 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1089 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1091 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1094 * predict.c (combine_predictions_for_bb): Fix up formatting.
1095 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1096 fill what it points to if non-NULL.
1097 (tree_predict_by_opcode): Adjust caller, use the predictor.
1098 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1100 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1102 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1103 proper constant for the store mode.
1105 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1107 * symtab.c (change_decl_assembler_name): Fix transparent alias
1110 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1112 * config/aarch64/aarch64.c: Correct the comments about the
1113 aarch64 stack layout.
1115 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1117 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1118 check for GF_OMP_FOR_KIND_FOR.
1120 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1122 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1123 ymm and zmm register names.
1125 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1128 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1129 note creation for the 2010-08-31 changes.
1131 2014-03-17 Marek Polacek <polacek@redhat.com>
1134 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1135 as -fno-tree-loop-vectorize.
1136 (expand_omp_simd): Likewise.
1138 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1140 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1141 (eligible_for_call_delay): New prototype.
1142 * config/sparc/sparc.c (tls_call_delay): Rename into...
1143 (eligible_for_call_delay): ...this. Return false if the instruction
1144 cannot be put in the delay slot of a branch.
1145 (eligible_for_restore_insn): Simplify.
1146 (eligible_for_return_delay): Return false if the instruction cannot be
1147 put in the delay slot of a branch and simplify.
1148 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1149 put in the delay slot of a branch.
1150 * config/sparc/sparc.md (fix_ut699): New attribute.
1151 (tls_call_delay): Delete.
1152 (in_call_delay): Reimplement.
1153 (eligible_for_sibcall_delay): Rename into...
1154 (in_sibcall_delay): ...this.
1155 (eligible_for_return_delay): Rename into...
1156 (in_return_delay): ...this.
1157 (in_branch_delay): Reimplement.
1158 (in_uncond_branch_delay): Delete.
1159 (in_annul_branch_delay): Delete.
1161 2014-03-14 Richard Henderson <rth@redhat.com>
1164 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1165 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1166 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1167 (floathi splitters): Remove.
1168 (float<SWI48x>xf2): New pattern.
1169 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1170 code that tried to handle DImode for 32-bit, but which was excluded
1171 by the pattern's condition. Drop allocation of stack temporary.
1172 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1173 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1174 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1175 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1176 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1177 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1178 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1179 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1180 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1181 (*float<SWI48x><X87MODEF>2_i387): Remove.
1182 (all float _with_temp splitters): Remove.
1183 (*float<SWI48x><MODEF>2_i387): New pattern.
1184 (*float<SWI48><MODEF>2_sse): New pattern.
1185 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1186 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1188 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1189 Marek Polacek <polacek@redhat.com>
1192 * common.opt (dump_base_name_prefixed): New Variable.
1193 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1194 if x_dump_base_name_prefixed is already set, set it at the end.
1196 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1198 PR rtl-optimization/60508
1199 * lra-constraints.c (get_reload_reg): Add new parameter
1201 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1202 Pass the new parameter values.
1204 2014-03-14 Richard Biener <rguenther@suse.de>
1206 * common.opt: Revert unintented changes from r205065.
1209 2014-03-14 Richard Biener <rguenther@suse.de>
1212 * cfghooks.c (split_block): Properly adjust all loops the
1213 block was a latch of.
1215 2014-03-14 Martin Jambor <mjambor@suse.cz>
1218 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1221 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1224 * config/avr/avr.c (avr_set_current_function): Pass function name
1225 through default_strip_name_encoding before sanity checking instead
1226 of skipping the first char of the assembler name.
1228 2014-03-13 Richard Henderson <rth@redhat.com>
1231 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1232 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1233 * config/i386/i386-protos.h: Likewise.
1234 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1235 in the expander instead of a splitter.
1236 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1237 any possibility of requiring a memory.
1238 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1239 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1240 (fp branch splitters): Update for ix86_split_fp_branch.
1241 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1242 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1243 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1244 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1245 (*fop_<MODEF>_3_i387): Likewise.
1246 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1247 (splitters for the fop_* register patterns): Remove.
1248 (fscalexf4_i387): Rename from *fscalexf4_i387.
1249 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1251 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1253 PR tree-optimization/59779
1254 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1255 type for bitsize and maxsize instead of HOST_WIDE_INT.
1257 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1259 PR rtl-optimization/57320
1260 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1261 the CFG after thread_prologue_and_epilogue_insns.
1263 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1265 PR rtl-optimization/57189
1266 * lra-constraints.c (process_alt_operands): Disfavor spilling
1269 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1271 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1273 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1275 PR tree-optimization/59025
1277 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1278 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1280 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1283 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1284 calls of avr_out_plus_1.
1286 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1288 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1289 BB's single pred and update the father loop's latch info later.
1291 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1293 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1297 (VEC_base): Likewise.
1298 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1299 registers, we need to swap double words in little endian mode.
1301 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1302 to be a container mode for 128-bit integer operations added in ISA
1303 2.07. Unlike TImode and PTImode, the preferred register set is
1304 the Altivec/VMX registers for the 128-bit operations.
1306 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1308 (rs6000_split_128bit_ok_p): Likewise.
1310 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1311 macros for creating ISA 2.07 normal and overloaded builtin
1312 functions with 3 arguments.
1313 (BU_P8V_OVERLOAD_3): Likewise.
1314 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1315 for use as overloaded functions.
1316 (VPERM_1TI_UNS): Likewise.
1317 (VSEL_1TI): Likewise.
1318 (VSEL_1TI_UNS): Likewise.
1319 (ST_INTERNAL_1ti): Likewise.
1320 (LD_INTERNAL_1ti): Likewise.
1321 (XXSEL_1TI): Likewise.
1322 (XXSEL_1TI_UNS): Likewise.
1323 (VPERM_1TI): Likewise.
1324 (VPERM_1TI_UNS): Likewise.
1325 (XXPERMDI_1TI): Likewise.
1326 (SET_1TI): Likewise.
1327 (LXVD2X_V1TI): Likewise.
1328 (STXVD2X_V1TI): Likewise.
1329 (VEC_INIT_V1TI): Likewise.
1330 (VEC_SET_V1TI): Likewise.
1331 (VEC_EXT_V1TI): Likewise.
1332 (EQV_V1TI): Likewise.
1333 (NAND_V1TI): Likewise.
1334 (ORC_V1TI): Likewise.
1335 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1336 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1338 (VADDUQM): Likewise.
1339 (VSUBCUQ): Likewise.
1340 (VADDEUQM): Likewise.
1341 (VADDECUQ): Likewise.
1342 (VSUBEUQM): Likewise.
1343 (VSUBECUQ): Likewise.
1345 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1346 __int128_t and __uint128_t types.
1347 (__uint128_type): Likewise.
1348 (altivec_categorize_keyword): Add support for vector __int128_t,
1349 vector __uint128_t, vector __int128, and vector unsigned __int128
1350 as a container type for TImode operations that need to be done in
1351 VSX/Altivec registers.
1352 (rs6000_macro_to_expand): Likewise.
1353 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1354 to support 128-bit integer instructions vaddcuq, vadduqm,
1355 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1356 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1358 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1359 for V1TImode, and set up preferences to use VSX/Altivec registers.
1360 Setup VSX reload handlers.
1361 (rs6000_debug_reg_global): Likewise.
1362 (rs6000_init_hard_regno_mode_ok): Likewise.
1363 (rs6000_preferred_simd_mode): Likewise.
1364 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1365 (easy_altivec_constant): Likewise.
1366 (output_vec_const_move): Likewise.
1367 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1369 (rs6000_expand_vector_extract): Likewise.
1370 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1372 (rs6000_const_vec): Add support for V1TImode.
1373 (rs6000_emit_le_vsx_load): Swap double words when loading or
1374 storing TImode/V1TImode.
1375 (rs6000_emit_le_vsx_store): Likewise.
1376 (rs6000_emit_le_vsx_move): Likewise.
1377 (rs6000_emit_move): Add support for V1TImode.
1378 (altivec_expand_ld_builtin): Likewise.
1379 (altivec_expand_st_builtin): Likewise.
1380 (altivec_expand_vec_init_builtin): Likewise.
1381 (altivec_expand_builtin): Likewise.
1382 (rs6000_init_builtins): Add support for V1TImode type. Add
1383 support for ISA 2.07 128-bit integer builtins. Define type names
1384 for the VSX/Altivec vector types.
1385 (altivec_init_builtins): Add support for overloaded vector
1386 functions with V1TImode type.
1387 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1388 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1390 (rs6000_split_128bit_ok_p): Likewise.
1391 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1392 __int128_t and vector __uint128_t.
1394 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1395 and mode attributes.
1402 (VS_scalar): Likewise.
1403 (VS_double): Likewise.
1404 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1406 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1407 we support the ISA 2.07 128-bit integer arithmetic instructions.
1408 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1409 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1410 and TImode types for use with the builtin functions.
1411 (V1TI_type_node): Likewise.
1412 (unsigned_V1TI_type_node): Likewise.
1413 (intTI_type_internal_node): Likewise.
1414 (uintTI_type_internal_node): Likewise.
1416 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1417 128-bit builtin functions.
1418 (UNSPEC_VADDEUQM): Likewise.
1419 (UNSPEC_VADDECUQ): Likewise.
1420 (UNSPEC_VSUBCUQ): Likewise.
1421 (UNSPEC_VSUBEUQM): Likewise.
1422 (UNSPEC_VSUBECUQ): Likewise.
1423 (VM): Add V1TImode to vector mode iterators.
1425 (VI_unit): Likewise.
1426 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1427 (altivec_vaddcuq): Likewise.
1428 (altivec_vsubuqm): Likewise.
1429 (altivec_vsubcuq): Likewise.
1430 (altivec_vaddeuqm): Likewise.
1431 (altivec_vaddecuq): Likewise.
1432 (altivec_vsubeuqm): Likewise.
1433 (altivec_vsubecuq): Likewise.
1435 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1437 (BOOL_128): Likewise.
1438 (BOOL_REGS_OUTPUT): Likewise.
1439 (BOOL_REGS_OP1): Likewise.
1440 (BOOL_REGS_OP2): Likewise.
1441 (BOOL_REGS_UNARY): Likewise.
1442 (BOOL_REGS_AND_CR0): Likewise.
1444 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1445 128-bit integer builtin support.
1446 (vec_vadduqm): Likewise.
1447 (vec_vaddecuq): Likewise.
1448 (vec_vaddeuqm): Likewise.
1449 (vec_vsubecuq): Likewise.
1450 (vec_vsubeuqm): Likewise.
1451 (vec_vsubcuq): Likewise.
1452 (vec_vsubuqm): Likewise.
1454 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1455 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1456 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1457 128-bit integer add/subtract to ISA 2.07.
1459 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1461 * config/arc/arc.c (arc_predicate_delay_insns):
1462 Fix third argument passed to conditionalize_nonjump.
1464 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1466 * config/aarch64/aarch64-builtins.c
1467 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1468 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1469 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1470 instead of __builtin_lfloor.
1471 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1473 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1475 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1476 (tree_ssa_ifcombine_bb_1): New function.
1477 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1478 is an empty forwarder block to then_bb or vice versa and then_bb
1479 and else_bb are effectively swapped.
1481 2014-03-12 Christian Bruel <christian.bruel@st.com>
1484 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1485 REG_CFA_DEF_CFA note.
1486 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1487 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1489 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1491 PR tree-optimization/60454
1492 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1494 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1496 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1497 Do not define target_cpu_default2 to generic.
1498 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1499 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1500 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1502 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1503 Marc Glisse <marc.glisse@inria.fr>
1505 PR tree-optimization/60502
1506 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1507 instead of build_low_bits_mask.
1509 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1512 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1513 if there are multiple uses, but op doesn't live on E edge.
1514 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1515 clobber stmts before __builtin_unreachable.
1517 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1519 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1520 hard_frame_pointer_rtx.
1521 * cse.c (cse_insn): Remove volatile check.
1522 * cselib.c (cselib_process_insn): Likewise.
1523 * dse.c (scan_insn): Likewise.
1525 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1527 * config/arc/arc.c (conditionalize_nonjump): New function,
1529 (arc_ifcvt): ... this.
1530 (arc_predicate_delay_insns): Use it.
1532 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1534 * config/arc/predicates.md (extend_operand): During/after reload,
1535 allow const_int_operand.
1536 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1537 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1538 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1540 (umulsi3_highpart_i): Likewise.
1542 2014-03-11 Richard Biener <rguenther@suse.de>
1544 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1545 Add asserts to guard possible wrong-code bugs.
1547 2014-03-11 Richard Biener <rguenther@suse.de>
1549 PR tree-optimization/60429
1550 PR tree-optimization/60485
1551 * tree-ssa-structalias.c (set_union_with_increment): Properly
1552 take into account all fields that overlap the shifted vars.
1553 (do_sd_constraint): Likewise.
1554 (do_ds_constraint): Likewise.
1555 (get_constraint_for_ptr_offset): Likewise.
1557 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1559 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1560 (nios2_compute_frame_layout):
1561 Add calculation of cfun->machine->fp_save_offset.
1562 (nios2_expand_prologue): Correct setting of frame pointer register
1564 (nios2_expand_epilogue): Update recovery of stack pointer from
1565 frame pointer accordingly.
1566 (nios2_initial_elimination_offset): Update calculation of offset
1567 for eliminating to HARD_FRAME_POINTER_REGNUM.
1569 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1572 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1573 cgraph_get_create_node on VAR_DECLs.
1575 2014-03-10 Richard Biener <rguenther@suse.de>
1578 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1580 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1582 * config/vms/vms.opt (vms_float_format): New variable.
1584 2014-03-08 Tobias Burnus <burnus@net-b.de>
1586 * doc/invoke.texi (-fcilkplus): Update implementation status.
1588 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1589 Richard Biener <rguenther@suse.de>
1591 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1592 consistently accross all TUs.
1593 (run_gcc): Enable -fshort-double automatically at link at link-time
1594 and disallow override.
1596 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1599 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1600 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1601 if they can't be used.
1603 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1605 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1606 for Solaris 11/x86 ld.
1607 * configure: Regenerate.
1609 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1611 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1612 (LIB_TLS_SPEC): Save as ld_tls_libs.
1613 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1614 (HAVE_AS_IX86_TLSLDM): New test.
1615 * configure, config.in: Regenerate.
1616 * config/i386/i386.c (legitimize_tls_address): Fall back to
1617 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1618 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1619 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1620 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1622 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1624 * common.opt (fira-loop-pressure): Mark as optimization.
1626 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1628 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1629 an OpenMP mappable type.
1631 2014-03-06 Matthias Klose <doko@ubuntu.com>
1633 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1634 MULTILIB_OSDIRNAMES is not defined.
1636 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1637 Meador Inge <meadori@codesourcery.com>
1640 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1641 (arm_legitimize_address): Call legitimize_tls_address for any
1642 arm_tls_referenced_p expression, handle constant addend. Call it
1643 before testing for !TARGET_ARM.
1644 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1646 2014-03-06 Richard Biener <rguenther@suse.de>
1652 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1654 * tree-streamer.c (record_common_node): Assert we don't record
1655 nodes with type double.
1656 (preload_common_node): Skip type double, complex double and double
1657 pointer since it is now frontend dependent due to fshort-double option.
1659 2014-03-06 Richard Biener <rguenther@suse.de>
1661 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1662 or -fno-lto is specified and the linker has full plugin support.
1663 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1664 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1665 * lto-wrapper.c (merge_and_complain): Merge compile-time
1666 optimization levels.
1667 (run_gcc): And pass it through to the link options.
1669 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1673 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1675 * cselib.c (remove_useless_values): Skip to avoid quadratic
1676 behavior if the condition moved from...
1677 (cselib_process_insn): ... here holds.
1679 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1682 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1683 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1686 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1687 (TM_H): Add x86-tune.def.
1689 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1691 * config/aarch64/aarch64.c (generic_tunings):
1692 Use cortexa57_extra_costs.
1694 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1697 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1698 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1699 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1702 2014-03-04 Heiher <r@hev.cc>
1704 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1705 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1707 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1709 * config/i386/predicates.md (const2356_operand): Change to ...
