1 2014-04-07 Jan Hubicka <hubcika@ucw.cz>
3 * ipa-devirt.c (maybe_record_node): When node is not recorded,
4 set completep to false rather than true.
6 2014-04-07 Douglas B Rupp <rupp@adacore.com>
9 * config/arm/arm.h (ASM_PREFERRED_EH_DATA_FORMAT): Expose from
10 ARM_TARGET2_DWARF_FORMAT.
12 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
15 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
16 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
19 2014-04-07 Richard Biener <rguenther@suse.de>
21 PR tree-optimization/60766
22 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an unsigned type.
23 (may_eliminate_iv): Convert cand_value_at result to desired type.
25 2014-04-07 Jason Merrill <jason@redhat.com>
28 * common.opt (-fno-gnu-unique): Add.
29 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
31 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33 * haifa-sched.c: Fix outdated function reference and minor
34 grammar errors in introductory comment.
36 2014-04-07 Richard Biener <rguenther@suse.de>
39 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
41 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
43 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
46 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
47 size accounting for thunks.
48 (pa_asm_output_mi_thunk): Use final_start_function() and
49 final_end_function() to output function start and end directives.
51 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
53 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
54 specific ISA/ feature information. Remove short_sp and errata_skip ds.
55 Add avr_device_specific_features enum to have device specific info.
56 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
57 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
58 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
59 updated device specific info.
60 * config/avr/avr-mcus.def: Merge device specific details to
62 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
64 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
65 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
66 assembler if RMW isa supported by current device.
67 * config/avr/genmultilib.awk: Update as device info structure changed.
68 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
70 2014-04-04 Cong Hou <congh@google.com>
72 PR tree-optimization/60656
73 * tree-vect-stmts.c (supportable_widening_operation):
74 Fix a bug that elements in a vector with vect_used_by_reduction
75 property are incorrectly reordered when the operation on it is not
76 consistant with the one in reduction operation.
78 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
80 PR rtl-optimization/60155
81 * gcse.c (record_set_data): New function.
82 (single_set_gcse): New function.
83 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
84 (hoist_code): Likewise.
85 (get_pressure_class_and_nregs): Likewise.
87 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
89 * explow.c (probe_stack_range): Emit a final optimization blockage.
91 2014-04-04 Anthony Green <green@moxielogic.com>
93 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
96 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
99 * lto-cgraph.c (input_overwrite_node): Check that partitioning
100 flags are set only during streaming.
101 * ipa.c (process_references, walk_polymorphic_call_targets,
102 symtab_remove_unreachable_nodes): Drop bodies of always inline
103 after early inlining.
104 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
106 2014-04-04 Jakub Jelinek <jakub@redhat.com>
107 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
110 * dwarf2out.c (const_ok_for_output_1): Reject expressions
113 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
116 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
118 (cortex_a53_fdivd): Likewise.
120 2014-04-04 Martin Jambor <mjambor@suse.cz>
123 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
125 * cgraph.c (clone_of_p): Also return true if thunks match.
126 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
127 cgraph_function_or_thunk_node and an obsolete comment.
128 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
130 (build_function_decl_skip_args): Likewise.
131 (set_new_clone_decl_and_node_flags): New function.
132 (duplicate_thunk_for_node): Likewise.
133 (redirect_edge_duplicating_thunks): Likewise.
134 (cgraph_clone_node): New parameter args_to_skip, pass it to
135 redirect_edge_duplicating_thunks which is called instead of
136 cgraph_redirect_edge_callee.
137 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
138 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
140 2014-04-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
143 * fortran/trans-types.c (gfc_get_function_type): In case of recursion
144 build a variadic function type with empty argument list instead of a
145 stdarg-like function type with incomplete argument list.
147 2014-04-04 Jeff Law <law@redhat.com>
150 * config/arm/predicates.md (const_int_I_operand): New predicate.
151 (const_int_M_operand): Similarly.
152 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
154 (insv_t2, extv_reg, extzv_t2): Likewise.
155 (load_multiple_with_writeback): Similarly for const_int_I_operand.
156 (pop_multiple_with_writeback_and_return): Likewise.
157 (vfp_pop_multiple_with_writeback): Likewise
159 2014-04-04 Richard Biener <rguenther@suse.de>
162 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
163 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
165 * gimplify.h (gimple_add_tmp_var_fn): Declare.
166 * gimplify.c (gimple_add_tmp_var_fn): New function.
167 * gimple-expr.h (create_tmp_reg_fn): Declare.
168 * gimple-expr.c (create_tmp_reg_fn): New function.
169 * gimple-low.c (record_vars_into): Don't change cfun.
170 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
171 code generation without cfun.
173 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
176 * Makefile.in (install-driver): Fix shell scripting.
178 2014-04-03 Cong Hou <congh@google.com>
180 PR tree-optimization/60505
181 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
182 threshold of number of iterations below which no vectorization will be
184 * tree-vect-loop.c (new_loop_vec_info):
185 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
186 * tree-vect-loop.c (vect_analyze_loop_operations):
187 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
188 * tree-vect-loop.c (vect_transform_loop):
189 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
190 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
191 of iterations of the loop and see if we should build the epilogue.
193 2014-04-03 Richard Biener <rguenther@suse.de>
195 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
197 (streamer_tree_cache_create): Adjust.
198 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
199 to allow optional nodes array.
200 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
201 (streamer_tree_cache_append): Likewise.
202 (streamer_tree_cache_create): Create nodes array optionally
203 as specified by parameter.
204 * lto-streamer-out.c (create_output_block): Avoid maintaining
205 the node array in the writer cache.
206 (DFS_write_tree): Remove assertion.
207 (produce_asm_for_decls): Free the out decl state hash table
209 * lto-streamer-in.c (lto_data_in_create): Adjust for
210 streamer_tree_cache_create prototype change.
212 2014-04-03 Richard Biener <rguenther@suse.de>
214 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
215 set TREE_CHAIN to NULL_TREE.
217 2014-04-03 Richard Biener <rguenther@suse.de>
219 PR tree-optimization/60740
220 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
221 over all GIMPLE_COND operands.
223 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
225 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
226 (Weffc++): Remove Scott's numbering, merge lists and reference
232 * class.c (accessible_nvdtor_p): New.
233 (check_bases): Don't check base destructor here ...
234 (check_bases_and_members): ... check them here. Trigger on
235 Wnon-virtual-dtor flag.
236 (finish_struct_1): Use accessible_nvdtor_p.
239 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
240 * g++.dg/warn/Wnvdtor-2.C: New.
241 * g++.dg/warn/Wnvdtor-3.C: New.
242 * g++.dg/warn/Wnvdtor-4.C: New.
243 * g++.dg/warn/Weff1.C: Delete.
244 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
245 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
247 2014-04-03 Nick Clifton <nickc@redhat.com>
249 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
252 2014-04-03 Martin Jambor <mjambor@suse.cz>
254 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
255 mention gcc_unreachable before failing.
256 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
259 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
262 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
263 code and instead mark the context inconsistent.
264 (possible_polymorphic_call_targets): For inconsistent contexts
265 return empty complete list.
267 2014-04-02 Anthony Green <green@moxielogic.com>
269 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
270 (extendqisi2, extendhisi2): Define.
271 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
272 (WCHAR_TYPE): Change to unsigned int.
274 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
276 PR tree-optimization/60733
277 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
278 insertion point for PHI candidates to be the end of the feeding
279 block for the PHI argument.
281 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
283 PR rtl-optimization/60650
284 * lra-constraints.c (process_alt_operands): Decrease reject for
285 earlyclobber matching.
287 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
289 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
291 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
293 * config/spu/spu.c (pad_bb): Do not crash when the last
294 insn is CODE_FOR_blockage.
296 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
298 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
299 lies outside the target mode.
301 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
304 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
305 software floating point or no floating point registers, do not
306 allow any type in the FPRs. Eliminate a test for SPE SIMD types
307 in GPRs that occurs after we tested for GPRs that would never be
310 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
311 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
312 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
313 specifically allow DDmode, since that does not use the SPE SIMD
316 2014-04-02 Richard Biener <rguenther@suse.de>
319 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
320 MODE_INTs. Properly use negv_optab.
321 (expand_abs): Likewise.
323 2014-04-02 Richard Biener <rguenther@suse.de>
326 * Makefile.in (install-driver): Guard extra installs with special
329 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
331 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
334 2014-04-01 Richard Henderson <rth@redhat.com>
337 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
338 alternative enabled before register allocation.
340 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
342 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
343 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
345 (nios2_large_got_address): Remove unneeded 'sym' parameter.
346 (nios2_got_address): Update nios2_large_got_address call site.
347 (nios2_delegitimize_address): New function.
348 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
349 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
350 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
352 2014-04-01 Martin Husemann <martin@duskware.de>
354 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
357 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
359 PR rtl-optimization/60604
360 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
361 check from register_operand.
362 (register_operand): Redefine in terms of general_operand.
363 (nonmemory_operand): Use register_operand for the non-constant cases.
365 2014-04-01 Richard Biener <rguenther@suse.de>
367 * gimple.h (struct gimple_statement_base): Align subcode to
370 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
372 * doc/invoke.texi (mapp-regs): Clarify.
374 2014-03-31 Ulrich Drepper <drepper@gmail.com>
376 * config/i386/avx512fintrin.h (__v32hi): Define type.
378 (_mm512_set1_epi8): Define.
379 (_mm512_set1_epi16): Define.
380 (_mm512_set4_epi32): Define.
381 (_mm512_set4_epi64): Define.
382 (_mm512_set4_pd): Define.
383 (_mm512_set4_ps): Define.
384 (_mm512_setr4_epi64): Define.
385 (_mm512_setr4_epi32): Define.
386 (_mm512_setr4_pd): Define.
387 (_mm512_setr4_ps): Define.
388 (_mm512_setzero_epi32): Define.
390 2014-03-31 Martin Jambor <mjambor@suse.cz>
393 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
394 callsite_arguments_match_p. Updated all callers. Also check types of
395 corresponding formal parameters and actual arguments.
396 (not_all_callers_have_enough_arguments_p) Renamed to
397 some_callers_have_mismatched_arguments_p.
399 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
401 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
403 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
406 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
409 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
411 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
412 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
414 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
415 Use FMAMODE_NOVF512 mode iterator.
416 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
417 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
418 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
419 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
421 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
422 Use VF_128_256 mode iterator.
423 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
426 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
428 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
429 static chain if needed.
431 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
434 * lra-constraints.c (index_part_to_reg): New.
435 (process_address): Use it.
437 2014-03-27 Jeff Law <law@redhat.com>
438 Jakub Jelinek <jakub@redhat.com>
441 * expr.c (do_tablejump): Use simplify_gen_binary rather than
442 gen_rtx_{PLUS,MULT} to build up the address expression.
444 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
445 creating non-canonical RTL.
447 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
450 * ipa-inline.c (want_inline_small_function_p): Short circuit large
451 functions; reorganize to make cheap checks first.
452 (inline_small_functions): Do not estimate growth when dumping;
454 * ipa-inline.h (inline_summary): Add min_size.
455 (growth_likely_positive): New function.
456 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
457 (set_cond_stmt_execution_predicate): Cleanup.
458 (estimate_edge_size_and_time): Compute min_size.
459 (estimate_calls_size_and_time): Likewise.
460 (estimate_node_size_and_time): Likewise.
461 (inline_update_overall_summary): Update min_size.
462 (do_estimate_edge_time): Likewise.
463 (do_estimate_edge_size): Update.
464 (do_estimate_edge_hints): Update.
465 (growth_likely_positive): New function.
467 2014-03-28 Jakub Jelinek <jakub@redhat.com>
470 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
471 also if addr has VOIDmode.
473 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
475 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
476 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
478 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
479 instructions as well as AdvancedSIMD loads.
481 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
483 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
484 Use crypto_aese type.
485 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
486 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
487 crypto_aese, crypto_aesmc. Move to types.md.
488 * config/arm/types.md (crypto_aes): Split into crypto_aese,
490 * config/arm/iterators.md (crypto_type): Likewise.
492 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
494 * cgraph.c: Include expr.h and tree-dfa.h.
495 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
498 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
501 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
502 regs from checking multi-reg pseudos.
504 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
506 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
508 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
510 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
511 if it would clobber the stack pointer, even temporarily.
513 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
515 * mode-switching.c: Make small adjustments to the top comment.
517 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
519 * config/rs6000/constraints.md (wD constraint): New constraint to
520 match the constant integer to get the top DImode/DFmode out of a
521 vector in a VSX register.
523 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
524 match the constant integer to get the top DImode/DFmode out of a
525 vector in a VSX register.
527 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
530 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
533 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
534 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
536 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
537 Optimize vec_extract of 64-bit values, where the value being
538 extracted is in the top word, where we can use scalar
539 instructions. Add direct move and store support. Combine the big
540 endian/little endian vector select load support into a single insn.
541 (vsx_extract_<mode>_internal1): Likewise.
542 (vsx_extract_<mode>_internal2): Likewise.
543 (vsx_extract_<mode>_load): Likewise.
544 (vsx_extract_<mode>_store): Likewise.
545 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
546 combined into vsx_extract_<mode>_load.
547 (vsx_extract_<mode>_one_le): Likewise.
549 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
550 define the top 64-bit vector element.
552 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
555 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
556 Document vec_vbpermq builtin.
559 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
560 enable use of xxsldwi and xxpermdi builtin functions.
561 (vec_xxpermdi): Likewise.