1710 (const2367_operand): ... this.
1711 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1713 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1714 (*avx512pf_scatterpf<mode>sf): Ditto.
1715 (avx512pf_scatterpf<mode>df): Ditto.
1716 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1717 (*avx512pf_scatterpf<mode>df): Ditto.
1718 * config/i386/i386.c (ix86_expand_builtin): Update
1719 incorrect hint operand error message.
1721 2014-03-04 Richard Biener <rguenther@suse.de>
1723 * lto-section-in.c (lto_get_section_data): Fix const cast.
1725 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1727 * tree-streamer.c (record_common_node): Assert we don't record
1728 nodes with type double.
1729 (preload_common_node): Skip type double, complex double and double
1730 pointer since it is now frontend dependent due to fshort-double option.
1732 2014-03-04 Richard Biener <rguenther@suse.de>
1735 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1736 (lto_input_toplevel_asms): Likewise.
1737 * lto-section-in.c (lto_get_section_data): Instead do it here
1740 2014-03-04 Richard Biener <rguenther@suse.de>
1742 PR tree-optimization/60382
1743 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1744 dead PHIs a reduction.
1746 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1748 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1750 (_mm_prefetch): Move out of GCC target("sse") pragma.
1751 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1752 GCC target("prfchw") pragma.
1753 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1755 * config/i386/i386.c (ix86_option_override_internal): Enable
1756 -mprfchw with -mprefetchwt1.
1758 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1760 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1763 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1765 * opts.h (CL_PCH_IGNORE): Define.
1766 * targhooks.c (option_affects_pch_p):
1767 Return false for options that have CL_PCH_IGNORE set.
1768 * opt-functions.awk: Process PchIgnore.
1769 * doc/options.texi: Document PchIgnore.
1771 * config/arc/arc.opt (misize): Add PchIgnore property.
1773 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1775 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1776 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1777 constraint on constants to permit them being loaded into
1778 GENERAL_REGS or BASE_REGS.
1780 2014-03-03 Nick Clifton <nickc@redhat.com>
1782 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1783 anti-cacnonical alternatives.
1784 (negandhi3_real): New pattern.
1785 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1787 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1789 * config/avr/avr-mcus.def: Remove atxmega16x1.
1790 * config/avr/avr-tables.opt: Regenerate.
1791 * config/avr/t-multilib: Regenerate.
1792 * doc/avr-mmcu.texi: Regenerate.
1794 2014-03-03 Tobias Grosser <tobias@grosser.es>
1795 Mircea Namolaru <mircea.namolaru@inria.fr>
1797 PR tree-optimization/58028
1798 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1801 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1803 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1804 not handled by recognizers.
1806 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1809 * function.c (expand_function_end): Don't emit
1810 clobber_return_register sequence if clobber_after is a BARRIER.
1811 * cfgexpand.c (construct_exit_block): Append instructions before
1812 return_label to prev_bb.
1814 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1816 * config/rs6000/constraints.md: Document reserved use of "wc".
1818 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1821 * ipa.c (function_and_variable_visibility): When dissolving comdat
1822 group, also set all symbols to local.
1824 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1829 2013-12-14 Jan Hubicka <jh@suse.cz>
1831 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1833 2014-03-02 Jon Beniston <jon@beniston.com>
1839 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1840 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1841 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1842 (simple_return, *simple_return): New patterns
1843 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1844 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1846 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1848 * dwarf2out.c (gen_subprogram_die): Tidy.
1850 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1853 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1854 (*mov_t_msb_neg_negc): ... this new insn.
1856 2014-02-28 Jason Merrill <jason@redhat.com>
1859 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1862 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1865 * dwarf2out.c (decltype_auto_die): New static.
1866 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1867 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1868 (is_cxx_auto): Likewise.
1870 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1872 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1873 we are not using general regs only.
1875 2014-02-28 Richard Biener <rguenther@suse.de>
1878 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1879 previous fix and only allow to remove trivial pre-headers
1880 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1881 (remove_forwarder_block): Properly update the latch of a loop.
1883 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1886 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1887 (cselib_preserved_hash_table): New.
1888 (preserve_constants_and_equivs): Move preserved vals to it.
1889 (cselib_find_slot): Look it up first.
1890 (cselib_init): Initialize it.
1891 (cselib_finish): Release it.
1892 (dump_cselib_table): Dump it.
1894 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1897 * cselib.c (remove_useless_values): Skip to avoid quadratic
1898 behavior if the condition moved from...
1899 (cselib_process_insn): ... here holds.
1901 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1904 * var-tracking.c (vt_initialize): Apply the same condition to
1905 preserve the CFA base value.
1907 2014-02-28 Joey Ye <joey.ye@arm.com>
1910 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1911 if reload in progress or completed.
1913 2014-02-28 Tobias Burnus <burnus@net-b.de>
1916 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1919 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1921 * doc/tm.texi.in (Condition Code Status): Update documention for
1922 relative locations of cc0-setter and cc0-user.
1924 2014-02-27 Jeff Law <law@redhat.com>
1926 PR rtl-optimization/52714
1927 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1928 into two independent simple sets, if I3 is a jump, ensure the
1929 pattern we place into I3 is a (set (pc) ...).
1931 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1932 Jeff Law <law@redhat.com>
1934 PR rtl-optimization/49847
1935 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1936 are in different blocks.
1937 * doc/tm.texi (Condition Code Status): Update documention for
1938 relative locations of cc0-setter and cc0-user.
1940 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1943 * lra.c (lra_emit_add): Check SUBREG too.
1945 2014-02-27 Andreas Schwab <schwab@suse.de>
1947 * config/m68k/m68k.c (m68k_option_override): Disable
1948 -flive-range-shrinkage for classic m68k.
1949 (m68k_override_options_after_change): Likewise.
1951 2014-02-27 Marek Polacek <polacek@redhat.com>
1954 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1955 -Wmaybe-uninitialized.
1957 2014-02-27 Alan Modra <amodra@gmail.com>
1960 * reload1.c (emit_input_reload_insns): When reload_override_in,
1961 set old to rl->in_reg when rl->in_reg is a subreg.
1963 2014-02-26 Richard Biener <rguenther@suse.de>
1966 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1968 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1970 * common/config/i386/predicates.md (const1256_operand): Remove.
1971 (const2356_operand): New.
1972 (const_1_to_2_operand): Remove.
1973 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1974 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1975 (*avx512pf_gatherpf<mode>sf): Ditto.
1976 (avx512pf_gatherpf<mode>df): Ditto.
1977 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1978 (*avx512pf_gatherpf<mode>df): Ditto.
1979 (avx512pf_scatterpf<mode>sf): Ditto.
1980 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1981 (*avx512pf_scatterpf<mode>sf): Ditto.
1982 (avx512pf_scatterpf<mode>df): Ditto.
1983 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1984 (*avx512pf_scatterpf<mode>df): Ditto.
1985 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1987 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1989 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1990 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1991 (_mm512_mask_testn_epi64_mask): Move to ...
1992 * config/i386/avx512cdintrin.h: Here.
1993 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1994 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1995 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1996 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1997 TARGET_AVX512F from TARGET_AVX512CD.
1999 2014-02-26 Richard Biener <rguenther@suse.de>
2002 * ipa.c (walk_polymorphic_call_targets): Properly guard
2003 call to inline_update_overall_summary.
2005 2014-02-26 Bin Cheng <bin.cheng@arm.com>
2008 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
2009 and latches only if requested. Fix latch if it is removed.
2010 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
2011 LOOPS_HAVE_PREHEADERS.
2013 2014-02-25 Andrew Pinski <apinski@cavium.com>
2015 * builtins.c (expand_builtin_thread_pointer): Create a new target
2016 when the target is NULL.
2018 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
2020 PR rtl-optimization/60317
2021 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2022 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
2023 * lra-assigns.c: Include params.h.
2024 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
2025 other reload pseudos considerations.
2027 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2029 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
2030 to use canonical form for nor<mode>3.
2032 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2035 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2038 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2040 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2041 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2042 (ix86_handle_option): Handle OPT_mprefetchwt1.
2043 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2044 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2046 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2047 OPTION_MASK_ISA_PREFETCHWT1.
2048 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2049 (PTA_PREFETCHWT1): New.
2050 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2051 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2052 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2053 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2054 (*prefetch_avx512pf_<mode>_: Change into ...
2055 (*prefetch_prefetchwt1_<mode>: This.
2056 * config/i386/i386.opt (mprefetchwt1): New.
2057 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2058 (_mm_prefetch): Handle intent to write.
2059 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2061 2014-02-25 Richard Biener <rguenther@suse.de>
2064 * emit-rtl.c (mem_attrs_htab): Remove.
2065 (mem_attrs_htab_hash): Likewise.
2066 (mem_attrs_htab_eq): Likewise.
2067 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2068 (init_emit_once): Do not allocate mem_attrs_htab.
2070 2014-02-25 Richard Biener <rguenther@suse.de>
2073 * lto-opts.c (lto_write_options): Output non-explicit conservative
2074 -fwrapv, -fno-trapv and -fno-strict-overflow.
2075 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2076 (run_gcc): And pass them through.
2078 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2080 * sel-sched.c (calculate_new_fences): New parameter ptime.
2081 Calculate it as a maximum over all fence cycles.
2082 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2083 Print the final schedule timing when sched_verbose.
2085 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2087 PR rtl-optimization/60292
2088 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2089 bit fot the fence instruction.
2091 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2093 * calls.h: Fix typo in comment.
2095 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2097 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2098 adjusting offsetable addresses.
2100 2014-02-24 Guozhi Wei <carrot@google.com>
2102 * sparseset.h (sparseset_pop): Fix the wrong index.
2104 2014-02-24 Walter Lee <walt@tilera.com>
2106 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2107 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2109 * common/config/tilegx/tilegx-common.c
2110 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2111 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2113 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2114 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2115 (tilegx_gimplify_va_arg_expr): Handle big endian.
2116 (tilegx_expand_unaligned_load): Ditto.
2117 (tilegx_expand_unaligned_store): Ditto.
2118 (TARGET_RETURN_IN_MSB): New.
2119 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2120 (TARGET_ENDIAN_DEFAULT): New.
2121 (TARGET_BIG_ENDIAN): Handle big endian.
2122 (BYTES_BIG_ENDIAN): Ditto.
2123 (WORDS_BIG_ENDIAN): Ditto.
2124 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2127 * config/tilegx/tilegx.md (extv): Handle big endian.
2129 (insn_st<n>): Ditto.
2130 (insn_st<n>_add<bitsuffix>): Ditto.
2131 (insn_stnt<n>): Ditto.
2132 (insn_stnt<n>_add<bitsuffix>):Ditto.
2133 (vec_interleave_highv8qi): Handle big endian.
2134 (vec_interleave_highv8qi_be): New.
2135 (vec_interleave_highv8qi_le): New.
2136 (insn_v1int_h): Handle big endian.
2137 (vec_interleave_lowv8qi): Handle big endian.
2138 (vec_interleave_lowv8qi_be): New.
2139 (vec_interleave_lowv8qi_le): New.
2140 (insn_v1int_l): Handle big endian.
2141 (vec_interleave_highv4hi): Handle big endian.
2142 (vec_interleave_highv4hi_be): New.
2143 (vec_interleave_highv4hi_le): New.
2144 (insn_v2int_h): Handle big endian.
2145 (vec_interleave_lowv4hi): Handle big endian.
2146 (vec_interleave_lowv4hi_be): New.
2147 (vec_interleave_lowv4hi_le): New.
2148 (insn_v2int_l): Handle big endian.
2149 (vec_interleave_highv2si): Handle big endian.
2150 (vec_interleave_highv2si_be): New.
2151 (vec_interleave_highv2si_le): New.
2152 (insn_v4int_h): Handle big endian.
2153 (vec_interleave_lowv2si): Handle big endian.
2154 (vec_interleave_lowv2si_be): New.
2155 (vec_interleave_lowv2si_le): New.
2156 (insn_v4int_l): Handle big endian.
2157 * config/tilegx/tilegx.opt (mbig-endian): New option.
2158 (mlittle-endian): New option.
2159 * doc/install.texi: Document tilegxbe-linux.
2160 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2162 2014-02-24 Martin Jambor <mjambor@suse.cz>
2165 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2166 there are no parameter descriptors.
2168 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2170 PR rtl-optimization/60268
2171 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2172 initialization to ...
2173 (sched_rgn_init): ... here.
2174 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2176 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2178 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2181 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2183 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2186 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2188 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2189 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2191 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2193 * config/microblaze/predicates.md: Add cmp_op predicate.
2194 * config/microblaze/microblaze.md: Add branch_compare instruction
2195 which uses cmp_op predicate and emits cmp insn before branch.
2196 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2197 to microblaze_expand_conditional_branch and consolidate logic.
2198 (microblaze_expand_conditional_branch): emit branch_compare
2199 insn instead of handling cmp op separate from branch insn.
2201 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2203 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2206 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2208 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2209 define_insn with define_expand and new define_insn
2210 *altivec_lve<VI_char>x_internal.
2211 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2212 and new define_insn *altivec_stve<VI_char>x_internal.
2213 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2215 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2217 (altivec_expand_stvex_be): New function.
2219 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2221 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2222 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2223 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2224 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2226 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2229 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2230 instead of emit_move_insn.
2232 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2234 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2236 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2237 gen_altivec_vsumsws.
2239 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2241 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2242 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2243 (altivec_lvxl_<mode>): New define_expand incorporating
2244 -maltivec=be semantics where needed.
2245 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2246 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2247 semantics where needed.
2248 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2249 (altivec_stvx_<mode>): New define_expand incorporating
2250 -maltivec=be semantics where needed.
2251 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2252 VM2 iterator instead of V4SI.
2253 (altivec_stvxl_<mode>): New define_expand incorporating
2254 -maltivec=be semantics where needed.
2255 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2256 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2257 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2258 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2259 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2260 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2261 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2262 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2263 ALTIVEC_BUILTIN_STVXL.
2264 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2265 (altivec_expand_stvx_be): Likewise.
2266 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2267 (altivec_expand_lvx_be): Likewise.
2268 (altivec_expand_stvx_be): Likewise.
2269 (altivec_expand_builtin): Add cases for
2270 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2271 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2272 (altivec_init_builtins): Add definitions for
2273 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2274 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2276 2014-02-21 Catherine Moore <clm@codesourcery.com>
2278 * doc/invoke.texi (mvirt, mno-virt): Document.
2279 * config/mips/mips.opt (mvirt): New option.
2280 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2282 2014-02-21 Richard Biener <rguenther@suse.de>
2284 PR tree-optimization/60276
2285 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2286 (STMT_VINFO_MIN_NEG_DIST): New macro.
2287 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2288 STMT_VINFO_MIN_NEG_DIST.
2289 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2290 made for negative dependence distances still hold.
2292 2014-02-21 Richard Biener <rguenther@suse.de>
2295 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2296 DECL_INITIAL for globals not in the current function context.
2298 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2300 PR tree-optimization/56490
2301 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2302 * tree-ssa-uninit.c: Include params.h.
2303 (compute_control_dep_chain): Add num_calls argument, return false
2304 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2305 num_calls to recursive call.
2306 (find_predicates): Change dep_chain into normal array,
2307 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2308 variable and adjust compute_control_dep_chain caller.
2309 (find_def_preds): Likewise.
2311 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2313 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2314 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2316 2014-02-21 Nick Clifton <nickc@redhat.com>
2318 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2319 (pushhi1): Likewise.
2320 (popqi1): Add mode to pre_dec.
2323 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2325 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2326 mode for mask of V8SFmode permutation.
2328 2014-02-20 Richard Henderson <rth@redhat.com>
2331 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2332 a new pseudo for OLDVAL.
2334 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2337 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2338 gen_reg_rtx if d->testing_p.
2339 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2340 if d->testing_p and we will certainly return true.
2341 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2344 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2346 * emit-rtl.c (gen_reg_rtx): Assert that
2347 crtl->emit.regno_pointer_align_length is non-zero.
2349 2014-02-20 Richard Henderson <rth@redhat.com>
2352 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2353 on failure the store back into EXPECT.