563 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
564 Document use of vec_xxsldwi and vec_xxpermdi builtins.
566 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
568 PR rtl-optimization/60650
569 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
571 (find_spills_for): New.
572 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
573 Spill all pseudos on the second iteration.
575 2014-03-27 Marek Polacek <polacek@redhat.com>
578 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
581 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
583 * config/s390/s390.c (s390_can_use_return_insn): Check for
584 call-saved FPRs on 31 bit.
586 2014-03-27 Jakub Jelinek <jakub@redhat.com>
589 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
590 if they need regimplification, just drop them instead of
591 calling gimple_regimplify_operands on them.
593 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
596 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
597 (aarch64_frame_pointer_required): Adjust logic.
598 (aarch64_can_eliminate): Adjust logic.
599 (aarch64_override_options_after_change): Adjust logic.
601 2014-03-27 Dehao Chen <dehao@google.com>
603 * ipa-inline.c (early_inliner): Update node's inline info.
605 2014-03-26 Dehao Chen <dehao@google.com>
607 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
608 compiler inserted conditional jumps for NAN float check.
610 2014-03-26 Jakub Jelinek <jakub@redhat.com>
612 * ubsan.h (ubsan_create_data): Change second argument's type
613 to const location_t *.
614 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
616 (ubsan_create_data): Change second argument to const location_t *PLOC.
617 Create Loc field whenever PLOC is non-NULL.
618 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
619 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
623 * real.c (real_to_integer2): Change type of low to UHWI.
625 2014-03-26 Tobias Burnus <burnus@net-b.de>
627 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
628 (CILK_SELF_SPECS): New define.
629 (driver_self_specs): Use it.
631 2014-03-26 Richard Biener <rguenther@suse.de>
633 * tree-pretty-print.c (percent_K_format): Implement special
634 case for LTO and its stripped down BLOCK tree.
636 2014-03-26 Jakub Jelinek <jakub@redhat.com>
639 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
641 * tree-vrp.c (simplify_internal_call_using_ranges): If only
642 one range is range_int_cst_p, but not both, at least optimize
643 addition/subtraction of 0 and multiplication by 0 or 1.
644 * gimple-fold.c (gimple_fold_call): Fold
645 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
646 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
647 INTEGER_CSTs, try to fold at least x * 0 and y - y.
649 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
651 PR rtl-optimization/60452
652 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
653 <case REG>: Return 1 for invalid offsets from the frame pointer.
655 2014-03-26 Marek Polacek <polacek@redhat.com>
658 * doc/extend.texi (C Extensions): Mention variable-length arrays in
661 2014-03-26 Marek Polacek <polacek@redhat.com>
664 * doc/extend.texi (Designated Inits): Describe what happens to omitted
667 2014-03-26 Marek Polacek <polacek@redhat.com>
670 * ira-color.c (update_conflict_hard_regno_costs): Perform the
671 multiplication in unsigned type.
673 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
675 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
677 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
679 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
681 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
684 * cif-code.def (UNREACHABLE) New code.
685 * ipa-inline.c (inline_small_functions): Skip edges to
686 __builtlin_unreachable.
687 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
688 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
689 predicate to __bulitin_unreachable.
690 (set_cond_stmt_execution_predicate): Fix issue when
691 invert_tree_comparison returns ERROR_MARK.
692 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
693 propagate to inline clones.
694 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
696 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
697 * cgraphclones.c (cgraph_clone_node): If call destination is already
698 ureachable, do not redirect it back.
699 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
702 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
704 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
705 Do not modify inline clones.
707 2014-03-25 Jakub Jelinek <jakub@redhat.com>
709 * config/i386/i386.md (general_sext_operand): New mode attr.
710 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
711 don't generate (sign_extend (const_int)).
712 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
713 operands[2]. Use We constraint instead of <i> and
714 <general_sext_operand> predicate instead of <general_operand>.
715 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
716 * config/i386/constraints.md (We): New constraint.
717 * config/i386/predicates.md (x86_64_sext_operand,
718 sext_operand): New predicates.
720 2014-03-25 Martin Jambor <mjambor@suse.cz>
723 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
724 inconsistent devirtualizations to __builtin_unreachable.
726 2014-03-25 Marek Polacek <polacek@redhat.com>
729 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
731 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
733 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
734 order of elements for big-endian.
736 2014-03-25 Richard Biener <rguenther@suse.de>
739 * gimplify-me.c (gimple_regimplify_operands): Update the
742 2014-03-25 Martin Jambor <mjambor@suse.cz>
745 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
746 (lto_output_varpool_node): Likewise.
747 (input_overwrite_node): Likewise.
748 (input_varpool_node): Likewise.
750 2014-03-25 Richard Biener <rguenther@suse.de>
752 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
755 2014-03-25 Jakub Jelinek <jakub@redhat.com>
757 * combine.c (simplify_compare_const): Add MODE argument.
758 Handle mode_width 0 as very large mode_width.
759 (try_combine, simplify_comparison): Adjust callers.
761 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
762 type to avoid signed integer overflow.
763 * explow.c (plus_constant): Likewise.
765 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
767 * doc/generic.texi: Correct typos.
769 2014-03-24 Tobias Burnus <burnus@net-b.de>
771 * doc/invoke.texi (-flto): Expand section about
772 using static libraries with LTO.
774 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
776 PR rtl-optimization/60501
777 * optabs.def (addptr3_optab): New optab.
778 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
779 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
780 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
782 * lra.c (emit_add3_insn): Use the addptr pattern if available.
784 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
786 2014-03-24 Ulrich Drepper <drepper@gmail.com>
788 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
791 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
792 (_mm256_undefined_ps): Define.
793 (_mm256_undefined_pd): Define.
794 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
795 (_mm_undefined_pd): Define.
796 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
797 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
798 (_mm512_undefined_ps): Define.
799 (_mm512_undefined_pd): Define.
800 Use _mm*_undefined_*.
801 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
803 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
805 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
806 (lshr_simd): DI mode added.
807 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
808 (aarch64_ushr_simddi): Likewise.
809 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
810 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
811 (vshrd_n_u64): Likewise.
813 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
815 * Makefile.in (s-macro_list): Depend on cc1.
817 2014-03-23 Teresa Johnson <tejohnson@google.com>
819 * ipa-utils.c (ipa_print_order): Use specified dump file.
821 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
823 PR rtl-optimization/60601
824 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
826 * gcc.c (eval_spec_function): Initialize save_growing_value.
828 2014-03-22 Jakub Jelinek <jakub@redhat.com>
831 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
832 code == MINUS_EXPR, never swap op0 with op1.
834 * toplev.c (init_local_tick): Avoid signed integer multiplication
836 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
837 shift by first operand's bitsize.
839 2014-03-21 Jakub Jelinek <jakub@redhat.com>
842 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
844 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
845 TARGET_ISA_64BIT_P(x).
847 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
849 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
850 pattern for vector nor instead of subtract from splat(-1).
851 (altivec_expand_vec_perm_const_le): Likewise.
853 2014-03-21 Richard Henderson <rth@twiddle.net>
856 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
857 related insns after epilogue_completed.
859 2014-03-21 Martin Jambor <mjambor@suse.cz>
862 * cgraph.h (symtab_node): New flag body_removed.
863 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
864 when removing bodies.
865 * symtab.c (dump_symtab_base): Dump body_removed flag.
866 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
867 had their bodies removed.
869 2014-03-21 Martin Jambor <mjambor@suse.cz>
872 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
875 2014-03-21 Richard Biener <rguenther@suse.de>
877 PR tree-optimization/60577
878 * tree-core.h (struct tree_base): Document nothrow_flag use
879 in VAR_DECL_NONALIASED.
880 * tree.h (VAR_DECL_NONALIASED): New.
881 (may_be_aliased): Adjust.
882 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
884 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
886 * expr.c (expand_expr_real_1): Remove outdated comment.
888 2014-03-20 Jakub Jelinek <jakub@redhat.com>
891 * ira.c (adjust_cleared_regs): Call copy_rtx on
892 *reg_equiv[REGNO (loc)].src_p before passing it to
893 simplify_replace_fn_rtx.
896 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
897 into CONST, put pic register as first operand of PLUS. Use
898 gen_const_mem for both 32-bit and 64-bit PIC got loads.
900 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
902 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
904 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
906 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
907 around for store forwarding issue in the FPU on the UT699.
908 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
909 loads and operations if -mfix-ut699 is specified.
910 (divtf3_hq): Tweak attribute.
911 (sqrttf2_hq): Likewise.
913 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
915 * calls.c (store_one_arg): Remove incorrect const qualification on the
916 type of the temporary.
917 * cfgexpand.c (expand_return): Likewise.
918 * expr.c (expand_constructor): Likewise.
919 (expand_expr_real_1): Likewise.
921 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
923 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
926 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
929 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
931 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
933 * config/arm/aarch-common-protos.h
934 (alu_cost_table): Fix spelling of "extend".
935 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
937 2014-03-19 Richard Biener <rguenther@suse.de>
940 * tree-core.h (tree_type_common): Re-order pointer members
941 to reduce recursion depth during GC walks.
943 2014-03-19 Marek Polacek <polacek@redhat.com>
946 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
949 2014-03-19 Richard Biener <rguenther@suse.de>
952 * lto-streamer-in.c (input_function): In WPA stage do not drop
955 2014-03-19 Jakub Jelinek <jakub@redhat.com>
957 PR tree-optimization/60559
958 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
959 with build_zero_cst assignment.
961 2014-03-18 Kai Tietz <ktietz@redhat.com>
963 PR rtl-optimization/56356
964 * sdbout.c (sdbout_parms): Verify that parms'
965 incoming argument is valid.
966 (sdbout_reg_parms): Likewise.
968 2014-03-18 Richard Henderson <rth@redhat.com>
971 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
972 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
973 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
975 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
977 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
978 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
979 Italicize plugin event names in description. Explain that
980 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
981 Remind that no GCC functions should be called after PLUGIN_FINISH.
982 Explain what pragmas with expansion are.
984 2014-03-18 Martin Liska <mliska@suse.cz>
986 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
987 gimple call statement is update.
988 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
989 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
991 2014-03-18 Jakub Jelinek <jakub@redhat.com>
994 * ubsan.c (ubsan_instrument_unreachable): Call
995 initialize_sanitizer_builtins.
996 (ubsan_pass): Likewise.
999 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
1000 varpool_finalize_decl instead of rest_of_decl_compilation.
1002 2014-03-18 Richard Biener <rguenther@suse.de>
1004 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
1005 by using bitmap_and_compl instead of bitmap_and_compl_into.
1006 (df_rd_transfer_function): Likewise.
1008 2014-03-18 Richard Biener <rguenther@suse.de>
1010 * doc/lto.texi (fresolution): Fix typo.
1012 2014-03-18 Richard Biener <rguenther@suse.de>
1014 * doc/invoke.texi (flto): Update for changes in 4.9.
1016 2014-03-18 Richard Biener <rguenther@suse.de>
1018 * doc/loop.texi: Remove section on the removed lambda framework.
1019 Update loop docs with recent changes in preserving loop structure.
1021 2014-03-18 Richard Biener <rguenther@suse.de>
1023 * doc/lto.texi (-fresolution): Document.
1025 2014-03-18 Richard Biener <rguenther@suse.de>
1027 * doc/contrib.texi: Adjust my name.
1029 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1032 * internal-fn.c: Include diagnostic-core.h.
1033 (expand_BUILTIN_EXPECT): New function.
1034 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1035 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1036 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1037 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1039 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1040 Revert 3 argument __builtin_expect code.
1041 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1042 * gimple-fold.c (gimple_fold_call): Likewise.
1043 * tree.h (fold_builtin_expect): New prototype.
1044 * builtins.c (build_builtin_expect_predicate): Add predictor
1045 argument, if non-NULL, create 3 argument __builtin_expect.
1046 (fold_builtin_expect): No longer static. Add ARG2 argument,
1047 pass it through to build_builtin_expect_predicate.
1048 (fold_builtin_2): Adjust caller.
1049 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1050 * internal-fn.def (BUILTIN_EXPECT): New.
1052 2014-03-18 Tobias Burnus <burnus@net-b.de>
1055 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1056 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1057 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1059 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1062 * predict.c (combine_predictions_for_bb): Fix up formatting.
1063 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1064 fill what it points to if non-NULL.
1065 (tree_predict_by_opcode): Adjust caller, use the predictor.
1066 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1068 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1070 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1071 proper constant for the store mode.
1073 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1075 * symtab.c (change_decl_assembler_name): Fix transparent alias
1078 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1080 * config/aarch64/aarch64.c: Correct the comments about the
1081 aarch64 stack layout.
1083 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1085 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1086 check for GF_OMP_FOR_KIND_FOR.
1088 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1090 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1091 ymm and zmm register names.
1093 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1096 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1097 note creation for the 2010-08-31 changes.
1099 2014-03-17 Marek Polacek <polacek@redhat.com>
1102 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1103 as -fno-tree-loop-vectorize.
1104 (expand_omp_simd): Likewise.
1106 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1108 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1109 (eligible_for_call_delay): New prototype.
1110 * config/sparc/sparc.c (tls_call_delay): Rename into...
1111 (eligible_for_call_delay): ...this. Return false if the instruction
1112 cannot be put in the delay slot of a branch.
1113 (eligible_for_restore_insn): Simplify.
1114 (eligible_for_return_delay): Return false if the instruction cannot be
1115 put in the delay slot of a branch and simplify.
1116 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1117 put in the delay slot of a branch.