2355 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2356 Sandra Loosemore <sandra@codesourcery.com>
2358 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2359 * config/nios2/nios2.c (nios2_function_profiler): Add
2360 -fPIC (flag_pic == 2) support.
2361 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2362 (nios2_large_offset_p): New function.
2363 (nios2_unspec_reloc_p): Move up position, update to use
2364 nios2_large_offset_p.
2365 (nios2_unspec_address): Remove function.
2366 (nios2_unspec_offset): New function.
2367 (nios2_large_got_address): New function.
2368 (nios2_got_address): Add large offset support.
2369 (nios2_legitimize_tls_address): Update usage of removed and new
2371 (nios2_symbol_binds_local_p): New function.
2372 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2373 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2374 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2375 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2376 processing for (const (unspec ...)).
2377 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2379 2014-02-20 Richard Biener <rguenther@suse.de>
2381 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2382 doing the substitution.
2383 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2385 2014-02-20 Martin Jambor <mjambor@suse.cz>
2388 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2389 info when checking whether lattices are bottom.
2391 2014-02-20 Richard Biener <rguenther@suse.de>
2394 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2397 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2400 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2401 parameter specifying the scaling.
2402 (inline_call): Update.
2403 (want_inline_recursively): Guard division by zero.
2404 (recursive_inlining): Update.
2405 * ipa-inline.h (clone_inlined_nodes): Update.
2407 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2410 * config/i386/i386.c (classify_argument): Pass structures of size
2411 64 bytes or less in register.
2413 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2414 Kirill Yukhin <kirill.yukhin@intel.com>
2416 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2417 (_mm_rcp28_round_ss): Ditto.
2418 (_mm_rsqrt28_round_sd): Ditto.
2419 (_mm_rsqrt28_round_ss): Ditto.
2420 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2421 (_mm_rcp14_round_ss): Ditto.
2422 (_mm_rsqrt14_round_sd): Ditto.
2423 (_mm_rsqrt14_round_ss): Ditto.
2424 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2425 the first input operand, get rid of match_dup.
2426 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2428 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2430 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2431 operand as the first input operand, set type attribute.
2432 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2434 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2435 operand as the first input operand, set type attribute.
2437 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2439 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2442 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2445 * config/i386/i386.c (construct_container): Remove TFmode check
2446 for X86_64_INTEGER_CLASS.
2448 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2451 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2452 only when -Wpsabi is enabled.
2454 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2457 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2458 passing arrays in registers are the same as for structs, so remove the
2459 special case for them.
2461 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2463 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2464 destination type, extract only the valid bits if the source type is not
2465 integral and has a different mode.
2467 2014-02-19 Richard Biener <rguenther@suse.de>
2470 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2473 2014-02-19 Richard Biener <rguenther@suse.de>
2476 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2477 (ipa_modify_call_arguments): Emit an argument load explicitely and
2478 preserve virtual SSA form there and for the replacement call.
2479 Do not update SSA form nor free dominance info.
2481 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2483 * ipa.c (function_and_variable_visibility): Also clear WEAK
2484 flag when disolving COMDAT_GROUP.
2486 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2488 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2489 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2491 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2492 do more sanity checks.
2493 (detect_type_change): Return true when giving up early.
2494 (compute_complex_assign_jump_func): Fix type parameter of
2495 ipa_set_ancestor_jf.
2496 (compute_complex_ancestor_jump_func): Likewise.
2497 (update_jump_functions_after_inlining): Fix updating of
2499 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2501 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2503 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2504 inline clones when edge disappears.
2506 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2509 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2510 Split 64-bit moves into 2 patterns. Do not allow the use of
2511 direct move for TDmode in little endian, since the decimal value
2512 has little endian bytes within a word, but the 64-bit pieces are
2513 ordered in a big endian fashion, and normal subreg's of TDmode are
2515 (mov<mode>_64bit_dm): Likewise.
2516 (movtd_64bit_nodm): Likewise.
2518 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2520 PR tree-optimization/60174
2521 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2522 statement of an SSA_NAME that occurs in an abnormal PHI node.
2524 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2527 * final.c (SEEN_BB): Remove.
2528 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2529 (final_scan_insn): Don't force_source_line on second
2530 NOTE_INSN_BASIC_BLOCK.
2532 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2535 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2536 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2537 (type_natural_mode): Warn ABI change when %zmm register is not
2538 available for AVX512F vector value passing.
2540 2014-02-18 Kai Tietz <ktietz@redhat.com>
2543 * config/i386/i386.c (ix86_expand_prologue): Use value in
2544 rax register as displacement when restoring %r10 or %rax.
2545 Fix wrong offset when restoring both registers.
2547 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2549 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2550 assertion with conditional return.
2552 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2553 Uros Bizjak <ubizjak@gmail.com>
2556 * config/i386/driver-i386.c (host_detect_local_cpu): If
2557 YMM state is not saved by the OS, also clear has_f16c. Move
2558 CPUID 0x80000001 handling before YMM state saving checking.
2560 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2562 PR rtl-optimization/58960
2563 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2564 factored out from ...
2565 (sched_init): ... here.
2566 (free_global_sched_pressure_data): New, factored out from ...
2567 (sched_finish): ... here.
2568 * sched-int.h (free_global_sched_pressure_data): Declare.
2569 * sched-rgn.c (nr_regions_initial): New static global.
2570 (haifa_find_rgns): Initialize it.
2571 (schedule_region): Disable sched-pressure for the newly
2574 2014-02-17 Richard Biener <rguenther@suse.de>
2576 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2577 release SSA defs of pattern stmts.
2579 2014-02-17 Richard Biener <rguenther@suse.de>
2581 * tree-inline.c (expand_call_inline): Release the virtual
2582 operand defined by the call we are about to inline.
2584 2014-02-17 Richard Biener <rguenther@suse.de>
2586 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2588 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2589 Ilya Tocar <ilya.tocar@intel.com>
2591 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2592 arguments order in builtin.
2593 (_mm512_permutexvar_epi64): Ditto.
2594 (_mm512_mask_permutexvar_epi64): Ditto
2595 (_mm512_maskz_permutexvar_epi32): Ditto
2596 (_mm512_permutexvar_epi32): Ditto
2597 (_mm512_mask_permutexvar_epi32): Ditto
2599 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2601 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2602 (p8_vmrgow): Likewise.
2604 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2606 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2609 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2612 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2613 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2614 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2615 using direct move instructions on ISA 2.07. Also adjust
2616 instruction length for 64-bit.
2617 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2618 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2620 2014-02-15 Alan Modra <amodra@gmail.com>
2624 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2625 find_replacement on parts of insn rtl that might be reloaded.
2627 2014-02-15 Richard Biener <rguenther@suse.de>
2629 PR tree-optimization/60183
2630 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2631 (tree_ssa_phiprop): Calculate and free post-dominators.
2633 2014-02-14 Jeff Law <law@redhat.com>
2635 PR rtl-optimization/60131
2636 * ree.c (get_extended_src_reg): New function.
2637 (combine_reaching_defs): Use it rather than assuming location of REG.
2638 (find_and_remove_re): Verify first operand of extension is
2639 a REG before adding the insns to the copy list.
2641 2014-02-14 Roland McGrath <mcgrathr@google.com>
2643 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2644 * configure: Regenerated.
2645 * config.in: Regenerated.
2646 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2647 instead of ASM_SHORT.
2649 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2650 Richard Earnshaw <rearnsha@arm.com>
2652 PR rtl-optimization/59535
2653 * lra-constraints.c (process_alt_operands): Encourage alternative
2654 when unassigned pseudo class is superset of the alternative class.
2655 (inherit_reload_reg): Don't inherit when optimizing for code size.
2656 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2657 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2658 modes not less than 4 for Thumb1.
2660 2014-02-14 Kyle McMartin <kyle@redhat.com>
2663 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2665 2014-02-14 Richard Biener <rguenther@suse.de>
2667 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2668 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2669 Do not strip INDIRECT_REFs.
2671 2014-02-14 Richard Biener <rguenther@suse.de>
2674 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2675 DECL_FUNCTION_SPECIFIC_TARGET.
2676 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2677 * tree-streamer-out.c (pack_ts_target_option): Remove.
2678 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2679 (write_ts_function_decl_tree_pointers): Do not stream
2680 DECL_FUNCTION_SPECIFIC_TARGET.
2681 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2682 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2683 (lto_input_ts_function_decl_tree_pointers): Do not stream
2684 DECL_FUNCTION_SPECIFIC_TARGET.
2686 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2688 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2689 (get_initial_def_for_induction, vectorizable_induction): Ignore
2690 debug stmts when looking for exit_phi.
2691 (vectorizable_live_operation): Fix up condition.
2693 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2695 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2696 nreverse() because it changes the content of original tree list.
2698 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2700 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2701 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2703 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2705 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2706 GNU coding standards.
2708 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2711 * dwarf2out.c (gen_subprogram_die): Don't call
2712 add_calling_convention_attribute if subr_die is old_die.
2714 2014-02-13 Sharad Singhai <singhai@google.com>
2716 * doc/optinfo.texi: Fix order of nodes.
2718 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2720 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2721 operands[2], not operands[3].
2723 2014-02-13 Richard Biener <rguenther@suse.de>
2726 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2727 mention the possibility of an in-tree build.
2728 (CLooG): Update recommended version to 0.18.1, mention the
2729 possibility of an in-tree build and clarify that the ISL
2730 bundled with CLooG does not work.
2732 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2735 * expr.c (compress_float_constant): If x is a hard register,
2736 extend into a pseudo and then move to x.
2738 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2740 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2741 caused by bad second argument to warning_at() with -mhotpatch and
2742 nested functions (e.g. with gfortran).
2744 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2746 * opts.c (option_name): Remove "enabled by default" rider.
2748 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2750 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2752 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2753 Uros Bizjak <ubizjak@gmail.com>
2756 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2757 * configure: Regenerated.
2759 2014-02-12 Richard Biener <rguenther@suse.de>
2761 * vec.c (vec_prefix::calculate_allocation): Move as
2762 inline variant to vec.h.
2763 (vec_prefix::calculate_allocation_1): New out-of-line version.
2764 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2765 (vec_prefix::m_has_auto_buf): Rename to ...
2766 (vec_prefix::m_using_auto_storage): ... this.
2767 (vec_prefix::calculate_allocation): Inline the easy cases
2768 and dispatch to calculate_allocation_1 which doesn't need the
2770 (va_heap::reserve): Use gcc_checking_assert.
2771 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2772 m_using_auto_storage.
2773 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2775 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2776 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2777 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2779 2014-02-12 Richard Biener <rguenther@suse.de>
2781 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2782 when we found a dependence.
2784 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2786 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2788 (maybe_fold_stmt): ... into this new function.
2789 * omp-low.c (lower_omp): Update comment.
2791 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2794 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2797 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2799 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2800 identifiers in comments.
2801 (cortexa53_extra_costs): Likewise.
2802 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2803 (cortexa7_extra_costs): Likewise.
2804 (cortexa12_extra_costs): Likewise.
2805 (cortexa15_extra_costs): Likewise.
2806 (v7m_extra_costs): Likewise.
2808 2014-02-12 Richard Biener <rguenther@suse.de>
2811 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2812 of posix_memalign being successful.
2813 (lower_stmt): Restrict lowering of posix_memalign to when
2814 -ftree-bit-ccp is enabled.
2816 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2818 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2820 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2822 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2824 PR rtl-optimization/60116
2825 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2826 other_insn once the combination has been validated.
2828 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2831 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2833 * ipa-devirt.c: Include demangle.h
2834 (odr_violation_reported): New static variable.
2835 (add_type_duplicate): Update odr_violations.
2836 (maybe_record_node): Add completep parameter; update it.
2837 (record_target_from_binfo): Add COMPLETEP parameter;
2838 update it as needed.
2839 (possible_polymorphic_call_targets_1): Likewise.
2840 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2841 rename FINAL to COMPLETE.
2842 (record_targets_from_bases): Sanity check we found the binfo;
2843 fix COMPLETEP updating.
2844 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2845 parameter, fix computing of COMPLETEP.
2846 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2847 at LTO time do demangling.
2848 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2849 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2851 (gimple_get_virt_method_for_binfo): Likewise.
2852 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2853 gimple_get_virt_method_for_vtable): Update prototypes.
2855 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2858 * genautomata.c (add_presence_absence): Fix typo with
2859 {final_}presence_list.
2861 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2864 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2865 for VSX/Altivec vectors that land in GPR registers.
2867 2014-02-11 Richard Henderson <rth@redhat.com>
2868 Jakub Jelinek <jakub@redhat.com>
2871 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2872 around drhs if type conversion to lacc->type is not useless.
2874 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2876 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2878 (cortex-a57.cortex-a53): Likewise.
2879 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2881 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2883 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2886 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2888 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2889 add_options_for_arm_vfp3.
2891 2014-02-11 Jeff Law <law@redhat.com>
2894 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2895 object with an undesirable mode.
2897 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2900 * config/i386/sol2-9.h: New file.
2901 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2902 *-*-solaris2.9*): Use it.
2904 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2906 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2907 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2909 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2911 * config/microblaze/microblaze.c: Extend mcpu version format
2913 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2915 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2917 2014-02-10 Richard Henderson <rth@redhat.com>
2920 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2921 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2922 ms-abi vs -mno-accumulate-outgoing-args.
2923 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2924 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2927 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2930 * cfgexpand.c (expand_asm_operands): Attach source location to
2931 ASM_INPUT rtx objects.
2932 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2934 2014-02-10 Nick Clifton <nickc@redhat.com>
2936 * config/mn10300/mn10300.c (popcount): New function.
2937 (mn10300_expand_prologue): Include saved registers in stack usage
2940 2014-02-10 Jeff Law <law@redhat.com>
2943 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2944 when changing the SET_DEST of a prior insn to avoid an input reload.
2946 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2948 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2949 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2950 -mcall-openbsd, or -mcall-linux.
2951 (CC1_ENDIAN_BIG_SPEC): Remove.
2952 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2953 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2954 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2955 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2956 and %cc1_endian_default.
2957 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2959 2014-02-10 Richard Biener <rguenther@suse.de>
2961 PR tree-optimization/60115
2962 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2963 MEM_REF handling. Properly verify that the accesses are not
2964 out of the objects bound.
2966 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2968 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2971 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2973 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2974 proper constants and fix formatting.
2975 (possible_polymorphic_call_targets): Fix formatting.
2977 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2978 Ilya Tocar <ilya.tocar@intel.com>
2980 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2981 (_mm512_loadu_epi32): Renamed into...
2982 (_mm512_loadu_si512): This.
2983 (_mm512_storeu_epi32): Renamed into...
2984 (_mm512_storeu_si512): This.
2985 (_mm512_maskz_ceil_ps): Removed.
2986 (_mm512_maskz_ceil_pd): Ditto.
2987 (_mm512_maskz_floor_ps): Ditto.
2988 (_mm512_maskz_floor_pd): Ditto.
2989 (_mm512_floor_round_ps): Ditto.
2990 (_mm512_floor_round_pd): Ditto.
2991 (_mm512_ceil_round_ps): Ditto.
2992 (_mm512_ceil_round_pd): Ditto.
2993 (_mm512_mask_floor_round_ps): Ditto.
2994 (_mm512_mask_floor_round_pd): Ditto.
2995 (_mm512_mask_ceil_round_ps): Ditto.
2996 (_mm512_mask_ceil_round_pd): Ditto.
2997 (_mm512_maskz_floor_round_ps): Ditto.
2998 (_mm512_maskz_floor_round_pd): Ditto.
2999 (_mm512_maskz_ceil_round_ps): Ditto.
3000 (_mm512_maskz_ceil_round_pd): Ditto.
3001 (_mm512_expand_pd): Ditto.
3002 (_mm512_expand_ps): Ditto.
3003 * config/i386/i386.c (ix86_builtins): Remove
3004 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
3005 (bdesc_args): Ditto.
3006 * config/i386/predicates.md (const1256_operand): New.
3007 (const_1_to_2_operand): Ditto.
3008 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
3009 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3010 (*avx512pf_gatherpf<mode>sf): Ditto.
3011 (avx512pf_gatherpf<mode>df): Ditto.
3012 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3013 (*avx512pf_gatherpf<mode>df): Ditto.