1118 * config/sparc/sparc.md (fix_ut699): New attribute.
1119 (tls_call_delay): Delete.
1120 (in_call_delay): Reimplement.
1121 (eligible_for_sibcall_delay): Rename into...
1122 (in_sibcall_delay): ...this.
1123 (eligible_for_return_delay): Rename into...
1124 (in_return_delay): ...this.
1125 (in_branch_delay): Reimplement.
1126 (in_uncond_branch_delay): Delete.
1127 (in_annul_branch_delay): Delete.
1129 2014-03-14 Richard Henderson <rth@redhat.com>
1132 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1133 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1134 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1135 (floathi splitters): Remove.
1136 (float<SWI48x>xf2): New pattern.
1137 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1138 code that tried to handle DImode for 32-bit, but which was excluded
1139 by the pattern's condition. Drop allocation of stack temporary.
1140 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1141 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1142 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1143 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1144 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1145 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1146 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1147 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1148 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1149 (*float<SWI48x><X87MODEF>2_i387): Remove.
1150 (all float _with_temp splitters): Remove.
1151 (*float<SWI48x><MODEF>2_i387): New pattern.
1152 (*float<SWI48><MODEF>2_sse): New pattern.
1153 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1154 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1156 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1157 Marek Polacek <polacek@redhat.com>
1160 * common.opt (dump_base_name_prefixed): New Variable.
1161 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1162 if x_dump_base_name_prefixed is already set, set it at the end.
1164 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1166 PR rtl-optimization/60508
1167 * lra-constraints.c (get_reload_reg): Add new parameter
1169 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1170 Pass the new parameter values.
1172 2014-03-14 Richard Biener <rguenther@suse.de>
1174 * common.opt: Revert unintented changes from r205065.
1177 2014-03-14 Richard Biener <rguenther@suse.de>
1180 * cfghooks.c (split_block): Properly adjust all loops the
1181 block was a latch of.
1183 2014-03-14 Martin Jambor <mjambor@suse.cz>
1186 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1189 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1192 * config/avr/avr.c (avr_set_current_function): Pass function name
1193 through default_strip_name_encoding before sanity checking instead
1194 of skipping the first char of the assembler name.
1196 2014-03-13 Richard Henderson <rth@redhat.com>
1199 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1200 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1201 * config/i386/i386-protos.h: Likewise.
1202 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1203 in the expander instead of a splitter.
1204 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1205 any possibility of requiring a memory.
1206 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1207 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1208 (fp branch splitters): Update for ix86_split_fp_branch.
1209 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1210 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1211 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1212 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1213 (*fop_<MODEF>_3_i387): Likewise.
1214 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1215 (splitters for the fop_* register patterns): Remove.
1216 (fscalexf4_i387): Rename from *fscalexf4_i387.
1217 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1219 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1221 PR tree-optimization/59779
1222 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1223 type for bitsize and maxsize instead of HOST_WIDE_INT.
1225 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1227 PR rtl-optimization/57320
1228 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1229 the CFG after thread_prologue_and_epilogue_insns.
1231 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1233 PR rtl-optimization/57189
1234 * lra-constraints.c (process_alt_operands): Disfavor spilling
1237 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1239 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1241 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1243 PR tree-optimization/59025
1245 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1246 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1248 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1251 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1252 calls of avr_out_plus_1.
1254 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1256 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1257 BB's single pred and update the father loop's latch info later.
1259 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1261 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1265 (VEC_base): Likewise.
1266 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1267 registers, we need to swap double words in little endian mode.
1269 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1270 to be a container mode for 128-bit integer operations added in ISA
1271 2.07. Unlike TImode and PTImode, the preferred register set is
1272 the Altivec/VMX registers for the 128-bit operations.
1274 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1276 (rs6000_split_128bit_ok_p): Likewise.
1278 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1279 macros for creating ISA 2.07 normal and overloaded builtin
1280 functions with 3 arguments.
1281 (BU_P8V_OVERLOAD_3): Likewise.
1282 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1283 for use as overloaded functions.
1284 (VPERM_1TI_UNS): Likewise.
1285 (VSEL_1TI): Likewise.
1286 (VSEL_1TI_UNS): Likewise.
1287 (ST_INTERNAL_1ti): Likewise.
1288 (LD_INTERNAL_1ti): Likewise.
1289 (XXSEL_1TI): Likewise.
1290 (XXSEL_1TI_UNS): Likewise.
1291 (VPERM_1TI): Likewise.
1292 (VPERM_1TI_UNS): Likewise.
1293 (XXPERMDI_1TI): Likewise.
1294 (SET_1TI): Likewise.
1295 (LXVD2X_V1TI): Likewise.
1296 (STXVD2X_V1TI): Likewise.
1297 (VEC_INIT_V1TI): Likewise.
1298 (VEC_SET_V1TI): Likewise.
1299 (VEC_EXT_V1TI): Likewise.
1300 (EQV_V1TI): Likewise.
1301 (NAND_V1TI): Likewise.
1302 (ORC_V1TI): Likewise.
1303 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1304 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1306 (VADDUQM): Likewise.
1307 (VSUBCUQ): Likewise.
1308 (VADDEUQM): Likewise.
1309 (VADDECUQ): Likewise.
1310 (VSUBEUQM): Likewise.
1311 (VSUBECUQ): Likewise.
1313 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1314 __int128_t and __uint128_t types.
1315 (__uint128_type): Likewise.
1316 (altivec_categorize_keyword): Add support for vector __int128_t,
1317 vector __uint128_t, vector __int128, and vector unsigned __int128
1318 as a container type for TImode operations that need to be done in
1319 VSX/Altivec registers.
1320 (rs6000_macro_to_expand): Likewise.
1321 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1322 to support 128-bit integer instructions vaddcuq, vadduqm,
1323 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1324 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1326 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1327 for V1TImode, and set up preferences to use VSX/Altivec registers.
1328 Setup VSX reload handlers.
1329 (rs6000_debug_reg_global): Likewise.
1330 (rs6000_init_hard_regno_mode_ok): Likewise.
1331 (rs6000_preferred_simd_mode): Likewise.
1332 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1333 (easy_altivec_constant): Likewise.
1334 (output_vec_const_move): Likewise.
1335 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1337 (rs6000_expand_vector_extract): Likewise.
1338 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1340 (rs6000_const_vec): Add support for V1TImode.
1341 (rs6000_emit_le_vsx_load): Swap double words when loading or
1342 storing TImode/V1TImode.
1343 (rs6000_emit_le_vsx_store): Likewise.
1344 (rs6000_emit_le_vsx_move): Likewise.
1345 (rs6000_emit_move): Add support for V1TImode.
1346 (altivec_expand_ld_builtin): Likewise.
1347 (altivec_expand_st_builtin): Likewise.
1348 (altivec_expand_vec_init_builtin): Likewise.
1349 (altivec_expand_builtin): Likewise.
1350 (rs6000_init_builtins): Add support for V1TImode type. Add
1351 support for ISA 2.07 128-bit integer builtins. Define type names
1352 for the VSX/Altivec vector types.
1353 (altivec_init_builtins): Add support for overloaded vector
1354 functions with V1TImode type.
1355 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1356 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1358 (rs6000_split_128bit_ok_p): Likewise.
1359 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1360 __int128_t and vector __uint128_t.
1362 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1363 and mode attributes.
1370 (VS_scalar): Likewise.
1371 (VS_double): Likewise.
1372 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1374 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1375 we support the ISA 2.07 128-bit integer arithmetic instructions.
1376 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1377 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1378 and TImode types for use with the builtin functions.
1379 (V1TI_type_node): Likewise.
1380 (unsigned_V1TI_type_node): Likewise.
1381 (intTI_type_internal_node): Likewise.
1382 (uintTI_type_internal_node): Likewise.
1384 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1385 128-bit builtin functions.
1386 (UNSPEC_VADDEUQM): Likewise.
1387 (UNSPEC_VADDECUQ): Likewise.
1388 (UNSPEC_VSUBCUQ): Likewise.
1389 (UNSPEC_VSUBEUQM): Likewise.
1390 (UNSPEC_VSUBECUQ): Likewise.
1391 (VM): Add V1TImode to vector mode iterators.
1393 (VI_unit): Likewise.
1394 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1395 (altivec_vaddcuq): Likewise.
1396 (altivec_vsubuqm): Likewise.
1397 (altivec_vsubcuq): Likewise.
1398 (altivec_vaddeuqm): Likewise.
1399 (altivec_vaddecuq): Likewise.
1400 (altivec_vsubeuqm): Likewise.
1401 (altivec_vsubecuq): Likewise.
1403 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1405 (BOOL_128): Likewise.
1406 (BOOL_REGS_OUTPUT): Likewise.
1407 (BOOL_REGS_OP1): Likewise.
1408 (BOOL_REGS_OP2): Likewise.
1409 (BOOL_REGS_UNARY): Likewise.
1410 (BOOL_REGS_AND_CR0): Likewise.
1412 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1413 128-bit integer builtin support.
1414 (vec_vadduqm): Likewise.
1415 (vec_vaddecuq): Likewise.
1416 (vec_vaddeuqm): Likewise.
1417 (vec_vsubecuq): Likewise.
1418 (vec_vsubeuqm): Likewise.
1419 (vec_vsubcuq): Likewise.
1420 (vec_vsubuqm): Likewise.
1422 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1423 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1424 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1425 128-bit integer add/subtract to ISA 2.07.
1427 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1429 * config/arc/arc.c (arc_predicate_delay_insns):
1430 Fix third argument passed to conditionalize_nonjump.
1432 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1434 * config/aarch64/aarch64-builtins.c
1435 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1436 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1437 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1438 instead of __builtin_lfloor.
1439 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1441 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1443 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1444 (tree_ssa_ifcombine_bb_1): New function.
1445 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1446 is an empty forwarder block to then_bb or vice versa and then_bb
1447 and else_bb are effectively swapped.
1449 2014-03-12 Christian Bruel <christian.bruel@st.com>
1452 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1453 REG_CFA_DEF_CFA note.
1454 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1455 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1457 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1459 PR tree-optimization/60454
1460 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1462 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1464 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1465 Do not define target_cpu_default2 to generic.
1466 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1467 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1468 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1470 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1471 Marc Glisse <marc.glisse@inria.fr>
1473 PR tree-optimization/60502
1474 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1475 instead of build_low_bits_mask.
1477 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1480 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1481 if there are multiple uses, but op doesn't live on E edge.
1482 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1483 clobber stmts before __builtin_unreachable.
1485 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1487 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1488 hard_frame_pointer_rtx.
1489 * cse.c (cse_insn): Remove volatile check.
1490 * cselib.c (cselib_process_insn): Likewise.
1491 * dse.c (scan_insn): Likewise.
1493 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1495 * config/arc/arc.c (conditionalize_nonjump): New function,
1497 (arc_ifcvt): ... this.
1498 (arc_predicate_delay_insns): Use it.
1500 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1502 * config/arc/predicates.md (extend_operand): During/after reload,
1503 allow const_int_operand.
1504 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1505 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1506 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1508 (umulsi3_highpart_i): Likewise.
1510 2014-03-11 Richard Biener <rguenther@suse.de>
1512 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1513 Add asserts to guard possible wrong-code bugs.
1515 2014-03-11 Richard Biener <rguenther@suse.de>
1517 PR tree-optimization/60429
1518 PR tree-optimization/60485
1519 * tree-ssa-structalias.c (set_union_with_increment): Properly
1520 take into account all fields that overlap the shifted vars.
1521 (do_sd_constraint): Likewise.
1522 (do_ds_constraint): Likewise.
1523 (get_constraint_for_ptr_offset): Likewise.
1525 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1527 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1528 (nios2_compute_frame_layout):
1529 Add calculation of cfun->machine->fp_save_offset.
1530 (nios2_expand_prologue): Correct setting of frame pointer register
1532 (nios2_expand_epilogue): Update recovery of stack pointer from
1533 frame pointer accordingly.
1534 (nios2_initial_elimination_offset): Update calculation of offset
1535 for eliminating to HARD_FRAME_POINTER_REGNUM.
1537 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1540 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1541 cgraph_get_create_node on VAR_DECLs.
1543 2014-03-10 Richard Biener <rguenther@suse.de>
1546 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1548 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1550 * config/vms/vms.opt (vms_float_format): New variable.
1552 2014-03-08 Tobias Burnus <burnus@net-b.de>
1554 * doc/invoke.texi (-fcilkplus): Update implementation status.
1556 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1557 Richard Biener <rguenther@suse.de>
1559 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1560 consistently accross all TUs.
1561 (run_gcc): Enable -fshort-double automatically at link at link-time
1562 and disallow override.
1564 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1567 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1568 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1569 if they can't be used.
1571 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1573 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1574 for Solaris 11/x86 ld.
1575 * configure: Regenerate.
1577 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1579 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1580 (LIB_TLS_SPEC): Save as ld_tls_libs.
1581 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1582 (HAVE_AS_IX86_TLSLDM): New test.
1583 * configure, config.in: Regenerate.
1584 * config/i386/i386.c (legitimize_tls_address): Fall back to
1585 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1586 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1587 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1588 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1590 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1592 * common.opt (fira-loop-pressure): Mark as optimization.
1594 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1596 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1597 an OpenMP mappable type.
1599 2014-03-06 Matthias Klose <doko@ubuntu.com>
1601 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1602 MULTILIB_OSDIRNAMES is not defined.
1604 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1605 Meador Inge <meadori@codesourcery.com>
1608 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1609 (arm_legitimize_address): Call legitimize_tls_address for any
1610 arm_tls_referenced_p expression, handle constant addend. Call it
1611 before testing for !TARGET_ARM.
1612 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1614 2014-03-06 Richard Biener <rguenther@suse.de>
1620 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1622 * tree-streamer.c (record_common_node): Assert we don't record
1623 nodes with type double.
1624 (preload_common_node): Skip type double, complex double and double
1625 pointer since it is now frontend dependent due to fshort-double option.
1627 2014-03-06 Richard Biener <rguenther@suse.de>
1629 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1630 or -fno-lto is specified and the linker has full plugin support.
1631 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1632 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1633 * lto-wrapper.c (merge_and_complain): Merge compile-time
1634 optimization levels.
1635 (run_gcc): And pass it through to the link options.
1637 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1641 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1643 * cselib.c (remove_useless_values): Skip to avoid quadratic
1644 behavior if the condition moved from...
1645 (cselib_process_insn): ... here holds.
1647 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1650 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1651 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1654 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1655 (TM_H): Add x86-tune.def.
1657 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1659 * config/aarch64/aarch64.c (generic_tunings):
1660 Use cortexa57_extra_costs.
1662 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1665 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1666 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1667 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1670 2014-03-04 Heiher <r@hev.cc>
1672 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1673 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1675 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1677 * config/i386/predicates.md (const2356_operand): Change to ...
1678 (const2367_operand): ... this.
1679 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1681 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1682 (*avx512pf_scatterpf<mode>sf): Ditto.
1683 (avx512pf_scatterpf<mode>df): Ditto.
1684 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1685 (*avx512pf_scatterpf<mode>df): Ditto.
1686 * config/i386/i386.c (ix86_expand_builtin): Update
1687 incorrect hint operand error message.
1689 2014-03-04 Richard Biener <rguenther@suse.de>
1691 * lto-section-in.c (lto_get_section_data): Fix const cast.
1693 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1695 * tree-streamer.c (record_common_node): Assert we don't record
1696 nodes with type double.
1697 (preload_common_node): Skip type double, complex double and double
1698 pointer since it is now frontend dependent due to fshort-double option.
1700 2014-03-04 Richard Biener <rguenther@suse.de>
1703 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1704 (lto_input_toplevel_asms): Likewise.
1705 * lto-section-in.c (lto_get_section_data): Instead do it here
1708 2014-03-04 Richard Biener <rguenther@suse.de>
1710 PR tree-optimization/60382
1711 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1712 dead PHIs a reduction.
1714 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1716 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1718 (_mm_prefetch): Move out of GCC target("sse") pragma.
1719 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1720 GCC target("prfchw") pragma.
1721 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1723 * config/i386/i386.c (ix86_option_override_internal): Enable
1724 -mprfchw with -mprefetchwt1.
1726 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1728 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1731 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1733 * opts.h (CL_PCH_IGNORE): Define.
1734 * targhooks.c (option_affects_pch_p):
1735 Return false for options that have CL_PCH_IGNORE set.
1736 * opt-functions.awk: Process PchIgnore.
1737 * doc/options.texi: Document PchIgnore.
1739 * config/arc/arc.opt (misize): Add PchIgnore property.
1741 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1743 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1744 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1745 constraint on constants to permit them being loaded into
1746 GENERAL_REGS or BASE_REGS.
1748 2014-03-03 Nick Clifton <nickc@redhat.com>
1750 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1751 anti-cacnonical alternatives.
1752 (negandhi3_real): New pattern.
1753 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1755 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1757 * config/avr/avr-mcus.def: Remove atxmega16x1.
1758 * config/avr/avr-tables.opt: Regenerate.
1759 * config/avr/t-multilib: Regenerate.
1760 * doc/avr-mmcu.texi: Regenerate.
1762 2014-03-03 Tobias Grosser <tobias@grosser.es>
1763 Mircea Namolaru <mircea.namolaru@inria.fr>
1765 PR tree-optimization/58028
1766 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1769 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1771 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1772 not handled by recognizers.
1774 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1777 * function.c (expand_function_end): Don't emit
1778 clobber_return_register sequence if clobber_after is a BARRIER.
1779 * cfgexpand.c (construct_exit_block): Append instructions before
1780 return_label to prev_bb.
1782 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1784 * config/rs6000/constraints.md: Document reserved use of "wc".
1786 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1789 * ipa.c (function_and_variable_visibility): When dissolving comdat
1790 group, also set all symbols to local.
1792 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1797 2013-12-14 Jan Hubicka <jh@suse.cz>
1799 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1801 2014-03-02 Jon Beniston <jon@beniston.com>
1807 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1808 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1809 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1810 (simple_return, *simple_return): New patterns
1811 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1812 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1814 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1816 * dwarf2out.c (gen_subprogram_die): Tidy.
1818 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1821 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1822 (*mov_t_msb_neg_negc): ... this new insn.
1824 2014-02-28 Jason Merrill <jason@redhat.com>
1827 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1830 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1833 * dwarf2out.c (decltype_auto_die): New static.
1834 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1835 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1836 (is_cxx_auto): Likewise.
1838 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1840 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1841 we are not using general regs only.
1843 2014-02-28 Richard Biener <rguenther@suse.de>
1846 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1847 previous fix and only allow to remove trivial pre-headers
1848 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1849 (remove_forwarder_block): Properly update the latch of a loop.
1851 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1854 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1855 (cselib_preserved_hash_table): New.
1856 (preserve_constants_and_equivs): Move preserved vals to it.
1857 (cselib_find_slot): Look it up first.
1858 (cselib_init): Initialize it.
1859 (cselib_finish): Release it.
1860 (dump_cselib_table): Dump it.
1862 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1865 * cselib.c (remove_useless_values): Skip to avoid quadratic
1866 behavior if the condition moved from...
1867 (cselib_process_insn): ... here holds.
1869 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1872 * var-tracking.c (vt_initialize): Apply the same condition to
1873 preserve the CFA base value.
1875 2014-02-28 Joey Ye <joey.ye@arm.com>
1878 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1879 if reload in progress or completed.
1881 2014-02-28 Tobias Burnus <burnus@net-b.de>
1884 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1887 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1889 * doc/tm.texi.in (Condition Code Status): Update documention for
1890 relative locations of cc0-setter and cc0-user.
1892 2014-02-27 Jeff Law <law@redhat.com>
1894 PR rtl-optimization/52714
1895 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1896 into two independent simple sets, if I3 is a jump, ensure the
1897 pattern we place into I3 is a (set (pc) ...).
1899 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1900 Jeff Law <law@redhat.com>
1902 PR rtl-optimization/49847
1903 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1904 are in different blocks.
1905 * doc/tm.texi (Condition Code Status): Update documention for
1906 relative locations of cc0-setter and cc0-user.
1908 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1911 * lra.c (lra_emit_add): Check SUBREG too.
1913 2014-02-27 Andreas Schwab <schwab@suse.de>
1915 * config/m68k/m68k.c (m68k_option_override): Disable
1916 -flive-range-shrinkage for classic m68k.
1917 (m68k_override_options_after_change): Likewise.
1919 2014-02-27 Marek Polacek <polacek@redhat.com>
1922 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1923 -Wmaybe-uninitialized.
1925 2014-02-27 Alan Modra <amodra@gmail.com>
1928 * reload1.c (emit_input_reload_insns): When reload_override_in,
1929 set old to rl->in_reg when rl->in_reg is a subreg.
1931 2014-02-26 Richard Biener <rguenther@suse.de>
1934 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1936 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1938 * common/config/i386/predicates.md (const1256_operand): Remove.
1939 (const2356_operand): New.
1940 (const_1_to_2_operand): Remove.
1941 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1942 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1943 (*avx512pf_gatherpf<mode>sf): Ditto.
1944 (avx512pf_gatherpf<mode>df): Ditto.
1945 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1946 (*avx512pf_gatherpf<mode>df): Ditto.
1947 (avx512pf_scatterpf<mode>sf): Ditto.
1948 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1949 (*avx512pf_scatterpf<mode>sf): Ditto.
1950 (avx512pf_scatterpf<mode>df): Ditto.
1951 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1952 (*avx512pf_scatterpf<mode>df): Ditto.
1953 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1955 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1957 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1958 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1959 (_mm512_mask_testn_epi64_mask): Move to ...
1960 * config/i386/avx512cdintrin.h: Here.
1961 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1962 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1963 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1964 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1965 TARGET_AVX512F from TARGET_AVX512CD.
1967 2014-02-26 Richard Biener <rguenther@suse.de>
1970 * ipa.c (walk_polymorphic_call_targets): Properly guard
1971 call to inline_update_overall_summary.
1973 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1976 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1977 and latches only if requested. Fix latch if it is removed.
1978 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1979 LOOPS_HAVE_PREHEADERS.
1981 2014-02-25 Andrew Pinski <apinski@cavium.com>
1983 * builtins.c (expand_builtin_thread_pointer): Create a new target
1984 when the target is NULL.
1986 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1988 PR rtl-optimization/60317
1989 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1990 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1991 * lra-assigns.c: Include params.h.
1992 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1993 other reload pseudos considerations.
1995 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1997 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1998 to use canonical form for nor<mode>3.
2000 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2003 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
2006 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
2008 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2009 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2010 (ix86_handle_option): Handle OPT_mprefetchwt1.
2011 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2012 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2014 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2015 OPTION_MASK_ISA_PREFETCHWT1.
2016 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2017 (PTA_PREFETCHWT1): New.
2018 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2019 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2020 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2021 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2022 (*prefetch_avx512pf_<mode>_: Change into ...
2023 (*prefetch_prefetchwt1_<mode>: This.
2024 * config/i386/i386.opt (mprefetchwt1): New.
2025 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2026 (_mm_prefetch): Handle intent to write.
2027 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2029 2014-02-25 Richard Biener <rguenther@suse.de>
2032 * emit-rtl.c (mem_attrs_htab): Remove.
2033 (mem_attrs_htab_hash): Likewise.
2034 (mem_attrs_htab_eq): Likewise.
2035 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2036 (init_emit_once): Do not allocate mem_attrs_htab.
2038 2014-02-25 Richard Biener <rguenther@suse.de>
2041 * lto-opts.c (lto_write_options): Output non-explicit conservative
2042 -fwrapv, -fno-trapv and -fno-strict-overflow.
2043 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2044 (run_gcc): And pass them through.
2046 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2048 * sel-sched.c (calculate_new_fences): New parameter ptime.
2049 Calculate it as a maximum over all fence cycles.
2050 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2051 Print the final schedule timing when sched_verbose.
2053 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2055 PR rtl-optimization/60292
2056 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2057 bit fot the fence instruction.
2059 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2061 * calls.h: Fix typo in comment.
2063 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2065 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2066 adjusting offsetable addresses.
2068 2014-02-24 Guozhi Wei <carrot@google.com>
2070 * sparseset.h (sparseset_pop): Fix the wrong index.
2072 2014-02-24 Walter Lee <walt@tilera.com>
2074 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2075 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2077 * common/config/tilegx/tilegx-common.c
2078 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2079 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2081 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2082 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2083 (tilegx_gimplify_va_arg_expr): Handle big endian.
2084 (tilegx_expand_unaligned_load): Ditto.
2085 (tilegx_expand_unaligned_store): Ditto.
2086 (TARGET_RETURN_IN_MSB): New.
2087 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2088 (TARGET_ENDIAN_DEFAULT): New.
2089 (TARGET_BIG_ENDIAN): Handle big endian.
2090 (BYTES_BIG_ENDIAN): Ditto.
2091 (WORDS_BIG_ENDIAN): Ditto.
2092 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2095 * config/tilegx/tilegx.md (extv): Handle big endian.
2097 (insn_st<n>): Ditto.
2098 (insn_st<n>_add<bitsuffix>): Ditto.
2099 (insn_stnt<n>): Ditto.
2100 (insn_stnt<n>_add<bitsuffix>):Ditto.
2101 (vec_interleave_highv8qi): Handle big endian.
2102 (vec_interleave_highv8qi_be): New.
2103 (vec_interleave_highv8qi_le): New.
2104 (insn_v1int_h): Handle big endian.
2105 (vec_interleave_lowv8qi): Handle big endian.
2106 (vec_interleave_lowv8qi_be): New.
2107 (vec_interleave_lowv8qi_le): New.
2108 (insn_v1int_l): Handle big endian.
2109 (vec_interleave_highv4hi): Handle big endian.
2110 (vec_interleave_highv4hi_be): New.
2111 (vec_interleave_highv4hi_le): New.
2112 (insn_v2int_h): Handle big endian.
2113 (vec_interleave_lowv4hi): Handle big endian.
2114 (vec_interleave_lowv4hi_be): New.
2115 (vec_interleave_lowv4hi_le): New.
2116 (insn_v2int_l): Handle big endian.
2117 (vec_interleave_highv2si): Handle big endian.
2118 (vec_interleave_highv2si_be): New.
2119 (vec_interleave_highv2si_le): New.
2120 (insn_v4int_h): Handle big endian.
2121 (vec_interleave_lowv2si): Handle big endian.
2122 (vec_interleave_lowv2si_be): New.
2123 (vec_interleave_lowv2si_le): New.
2124 (insn_v4int_l): Handle big endian.
2125 * config/tilegx/tilegx.opt (mbig-endian): New option.
2126 (mlittle-endian): New option.
2127 * doc/install.texi: Document tilegxbe-linux.
2128 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2130 2014-02-24 Martin Jambor <mjambor@suse.cz>
2133 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2134 there are no parameter descriptors.