3014 (avx512pf_scatterpf<mode>sf): Ditto.
3015 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3016 (*avx512pf_scatterpf<mode>sf): Ditto.
3017 (avx512pf_scatterpf<mode>df): Ditto.
3018 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3019 (*avx512pf_scatterpf<mode>df): Ditto.
3020 (avx512f_expand<mode>): Removed.
3021 (<shift_insn><mode>3<mask_name>): Change predicate type.
3023 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3025 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
3026 not at the end of datarefs vector use ordered_remove to avoid
3027 reordering datarefs vector.
3030 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
3031 mark local addressable non-static vars as GOVD_PRIVATE
3032 instead of GOVD_LOCAL.
3033 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3034 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3038 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3039 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3040 assume_aligned or alloc_align attributes.
3041 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3042 arguments. Handle also assume_aligned and alloc_align attributes.
3043 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3044 calls to functions with assume_aligned or alloc_align attributes.
3045 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3047 2014-02-08 Terry Guo <terry.guo@arm.com>
3049 * doc/invoke.texi: Document ARM -march=armv7e-m.
3051 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3053 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3054 flag on __cilkrts_rethrow builtin.
3057 * ipa-cp.c (determine_versionability): Fail at -O0
3058 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3059 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3062 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3065 * tree-inline.c (copy_forbidden): Fail for
3066 __attribute__((optimize (0))) functions.
3068 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3070 * varpool.c: Include pointer-set.h.
3071 (varpool_remove_unreferenced_decls): Variables in other partitions
3072 will not be output; be however careful to not lose information
3075 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3077 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3078 lookup in the vtable constructor.
3080 2014-02-07 Jeff Law <law@redhat.com>
3083 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3084 define_insn_and_split.
3086 * ipa-inline.c (inline_small_functions): Fix typos.
3088 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3090 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3091 (s390_can_use_return_insn): Declare.
3092 * config/s390/s390.h (EPILOGUE_USES): Define.
3093 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3095 (s390_chunkify_start): Handle return JUMP_LABELs.
3096 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3097 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3098 (s390_can_use_return_insn): New functions.
3099 (s390_fix_long_loop_prediction): Handle conditional returns.
3100 (TARGET_SET_UP_BY_PROLOGUE): Define.
3101 * config/s390/s390.md (ANY_RETURN): New code iterator.
3102 (*creturn, *csimple_return, return, simple_return): New patterns.
3104 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3106 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3107 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3108 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3109 REG_CFA_RESTORE list when deciding not to restore a register.
3111 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3113 * config/s390/s390.c: Include tree-pass.h and context.h.
3114 (s390_early_mach): New function, split out from...
3115 (s390_emit_prologue): ...here.
3116 (pass_data_s390_early_mach): New pass structure.
3117 (pass_s390_early_mach): New class.
3118 (s390_option_override): Create and register early_mach pass.
3119 Move to end of file.
3121 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3123 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3124 to match for the exit block.
3126 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3128 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3129 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3130 Reject misaligned operands.
3132 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3134 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3136 2014-02-07 Richard Biener <rguenther@suse.de>
3139 * gimple-low.c (lower_builtin_posix_memalign): New function.
3140 (lower_stmt): Call it to lower posix_memalign in a way
3141 to make alignment info accessible.
3143 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3146 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3147 __builtin_setjmp_receiver.
3149 2014-02-07 Richard Biener <rguenther@suse.de>
3152 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3153 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3154 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3155 Handle BUILT_IN_POSIX_MEMALIGN.
3156 (find_func_clobbers): Likewise.
3157 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3158 (call_may_clobber_ref_p_1): Likewise.
3160 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3163 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3166 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3169 * lto-cgraph.c (lto_output_node): Use
3170 symtab_get_symbol_partitioning_class.
3171 (lto_output_varpool_node): likewise.
3172 (symtab_get_symbol_partitioning_class): Move here from
3174 * cgraph.h (symbol_partitioning_class): Likewise.
3175 (symtab_get_symbol_partitioning_class): Declare.
3177 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3179 * ggc.h (ggc_internal_cleared_alloc): New macro.
3180 * vec.h (vec_safe_copy): Handle memory stats.
3181 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3182 * target-globals.c (save_target_globals): Likewise.
3184 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3187 * expr.c (emit_move_resolve_push): Export; be bit more selective
3188 on when to clear alias set.
3189 * expr.h (emit_move_resolve_push): Declare.
3190 * function.h (struct function): Add tail_call_marked.
3191 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3192 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3193 * config/i386/i386.md (TImode move expander): De not call
3195 (FP push expanders): Preserve memory attributes.
3196 * config/i386/sse.md (push<mode>1): Remove.
3197 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3198 (ix86_expand_push): Remove.
3199 * config/i386/mmx.md (push<mode>1): Remove.
3201 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3203 PR rtl-optimization/60030
3204 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3205 lopart with paradoxical subreg before shifting it up by hprec.
3207 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3209 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3210 Remove extra newline at end of file.
3211 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3212 (arm_issue_rate): Handle cortexa57.
3213 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3214 (cortex-a57.cortex-a53): Likewise.
3216 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3219 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3220 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3222 (arm_expand_prologue): Adjust all callers.
3223 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3224 info, registers also at the lowest numbered registers side. Use
3225 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3229 * var-tracking.c (adjust_mems): Before adding a SET to
3230 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3232 2014-02-06 Alan Modra <amodra@gmail.com>
3235 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3236 change SDmode to DDmode when lra_in_progress.
3238 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3241 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3242 free_data_ref on the dr first, and before goto again also set dr
3243 to the next dr. For simd_lane_access, free old datarefs[i] before
3244 overwriting it. For get_vectype_for_scalar_type failure, don't
3245 free_data_ref if simd_lane_access.
3247 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3250 * tree.h (opts_for_fn): New inline function.
3251 (opt_for_fn): Define.
3252 * config/i386/i386.c (ix86_function_regparm): Use
3253 opt_for_fn (decl, optimize) instead of optimize.
3255 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3257 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3258 for SYMBOL_REF in large memory model.
3260 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3262 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3264 (cortex-a57): Likewise.
3265 (cortex-a57.cortex-a53): Likewise.
3267 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3268 Kugan Vivekanandarajah <kuganv@linaro.org>
3270 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3272 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3274 2014-02-06 Richard Biener <rguenther@suse.de>
3276 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3277 set_loop_copy and initialize_original_copy_tables.
3279 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3281 * config/aarch64/aarch64-simd.md
3282 (aarch64_ashr_simddi): Change QI to SI.
3284 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3285 Jakub Jelinek <jakub@redhat.com>
3288 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3291 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3293 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3294 CODE_FOR_altivec_vpku[hw]um to
3295 CODE_FOR_altivec_vpku[hw]um_direct.
3296 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3297 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3298 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3299 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3301 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3303 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3304 generation for -maltivec=be.
3305 (altivec_vsumsws): Simplify redundant test.
3307 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3309 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3310 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3311 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3312 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3313 gen_altivec_vpkuwum.
3314 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3316 (altivec_vpks<VI_char>ss): Likewise.
3317 (altivec_vpks<VI_char>us): Likewise.
3318 (altivec_vpku<VI_char>us): Likewise.
3319 (altivec_vpku<VI_char>um): Likewise.
3320 (altivec_vpku<VI_char>um_direct): New (copy of
3321 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3323 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3324 target is little endian and -maltivec=be is not specified.
3325 (*altivec_vupkhs<VU_char>_direct): New (copy of
3326 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3327 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3328 target is little endian and -maltivec=be is not specified.
3329 (*altivec_vupkls<VU_char>_direct): New (copy of
3330 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3331 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3332 little endian and -maltivec=be is not specified.
3333 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3334 little endian and -maltivec=be is not specified.
3336 2014-02-05 Richard Henderson <rth@redhat.com>
3339 * combine-stack-adj.c: Revert r206943.
3340 * sched-int.h (struct deps_desc): Add last_args_size.
3341 * sched-deps.c (init_deps): Initialize it.
3342 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3343 contain REG_ARGS_SIZE notes.
3345 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3347 * lto-cgraph.c (asm_nodes_output): Make global.
3348 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3349 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3350 (driver_handle_option): Handle OPT_fwpa.
3352 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3355 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3356 a comment typo and formatting issue. If odr_hash hasn't been
3357 created, return vNULL and set *completep to false.
3360 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3361 bb with no successors.
3363 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3366 * doc/invoke.texi (-march): Clarify documentation for ARM.
3370 2014-02-05 Richard Biener <rguenther@suse.de>
3372 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3373 when not vectorizing because of too many alias checks.
3374 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3375 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3377 2014-02-05 Nick Clifton <nickc@redhat.com>
3379 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3380 accept extended registers in any mode when compiling for the MN10300.
3382 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3384 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3385 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3386 sanitization attributes.
3387 (can_inline_edge_p): Likewise.
3388 (sanitize_attrs_match_for_inline_p): New function.
3390 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3392 * ipa-prop.c (detect_type_change): Shor circuit testing of
3393 type changes on THIS pointer.
3395 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3398 * config/pa/pa.c (legitimize_tls_address): Return original address
3399 if not passed a SYMBOL_REF rtx.
3400 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3402 (pa_emit_move_sequence): Simplify TLS source operands.
3403 (pa_legitimate_constant_p): Reject all TLS constants.
3404 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3405 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3407 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3409 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3410 groups when we know they are controlled by LTO.
3411 * varasm.c (default_binds_local_p_1): If object is in other partition,
3412 it will be resolved locally.
3414 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3416 * config/host-linux.c (linux_gt_pch_use_address): Don't
3417 use SSIZE_MAX because it is not always defined.
3419 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3422 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3423 threshold for pseudo splitting.
3424 (update_ebb_live_info): Process call argument hard registers and
3425 hard registers from insn definition too.
3426 (max_small_class_regs_num): New constant.
3427 (inherit_in_ebb): Update live hard regs through EBBs. Update
3428 reloads_num only for small register classes. Don't split for
3431 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3434 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3437 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3439 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3442 2014-02-04 Marek Polacek <polacek@redhat.com>
3444 * gdbinit.in (pel): Define.
3446 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3448 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3451 2014-02-04 Richard Biener <rguenther@suse.de>
3454 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3455 in function context local.
3456 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3457 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3458 similar to LTO_imported_decl_ref.
3460 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3462 PR tree-optimization/60002
3463 * cgraphclones.c (build_function_decl_skip_args): Clear
3466 PR tree-optimization/60023
3467 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3468 false to gsi_replace.
3469 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3470 has been in some EH region and vec_stmt could throw, add
3471 vec_stmt into the same EH region.
3472 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3473 has no lhs, ignore it.
3474 * internal-fn.c (expand_MASK_LOAD): Likewise.
3477 * tree-inline.c (copy_forbidden): Fail for
3478 __attribute__((optimize (0))) functions.
3481 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3482 is set, copy one less argument.
3483 (expand_simd_clones): Don't subtract clone_info->inbranch
3484 from simd_clone_struct_alloc argument.
3486 PR rtl-optimization/57915
3487 * recog.c (simplify_while_replacing): If all unary/binary/relational
3488 operation arguments are constant, attempt to simplify those.
3491 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3492 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3494 2014-02-04 Richard Biener <rguenther@suse.de>
3496 PR tree-optimization/60012
3497 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3498 TBAA disambiguation to all DDRs.
3500 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3503 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3504 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3506 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3509 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3511 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3513 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3514 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3516 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3519 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3520 to figure out targets of polymorphic calls with known decl.
3521 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3522 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3523 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3524 (get_polymorphic_call_info): ... here.
3525 (get_polymorphic_call_info_from_invariant): New function.
3527 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3529 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3530 lookup via vtable pointer; check for type consistency
3531 and turn inconsitent facts into UNREACHABLE.
3532 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3533 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3534 type inconsistent querries; return UNREACHABLE instead.
3536 2014-02-03 Richard Henderson <rth@twiddle.net>
3539 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3540 already processed this node.
3541 (normalize_one_pred_1): Pass along mark_set.
3542 (normalize_one_pred): Create and destroy a pointer_set_t.
3543 (normalize_one_pred_chain): Likewise.
3545 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3547 PR gcov-profile/58602
3548 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3550 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3553 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3554 -fno-devirtualize; try to devirtualize by the knowledge of
3555 virtual table pointer given by aggregate propagation.
3556 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3557 (ipa_print_node_jump_functions): Dump also offset that
3558 is relevant for polymorphic calls.
3559 (determine_known_aggregate_parts): Add arg_type parameter; use it
3560 instead of determining the type from pointer type.
3561 (ipa_compute_jump_functions_for_edge): Update call of
3562 determine_known_aggregate_parts.
3563 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3564 (gimple_get_virt_method_for_binfo): ... here; simplify using
3565 vtable_pointer_value_to_vtable.
3566 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3567 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3568 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3569 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3571 (vtable_pointer_value_to_binfo): ... here.
3572 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3574 2014-02-03 Teresa Johnson <tejohnson@google.com>
3576 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3577 redef of outer loop index variable.
3579 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3583 * doc/extend.texi (Function Attributes): Typo.
3585 2014-02-03 Cong Hou <congh@google.com>
3587 PR tree-optimization/60000
3588 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3589 if the vectorized statement is a store. A store statement can only
3590 appear at the end of pattern statements.
3592 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3594 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3595 (ix86_option_override_internal): Default long double to 64-bit for
3596 32-bit Bionic and to 128-bit for 64-bit Bionic.
3598 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3599 TARGET_LONG_DOUBLE_128 is true.
3600 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3602 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3603 (mlong-double-64): Negate -mlong-double-128.
3604 (mlong-double-128): New option.
3606 * config/i386/i386-c.c (ix86_target_macros): Define
3607 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3609 * doc/invoke.texi: Document -mlong-double-128.
3611 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3613 PR rtl-optimization/60024
3614 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3616 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3618 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3620 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3622 PR rtl-optimization/57662
3623 * sel-sched.c (code_motion_path_driver): Do not mark already not
3624 existing blocks in the visiting bitmap.
3626 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3628 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3629 on the insn being emitted.
3631 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3632 Will Deacon <will.deacon@arm.com>
3634 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3636 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3638 * config/arm/arm-tables.opt: Regenerate.
3640 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3642 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3643 for vector types other than V16QImode.
3644 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3645 define_expand, and call altivec_expand_vec_perm_le when producing
3646 code with little endian element order.
3647 (*altivec_vperm_<mode>_internal): New insn having previous
3648 behavior of altivec_vperm_<mode>.
3649 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3650 altivec_expand_vec_perm_le when producing code with little endian
3652 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3653 behavior of altivec_vperm_<mode>_uns.
3655 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3657 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3658 (altivec_vsumsws): Add handling for -maltivec=be with a little
3660 (altivec_vsumsws_direct): New.
3661 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3662 gen_altivec_vsumsws.
3664 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3666 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3667 vtable_pointer_value_to_binfo): New functions.
3668 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3669 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3671 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3673 * config/nios2/nios2.md (load_got_register): Initialize GOT
3674 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3675 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3677 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3679 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3680 preserverd by passthrough, do not propagate the type.
3682 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3684 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3685 (mips_atomic_assign_expand_fenv): New function.
3686 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3688 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3690 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3691 (__builtin_mips_set_fcsr): Likewise.
3692 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3693 MIPS_USI_FTYPE_VOID.
3694 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3695 (mips16_expand_set_fcsr): Likewise.
3696 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3697 (mips16_set_fcsr_stub): Likewise.
3698 (mips16_get_fcsr_one_only_stub): New class.
3699 (mips16_set_fcsr_one_only_stub): Likewise.
3700 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3701 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3702 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3703 (hard_float): New availability predicate.
3704 (mips_builtins): Add get_fcsr and set_fcsr.
3705 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3706 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3707 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3708 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3709 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3712 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3714 * config/mips/mips.c (mips_one_only_stub): New class.
3715 (mips_need_mips16_rdhwr_p): Replace with...
3716 (mips16_rdhwr_stub): ...this new variable.
3717 (mips16_stub_call_address): New function.
3718 (mips16_rdhwr_one_only_stub): New class.
3719 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3720 (mips_output_mips16_rdhwr): Delete.