2136 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2138 PR rtl-optimization/60268
2139 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2140 initialization to ...
2141 (sched_rgn_init): ... here.
2142 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2144 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2146 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2149 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2151 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2154 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2156 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2157 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2159 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2161 * config/microblaze/predicates.md: Add cmp_op predicate.
2162 * config/microblaze/microblaze.md: Add branch_compare instruction
2163 which uses cmp_op predicate and emits cmp insn before branch.
2164 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2165 to microblaze_expand_conditional_branch and consolidate logic.
2166 (microblaze_expand_conditional_branch): emit branch_compare
2167 insn instead of handling cmp op separate from branch insn.
2169 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2171 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2174 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2176 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2177 define_insn with define_expand and new define_insn
2178 *altivec_lve<VI_char>x_internal.
2179 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2180 and new define_insn *altivec_stve<VI_char>x_internal.
2181 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2183 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2185 (altivec_expand_stvex_be): New function.
2187 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2189 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2190 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2191 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2192 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2194 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2197 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2198 instead of emit_move_insn.
2200 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2202 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2204 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2205 gen_altivec_vsumsws.
2207 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2209 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2210 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2211 (altivec_lvxl_<mode>): New define_expand incorporating
2212 -maltivec=be semantics where needed.
2213 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2214 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2215 semantics where needed.
2216 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2217 (altivec_stvx_<mode>): New define_expand incorporating
2218 -maltivec=be semantics where needed.
2219 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2220 VM2 iterator instead of V4SI.
2221 (altivec_stvxl_<mode>): New define_expand incorporating
2222 -maltivec=be semantics where needed.
2223 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2224 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2225 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2226 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2227 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2228 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2229 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2230 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2231 ALTIVEC_BUILTIN_STVXL.
2232 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2233 (altivec_expand_stvx_be): Likewise.
2234 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2235 (altivec_expand_lvx_be): Likewise.
2236 (altivec_expand_stvx_be): Likewise.
2237 (altivec_expand_builtin): Add cases for
2238 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2239 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2240 (altivec_init_builtins): Add definitions for
2241 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2242 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2244 2014-02-21 Catherine Moore <clm@codesourcery.com>
2246 * doc/invoke.texi (mvirt, mno-virt): Document.
2247 * config/mips/mips.opt (mvirt): New option.
2248 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2250 2014-02-21 Richard Biener <rguenther@suse.de>
2252 PR tree-optimization/60276
2253 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2254 (STMT_VINFO_MIN_NEG_DIST): New macro.
2255 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2256 STMT_VINFO_MIN_NEG_DIST.
2257 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2258 made for negative dependence distances still hold.
2260 2014-02-21 Richard Biener <rguenther@suse.de>
2263 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2264 DECL_INITIAL for globals not in the current function context.
2266 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2268 PR tree-optimization/56490
2269 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2270 * tree-ssa-uninit.c: Include params.h.
2271 (compute_control_dep_chain): Add num_calls argument, return false
2272 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2273 num_calls to recursive call.
2274 (find_predicates): Change dep_chain into normal array,
2275 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2276 variable and adjust compute_control_dep_chain caller.
2277 (find_def_preds): Likewise.
2279 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2281 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2282 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2284 2014-02-21 Nick Clifton <nickc@redhat.com>
2286 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2287 (pushhi1): Likewise.
2288 (popqi1): Add mode to pre_dec.
2291 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2293 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2294 mode for mask of V8SFmode permutation.
2296 2014-02-20 Richard Henderson <rth@redhat.com>
2299 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2300 a new pseudo for OLDVAL.
2302 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2305 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2306 gen_reg_rtx if d->testing_p.
2307 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2308 if d->testing_p and we will certainly return true.
2309 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2312 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2314 * emit-rtl.c (gen_reg_rtx): Assert that
2315 crtl->emit.regno_pointer_align_length is non-zero.
2317 2014-02-20 Richard Henderson <rth@redhat.com>
2320 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2321 on failure the store back into EXPECT.
2323 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2324 Sandra Loosemore <sandra@codesourcery.com>
2326 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2327 * config/nios2/nios2.c (nios2_function_profiler): Add
2328 -fPIC (flag_pic == 2) support.
2329 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2330 (nios2_large_offset_p): New function.
2331 (nios2_unspec_reloc_p): Move up position, update to use
2332 nios2_large_offset_p.
2333 (nios2_unspec_address): Remove function.
2334 (nios2_unspec_offset): New function.
2335 (nios2_large_got_address): New function.
2336 (nios2_got_address): Add large offset support.
2337 (nios2_legitimize_tls_address): Update usage of removed and new
2339 (nios2_symbol_binds_local_p): New function.
2340 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2341 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2342 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2343 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2344 processing for (const (unspec ...)).
2345 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2347 2014-02-20 Richard Biener <rguenther@suse.de>
2349 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2350 doing the substitution.
2351 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2353 2014-02-20 Martin Jambor <mjambor@suse.cz>
2356 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2357 info when checking whether lattices are bottom.
2359 2014-02-20 Richard Biener <rguenther@suse.de>
2362 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2365 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2368 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2369 parameter specifying the scaling.
2370 (inline_call): Update.
2371 (want_inline_recursively): Guard division by zero.
2372 (recursive_inlining): Update.
2373 * ipa-inline.h (clone_inlined_nodes): Update.
2375 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2378 * config/i386/i386.c (classify_argument): Pass structures of size
2379 64 bytes or less in register.
2381 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2382 Kirill Yukhin <kirill.yukhin@intel.com>
2384 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2385 (_mm_rcp28_round_ss): Ditto.
2386 (_mm_rsqrt28_round_sd): Ditto.
2387 (_mm_rsqrt28_round_ss): Ditto.
2388 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2389 (_mm_rcp14_round_ss): Ditto.
2390 (_mm_rsqrt14_round_sd): Ditto.
2391 (_mm_rsqrt14_round_ss): Ditto.
2392 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2393 the first input operand, get rid of match_dup.
2394 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2396 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2398 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2399 operand as the first input operand, set type attribute.
2400 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2402 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2403 operand as the first input operand, set type attribute.
2405 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2407 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2410 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2413 * config/i386/i386.c (construct_container): Remove TFmode check
2414 for X86_64_INTEGER_CLASS.
2416 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2419 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2420 only when -Wpsabi is enabled.
2422 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2425 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2426 passing arrays in registers are the same as for structs, so remove the
2427 special case for them.
2429 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2431 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2432 destination type, extract only the valid bits if the source type is not
2433 integral and has a different mode.
2435 2014-02-19 Richard Biener <rguenther@suse.de>
2438 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2441 2014-02-19 Richard Biener <rguenther@suse.de>
2444 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2445 (ipa_modify_call_arguments): Emit an argument load explicitely and
2446 preserve virtual SSA form there and for the replacement call.
2447 Do not update SSA form nor free dominance info.
2449 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2451 * ipa.c (function_and_variable_visibility): Also clear WEAK
2452 flag when disolving COMDAT_GROUP.
2454 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2456 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2457 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2459 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2460 do more sanity checks.
2461 (detect_type_change): Return true when giving up early.
2462 (compute_complex_assign_jump_func): Fix type parameter of
2463 ipa_set_ancestor_jf.
2464 (compute_complex_ancestor_jump_func): Likewise.
2465 (update_jump_functions_after_inlining): Fix updating of
2467 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2469 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2471 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2472 inline clones when edge disappears.
2474 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2477 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2478 Split 64-bit moves into 2 patterns. Do not allow the use of
2479 direct move for TDmode in little endian, since the decimal value
2480 has little endian bytes within a word, but the 64-bit pieces are
2481 ordered in a big endian fashion, and normal subreg's of TDmode are
2483 (mov<mode>_64bit_dm): Likewise.
2484 (movtd_64bit_nodm): Likewise.
2486 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2488 PR tree-optimization/60174
2489 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2490 statement of an SSA_NAME that occurs in an abnormal PHI node.
2492 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2495 * final.c (SEEN_BB): Remove.
2496 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2497 (final_scan_insn): Don't force_source_line on second
2498 NOTE_INSN_BASIC_BLOCK.
2500 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2503 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2504 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2505 (type_natural_mode): Warn ABI change when %zmm register is not
2506 available for AVX512F vector value passing.
2508 2014-02-18 Kai Tietz <ktietz@redhat.com>
2511 * config/i386/i386.c (ix86_expand_prologue): Use value in
2512 rax register as displacement when restoring %r10 or %rax.
2513 Fix wrong offset when restoring both registers.
2515 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2517 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2518 assertion with conditional return.
2520 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2521 Uros Bizjak <ubizjak@gmail.com>
2524 * config/i386/driver-i386.c (host_detect_local_cpu): If
2525 YMM state is not saved by the OS, also clear has_f16c. Move
2526 CPUID 0x80000001 handling before YMM state saving checking.
2528 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2530 PR rtl-optimization/58960
2531 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2532 factored out from ...
2533 (sched_init): ... here.
2534 (free_global_sched_pressure_data): New, factored out from ...
2535 (sched_finish): ... here.
2536 * sched-int.h (free_global_sched_pressure_data): Declare.
2537 * sched-rgn.c (nr_regions_initial): New static global.
2538 (haifa_find_rgns): Initialize it.
2539 (schedule_region): Disable sched-pressure for the newly
2542 2014-02-17 Richard Biener <rguenther@suse.de>
2544 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2545 release SSA defs of pattern stmts.
2547 2014-02-17 Richard Biener <rguenther@suse.de>
2549 * tree-inline.c (expand_call_inline): Release the virtual
2550 operand defined by the call we are about to inline.
2552 2014-02-17 Richard Biener <rguenther@suse.de>
2554 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2556 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2557 Ilya Tocar <ilya.tocar@intel.com>
2559 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2560 arguments order in builtin.
2561 (_mm512_permutexvar_epi64): Ditto.
2562 (_mm512_mask_permutexvar_epi64): Ditto
2563 (_mm512_maskz_permutexvar_epi32): Ditto
2564 (_mm512_permutexvar_epi32): Ditto
2565 (_mm512_mask_permutexvar_epi32): Ditto
2567 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2569 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2570 (p8_vmrgow): Likewise.
2572 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2574 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2577 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2580 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2581 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2582 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2583 using direct move instructions on ISA 2.07. Also adjust
2584 instruction length for 64-bit.
2585 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2586 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2588 2014-02-15 Alan Modra <amodra@gmail.com>
2592 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2593 find_replacement on parts of insn rtl that might be reloaded.
2595 2014-02-15 Richard Biener <rguenther@suse.de>
2597 PR tree-optimization/60183
2598 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2599 (tree_ssa_phiprop): Calculate and free post-dominators.
2601 2014-02-14 Jeff Law <law@redhat.com>
2603 PR rtl-optimization/60131
2604 * ree.c (get_extended_src_reg): New function.
2605 (combine_reaching_defs): Use it rather than assuming location of REG.
2606 (find_and_remove_re): Verify first operand of extension is
2607 a REG before adding the insns to the copy list.
2609 2014-02-14 Roland McGrath <mcgrathr@google.com>
2611 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2612 * configure: Regenerated.
2613 * config.in: Regenerated.
2614 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2615 instead of ASM_SHORT.
2617 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2618 Richard Earnshaw <rearnsha@arm.com>
2620 PR rtl-optimization/59535
2621 * lra-constraints.c (process_alt_operands): Encourage alternative
2622 when unassigned pseudo class is superset of the alternative class.
2623 (inherit_reload_reg): Don't inherit when optimizing for code size.
2624 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2625 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2626 modes not less than 4 for Thumb1.
2628 2014-02-14 Kyle McMartin <kyle@redhat.com>
2631 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2633 2014-02-14 Richard Biener <rguenther@suse.de>
2635 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2636 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2637 Do not strip INDIRECT_REFs.
2639 2014-02-14 Richard Biener <rguenther@suse.de>
2642 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2643 DECL_FUNCTION_SPECIFIC_TARGET.
2644 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2645 * tree-streamer-out.c (pack_ts_target_option): Remove.
2646 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2647 (write_ts_function_decl_tree_pointers): Do not stream
2648 DECL_FUNCTION_SPECIFIC_TARGET.
2649 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2650 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2651 (lto_input_ts_function_decl_tree_pointers): Do not stream
2652 DECL_FUNCTION_SPECIFIC_TARGET.
2654 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2656 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2657 (get_initial_def_for_induction, vectorizable_induction): Ignore
2658 debug stmts when looking for exit_phi.
2659 (vectorizable_live_operation): Fix up condition.
2661 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2663 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2664 nreverse() because it changes the content of original tree list.
2666 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2668 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2669 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2671 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2673 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2674 GNU coding standards.
2676 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2679 * dwarf2out.c (gen_subprogram_die): Don't call
2680 add_calling_convention_attribute if subr_die is old_die.
2682 2014-02-13 Sharad Singhai <singhai@google.com>
2684 * doc/optinfo.texi: Fix order of nodes.
2686 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2688 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2689 operands[2], not operands[3].
2691 2014-02-13 Richard Biener <rguenther@suse.de>
2694 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2695 mention the possibility of an in-tree build.
2696 (CLooG): Update recommended version to 0.18.1, mention the
2697 possibility of an in-tree build and clarify that the ISL
2698 bundled with CLooG does not work.
2700 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2703 * expr.c (compress_float_constant): If x is a hard register,
2704 extend into a pseudo and then move to x.
2706 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2708 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2709 caused by bad second argument to warning_at() with -mhotpatch and
2710 nested functions (e.g. with gfortran).
2712 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2714 * opts.c (option_name): Remove "enabled by default" rider.
2716 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2718 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2720 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2721 Uros Bizjak <ubizjak@gmail.com>
2724 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2725 * configure: Regenerated.
2727 2014-02-12 Richard Biener <rguenther@suse.de>
2729 * vec.c (vec_prefix::calculate_allocation): Move as
2730 inline variant to vec.h.
2731 (vec_prefix::calculate_allocation_1): New out-of-line version.
2732 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2733 (vec_prefix::m_has_auto_buf): Rename to ...
2734 (vec_prefix::m_using_auto_storage): ... this.
2735 (vec_prefix::calculate_allocation): Inline the easy cases
2736 and dispatch to calculate_allocation_1 which doesn't need the
2738 (va_heap::reserve): Use gcc_checking_assert.
2739 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2740 m_using_auto_storage.
2741 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2743 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2744 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2745 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2747 2014-02-12 Richard Biener <rguenther@suse.de>
2749 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2750 when we found a dependence.
2752 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2754 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2756 (maybe_fold_stmt): ... into this new function.
2757 * omp-low.c (lower_omp): Update comment.
2759 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2762 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2765 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2767 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2768 identifiers in comments.
2769 (cortexa53_extra_costs): Likewise.
2770 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2771 (cortexa7_extra_costs): Likewise.
2772 (cortexa12_extra_costs): Likewise.
2773 (cortexa15_extra_costs): Likewise.
2774 (v7m_extra_costs): Likewise.
2776 2014-02-12 Richard Biener <rguenther@suse.de>
2779 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2780 of posix_memalign being successful.
2781 (lower_stmt): Restrict lowering of posix_memalign to when
2782 -ftree-bit-ccp is enabled.
2784 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2786 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2788 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2790 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2792 PR rtl-optimization/60116
2793 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2794 other_insn once the combination has been validated.
2796 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2799 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2801 * ipa-devirt.c: Include demangle.h
2802 (odr_violation_reported): New static variable.
2803 (add_type_duplicate): Update odr_violations.
2804 (maybe_record_node): Add completep parameter; update it.
2805 (record_target_from_binfo): Add COMPLETEP parameter;
2806 update it as needed.
2807 (possible_polymorphic_call_targets_1): Likewise.
2808 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2809 rename FINAL to COMPLETE.
2810 (record_targets_from_bases): Sanity check we found the binfo;
2811 fix COMPLETEP updating.
2812 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2813 parameter, fix computing of COMPLETEP.
2814 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2815 at LTO time do demangling.
2816 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2817 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2819 (gimple_get_virt_method_for_binfo): Likewise.
2820 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2821 gimple_get_virt_method_for_vtable): Update prototypes.
2823 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2826 * genautomata.c (add_presence_absence): Fix typo with
2827 {final_}presence_list.
2829 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2832 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2833 for VSX/Altivec vectors that land in GPR registers.
2835 2014-02-11 Richard Henderson <rth@redhat.com>
2836 Jakub Jelinek <jakub@redhat.com>
2839 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2840 around drhs if type conversion to lacc->type is not useless.
2842 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2844 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2846 (cortex-a57.cortex-a53): Likewise.
2847 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2849 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2851 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2854 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2856 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2857 add_options_for_arm_vfp3.
2859 2014-02-11 Jeff Law <law@redhat.com>
2862 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2863 object with an undesirable mode.
2865 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2868 * config/i386/sol2-9.h: New file.
2869 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2870 *-*-solaris2.9*): Use it.
2872 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2874 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2875 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2877 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2879 * config/microblaze/microblaze.c: Extend mcpu version format
2881 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2883 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2885 2014-02-10 Richard Henderson <rth@redhat.com>
2888 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2889 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2890 ms-abi vs -mno-accumulate-outgoing-args.
2891 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2892 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2895 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2898 * cfgexpand.c (expand_asm_operands): Attach source location to
2899 ASM_INPUT rtx objects.
2900 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2902 2014-02-10 Nick Clifton <nickc@redhat.com>
2904 * config/mn10300/mn10300.c (popcount): New function.
2905 (mn10300_expand_prologue): Include saved registers in stack usage
2908 2014-02-10 Jeff Law <law@redhat.com>
2911 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2912 when changing the SET_DEST of a prior insn to avoid an input reload.
2914 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2916 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2917 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2918 -mcall-openbsd, or -mcall-linux.
2919 (CC1_ENDIAN_BIG_SPEC): Remove.
2920 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2921 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2922 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2923 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2924 and %cc1_endian_default.
2925 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2927 2014-02-10 Richard Biener <rguenther@suse.de>
2929 PR tree-optimization/60115
2930 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2931 MEM_REF handling. Properly verify that the accesses are not
2932 out of the objects bound.
2934 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2936 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2939 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2941 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2942 proper constants and fix formatting.
2943 (possible_polymorphic_call_targets): Fix formatting.
2945 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2946 Ilya Tocar <ilya.tocar@intel.com>
2948 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2949 (_mm512_loadu_epi32): Renamed into...
2950 (_mm512_loadu_si512): This.
2951 (_mm512_storeu_epi32): Renamed into...
2952 (_mm512_storeu_si512): This.
2953 (_mm512_maskz_ceil_ps): Removed.
2954 (_mm512_maskz_ceil_pd): Ditto.
2955 (_mm512_maskz_floor_ps): Ditto.
2956 (_mm512_maskz_floor_pd): Ditto.
2957 (_mm512_floor_round_ps): Ditto.
2958 (_mm512_floor_round_pd): Ditto.
2959 (_mm512_ceil_round_ps): Ditto.
2960 (_mm512_ceil_round_pd): Ditto.
2961 (_mm512_mask_floor_round_ps): Ditto.
2962 (_mm512_mask_floor_round_pd): Ditto.
2963 (_mm512_mask_ceil_round_ps): Ditto.
2964 (_mm512_mask_ceil_round_pd): Ditto.
2965 (_mm512_maskz_floor_round_ps): Ditto.
2966 (_mm512_maskz_floor_round_pd): Ditto.
2967 (_mm512_maskz_ceil_round_ps): Ditto.
2968 (_mm512_maskz_ceil_round_pd): Ditto.
2969 (_mm512_expand_pd): Ditto.
2970 (_mm512_expand_ps): Ditto.
2971 * config/i386/i386.c (ix86_builtins): Remove
2972 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2973 (bdesc_args): Ditto.
2974 * config/i386/predicates.md (const1256_operand): New.
2975 (const_1_to_2_operand): Ditto.
2976 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2977 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2978 (*avx512pf_gatherpf<mode>sf): Ditto.
2979 (avx512pf_gatherpf<mode>df): Ditto.
2980 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2981 (*avx512pf_gatherpf<mode>df): Ditto.
2982 (avx512pf_scatterpf<mode>sf): Ditto.
2983 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2984 (*avx512pf_scatterpf<mode>sf): Ditto.
2985 (avx512pf_scatterpf<mode>df): Ditto.
2986 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2987 (*avx512pf_scatterpf<mode>df): Ditto.
2988 (avx512f_expand<mode>): Removed.
2989 (<shift_insn><mode>3<mask_name>): Change predicate type.
2991 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2993 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2994 not at the end of datarefs vector use ordered_remove to avoid
2995 reordering datarefs vector.
2998 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2999 mark local addressable non-static vars as GOVD_PRIVATE
3000 instead of GOVD_LOCAL.
3001 * omp-low.c (lower_omp_for): Move gimple_bind_vars
3002 and BLOCK_VARS of gimple_bind_block to new_stmt rather
3006 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
3007 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
3008 assume_aligned or alloc_align attributes.
3009 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3010 arguments. Handle also assume_aligned and alloc_align attributes.
3011 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3012 calls to functions with assume_aligned or alloc_align attributes.
3013 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3015 2014-02-08 Terry Guo <terry.guo@arm.com>
3017 * doc/invoke.texi: Document ARM -march=armv7e-m.
3019 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3021 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3022 flag on __cilkrts_rethrow builtin.
3025 * ipa-cp.c (determine_versionability): Fail at -O0
3026 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3027 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3030 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3033 * tree-inline.c (copy_forbidden): Fail for
3034 __attribute__((optimize (0))) functions.
3036 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3038 * varpool.c: Include pointer-set.h.
3039 (varpool_remove_unreferenced_decls): Variables in other partitions
3040 will not be output; be however careful to not lose information
3043 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3045 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3046 lookup in the vtable constructor.
3048 2014-02-07 Jeff Law <law@redhat.com>
3051 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3052 define_insn_and_split.
3054 * ipa-inline.c (inline_small_functions): Fix typos.
3056 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3058 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3059 (s390_can_use_return_insn): Declare.
3060 * config/s390/s390.h (EPILOGUE_USES): Define.
3061 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3063 (s390_chunkify_start): Handle return JUMP_LABELs.
3064 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3065 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3066 (s390_can_use_return_insn): New functions.
3067 (s390_fix_long_loop_prediction): Handle conditional returns.
3068 (TARGET_SET_UP_BY_PROLOGUE): Define.
3069 * config/s390/s390.md (ANY_RETURN): New code iterator.
3070 (*creturn, *csimple_return, return, simple_return): New patterns.
3072 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3074 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3075 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3076 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3077 REG_CFA_RESTORE list when deciding not to restore a register.
3079 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3081 * config/s390/s390.c: Include tree-pass.h and context.h.
3082 (s390_early_mach): New function, split out from...
3083 (s390_emit_prologue): ...here.
3084 (pass_data_s390_early_mach): New pass structure.
3085 (pass_s390_early_mach): New class.
3086 (s390_option_override): Create and register early_mach pass.
3087 Move to end of file.
3089 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3091 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3092 to match for the exit block.
3094 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3096 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3097 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3098 Reject misaligned operands.
3100 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3102 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3104 2014-02-07 Richard Biener <rguenther@suse.de>
3107 * gimple-low.c (lower_builtin_posix_memalign): New function.
3108 (lower_stmt): Call it to lower posix_memalign in a way
3109 to make alignment info accessible.
3111 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3114 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3115 __builtin_setjmp_receiver.
3117 2014-02-07 Richard Biener <rguenther@suse.de>
3120 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3121 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3122 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3123 Handle BUILT_IN_POSIX_MEMALIGN.
3124 (find_func_clobbers): Likewise.
3125 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3126 (call_may_clobber_ref_p_1): Likewise.
3128 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3131 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3134 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3137 * lto-cgraph.c (lto_output_node): Use
3138 symtab_get_symbol_partitioning_class.
3139 (lto_output_varpool_node): likewise.
3140 (symtab_get_symbol_partitioning_class): Move here from
3142 * cgraph.h (symbol_partitioning_class): Likewise.
3143 (symtab_get_symbol_partitioning_class): Declare.
3145 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3147 * ggc.h (ggc_internal_cleared_alloc): New macro.
3148 * vec.h (vec_safe_copy): Handle memory stats.
3149 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3150 * target-globals.c (save_target_globals): Likewise.
3152 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3155 * expr.c (emit_move_resolve_push): Export; be bit more selective
3156 on when to clear alias set.
3157 * expr.h (emit_move_resolve_push): Declare.
3158 * function.h (struct function): Add tail_call_marked.
3159 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3160 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3161 * config/i386/i386.md (TImode move expander): De not call
3163 (FP push expanders): Preserve memory attributes.
3164 * config/i386/sse.md (push<mode>1): Remove.
3165 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3166 (ix86_expand_push): Remove.
3167 * config/i386/mmx.md (push<mode>1): Remove.
3169 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3171 PR rtl-optimization/60030
3172 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3173 lopart with paradoxical subreg before shifting it up by hprec.
3175 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3177 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3178 Remove extra newline at end of file.
3179 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3180 (arm_issue_rate): Handle cortexa57.
3181 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3182 (cortex-a57.cortex-a53): Likewise.
3184 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3187 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3188 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3190 (arm_expand_prologue): Adjust all callers.
3191 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3192 info, registers also at the lowest numbered registers side. Use
3193 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3197 * var-tracking.c (adjust_mems): Before adding a SET to
3198 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3200 2014-02-06 Alan Modra <amodra@gmail.com>
3203 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3204 change SDmode to DDmode when lra_in_progress.
3206 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3209 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3210 free_data_ref on the dr first, and before goto again also set dr
3211 to the next dr. For simd_lane_access, free old datarefs[i] before
3212 overwriting it. For get_vectype_for_scalar_type failure, don't
3213 free_data_ref if simd_lane_access.
3215 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3218 * tree.h (opts_for_fn): New inline function.
3219 (opt_for_fn): Define.
3220 * config/i386/i386.c (ix86_function_regparm): Use
3221 opt_for_fn (decl, optimize) instead of optimize.
3223 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3225 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3226 for SYMBOL_REF in large memory model.
3228 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3230 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3232 (cortex-a57): Likewise.
3233 (cortex-a57.cortex-a53): Likewise.
3235 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3236 Kugan Vivekanandarajah <kuganv@linaro.org>
3238 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3240 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3242 2014-02-06 Richard Biener <rguenther@suse.de>
3244 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3245 set_loop_copy and initialize_original_copy_tables.
3247 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3249 * config/aarch64/aarch64-simd.md
3250 (aarch64_ashr_simddi): Change QI to SI.
3252 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3253 Jakub Jelinek <jakub@redhat.com>
3256 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3259 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3261 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3262 CODE_FOR_altivec_vpku[hw]um to
3263 CODE_FOR_altivec_vpku[hw]um_direct.
3264 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3265 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3266 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3267 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3269 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3271 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3272 generation for -maltivec=be.