3721 (mips_finish_stub): New function.
3722 (mips_code_end): Use it to handle rdhwr stubs.
3724 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3727 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3728 when calculating size of integer atomic types.
3730 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3732 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3734 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3736 PR tree-optimization/60003
3737 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3738 * profile.c (branch_prob): Use gimple_call_builtin_p
3739 to check for BUILT_IN_SETJMP_RECEIVER.
3740 * tree-inline.c (copy_bb): Call notice_special_calls.
3742 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3745 * lra-constraints.c (process_alt_operands): Update reload_sum only
3748 2014-01-31 Richard Henderson <rth@redhat.com>
3751 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3752 until after else_eh is processed.
3754 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3756 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3757 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3758 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3759 in smmintrin.h, remove them.
3760 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3761 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3762 * config/i386/i386.md (ROUND_SAE): Fix value.
3763 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3764 (const48_operand): New.
3765 * config/i386/subst.md (round), (round_expand): Use
3766 const_4_or_8_to_11_operand.
3767 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3769 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3771 * config/i386/constraints.md (Yk): Swap meaning with k.
3772 * config/i386/i386.md (movhi_internal): Change Yk to k.
3773 (movqi_internal): Ditto.
3774 (*k<logic><mode>): Ditto.
3777 (kandn<mode>): Ditto.
3778 (*<code>hi_1): Ditto.
3779 (*<code>qi_1): Ditto.
3780 (kxnor<mode>): Ditto.
3781 (kortestzhi): Ditto.
3782 (kortestchi): Ditto.
3784 (*one_cmplhi2_1): Ditto.
3785 (*one_cmplqi2_1): Ditto.
3786 * config/i386/sse.md (): Change k to Yk.
3787 (avx512f_load<mode>_mask): Ditto.
3788 (avx512f_blendm<mode>): Ditto.
3789 (avx512f_store<mode>_mask): Ditto.
3790 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3791 (avx512f_storedqu<mode>_mask): Ditto.
3792 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3794 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3795 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3796 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3797 (avx512f_maskcmp<mode>3): Ditto.
3798 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3799 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3800 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3801 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3802 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3803 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3804 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3805 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3806 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3807 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3808 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3809 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3810 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3811 (vec_extract_lo_<mode>_maskm): Ditto.
3812 (vec_extract_hi_<mode>_maskm): Ditto.
3813 (avx512f_vternlog<mode>_mask): Ditto.
3814 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3815 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3816 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3817 (avx512f_<code>v8div16qi2_mask): Ditto.
3818 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3819 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3820 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3821 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3822 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3823 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3824 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3825 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3826 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3827 (avx512cd_maskb_vec_dupv8di): Ditto.
3828 (avx512cd_maskw_vec_dupv16si): Ditto.
3829 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3830 (avx512f_vpermi2var<mode>3_mask): Ditto.
3831 (avx512f_vpermi2var<mode>3_mask): Ditto.
3832 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3833 (*avx512f_gathersi<mode>): Ditto.
3834 (*avx512f_gathersi<mode>_2): Ditto.
3835 (*avx512f_gatherdi<mode>): Ditto.
3836 (*avx512f_gatherdi<mode>_2): Ditto.
3837 (*avx512f_scattersi<mode>): Ditto.
3838 (*avx512f_scatterdi<mode>): Ditto.
3839 (avx512f_compress<mode>_mask): Ditto.
3840 (avx512f_compressstore<mode>_mask): Ditto.
3841 (avx512f_expand<mode>_mask): Ditto.
3842 * config/i386/subst.md (mask): Change k to Yk.
3843 (mask_scalar_merge): Ditto.
3846 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3848 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3850 2014-01-31 Richard Biener <rguenther@suse.de>
3853 * builtins.c (fold_builtin_memory_op): Make sure to not
3854 use a floating-point mode or a boolean or enumeral type for
3857 2014-01-30 DJ Delorie <dj@redhat.com>
3859 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3860 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3861 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3862 whenever main() has an epilogue.
3864 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3866 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3867 unused variable "field".
3868 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3869 (vsx_mergeh_<mode>): Likewise.
3870 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3871 (altivec_vmrghh): Likewise.
3872 (altivec_vmrghw): Likewise.
3873 (altivec_vmrglb): Likewise.
3874 (altivec_vmrglh): Likewise.
3875 (altivec_vmrglw): Likewise.
3876 (altivec_vspltb): Add missing uses.
3877 (altivec_vsplth): Likewise.
3878 (altivec_vspltw): Likewise.
3879 (altivec_vspltsf): Likewise.
3881 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3884 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3885 frame related instructions.
3887 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3889 PR rtl-optimization/59959
3890 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3891 any reload of register whose subreg is invalid.
3893 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3895 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3896 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3897 Add missing return type - void.
3899 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3901 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3902 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3903 remove element index adjustment for endian (now handled in vsx.md
3905 (altivec_expand_vec_perm_const): Use
3906 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3907 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3908 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3909 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3910 define_expand and a new define_insn *altivec_vspltb_internal;
3911 adjust for -maltivec=be on a little endian target.
3912 (altivec_vspltb_direct): New.
3913 (altivec_vsplth): Divide into a define_expand and a new
3914 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3915 little endian target.
3916 (altivec_vsplth_direct): New.
3917 (altivec_vspltw): Divide into a define_expand and a new
3918 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3919 little endian target.
3920 (altivec_vspltw_direct): New.
3921 (altivec_vspltsf): Divide into a define_expand and a new
3922 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3923 a little endian target.
3925 2014-01-30 Richard Biener <rguenther@suse.de>
3927 PR tree-optimization/59993
3928 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3929 can propagate form the earlier stmt and avoid the transform
3930 when the intermediate result is needed.
3932 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3934 * README.Portability: Fix typo.
3936 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3938 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3939 comparison_operator with ordered_comparison_operator.
3941 2014-01-30 Nick Clifton <nickc@redhat.com>
3943 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3944 Rename to mn10300_store_multiple_regs.
3945 * config/mn10300/mn10300.c: Likewise.
3946 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3947 store_multiple_regs.
3948 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3949 Call mn10300_store_multiple_regs.
3951 2014-01-30 Nick Clifton <nickc@redhat.com>
3952 DJ Delorie <dj@redhat.com>
3954 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3955 %fp 2 to keep registers after it properly word-aligned.
3956 (rl78_alloc_physical_registers_umul): Handle the case where both
3957 input operands are the same.
3959 2014-01-30 Richard Biener <rguenther@suse.de>
3961 PR tree-optimization/59903
3962 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3965 2014-01-30 Jason Merrill <jason@redhat.com>
3968 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3971 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3973 2014-01-30 Richard Biener <rguenther@suse.de>
3975 PR tree-optimization/59951
3976 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3978 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3981 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3982 SFmode to DFmode case.
3984 2014-01-29 DJ Delorie <dj@redhat.com>
3986 * config/msp430/msp430.opt (-minrt): New.
3987 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3989 (ENDFILE_SPEC): Likewise.
3991 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3993 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3994 (estimate_function_body_sizes): Use it.
3996 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3999 * dwarf2out.c (is_cxx_auto): New.
4000 (is_base_type): Use it.
4001 (gen_type_die_with_usage): Likewise.
4003 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4005 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
4006 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
4007 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
4008 -maltivec=be with LE targets.
4009 (vsx_mergeh_<mode>): Likewise.
4010 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
4011 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
4012 (altivec_vmrghb): Replace with define_expand and new
4013 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
4014 (altivec_vmrghb_direct): New define_insn.
4015 (altivec_vmrghh): Replace with define_expand and new
4016 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
4017 (altivec_vmrghh_direct): New define_insn.
4018 (altivec_vmrghw): Replace with define_expand and new
4019 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
4020 (altivec_vmrghw_direct): New define_insn.
4021 (*altivec_vmrghsf): Adjust for endianness.
4022 (altivec_vmrglb): Replace with define_expand and new
4023 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
4024 (altivec_vmrglb_direct): New define_insn.
4025 (altivec_vmrglh): Replace with define_expand and new
4026 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
4027 (altivec_vmrglh_direct): New define_insn.
4028 (altivec_vmrglw): Replace with define_expand and new
4029 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
4030 (altivec_vmrglw_direct): New define_insn.
4031 (*altivec_vmrglsf): Adjust for endianness.
4032 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4033 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4034 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4035 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4036 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4037 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4038 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4039 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4041 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4043 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4044 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4047 2014-01-29 Richard Biener <rguenther@suse.de>
4049 PR tree-optimization/58742
4050 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4051 associate_pointerplus_align.
4052 (associate_pointerplus_diff): New function.
4053 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4054 and associate_pointerplus_diff.
4056 2014-01-29 Richard Biener <rguenther@suse.de>
4058 * lto-streamer.h (LTO_major_version): Bump to 3.
4059 (LTO_minor_version): Reset to 0.
4061 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4063 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4064 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4065 (arm_file_start): Generate correct asm header for armv7ve.
4066 * config/arm/bpabi.h: Add multilib support for armv7ve.
4067 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4068 and cortex-a15 to armv7ve.
4069 * config/arm/t-aprofile: Add multilib support for armv7ve.
4070 * doc/invoke.texi: Document -march=armv7ve.
4072 2014-01-29 Richard Biener <rguenther@suse.de>
4074 PR tree-optimization/58742
4075 * tree-ssa-forwprop.c (associate_plusminus): Return true
4076 if we changed sth, defer EH cleanup to ...
4077 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4078 (simplify_mult): New function.
4080 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4083 PR tree-optimization/59920
4084 * tree.c (build_common_builtin_nodes): Remove
4085 __builtin_setjmp_dispatcher initialization.
4086 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4087 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4088 instead of gsi_after_labels + manually skipping debug stmts.
4089 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4090 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4091 * tree-inline.c (copy_edges_for_bb): Remove
4092 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4093 argument. Ignore computed_goto_p stmts. Don't call
4094 make_abnormal_goto_edges. If a call might need abnormal edges
4095 for non-local gotos, see if it already has an edge to
4096 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4097 with true argument, don't do anything then, otherwise add
4098 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4099 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4101 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4102 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4103 (lower_stmt): Don't set data->calls_builtin_setjmp.
4104 (lower_builtin_setjmp): Adjust comment.
4105 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4106 * tree-cfg.c (found_computed_goto): Remove.
4107 (factor_computed_gotos): Remove.
4108 (make_goto_expr_edges): Return bool, true for computed gotos.
4109 Don't call make_abnormal_goto_edges.
4110 (build_gimple_cfg): Don't set found_computed_goto, don't call
4111 factor_computed_gotos.
4112 (computed_goto_p): No longer static.
4113 (make_blocks): Don't set found_computed_goto.
4114 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4115 (make_edges): If make_goto_expr_edges returns true, push bb
4116 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4117 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4118 vector. Record mapping between bbs and OpenMP regions if there
4119 are any, adjust make_gimple_omp_edges caller. Call
4120 handle_abnormal_edges.
4121 (make_abnormal_goto_edges): Remove.
4122 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4123 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4124 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4125 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4126 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4127 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4128 filling *region also set *region_idx to (*region)->entry->index.
4131 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4132 For REGs set ORIGINAL_REGNO.
4134 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4136 * doc/md.texi: Mention that a target shouldn't implement
4137 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4140 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4142 PR tree-optimization/59594
4143 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4144 a copy of the datarefs vector rather than the vector itself.
4146 2014-01-28 Jason Merrill <jason@redhat.com>
4149 * dwarf2out.c (auto_die): New static.
4150 (gen_type_die_with_usage): Handle C++1y 'auto'.
4151 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4154 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4157 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4158 (SPEC_X32): Likewise.
4159 (SPEC_64): Likewise.
4160 * config/i386/i386.c (ix86_option_override_internal): Turn off
4161 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4163 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4164 * config/i386/i386.h (TARGET_16BIT): New macro.
4165 (TARGET_16BIT_P): Likewise.
4166 * config/i386/i386.opt: Add m16.
4167 * doc/invoke.texi: Document -m16.
4169 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4171 PR preprocessor/59935
4172 * input.c (location_get_source_line): Bail out on when line number
4173 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4175 2014-01-28 Richard Biener <rguenther@suse.de>
4177 PR tree-optimization/58742
4178 * tree-ssa-forwprop.c (associate_plusminus): Handle
4179 pointer subtraction of the form (T)(P + A) - (T)P.
4181 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4183 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4186 2014-01-28 Richard Biener <rguenther@suse.de>
4189 2014-01-28 Richard Biener <rguenther@suse.de>
4191 PR rtl-optimization/45364
4192 PR rtl-optimization/59890
4193 * var-tracking.c (local_get_addr_clear_given_value): Handle
4194 already cleared slot.
4195 (val_reset): Handle not allocated local_get_addr_cache.
4196 (vt_find_locations): Use post-order on the inverted CFG.
4198 2014-01-28 Richard Biener <rguenther@suse.de>
4200 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4202 2014-01-28 Richard Biener <rguenther@suse.de>
4204 PR rtl-optimization/45364
4205 PR rtl-optimization/59890
4206 * var-tracking.c (local_get_addr_clear_given_value): Handle
4207 already cleared slot.
4208 (val_reset): Handle not allocated local_get_addr_cache.
4209 (vt_find_locations): Use post-order on the inverted CFG.
4211 2014-01-28 Alan Modra <amodra@gmail.com>
4213 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4214 * configure.ac <recursive call for build != host>: Define
4215 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4216 and LD_FOR_BUILD too.
4217 * configure: Regenerate.
4219 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4221 * config/i386/i386.c (get_builtin_code_for_version): Separate
4222 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4223 Broadwell from Haswell.
4225 2014-01-27 Steve Ellcey <sellcey@mips.com>
4227 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4228 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4229 * config/mips/mips.c (mips_option_override): Change setting
4231 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4232 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4233 Change from Mask to Var.
4235 2014-01-27 Jeff Law <law@redhat.com>
4237 * ipa-inline.c (inline_small_functions): Fix typo.
4239 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4241 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4242 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4243 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4244 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4245 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4246 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4247 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4248 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4249 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4250 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4251 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4252 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4253 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4254 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4255 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4256 (_mm512_storeu_epi64): Ditto.
4257 (_mm512_cmpge_epi32_mask): Ditto.
4258 (_mm512_cmpge_epu32_mask): Ditto.
4259 (_mm512_cmpge_epi64_mask): Ditto.
4260 (_mm512_cmpge_epu64_mask): Ditto.
4261 (_mm512_cmple_epi32_mask): Ditto.
4262 (_mm512_cmple_epu32_mask): Ditto.
4263 (_mm512_cmple_epi64_mask): Ditto.
4264 (_mm512_cmple_epu64_mask): Ditto.
4265 (_mm512_cmplt_epi32_mask): Ditto.
4266 (_mm512_cmplt_epu32_mask): Ditto.
4267 (_mm512_cmplt_epi64_mask): Ditto.
4268 (_mm512_cmplt_epu64_mask): Ditto.
4269 (_mm512_cmpneq_epi32_mask): Ditto.
4270 (_mm512_cmpneq_epu32_mask): Ditto.
4271 (_mm512_cmpneq_epi64_mask): Ditto.
4272 (_mm512_cmpneq_epu64_mask): Ditto.
4273 (_mm512_expand_pd): Ditto.
4274 (_mm512_expand_ps): Ditto.
4275 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4276 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4277 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4278 * config/i386/i386.c (ix86_builtins): Add
4279 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4280 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4281 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4282 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4283 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4284 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4285 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4286 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4287 IX86_BUILTIN_PMOVUSQW512_MEM.
4288 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4289 __builtin_ia32_pmovsqd512mem_mask,
4290 __builtin_ia32_pmovqd512mem_mask,
4291 __builtin_ia32_pmovusqw512mem_mask,
4292 __builtin_ia32_pmovsqw512mem_mask,
4293 __builtin_ia32_pmovqw512mem_mask,
4294 __builtin_ia32_pmovusdw512mem_mask,
4295 __builtin_ia32_pmovsdw512mem_mask,
4296 __builtin_ia32_pmovdw512mem_mask,
4297 __builtin_ia32_pmovqb512mem_mask,
4298 __builtin_ia32_pmovusqb512mem_mask,
4299 __builtin_ia32_pmovsqb512mem_mask,
4300 __builtin_ia32_pmovusdb512mem_mask,
4301 __builtin_ia32_pmovsdb512mem_mask,
4302 __builtin_ia32_pmovdb512mem_mask.