3273 (altivec_vsumsws): Simplify redundant test.
3275 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3277 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3278 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3279 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3280 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3281 gen_altivec_vpkuwum.
3282 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3284 (altivec_vpks<VI_char>ss): Likewise.
3285 (altivec_vpks<VI_char>us): Likewise.
3286 (altivec_vpku<VI_char>us): Likewise.
3287 (altivec_vpku<VI_char>um): Likewise.
3288 (altivec_vpku<VI_char>um_direct): New (copy of
3289 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3291 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3292 target is little endian and -maltivec=be is not specified.
3293 (*altivec_vupkhs<VU_char>_direct): New (copy of
3294 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3295 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3296 target is little endian and -maltivec=be is not specified.
3297 (*altivec_vupkls<VU_char>_direct): New (copy of
3298 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3299 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3300 little endian and -maltivec=be is not specified.
3301 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3302 little endian and -maltivec=be is not specified.
3304 2014-02-05 Richard Henderson <rth@redhat.com>
3307 * combine-stack-adj.c: Revert r206943.
3308 * sched-int.h (struct deps_desc): Add last_args_size.
3309 * sched-deps.c (init_deps): Initialize it.
3310 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3311 contain REG_ARGS_SIZE notes.
3313 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3315 * lto-cgraph.c (asm_nodes_output): Make global.
3316 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3317 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3318 (driver_handle_option): Handle OPT_fwpa.
3320 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3323 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3324 a comment typo and formatting issue. If odr_hash hasn't been
3325 created, return vNULL and set *completep to false.
3328 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3329 bb with no successors.
3331 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3334 * doc/invoke.texi (-march): Clarify documentation for ARM.
3338 2014-02-05 Richard Biener <rguenther@suse.de>
3340 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3341 when not vectorizing because of too many alias checks.
3342 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3343 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3345 2014-02-05 Nick Clifton <nickc@redhat.com>
3347 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3348 accept extended registers in any mode when compiling for the MN10300.
3350 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3352 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3353 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3354 sanitization attributes.
3355 (can_inline_edge_p): Likewise.
3356 (sanitize_attrs_match_for_inline_p): New function.
3358 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3360 * ipa-prop.c (detect_type_change): Shor circuit testing of
3361 type changes on THIS pointer.
3363 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3366 * config/pa/pa.c (legitimize_tls_address): Return original address
3367 if not passed a SYMBOL_REF rtx.
3368 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3370 (pa_emit_move_sequence): Simplify TLS source operands.
3371 (pa_legitimate_constant_p): Reject all TLS constants.
3372 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3373 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3375 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3377 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3378 groups when we know they are controlled by LTO.
3379 * varasm.c (default_binds_local_p_1): If object is in other partition,
3380 it will be resolved locally.
3382 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3384 * config/host-linux.c (linux_gt_pch_use_address): Don't
3385 use SSIZE_MAX because it is not always defined.
3387 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3390 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3391 threshold for pseudo splitting.
3392 (update_ebb_live_info): Process call argument hard registers and
3393 hard registers from insn definition too.
3394 (max_small_class_regs_num): New constant.
3395 (inherit_in_ebb): Update live hard regs through EBBs. Update
3396 reloads_num only for small register classes. Don't split for
3399 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3402 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3405 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3407 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3410 2014-02-04 Marek Polacek <polacek@redhat.com>
3412 * gdbinit.in (pel): Define.
3414 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3416 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3419 2014-02-04 Richard Biener <rguenther@suse.de>
3422 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3423 in function context local.
3424 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3425 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3426 similar to LTO_imported_decl_ref.
3428 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3430 PR tree-optimization/60002
3431 * cgraphclones.c (build_function_decl_skip_args): Clear
3434 PR tree-optimization/60023
3435 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3436 false to gsi_replace.
3437 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3438 has been in some EH region and vec_stmt could throw, add
3439 vec_stmt into the same EH region.
3440 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3441 has no lhs, ignore it.
3442 * internal-fn.c (expand_MASK_LOAD): Likewise.
3445 * tree-inline.c (copy_forbidden): Fail for
3446 __attribute__((optimize (0))) functions.
3449 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3450 is set, copy one less argument.
3451 (expand_simd_clones): Don't subtract clone_info->inbranch
3452 from simd_clone_struct_alloc argument.
3454 PR rtl-optimization/57915
3455 * recog.c (simplify_while_replacing): If all unary/binary/relational
3456 operation arguments are constant, attempt to simplify those.
3459 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3460 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3462 2014-02-04 Richard Biener <rguenther@suse.de>
3464 PR tree-optimization/60012
3465 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3466 TBAA disambiguation to all DDRs.
3468 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3471 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3472 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3474 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3477 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3479 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3481 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3482 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3484 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3487 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3488 to figure out targets of polymorphic calls with known decl.
3489 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3490 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3491 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3492 (get_polymorphic_call_info): ... here.
3493 (get_polymorphic_call_info_from_invariant): New function.
3495 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3497 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3498 lookup via vtable pointer; check for type consistency
3499 and turn inconsitent facts into UNREACHABLE.
3500 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3501 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3502 type inconsistent querries; return UNREACHABLE instead.
3504 2014-02-03 Richard Henderson <rth@twiddle.net>
3507 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3508 already processed this node.
3509 (normalize_one_pred_1): Pass along mark_set.
3510 (normalize_one_pred): Create and destroy a pointer_set_t.
3511 (normalize_one_pred_chain): Likewise.
3513 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3515 PR gcov-profile/58602
3516 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3518 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3521 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3522 -fno-devirtualize; try to devirtualize by the knowledge of
3523 virtual table pointer given by aggregate propagation.
3524 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3525 (ipa_print_node_jump_functions): Dump also offset that
3526 is relevant for polymorphic calls.
3527 (determine_known_aggregate_parts): Add arg_type parameter; use it
3528 instead of determining the type from pointer type.
3529 (ipa_compute_jump_functions_for_edge): Update call of
3530 determine_known_aggregate_parts.
3531 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3532 (gimple_get_virt_method_for_binfo): ... here; simplify using
3533 vtable_pointer_value_to_vtable.
3534 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3535 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3536 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3537 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3539 (vtable_pointer_value_to_binfo): ... here.
3540 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3542 2014-02-03 Teresa Johnson <tejohnson@google.com>
3544 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3545 redef of outer loop index variable.
3547 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3551 * doc/extend.texi (Function Attributes): Typo.
3553 2014-02-03 Cong Hou <congh@google.com>
3555 PR tree-optimization/60000
3556 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3557 if the vectorized statement is a store. A store statement can only
3558 appear at the end of pattern statements.
3560 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3562 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3563 (ix86_option_override_internal): Default long double to 64-bit for
3564 32-bit Bionic and to 128-bit for 64-bit Bionic.
3566 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3567 TARGET_LONG_DOUBLE_128 is true.
3568 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3570 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3571 (mlong-double-64): Negate -mlong-double-128.
3572 (mlong-double-128): New option.
3574 * config/i386/i386-c.c (ix86_target_macros): Define
3575 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3577 * doc/invoke.texi: Document -mlong-double-128.
3579 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3581 PR rtl-optimization/60024
3582 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3584 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3586 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3588 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3590 PR rtl-optimization/57662
3591 * sel-sched.c (code_motion_path_driver): Do not mark already not
3592 existing blocks in the visiting bitmap.
3594 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3596 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3597 on the insn being emitted.
3599 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3600 Will Deacon <will.deacon@arm.com>
3602 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3604 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3606 * config/arm/arm-tables.opt: Regenerate.
3608 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3610 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3611 for vector types other than V16QImode.
3612 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3613 define_expand, and call altivec_expand_vec_perm_le when producing
3614 code with little endian element order.
3615 (*altivec_vperm_<mode>_internal): New insn having previous
3616 behavior of altivec_vperm_<mode>.
3617 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3618 altivec_expand_vec_perm_le when producing code with little endian
3620 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3621 behavior of altivec_vperm_<mode>_uns.
3623 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3625 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3626 (altivec_vsumsws): Add handling for -maltivec=be with a little
3628 (altivec_vsumsws_direct): New.
3629 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3630 gen_altivec_vsumsws.
3632 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3634 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3635 vtable_pointer_value_to_binfo): New functions.
3636 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3637 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3639 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3641 * config/nios2/nios2.md (load_got_register): Initialize GOT
3642 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3643 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3645 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3647 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3648 preserverd by passthrough, do not propagate the type.
3650 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3652 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3653 (mips_atomic_assign_expand_fenv): New function.
3654 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3656 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3658 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3659 (__builtin_mips_set_fcsr): Likewise.
3660 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3661 MIPS_USI_FTYPE_VOID.
3662 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3663 (mips16_expand_set_fcsr): Likewise.
3664 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3665 (mips16_set_fcsr_stub): Likewise.
3666 (mips16_get_fcsr_one_only_stub): New class.
3667 (mips16_set_fcsr_one_only_stub): Likewise.
3668 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3669 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3670 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3671 (hard_float): New availability predicate.
3672 (mips_builtins): Add get_fcsr and set_fcsr.
3673 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3674 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3675 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3676 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3677 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3680 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3682 * config/mips/mips.c (mips_one_only_stub): New class.
3683 (mips_need_mips16_rdhwr_p): Replace with...
3684 (mips16_rdhwr_stub): ...this new variable.
3685 (mips16_stub_call_address): New function.
3686 (mips16_rdhwr_one_only_stub): New class.
3687 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3688 (mips_output_mips16_rdhwr): Delete.
3689 (mips_finish_stub): New function.
3690 (mips_code_end): Use it to handle rdhwr stubs.
3692 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3695 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3696 when calculating size of integer atomic types.
3698 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3700 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3702 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3704 PR tree-optimization/60003
3705 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3706 * profile.c (branch_prob): Use gimple_call_builtin_p
3707 to check for BUILT_IN_SETJMP_RECEIVER.
3708 * tree-inline.c (copy_bb): Call notice_special_calls.
3710 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3713 * lra-constraints.c (process_alt_operands): Update reload_sum only
3716 2014-01-31 Richard Henderson <rth@redhat.com>
3719 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3720 until after else_eh is processed.
3722 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3724 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3725 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3726 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3727 in smmintrin.h, remove them.
3728 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3729 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3730 * config/i386/i386.md (ROUND_SAE): Fix value.
3731 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3732 (const48_operand): New.
3733 * config/i386/subst.md (round), (round_expand): Use
3734 const_4_or_8_to_11_operand.
3735 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3737 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3739 * config/i386/constraints.md (Yk): Swap meaning with k.
3740 * config/i386/i386.md (movhi_internal): Change Yk to k.
3741 (movqi_internal): Ditto.
3742 (*k<logic><mode>): Ditto.
3745 (kandn<mode>): Ditto.
3746 (*<code>hi_1): Ditto.
3747 (*<code>qi_1): Ditto.
3748 (kxnor<mode>): Ditto.
3749 (kortestzhi): Ditto.
3750 (kortestchi): Ditto.
3752 (*one_cmplhi2_1): Ditto.
3753 (*one_cmplqi2_1): Ditto.
3754 * config/i386/sse.md (): Change k to Yk.
3755 (avx512f_load<mode>_mask): Ditto.
3756 (avx512f_blendm<mode>): Ditto.
3757 (avx512f_store<mode>_mask): Ditto.
3758 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3759 (avx512f_storedqu<mode>_mask): Ditto.
3760 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3762 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3763 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3764 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3765 (avx512f_maskcmp<mode>3): Ditto.
3766 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3767 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3768 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3769 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3770 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3771 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3772 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3773 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3774 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3775 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3776 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3777 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3778 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3779 (vec_extract_lo_<mode>_maskm): Ditto.
3780 (vec_extract_hi_<mode>_maskm): Ditto.
3781 (avx512f_vternlog<mode>_mask): Ditto.
3782 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3783 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3784 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3785 (avx512f_<code>v8div16qi2_mask): Ditto.
3786 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3787 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3788 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3789 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3790 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3791 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3792 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3793 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3794 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3795 (avx512cd_maskb_vec_dupv8di): Ditto.
3796 (avx512cd_maskw_vec_dupv16si): Ditto.
3797 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3798 (avx512f_vpermi2var<mode>3_mask): Ditto.
3799 (avx512f_vpermi2var<mode>3_mask): Ditto.
3800 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3801 (*avx512f_gathersi<mode>): Ditto.
3802 (*avx512f_gathersi<mode>_2): Ditto.
3803 (*avx512f_gatherdi<mode>): Ditto.
3804 (*avx512f_gatherdi<mode>_2): Ditto.
3805 (*avx512f_scattersi<mode>): Ditto.
3806 (*avx512f_scatterdi<mode>): Ditto.
3807 (avx512f_compress<mode>_mask): Ditto.
3808 (avx512f_compressstore<mode>_mask): Ditto.
3809 (avx512f_expand<mode>_mask): Ditto.
3810 * config/i386/subst.md (mask): Change k to Yk.
3811 (mask_scalar_merge): Ditto.
3814 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3816 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3818 2014-01-31 Richard Biener <rguenther@suse.de>
3821 * builtins.c (fold_builtin_memory_op): Make sure to not
3822 use a floating-point mode or a boolean or enumeral type for
3825 2014-01-30 DJ Delorie <dj@redhat.com>
3827 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3828 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3829 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3830 whenever main() has an epilogue.