4303 (bdesc_args): Add __builtin_ia32_expanddf512,
4304 __builtin_ia32_expandsf512.
4305 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4306 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4307 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4308 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4309 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4310 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4311 (avx512f_<code>v8div16qi2_mask_store): This.
4312 (avx512f_expand<mode>): New.
4314 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4316 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4318 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4319 (_mm512_prefetch_i32scatter_pd): Ditto.
4320 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4321 (_mm512_prefetch_i64scatter_pd): Ditto.
4322 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4323 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4324 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4325 (_mm512_prefetch_i32scatter_ps): Ditto.
4326 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4327 (_mm512_prefetch_i64scatter_ps): Ditto.
4328 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4329 * config/i386/i386-builtin-types.def: Define
4330 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4331 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4332 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4333 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4334 IX86_BUILTIN_SCATTERPFQPD.
4335 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4336 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4337 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4338 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4339 __builtin_ia32_scatterpfqps.
4340 (ix86_expand_builtin): Expand new built-ins.
4341 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4342 fix memory access data type.
4343 (*avx512pf_gatherpf<mode>_mask): Ditto.
4344 (*avx512pf_gatherpf<mode>): Ditto.
4345 (avx512pf_scatterpf<mode>): Ditto.
4346 (*avx512pf_scatterpf<mode>_mask): Ditto.
4347 (*avx512pf_scatterpf<mode>): Ditto.
4348 (GATHER_SCATTER_SF_MEM_MODE): New.
4349 (avx512pf_gatherpf<mode>df): Ditto.
4350 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4351 (*avx512pf_scatterpf<mode>df): Ditto.
4353 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4356 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4357 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4360 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4362 * common/config/arm/arm-common.c
4363 (arm_rewrite_mcpu): Handle multiple names.
4365 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4367 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4369 * gimple-builder.h (create_gimple_tmp): Delete.
4371 2014-01-27 Christian Bruel <christian.bruel@st.com>
4373 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4376 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4378 * config/pa/pa.md (call): Generate indirect long calls to non-local
4379 functions when outputing 32-bit code.
4380 (call_value): Likewise except for special call to buggy powf function.
4382 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4383 portable runtime and PIC indirect calls.
4384 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4385 and PIC call sequences. Use ldo instead of blr to set return register
4386 in PIC call sequence.
4388 2014-01-25 Walter Lee <walt@tilera.com>
4390 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4391 avoid clobbering a live register.
4393 2014-01-25 Walter Lee <walt@tilera.com>
4395 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4396 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4397 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4398 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4400 2014-01-25 Walter Lee <walt@tilera.com>
4402 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4403 arguments on even registers.
4404 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4406 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4407 (BIGGEST_ALIGNMENT): Ditto.
4408 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4410 2014-01-25 Walter Lee <walt@tilera.com>
4412 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4413 insns before bundling.
4414 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4416 2014-01-25 Walter Lee <walt@tilera.com>
4418 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4419 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4420 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4422 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4424 * config/mips/constraints.md (kl): Delete.
4425 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4426 define expands, using...
4427 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4428 instructions for MIPS16.
4429 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4430 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4432 2014-01-25 Walter Lee <walt@tilera.com>
4434 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4438 2014-01-25 Walter Lee <walt@tilera.com>
4440 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4441 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4443 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4445 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4448 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4450 * print-rtl.c (in_call_function_usage): New var.
4451 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4452 EXPR_LIST mode as mode and not as reg note name.
4455 * cfgloopmanip.c (copy_loop_info): If
4456 loop->warned_aggressive_loop_optimizations, make sure
4457 the flag is set in target loop too.
4459 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4461 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4463 * builtins.def: Likewise.
4464 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4465 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4466 * ira.c (ira_setup_eliminable_regset): Likewise.
4467 * omp-low.c (gate_expand_omp): Likewise.
4468 (execute_lower_omp): Likewise.
4469 (diagnose_sb_0): Likewise.
4470 (gate_diagnose_omp_blocks): Likewise.
4471 (simd_clone_clauses_extract): Likewise.
4474 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4476 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4477 correction for little endian...
4478 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4481 2014-01-24 Jeff Law <law@redhat.com>
4483 PR tree-optimization/59919
4484 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4485 for non-returning calls.
4487 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4489 * common/config/aarch64/aarch64-common.c
4490 (aarch64_rewrite_mcpu): Handle multiple names.
4491 * config/aarch64/aarch64.h
4492 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4494 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4496 * input.c (add_file_to_cache_tab): Handle the case where fopen
4499 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4502 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4503 from push operand if code of push isn't PRE_DEC.
4505 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4508 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4509 -mquad-memory-atomic. Update -mquad-memory documentation to say
4510 it is only used for non-atomic loads/stores.
4512 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4513 -mquad-memory or -mquad-memory-atomic switches.
4515 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4516 -mquad-memory-atomic to ISA 2.07 support.
4518 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4519 to separate support of normal quad word memory operations (ldq, stq)
4520 from the atomic quad word memory operations.
4522 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4523 support to separate non-atomic quad word operations from atomic
4524 quad word operations. Disable non-atomic quad word operations in
4525 little endian mode so that we don't have to swap words after the
4526 load and before the store.
4527 (quad_load_store_p): Add comment about atomic quad word support.
4528 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4529 options printed with -mdebug=reg.
4531 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4532 -mquad-memory-atomic as the test for whether we have quad word
4533 atomic instructions.
4534 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4535 or -mp8-vector are used, allow byte/half-word atomic operations.
4537 * config/rs6000/sync.md (load_lockedti): Insure that the address
4538 is a proper indexed or indirect address for the lqarx instruction.
4539 On little endian systems, swap the hi/lo registers after the lqarx
4541 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4542 insure the address is valid for the lqarx instruction.
4543 (store_conditionalti): Insure that the address is a proper indexed
4544 or indirect address for the stqcrx. instruction. On little endian
4545 systems, swap the hi/lo registers before doing the stqcrx.
4547 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4548 insure the address is valid for the stqcrx. instruction.
4550 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4551 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4552 type of quad memory support is available.
4554 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4557 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4558 there is a danger of looping.
4560 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4562 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4563 force flag_ira_loop_pressure if set via command line.
4565 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4567 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4568 (ashr_simd): New builtin handling DI mode.
4569 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4570 (aarch64_sshr_simddi): New match pattern.
4571 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4572 (vshrd_n_s64): Likewise.
4573 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4575 2014-01-23 Nick Clifton <nickc@redhat.com>
4577 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4578 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4579 favour of mcu specific scripts.
4580 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4583 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4584 Alex Velenko <Alex.Velenko@arm.com>
4586 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4587 (vaddv_s16): Likewise.
4588 (vaddv_s32): Likewise.
4589 (vaddv_u8): Likewise.
4590 (vaddv_u16): Likewise.
4591 (vaddv_u32): Likewise.
4592 (vaddvq_s8): Likewise.
4593 (vaddvq_s16): Likewise.
4594 (vaddvq_s32): Likewise.
4595 (vaddvq_s64): Likewise.
4596 (vaddvq_u8): Likewise.
4597 (vaddvq_u16): Likewise.
4598 (vaddvq_u32): Likewise.
4599 (vaddvq_u64): Likewise.
4600 (vaddv_f32): Likewise.
4601 (vaddvq_f32): Likewise.
4602 (vaddvq_f64): Likewise.
4603 (vmaxv_f32): Likewise.
4604 (vmaxv_s8): Likewise.
4605 (vmaxv_s16): Likewise.
4606 (vmaxv_s32): Likewise.
4607 (vmaxv_u8): Likewise.
4608 (vmaxv_u16): Likewise.
4609 (vmaxv_u32): Likewise.
4610 (vmaxvq_f32): Likewise.
4611 (vmaxvq_f64): Likewise.
4612 (vmaxvq_s8): Likewise.
4613 (vmaxvq_s16): Likewise.
4614 (vmaxvq_s32): Likewise.
4615 (vmaxvq_u8): Likewise.
4616 (vmaxvq_u16): Likewise.
4617 (vmaxvq_u32): Likewise.
4618 (vmaxnmv_f32): Likewise.
4619 (vmaxnmvq_f32): Likewise.
4620 (vmaxnmvq_f64): Likewise.
4621 (vminv_f32): Likewise.
4622 (vminv_s8): Likewise.
4623 (vminv_s16): Likewise.
4624 (vminv_s32): Likewise.
4625 (vminv_u8): Likewise.
4626 (vminv_u16): Likewise.
4627 (vminv_u32): Likewise.
4628 (vminvq_f32): Likewise.
4629 (vminvq_f64): Likewise.
4630 (vminvq_s8): Likewise.
4631 (vminvq_s16): Likewise.
4632 (vminvq_s32): Likewise.
4633 (vminvq_u8): Likewise.
4634 (vminvq_u16): Likewise.
4635 (vminvq_u32): Likewise.
4636 (vminnmv_f32): Likewise.
4637 (vminnmvq_f32): Likewise.
4638 (vminnmvq_f64): Likewise.
4640 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4642 * config/aarch64/aarch64-simd.md
4643 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4644 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4645 (*aarch64_mul3_elt<mode>): Likewise.
4646 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4647 (*aarch64_mul3_elt_to_64v2df): Likewise.
4648 (*aarch64_mla_elt<mode>): Likewise.
4649 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4650 (*aarch64_mls_elt<mode>): Likewise.
4651 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4652 (*aarch64_fma4_elt<mode>): Likewise.
4653 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4654 (*aarch64_fma4_elt_to_64v2df): Likewise.
4655 (*aarch64_fnma4_elt<mode>): Likewise.
4656 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4657 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4658 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4659 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4660 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4661 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4662 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4663 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4664 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4666 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4668 * config/aarch64/aarch64-simd.md
4669 (aarch64_be_checked_get_lane<mode>): New define_expand.
4670 * config/aarch64/aarch64-simd-builtins.def
4671 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4672 New builtin definition.
4673 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4674 Use new safe be builtin.
4676 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4678 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4680 (aarch64_be_st1<mode>): Likewise.
4681 (aarch_ld1<VALL:mode>): Define_expand modified.
4682 (aarch_st1<VALL:mode>): Likewise.
4683 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4684 (UNSPEC_ST1): Likewise.
4686 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4688 * config/microblaze/microblaze.md: Add trap insn and attribute
4690 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4692 PR preprocessor/58580
4693 * input.h (location_get_source_line): Take an additional line_size
4695 (void diagnostics_file_cache_fini): Declare new function.
4696 * input.c (struct fcache): New type.
4697 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4698 New static constants.
4699 (diagnostic_file_cache_init, total_lines_num)
4700 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4701 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4702 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4703 (get_next_line, read_next_line, goto_next_line, read_line_num):
4704 New static function definitions.
4705 (diagnostic_file_cache_fini): New function.
4706 (location_get_source_line): Take an additional output line_len
4707 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4709 * diagnostic.c (diagnostic_finish): Call
4710 diagnostic_file_cache_fini.
4711 (adjust_line): Take an additional input parameter for the length
4712 of the line, rather than calculating it with strlen.
4713 (diagnostic_show_locus): Adjust the use of
4714 location_get_source_line and adjust_line with respect to their new
4715 signature. While displaying a line now, do not stop at the first
4716 null byte. Rather, display the zero byte as a space and keep
4717 going until we reach the size of the line.
4718 * Makefile.in: Add vec.o to OBJS-libcommon
4720 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4721 Ilya Tocar <ilya.tocar@intel.com>
4723 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4724 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4725 (__builtin_ia32_kmov16): Ditto.
4726 * config/i386/i386.md (UNSPEC_KMOV): New.
4729 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4731 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4732 (_mm512_storeu_si512): Ditto.
4734 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4737 * rtl.h (get_referenced_operands): Declare.
4738 * recog.c (get_referenced_operands): New function.
4739 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4740 operands have been referenced when recording LO_SUM references.
4742 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4744 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4746 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4748 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4749 Enable for generic and recent AMD targets.
4751 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4753 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4754 ARG_SIZE note when adjustment was eliminated.
4756 2014-01-22 Jeff Law <law@redhat.com>
4758 PR tree-optimization/59597
4759 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4760 in file. Accept new argument REGISTERING and use it to modify
4761 dump output appropriately.
4762 (register_jump_thread): Corresponding changes.
4763 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4764 thread paths involving joiner blocks. Add code to dump cancelled
4765 jump threading paths.
4767 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4769 PR rtl-optimization/59477
4770 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4771 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4773 2014-01-22 Tom Tromey <tromey@redhat.com>
4775 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4777 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4779 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4781 PR rtl-optimization/59896
4782 * lra-constraints.c (process_alt_operands): Check unused note for
4783 matched operands of insn with no output reloads.
4785 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4787 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4788 (mips_move_from_gpr_cost): Likewise.
4790 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4792 PR rtl-optimization/59858
4793 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4794 ira_class_hard_regs_num.
4795 (process_alt_operands): Increase reject for dying matched operand.
4797 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4800 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4801 smaller than size, perform several stores or loads and stores
4802 at dst + count - size to store or copy all of size bytes, rather
4803 than just last modesize bytes.
4805 2014-01-20 DJ Delorie <dj@redhat.com>
4807 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4808 that CLOBBERs are REGs before propogating their values.
4810 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4813 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4814 (cgraph_inline_failed_type): New function.
4815 * cgraph.h (DEFCIFCODE): Add type.
4816 (cgraph_inline_failed_type_t): New enum.
4817 (cgraph_inline_failed_type): New prototype.
4818 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4819 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4820 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4821 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4822 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4823 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4824 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4825 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4826 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4827 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4828 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4829 OPTIMIZATION_MISMATCH.
4830 * tree-inline.c (expand_call_inline): Emit errors during
4831 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4833 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4836 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4837 mode attribute in insn output.
4839 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4841 * output.h (output_constant): Delete.
4842 * varasm.c (output_constant): Make private.
4844 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4846 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4848 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4851 * tree.h (fold_builtin_strcat): New prototype.
4852 * builtins.c (fold_builtin_strcat): No longer static. Add len
4853 argument, if non-NULL, don't call c_strlen. Optimize
4854 directly into __builtin_memcpy instead of __builtin_strcpy.
4855 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4856 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4858 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4860 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4861 for SImode_address_operand operands, having only a REG argument.
4863 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4865 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4866 loader name using mbig-endian.
4867 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4869 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4871 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4875 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4877 * config/aarch64/aarch64-protos.h
4878 (aarch64_cannot_change_mode_class_ptr): Declare.
4879 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4880 aarch64_cannot_change_mode_class_ptr): New.
4881 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4882 backend hook aarch64_cannot_change_mode_class.
4884 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4886 * common/config/aarch64/aarch64-common.c
4887 (aarch64_handle_option): Don't handle any option order logic here.
4888 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4889 selected_cpu, warn on architecture version mismatch.
4890 (aarch64_override_options): Fix parsing order for option strings.
4892 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4893 Iain Sandoe <iain@codesourcery.com>
4896 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4897 warning. Amend comment to reflect current functionality.
4899 2014-01-20 Richard Biener <rguenther@suse.de>
4902 * builtins.c (fold_builtin_strcat): Remove case better handled
4903 by tree-ssa-strlen.c.
4905 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4907 * config/aarch64/aarch64.opt
4908 (mcpu, march, mtune): Make case-insensitive.
4910 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4913 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4914 if operands[1] is a REG or ZERO_EXTEND of a REG.
4916 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4918 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4920 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4922 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4923 long non-pic millicode calls.
4925 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4927 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4929 2014-01-19 Kito Cheng <kito@0xlab.org>
4931 * builtins.c (expand_movstr): Check movstr expand done or fail.
4933 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4934 H.J. Lu <hongjiu.lu@intel.com>
4937 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4938 to DImode for zero-extended addresses.
4940 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4942 PR rtl-optimization/57763
4943 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4944 on the new indirect jump_insn and increment LABEL_NUSES (label).
4946 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4950 * config.gcc (x86_archs): New variable.
4951 (x86_64_archs): Likewise.
4952 (x86_cpus): Likewise.
4953 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4954 --with-arch/--with-cpu= options.
4955 Support --with-arch=/--with-cpu={nehalem,westmere,
4956 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4958 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4960 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4961 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4963 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4965 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4967 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4970 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4971 clear cpp_get_options (parse_in)->warn_unused_macros for
4972 ix86_target_macros_internal with cpp_define.
4974 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4976 * jump.c (delete_related_insns): Keep (use (insn))s.
4977 * reorg.c (redundant_insn): Check for barriers too.
4979 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4981 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4983 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4985 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4986 call to $$dyncall when TARGET_LONG_CALLS is true.
4988 2014-01-17 Jeff Law <law@redhat.com>
4990 * ree.c (combine_set_extension): Temporarily disable test for
4991 changing number of hard registers.
4993 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4996 * ipa-inline-analysis.c (inline_free_summary):
4997 Do not free summary of aliases.
4999 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5002 * gimplify.c (gimplify_expr): Use create_tmp_var
5003 instead of create_tmp_var_raw. If cond doesn't have
5004 integral type, don't add the IFN_ANNOTATE builtin at all.
5006 2014-01-17 Martin Jambor <mjambor@suse.cz>
5009 * ipa-cp.c (prev_edge_clone): New variable.
5010 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
5011 Also resize prev_edge_clone vector.
5012 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
5013 (ipcp_edge_removal_hook): New function.
5014 (ipcp_driver): Register ipcp_edge_removal_hook.
5016 2014-01-17 Andrew Pinski <apinski@cavium.com>
5017 Steve Ellcey <sellcey@mips.com>
5020 * config/mips/mips.c (mips_print_operand): Check operand mode instead
5023 2014-01-17 Jeff Law <law@redhat.com>
5026 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
5027 so that pass_ccp runs first.
5029 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5031 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5032 (ix86_adjust_cost): Use !TARGET_XXX.
5033 (do_reorder_for_imul): Likewise.
5034 (swap_top_of_ready_list): Likewise.
5035 (ix86_sched_reorder): Likewise.
5037 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5039 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5040 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5041 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5042 (intel_memset): New. Duplicate slm_memset.
5043 (intel_cost): New. Duplicate slm_cost.
5044 (m_INTEL): New macro.
5045 (processor_target_table): Add "intel".
5046 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5047 with PROCESSOR_INTEL for "intel".
5048 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5049 PROCESSOR_SILVERMONT.
5050 (ix86_issue_rate): Likewise.
5051 (ix86_adjust_cost): Likewise.
5052 (ia32_multipass_dfa_lookahead): Likewise.
5053 (swap_top_of_ready_list): Likewise.
5054 (ix86_sched_reorder): Likewise.
5055 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5056 instead of TARGET_OPT_AGU.
5057 * config/i386/i386.h (TARGET_INTEL): New.
5058 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5059 (processor_type): Add PROCESSOR_INTEL.
5060 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5061 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5063 2014-01-17 Marek Polacek <polacek@redhat.com>
5066 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5069 2014-01-17 Richard Biener <rguenther@suse.de>
5071 PR tree-optimization/46590
5072 * opts.c (default_options_table): Add entries for
5073 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5074 all enabled at -O1 but not for -Og.
5075 * common.opt (fbranch-count-reg): Remove Init(1).
5076 (fmove-loop-invariants): Likewise.
5077 (ftree-pta): Likewise.
5079 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5081 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5082 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5083 decls to at least the GCC 4.8 used alignments.
5086 * tree-nested.c (convert_nonlocal_reference_stmt,
5087 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5088 of GIMPLE_BIND stmts, adjust associated decls.
5090 2014-01-17 Richard Biener <rguenther@suse.de>
5092 PR tree-optimization/46590
5093 * vec.h (vec<>::bseach): New member function implementing
5094 binary search according to C89 bsearch.
5095 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5096 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5097 bitmap pointer again. Make accesses_in_loop a flat array.
5098 (mem_ref_obstack): New global.
5099 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5100 (mark_ref_stored): Likewise.
5101 (ref_indep_loop_p_2): Likewise.
5102 (set_ref_stored_in_loop): New helper function.
5103 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5104 (memref_free): Adjust.
5105 (record_mem_ref_loc): Simplify.
5106 (gather_mem_refs_stmt): Adjust.
5107 (sort_locs_in_loop_postorder_cmp): New function.
5108 (analyze_memory_references): Sort accesses_in_loop after
5109 loop postorder number.
5110 (find_ref_loc_in_loop_cmp): New function.
5111 (for_all_locs_in_loop): Find relevant cluster of locs in
5112 accesses_in_loop and iterate without recursion.
5113 (execute_sm): Avoid uninit warning.
5114 (struct ref_always_accessed): Simplify.
5115 (ref_always_accessed::operator ()): Likewise.
5116 (ref_always_accessed_p): Likewise.
5117 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5118 loop postorder numbers here.
5119 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5122 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5125 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5126 on decls for which assemble_alias has been called.
5128 2014-01-17 Nick Clifton <nickc@redhat.com>
5130 * config/msp430/msp430.opt: (mcpu): New option.
5131 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5132 (msp430_option_override): Parse target_cpu. If the MCU name
5133 matches a generic string, clear target_mcu.
5134 (msp430_attr): Allow numeric interrupt values up to 63.
5135 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5136 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5138 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5140 * config/msp430/msp430.md (popm): Use %J rather than %I.
5141 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5142 (addhi_cy_i): Use immediate_operand for operand 2.
5143 * doc/invoke.texi: Document -mcpu option.
5145 2014-01-17 Richard Biener <rguenther@suse.de>
5147 PR rtl-optimization/38518
5148 * df.h (df_analyze_loop): Declare.
5149 * df-core.c: Include cfgloop.h.
5150 (df_analyze_1): Split out main part of df_analyze.
5151 (df_analyze): Adjust.
5152 (loop_inverted_post_order_compute): New function.
5153 (loop_post_order_compute): Likewise.
5154 (df_analyze_loop): New function avoiding whole-function
5156 * loop-invariant.c (find_defs): Use df_analyze_loop.
5157 (find_invariants): Adjust.
5158 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5160 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5162 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5163 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5165 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5167 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5168 traversal when removing references.
5170 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5173 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5175 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5178 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5179 pushing a reload for an autoinc when we had previously reloaded an
5180 inner part of the address.
5182 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5184 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5186 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5187 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5188 when not giving up or versioning for alias only because of
5190 (vect_analyze_data_ref_dependences): Set to true.
5191 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5193 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5194 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5198 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5201 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5202 operand 0 predicate for gathers, use a new pseudo as subtarget.
5204 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5207 * lra-constraints.c (process_alt_operands): Add printing debug info.
5208 Check absence of input/output reloads for matched operands too.
5210 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5212 PR rtl-optimization/59835
5213 * ira.c (ira_init_register_move_cost): Increase cost for
5216 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5218 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5220 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5223 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5224 non-register objects. Use gen_(high/low)part more consistently.
5227 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5230 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5231 endian support, remove tests for WORDS_BIG_ENDIAN.
5232 (p8_mfvsrd_3_<mode>): Likewise.
5233 (reload_gpr_from_vsx<mode>): Likewise.
5234 (reload_gpr_from_vsxsf): Likewise.
5235 (p8_mfvsrd_4_disf): Likewise.
5237 2014-01-16 Richard Biener <rguenther@suse.de>
5239 PR rtl-optimization/46590
5240 * lcm.c (compute_antinout_edge): Use postorder iteration.
5241 (compute_laterin): Use inverted postorder iteration.
5243 2014-01-16 Nick Clifton <nickc@redhat.com>
5246 * varasm.c (output_constant): Return the number of bytes actually
5248 (output_constructor_array_range): Update the field size with the
5249 number of bytes emitted by output_constant.
5250 (output_constructor_regular_field): Likewise. Also do not
5251 complain if the total number of bytes emitted is now greater
5252 than the expected fieldpos.
5253 * output.h (output_constant): Update prototype and descriptive comment.
5255 2014-01-16 Marek Polacek <polacek@redhat.com>
5258 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5259 it is error_mark_node.
5261 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5263 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5264 VALID_AVX256_REG_OR_OI_MODE.
5266 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5268 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5269 current procedure should be profiled.
5271 2014-01-15 Andrew Pinski <apinski@cavium.com>
5273 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5274 of moving from/to the STACK_REG register class.
5276 2014-01-15 Richard Henderson <rth@redhat.com>
5279 * reginfo.c (global_regs_decl): Globalize.
5280 * rtl.h (global_regs_decl): Declare.
5281 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5282 reserved via global_regs.
5284 2014-01-15 Teresa Johnson <tejohnson@google.com>
5286 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5288 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5290 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5291 and vmulosh rather than call gen_vec_widen_smult_*.
5292 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5293 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5294 (vec_widen_smult_even_v16qi): Likewise.
5295 (vec_widen_umult_even_v8hi): Likewise.
5296 (vec_widen_smult_even_v8hi): Likewise.
5297 (vec_widen_umult_odd_v16qi): Likewise.
5298 (vec_widen_smult_odd_v16qi): Likewise.
5299 (vec_widen_umult_odd_v8hi): Likewise.
5300 (vec_widen_smult_odd_v8hi): Likewise.
5301 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5302 vmuloub rather than call gen_vec_widen_umult_*.
5303 (vec_widen_umult_lo_v16qi): Likewise.
5304 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5305 vmulosb rather than call gen_vec_widen_smult_*.
5306 (vec_widen_smult_lo_v16qi): Likewise.
5307 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5308 rather than call gen_vec_widen_umult_*.
5309 (vec_widen_umult_lo_v8hi): Likewise.
5310 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5311 rather than call gen_vec_widen_smult_*.
5312 (vec_widen_smult_lo_v8hi): Likewise.
5314 2014-01-15 Jeff Law <law@redhat.com>
5316 PR tree-optimization/59747
5317 * ree.c (find_and_remove_re): Properly handle case where a second
5318 eliminated extension requires widening a copy created for elimination
5319 of a prior extension.
5320 (combine_set_extension): Ensure that the number of hard regs needed
5321 for a destination register does not change when we widen it.
5323 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5325 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5326 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5327 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5328 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5329 (avr-*-rtems*): Likewise.
5330 (bfin*-rtems*): Likewise.
5331 (moxie-*-rtems*): Likewise.
5332 (h8300-*-rtems*): Likewise.
5333 (i[34567]86-*-rtems*): Likewise.
5334 (lm32-*-rtems*): Likewise.
5335 (m32r-*-rtems*): Likewise.
5336 (m68k-*-rtems*): Likewise.
5337 (microblaze*-*-rtems*): Likewise.
5338 (mips*-*-rtems*): Likewise.
5339 (powerpc-*-rtems*): Likewise.
5340 (sh-*-rtems*): Likewise.
5341 (sparc-*-rtems*): Likewise.
5342 (sparc64-*-rtems*): Likewise.
5343 (v850-*-rtems*): Likewise.
5344 (m32c-*-rtems*): Likewise.
5346 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5348 PR rtl-optimization/59511
5349 * ira.c (ira_init_register_move_cost): Use memory costs for some
5350 cases of register move cost calculations.
5351 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5352 instead of BB frequency.
5353 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5354 * lra-assigns.c (find_hard_regno_for): Ditto.
5356 2014-01-15 Richard Biener <rguenther@suse.de>
5358 PR tree-optimization/59822
5359 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5360 (vectorizable_load): Use it to hoist defs of uses of invariant
5361 loads out of the loop.
5363 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5364 Kugan Vivekanandarajah <kuganv@linaro.org>
5367 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5370 2014-01-15 Richard Biener <rguenther@suse.de>
5372 PR rtl-optimization/59802
5373 * lcm.c (compute_available): Use inverted postorder to seed
5374 the initial worklist.
5376 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5379 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5380 ADDR_REGS for invalid symrefs in non-PIC code.
5382 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5385 * builtins.c (determine_block_size): Initialize *probable_max_size
5386 even if len_rtx is CONST_INT.
5388 2014-01-14 Andrew Pinski <apinski@cavium.com>
5390 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5391 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5392 (cortexa53_tunings): Likewise.
5393 (aarch64_sched_issue_rate): New function.
5394 (TARGET_SCHED_ISSUE_RATE): Define.
5396 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5398 * ira-costs.c (find_costs_and_classes): Add missed
5399 ira_init_register_move_cost_if_necessary.
5401 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5404 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5406 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5409 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5410 to indicate if type is used for function return value. Warn ABI
5411 change if the vector mode isn't available for function return value.
5412 (ix86_function_arg_advance): Pass false to type_natural_mode.
5413 (ix86_function_arg): Likewise.
5414 (ix86_gimplify_va_arg): Likewise.
5415 (function_arg_32): Don't warn ABI change.
5416 (ix86_function_value): Pass true to type_natural_mode.
5417 (ix86_return_in_memory): Likewise.
5418 (ix86_struct_value_rtx): Removed.
5419 (TARGET_STRUCT_VALUE_RTX): Likewise.
5421 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5423 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5424 converting a conditional jump into a conditional return.
5426 2014-01-14 Richard Biener <rguenther@suse.de>
5428 PR tree-optimization/58921
5429 PR tree-optimization/59006
5430 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5431 hoisting invariant stmts.
5432 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5433 invariant loads on the preheader edge if possible.
5435 2014-01-14 Joey Ye <joey.ye@arm.com>
5437 * doc/plugin.texi (Building GCC plugins): Update to C++.
5439 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5441 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5442 (_mm_rcp28_round_ss): Ditto.
5443 (_mm_rsqrt28_round_sd): Ditto.
5444 (_mm_rsqrt28_round_ss): Ditto.
5445 (_mm_rcp28_sd): Ditto.
5446 (_mm_rcp28_ss): Ditto.
5447 (_mm_rsqrt28_sd): Ditto.
5448 (_mm_rsqrt28_ss): Ditto.
5449 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5450 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5451 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5452 (IX86_BUILTIN_RCP28SD): Ditto.
5453 (IX86_BUILTIN_RCP28SS): Ditto.
5454 (IX86_BUILTIN_RSQRT28SD): Ditto.
5455 (IX86_BUILTIN_RSQRT28SS): Ditto.
5456 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5457 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5458 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5459 (ix86_expand_special_args_builtin): Expand new FTYPE.
5460 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5461 (srcp14<mode>): Make insn unary.
5462 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5463 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5464 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5465 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5466 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5467 Fix rounding: make it SAE only.
5468 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5470 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5472 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5473 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5474 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5475 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5476 (round_saeonly_mask_scalar_operand4): Ditto.
5477 (round_saeonly_mask_scalar_op3): Ditto.
5478 (round_saeonly_mask_scalar_op4): Ditto.
5480 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5482 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5483 Implement -maltivec=be for vec_insert and vec_extract.
5485 2014-01-10 DJ Delorie <dj@redhat.com>
5487 * config/msp430/msp430.md (call_internal): Don't allow memory
5488 references with SP as the base register.
5489 (call_value_internal): Likewise.
5490 * config/msp430/constraints.md (Yc): New. For memory references
5491 that don't use SP as a base register.
5493 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5494 "an integer without a # prefix"
5495 * config/msp430/msp430.md (epilogue_helper): Use it.
5497 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5500 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5501 AVX512F gather builtins.
5502 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5503 on gather decls with INTEGER_TYPE masktype.
5504 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5505 directly into the builtin rather than hoisting it before loop.
5507 PR tree-optimization/59387
5508 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5509 (scev_const_prop): If folded_casts and type has undefined overflow,
5510 use force_gimple_operand instead of force_gimple_operand_gsi and
5511 for each added stmt if it is assign with
5512 arith_code_with_undefined_signed_overflow, call
5513 rewrite_to_defined_overflow.
5514 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5515 gimple-fold.h instead.
5516 (arith_code_with_undefined_signed_overflow,
5517 rewrite_to_defined_overflow): Moved to ...
5518 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5519 rewrite_to_defined_overflow): ... here. No longer static.
5520 Include gimplify-me.h.
5521 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5522 rewrite_to_defined_overflow): New prototypes.
5524 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5526 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5528 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5530 * builtins.c (get_object_alignment_2): Minor tweak.
5531 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5533 2014-01-13 Christian Bruel <christian.bruel@st.com>
5535 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5536 optimized non constant lengths.
5538 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5541 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5542 load as __atomic_load_N if possible.
5544 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5546 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5548 (rs6000_expand_builtin): Adjust call.
5550 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5553 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5554 * config/rs6000/rs6000.c: Include target-globals.h.
5555 (rs6000_set_current_function): Instead of doing target_reinit
5556 unconditionally, use save_target_globals_default_opts and
5557 restore_target_globals.
5559 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5561 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5562 (rs6000_expand_builtin): Handle mffs and mtfsf.
5563 (rs6000_init_builtins): Define mffs and mtfsf.
5564 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5565 (rs6000_mffs): New pattern.
5566 (rs6000_mtfsf): New pattern.
5568 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5570 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5571 Start narrowing with START. Apply candidate-use pair
5572 and check overall cost in narrowing.
5573 (iv_ca_prune): Pass new argument.
5575 2014-01-10 Jeff Law <law@redhat.com>
5578 * ree.c (combine_reaching_defs): Ensure the defining statement
5579 occurs before the extension when optimizing extensions with
5580 different source and destination hard registers.
5582 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5585 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5586 vtables into the type inheritance graph.
5588 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5590 PR rtl-optimization/59754
5591 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5592 modes in the REGNO != REGNO case.
5594 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5596 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5598 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5600 PR tree-optimization/59745
5601 * tree-predcom.c (tree_predictive_commoning_loop): Call
5602 free_affine_expand_cache if giving up because components is NULL.
5604 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5605 GC in payload of target_globals struct instead of allocating them on
5606 the heap and the larger structs separately using GC.
5607 * target-globals.h (struct target_globals): Make regs, hard_regs,
5608 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5609 of GTY((skip)) and change type to void *.
5610 (reset_target_globals): Cast loads from those fields to corresponding
5613 2014-01-10 Steve Ellcey <sellcey@mips.com>
5616 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5617 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5618 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5620 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5623 * aarch64-modes.def (CC_Zmode): New flags mode.
5624 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5625 represents an equality.
5626 (aarch64_get_condition_code): Handle CC_Zmode.
5627 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5629 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5631 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5632 extraction in good case.
5634 2014-01-10 Richard Biener <rguenther@suse.de>
5636 PR tree-optimization/59374
5637 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5638 checking after SLP discovery. Mark stmts not participating
5639 in any SLP instance properly.
5641 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5643 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5644 when handling a SET rtx.
5646 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5648 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5649 (cortex-a57): Likewise.
5650 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5652 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5654 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5655 non-iwmmxt builtins.
5657 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5661 * ipa-devirt.c record_target_from_binfo): Take as argument
5662 stack of binfos and lookup matching one for virtual inheritance.
5663 (possible_polymorphic_call_targets_1): Update.
5665 2014-01-10 Huacai Chen <chenhc@lemote.com>
5667 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5668 kernel strings for Loongson-2E/2F/3A.
5670 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5673 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5674 is_gimple_call before calling gimple_call_internal_p.
5676 2014-01-09 Steve Ellcey <sellcey@mips.com>
5678 * Makefile.in (TREE_FLOW_H): Remove.
5679 (TREE_SSA_H): Add file names from tree-flow.h.
5680 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5681 * tree.h: Remove tree-flow.h reference.
5682 * hash-table.h: Remove tree-flow.h reference.
5683 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5684 reference with tree-ssa-loop.h.
5686 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5688 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5689 default element-order behavior for -maltivec.
5690 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5691 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5692 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5693 when targeting big endian, at least for now.
5694 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5696 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5699 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5700 var satisfies use_register_for_decl, just take into account type
5701 alignment, rather than decl alignment.
5703 PR tree-optimization/59622
5704 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5705 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5706 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5707 Don't devirtualize for inplace at all. For targets.length () == 1,
5708 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5710 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5712 * config/i386/i386.md (cpu): Remove the unused btver1.
5714 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5716 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5718 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5721 * tree-core.h (struct target_globals): New forward declaration.
5722 (struct tree_target_option): Add globals field.
5723 * tree.h (TREE_TARGET_GLOBALS): Define.
5724 (prepare_target_option_nodes_for_pch): New prototype.
5725 * target-globals.h (struct target_globals): Define even if
5727 * tree.c (prepare_target_option_node_for_pch,
5728 prepare_target_option_nodes_for_pch): New functions.
5729 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5730 * config/i386/i386.c: Include target-globals.h.
5731 (ix86_set_current_function): Instead of doing target_reinit
5732 unconditionally, use save_target_globals_default_opts and
5733 restore_target_globals.
5735 2014-01-09 Richard Biener <rguenther@suse.de>
5737 PR tree-optimization/59715
5738 * tree-cfg.h (split_critical_edges): Declare.
5739 * tree-cfg.c (split_critical_edges): Export.
5740 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5742 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5744 * cfgexpand.c (expand_stack_vars): Optionally disable
5745 asan stack protection.
5746 (expand_used_vars): Likewise.
5747 (partition_stack_vars): Likewise.
5748 * asan.c (asan_emit_stack_protection): Optionally disable
5749 after return stack usage.
5750 (instrument_derefs): Optionally disable memory access instrumentation.
5751 (instrument_builtin_call): Likewise.
5752 (instrument_strlen_call): Likewise.
5753 (asan_protect_global): Optionally disable global variables protection.
5754 * doc/invoke.texi: Added doc for new options.
5755 * params.def: Added new options.
5756 * params.h: Likewise.
5758 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5760 PR rtl-optimization/59724
5761 * ifcvt.c (cond_exec_process_if_block): Don't call
5762 flow_find_head_matching_sequence with 0 longest_match.
5763 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5764 non-active insns if !stop_after.
5765 (try_head_merge_bb): Revert 2014-01-07 changes.
5767 2014-01-08 Jeff Law <law@redhat.com>
5769 * ree.c (get_sub_rtx): New function, extracted from...
5770 (merge_def_and_ext): Here.
5771 (combine_reaching_defs): Use get_sub_rtx.
5773 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5775 * cgraph.h (varpool_variable_node): Do not choke on null node.
5777 2014-01-08 Catherine Moore <clm@codesourcery.com>
5779 * config/mips/mips.md (simple_return): Attempt to use JRC
5781 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5783 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5785 PR rtl-optimization/59137
5786 * reorg.c (steal_delay_list_from_target): Call update_block for
5788 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5790 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5792 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5793 two duplicate entries.
5795 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5798 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5800 * config/mips/mips.c (mips_truncated_op_cost): New function.
5801 (mips_rtx_costs): Adjust test for BADDU.
5802 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5804 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5806 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5807 (*baddu_si): ...this new pattern.
5809 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5812 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5814 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5817 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5819 (expand_expr, expand_normal): Adjust.
5820 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5821 inner_reference_p. Use inner_reference_p to expand inner references.
5822 (store_expr): Adjust.
5823 * cfgexpand.c (expand_call_stmt): Adjust.
5825 2014-01-08 Rong Xu <xur@google.com>
5827 * gcov-io.c (gcov_var): Move from gcov-io.h.
5828 (gcov_position): Ditto.
5829 (gcov_is_error): Ditto.
5830 (gcov_rewrite): Ditto.
5831 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5832 only part to libgcc/libgcov.h.
5834 2014-01-08 Marek Polacek <polacek@redhat.com>
5837 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5839 2014-01-08 Marek Polacek <polacek@redhat.com>
5842 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5844 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5846 PR rtl-optimization/59649
5847 * stor-layout.c (get_mode_bounds): For BImode return
5848 0 and STORE_FLAG_VALUE.
5850 2014-01-08 Richard Biener <rguenther@suse.de>
5853 * gimple.h (is_gimple_builtin_call): Remove.
5854 (gimple_builtin_call_types_compatible_p): New.
5855 (gimple_call_builtin_p): New overload.
5856 * gimple.c (is_gimple_builtin_call): Remove.
5857 (validate_call): Rename to ...
5858 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5860 (validate_type): New static function.
5861 (gimple_call_builtin_p): New overload and adjust.
5862 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5863 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5864 (gimple_fold_stmt_to_constant_1): Likewise.
5865 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5867 2014-01-08 Richard Biener <rguenther@suse.de>
5870 * gimplify.c (gimplify_expr): Gimplify register-register type
5871 VIEW_CONVERT_EXPRs to separate stmts.
5873 2014-01-07 Jeff Law <law@redhat.com>
5876 * ree.c (combine_set_extension): Handle case where source
5877 and destination registers in an extension insn are different.
5878 (combine_reaching_defs): Allow source and destination registers
5879 in extension to be different under limited circumstances.
5880 (add_removable_extension): Remove restriction that the
5881 source and destination registers in the extension are the same.
5882 (find_and_remove_re): Emit a copy from the extension's
5883 destination to its source after the defining insn if
5884 the source and destination registers are different.
5887 * ifcvt.c (merge_if_block): If we are merging a block with more than
5888 one successor with a block with no successors, remove any BARRIER
5889 after the second block.
5891 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5893 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5895 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5898 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5899 for 14-bit register offsets when INT14_OK_STRICT is false.
5901 2014-01-07 Roland Stigge <stigge@antcom.de>
5902 Michael Meissner <meissner@linux.vnet.ibm.com>
5905 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5906 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5908 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5910 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5913 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5915 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5916 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5917 rtx is const0_rtx or not.
5919 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5922 * target-globals.c (save_target_globals): Remove this_fn_optab
5924 * toplev.c: Include optabs.h.
5925 (target_reinit): Temporarily restore the global options if another
5926 set of options are in force.
5928 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5930 PR rtl-optimization/58668
5931 * cfgcleanup.c (flow_find_cross_jump): Don't count
5932 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5933 to determine what is counted.
5934 (flow_find_head_matching_sequence): Use active_insn_p to determine
5936 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5938 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5939 determine what is counted.
5941 PR tree-optimization/59643
5942 * tree-predcom.c (split_data_refs_to_components): If one dr is
5943 read and one write, determine_offset fails and the write isn't
5944 in the bad component, just put the read into the bad component.
5946 2014-01-07 Mike Stump <mikestump@comcast.net>
5947 Jakub Jelinek <jakub@redhat.com>
5950 * tree-core.h (struct tree_optimization_option): Change optabs
5951 type from unsigned char * to void *.
5952 * optabs.c (init_tree_optimization_optabs): Adjust
5953 TREE_OPTIMIZATION_OPTABS initialization.
5955 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5958 * config/i386/i386.h (struct machine_function): Add
5959 no_drap_save_restore field.
5960 * config/i386/i386.c (ix86_save_reg): Use
5961 !cfun->machine->no_drap_save_restore instead of
5962 crtl->stack_realign_needed.
5963 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5964 this function clears frame_pointer_needed. Set
5965 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5966 and DRAP reg is needed.
5968 2014-01-06 Marek Polacek <polacek@redhat.com>
5971 * doc/implement-c.texi: Mention that other integer types are
5972 permitted as bit-field types in strictly conforming mode.
5974 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5976 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5979 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5981 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5983 2014-01-06 Martin Jambor <mjambor@suse.cz>
5986 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5988 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5990 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5994 * var-tracking.c (add_stores): Preserve the value of the source even if
5995 we don't record the store.
5997 2014-01-06 Terry Guo <terry.guo@arm.com>
5999 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
6001 2014-01-05 Iain Sandoe <iain@codesourcery.com>
6004 * config/darwin.c (darwin_function_section): Adjust return values to
6005 correspond to optimisation changes made in r206070.
6007 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
6009 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
6010 from prefetch_block tune setting.
6011 (nocona_cost): Correct size of prefetch block to 64.
6013 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
6015 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
6016 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
6017 used to save the static chain register in the computation of the offset
6018 from which the FP registers need to be restored.
6020 2014-01-04 Jakub Jelinek <jakub@redhat.com>
6022 PR tree-optimization/59519
6023 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
6024 ICE if get_current_def (current_new_name) is already non-NULL, as long
6025 as it is a phi result of some other phi in *new_exit_bb that has
6028 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
6029 or vmovdqu* for misaligned_operand.
6030 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
6031 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6032 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6033 aligned_mem for AVX512F masked aligned load and store builtins and for
6036 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6038 PR tree-optimization/59651
6039 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6040 Address range for negative step should be added by TYPE_SIZE_UNIT.
6042 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6044 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6045 overlapping registers also for registers other than the stack pointer.
6047 2014-01-03 Marek Polacek <polacek@redhat.com>
6050 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6053 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6056 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6059 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6060 (push splitter): Use <P:MODE_SIZE> instead of
6061 GET_MODE_SIZE (<P:MODE>mode).
6062 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6063 (mov -1, reg peephole2): Likewise.
6064 * config/i386/sse.md (*mov<mode>_internal,
6065 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6066 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6067 *<code><mode>3, *andnot<mode>3<mask_name>,
6068 <mask_codefor><code><mode>3<mask_name>): Likewise.
6069 * config/i386/subst.md (mask_mode512bit_condition,
6070 sd_mask_mode512bit_condition): Likewise.
6072 2014-01-02 Xinliang David Li <davidxl@google.com>
6074 PR tree-optimization/59303
6075 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6076 (dump_predicates): Better output format.
6077 (pred_equal_p): New function.
6078 (is_neq_relop_p): Ditto.
6079 (is_neq_zero_form_p): Ditto.
6080 (pred_expr_equal_p): Ditto.
6081 (pred_neg_p): Ditto.
6082 (simplify_pred): Ditto.
6083 (simplify_preds_2): Ditto.
6084 (simplify_preds_3): Ditto.
6085 (simplify_preds_4): Ditto.
6086 (simplify_preds): Ditto.
6088 (push_to_worklist): Ditto.
6089 (get_pred_info_from_cmp): Ditto.
6090 (is_degenerated_phi): Ditto.
6091 (normalize_one_pred_1): Ditto.
6092 (normalize_one_pred): Ditto.
6093 (normalize_one_pred_chain): Ditto.
6094 (normalize_preds): Ditto.
6095 (normalize_cond_1): Remove function.
6096 (normalize_cond): Ditto.
6097 (is_gcond_subset_of): Ditto.
6098 (is_subset_of_any): Ditto.
6099 (is_or_set_subset_of): Ditto.
6100 (is_and_set_subset_of): Ditto.
6101 (is_norm_cond_subset_of): Ditto.
6102 (pred_chain_length_cmp): Ditto.
6103 (convert_control_dep_chain_into_preds): Type change.
6104 (find_predicates): Ditto.
6105 (find_def_preds): Ditto.
6106 (destroy_predicates_vecs): Ditto.
6107 (find_matching_predicates_in_rest_chains): Ditto.
6108 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6109 (is_pred_expr_subset): Ditto.
6110 (is_pred_chain_subset_of): Ditto.
6111 (is_included_in): Ditto.
6112 (is_superset_of): Ditto.
6114 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6116 Update copyright years.
6118 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6120 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6121 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6122 config/arc/arc.md, config/arc/arc.opt,
6123 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6124 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6125 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6126 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6127 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6128 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6129 vtable-verify.c, vtable-verify.h: Use the standard form for the
6132 2014-01-02 Tobias Burnus <burnus@net-b.de>
6134 * gcc.c (process_command): Update copyright notice dates.
6135 * gcov-dump.c: Ditto.
6137 * doc/cpp.texi: Bump @copying's copyright year.
6138 * doc/cppinternals.texi: Ditto.
6139 * doc/gcc.texi: Ditto.
6140 * doc/gccint.texi: Ditto.
6141 * doc/gcov.texi: Ditto.
6142 * doc/install.texi: Ditto.
6143 * doc/invoke.texi: Ditto.
6145 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6147 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6149 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6151 * config/i386/sse.md (*mov<mode>_internal): Guard
6152 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6154 PR rtl-optimization/59647
6155 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6156 new_rtx into UNSIGNED_FLOAT rtxes.
6158 Copyright (C) 2014 Free Software Foundation, Inc.
6160 Copying and distribution of this file, with or without modification,
6161 are permitted in any medium without royalty provided the copyright
6162 notice and this notice are preserved.