3832 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3834 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3835 unused variable "field".
3836 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3837 (vsx_mergeh_<mode>): Likewise.
3838 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3839 (altivec_vmrghh): Likewise.
3840 (altivec_vmrghw): Likewise.
3841 (altivec_vmrglb): Likewise.
3842 (altivec_vmrglh): Likewise.
3843 (altivec_vmrglw): Likewise.
3844 (altivec_vspltb): Add missing uses.
3845 (altivec_vsplth): Likewise.
3846 (altivec_vspltw): Likewise.
3847 (altivec_vspltsf): Likewise.
3849 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3852 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3853 frame related instructions.
3855 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3857 PR rtl-optimization/59959
3858 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3859 any reload of register whose subreg is invalid.
3861 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3863 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3864 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3865 Add missing return type - void.
3867 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3869 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3870 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3871 remove element index adjustment for endian (now handled in vsx.md
3873 (altivec_expand_vec_perm_const): Use
3874 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3875 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3876 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3877 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3878 define_expand and a new define_insn *altivec_vspltb_internal;
3879 adjust for -maltivec=be on a little endian target.
3880 (altivec_vspltb_direct): New.
3881 (altivec_vsplth): Divide into a define_expand and a new
3882 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3883 little endian target.
3884 (altivec_vsplth_direct): New.
3885 (altivec_vspltw): Divide into a define_expand and a new
3886 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3887 little endian target.
3888 (altivec_vspltw_direct): New.
3889 (altivec_vspltsf): Divide into a define_expand and a new
3890 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3891 a little endian target.
3893 2014-01-30 Richard Biener <rguenther@suse.de>
3895 PR tree-optimization/59993
3896 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3897 can propagate form the earlier stmt and avoid the transform
3898 when the intermediate result is needed.
3900 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3902 * README.Portability: Fix typo.
3904 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3906 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3907 comparison_operator with ordered_comparison_operator.
3909 2014-01-30 Nick Clifton <nickc@redhat.com>
3911 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3912 Rename to mn10300_store_multiple_regs.
3913 * config/mn10300/mn10300.c: Likewise.
3914 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3915 store_multiple_regs.
3916 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3917 Call mn10300_store_multiple_regs.
3919 2014-01-30 Nick Clifton <nickc@redhat.com>
3920 DJ Delorie <dj@redhat.com>
3922 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3923 %fp 2 to keep registers after it properly word-aligned.
3924 (rl78_alloc_physical_registers_umul): Handle the case where both
3925 input operands are the same.
3927 2014-01-30 Richard Biener <rguenther@suse.de>
3929 PR tree-optimization/59903
3930 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3933 2014-01-30 Jason Merrill <jason@redhat.com>
3936 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3939 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3941 2014-01-30 Richard Biener <rguenther@suse.de>
3943 PR tree-optimization/59951
3944 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3946 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3949 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3950 SFmode to DFmode case.
3952 2014-01-29 DJ Delorie <dj@redhat.com>
3954 * config/msp430/msp430.opt (-minrt): New.
3955 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3957 (ENDFILE_SPEC): Likewise.
3959 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3961 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3962 (estimate_function_body_sizes): Use it.
3964 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3967 * dwarf2out.c (is_cxx_auto): New.
3968 (is_base_type): Use it.
3969 (gen_type_die_with_usage): Likewise.
3971 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3973 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3974 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3975 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3976 -maltivec=be with LE targets.
3977 (vsx_mergeh_<mode>): Likewise.
3978 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3979 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3980 (altivec_vmrghb): Replace with define_expand and new
3981 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3982 (altivec_vmrghb_direct): New define_insn.
3983 (altivec_vmrghh): Replace with define_expand and new
3984 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3985 (altivec_vmrghh_direct): New define_insn.
3986 (altivec_vmrghw): Replace with define_expand and new
3987 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3988 (altivec_vmrghw_direct): New define_insn.
3989 (*altivec_vmrghsf): Adjust for endianness.
3990 (altivec_vmrglb): Replace with define_expand and new
3991 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3992 (altivec_vmrglb_direct): New define_insn.
3993 (altivec_vmrglh): Replace with define_expand and new
3994 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3995 (altivec_vmrglh_direct): New define_insn.
3996 (altivec_vmrglw): Replace with define_expand and new
3997 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3998 (altivec_vmrglw_direct): New define_insn.
3999 (*altivec_vmrglsf): Adjust for endianness.
4000 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4001 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4002 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
4003 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
4004 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4005 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4006 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
4007 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4009 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4011 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4012 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4015 2014-01-29 Richard Biener <rguenther@suse.de>
4017 PR tree-optimization/58742
4018 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4019 associate_pointerplus_align.
4020 (associate_pointerplus_diff): New function.
4021 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4022 and associate_pointerplus_diff.
4024 2014-01-29 Richard Biener <rguenther@suse.de>
4026 * lto-streamer.h (LTO_major_version): Bump to 3.
4027 (LTO_minor_version): Reset to 0.
4029 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4031 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4032 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4033 (arm_file_start): Generate correct asm header for armv7ve.
4034 * config/arm/bpabi.h: Add multilib support for armv7ve.
4035 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4036 and cortex-a15 to armv7ve.
4037 * config/arm/t-aprofile: Add multilib support for armv7ve.
4038 * doc/invoke.texi: Document -march=armv7ve.
4040 2014-01-29 Richard Biener <rguenther@suse.de>
4042 PR tree-optimization/58742
4043 * tree-ssa-forwprop.c (associate_plusminus): Return true
4044 if we changed sth, defer EH cleanup to ...
4045 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4046 (simplify_mult): New function.
4048 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4051 PR tree-optimization/59920
4052 * tree.c (build_common_builtin_nodes): Remove
4053 __builtin_setjmp_dispatcher initialization.
4054 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4055 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4056 instead of gsi_after_labels + manually skipping debug stmts.
4057 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4058 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4059 * tree-inline.c (copy_edges_for_bb): Remove
4060 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4061 argument. Ignore computed_goto_p stmts. Don't call
4062 make_abnormal_goto_edges. If a call might need abnormal edges
4063 for non-local gotos, see if it already has an edge to
4064 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4065 with true argument, don't do anything then, otherwise add
4066 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4067 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4069 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4070 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4071 (lower_stmt): Don't set data->calls_builtin_setjmp.
4072 (lower_builtin_setjmp): Adjust comment.
4073 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4074 * tree-cfg.c (found_computed_goto): Remove.
4075 (factor_computed_gotos): Remove.
4076 (make_goto_expr_edges): Return bool, true for computed gotos.
4077 Don't call make_abnormal_goto_edges.
4078 (build_gimple_cfg): Don't set found_computed_goto, don't call
4079 factor_computed_gotos.
4080 (computed_goto_p): No longer static.
4081 (make_blocks): Don't set found_computed_goto.
4082 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4083 (make_edges): If make_goto_expr_edges returns true, push bb
4084 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4085 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4086 vector. Record mapping between bbs and OpenMP regions if there
4087 are any, adjust make_gimple_omp_edges caller. Call
4088 handle_abnormal_edges.
4089 (make_abnormal_goto_edges): Remove.
4090 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4091 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4092 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4093 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4094 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4095 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4096 filling *region also set *region_idx to (*region)->entry->index.
4099 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4100 For REGs set ORIGINAL_REGNO.
4102 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4104 * doc/md.texi: Mention that a target shouldn't implement
4105 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4108 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4110 PR tree-optimization/59594
4111 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4112 a copy of the datarefs vector rather than the vector itself.
4114 2014-01-28 Jason Merrill <jason@redhat.com>
4117 * dwarf2out.c (auto_die): New static.
4118 (gen_type_die_with_usage): Handle C++1y 'auto'.
4119 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4122 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4125 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4126 (SPEC_X32): Likewise.
4127 (SPEC_64): Likewise.
4128 * config/i386/i386.c (ix86_option_override_internal): Turn off
4129 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4131 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4132 * config/i386/i386.h (TARGET_16BIT): New macro.
4133 (TARGET_16BIT_P): Likewise.
4134 * config/i386/i386.opt: Add m16.
4135 * doc/invoke.texi: Document -m16.
4137 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4139 PR preprocessor/59935
4140 * input.c (location_get_source_line): Bail out on when line number
4141 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4143 2014-01-28 Richard Biener <rguenther@suse.de>
4145 PR tree-optimization/58742
4146 * tree-ssa-forwprop.c (associate_plusminus): Handle
4147 pointer subtraction of the form (T)(P + A) - (T)P.
4149 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4151 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4154 2014-01-28 Richard Biener <rguenther@suse.de>
4157 2014-01-28 Richard Biener <rguenther@suse.de>
4159 PR rtl-optimization/45364
4160 PR rtl-optimization/59890
4161 * var-tracking.c (local_get_addr_clear_given_value): Handle
4162 already cleared slot.
4163 (val_reset): Handle not allocated local_get_addr_cache.
4164 (vt_find_locations): Use post-order on the inverted CFG.
4166 2014-01-28 Richard Biener <rguenther@suse.de>
4168 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4170 2014-01-28 Richard Biener <rguenther@suse.de>
4172 PR rtl-optimization/45364
4173 PR rtl-optimization/59890
4174 * var-tracking.c (local_get_addr_clear_given_value): Handle
4175 already cleared slot.
4176 (val_reset): Handle not allocated local_get_addr_cache.
4177 (vt_find_locations): Use post-order on the inverted CFG.
4179 2014-01-28 Alan Modra <amodra@gmail.com>
4181 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4182 * configure.ac <recursive call for build != host>: Define
4183 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4184 and LD_FOR_BUILD too.
4185 * configure: Regenerate.
4187 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4189 * config/i386/i386.c (get_builtin_code_for_version): Separate
4190 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4191 Broadwell from Haswell.
4193 2014-01-27 Steve Ellcey <sellcey@mips.com>
4195 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4196 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4197 * config/mips/mips.c (mips_option_override): Change setting
4199 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4200 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4201 Change from Mask to Var.
4203 2014-01-27 Jeff Law <law@redhat.com>
4205 * ipa-inline.c (inline_small_functions): Fix typo.
4207 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4209 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4210 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4211 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4212 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4213 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4214 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4215 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4216 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4217 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4218 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4219 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4220 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4221 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4222 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4223 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4224 (_mm512_storeu_epi64): Ditto.
4225 (_mm512_cmpge_epi32_mask): Ditto.
4226 (_mm512_cmpge_epu32_mask): Ditto.
4227 (_mm512_cmpge_epi64_mask): Ditto.
4228 (_mm512_cmpge_epu64_mask): Ditto.
4229 (_mm512_cmple_epi32_mask): Ditto.
4230 (_mm512_cmple_epu32_mask): Ditto.
4231 (_mm512_cmple_epi64_mask): Ditto.
4232 (_mm512_cmple_epu64_mask): Ditto.
4233 (_mm512_cmplt_epi32_mask): Ditto.
4234 (_mm512_cmplt_epu32_mask): Ditto.
4235 (_mm512_cmplt_epi64_mask): Ditto.
4236 (_mm512_cmplt_epu64_mask): Ditto.
4237 (_mm512_cmpneq_epi32_mask): Ditto.
4238 (_mm512_cmpneq_epu32_mask): Ditto.
4239 (_mm512_cmpneq_epi64_mask): Ditto.
4240 (_mm512_cmpneq_epu64_mask): Ditto.
4241 (_mm512_expand_pd): Ditto.
4242 (_mm512_expand_ps): Ditto.
4243 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4244 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4245 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4246 * config/i386/i386.c (ix86_builtins): Add
4247 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4248 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4249 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4250 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4251 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4252 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4253 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4254 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4255 IX86_BUILTIN_PMOVUSQW512_MEM.
4256 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4257 __builtin_ia32_pmovsqd512mem_mask,
4258 __builtin_ia32_pmovqd512mem_mask,
4259 __builtin_ia32_pmovusqw512mem_mask,
4260 __builtin_ia32_pmovsqw512mem_mask,
4261 __builtin_ia32_pmovqw512mem_mask,
4262 __builtin_ia32_pmovusdw512mem_mask,
4263 __builtin_ia32_pmovsdw512mem_mask,
4264 __builtin_ia32_pmovdw512mem_mask,
4265 __builtin_ia32_pmovqb512mem_mask,
4266 __builtin_ia32_pmovusqb512mem_mask,
4267 __builtin_ia32_pmovsqb512mem_mask,
4268 __builtin_ia32_pmovusdb512mem_mask,
4269 __builtin_ia32_pmovsdb512mem_mask,
4270 __builtin_ia32_pmovdb512mem_mask.
4271 (bdesc_args): Add __builtin_ia32_expanddf512,
4272 __builtin_ia32_expandsf512.
4273 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4274 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4275 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4276 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4277 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4278 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4279 (avx512f_<code>v8div16qi2_mask_store): This.
4280 (avx512f_expand<mode>): New.
4282 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4284 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4286 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4287 (_mm512_prefetch_i32scatter_pd): Ditto.
4288 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4289 (_mm512_prefetch_i64scatter_pd): Ditto.
4290 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4291 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4292 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4293 (_mm512_prefetch_i32scatter_ps): Ditto.
4294 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4295 (_mm512_prefetch_i64scatter_ps): Ditto.
4296 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4297 * config/i386/i386-builtin-types.def: Define
4298 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4299 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4300 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4301 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4302 IX86_BUILTIN_SCATTERPFQPD.
4303 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4304 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4305 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4306 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4307 __builtin_ia32_scatterpfqps.
4308 (ix86_expand_builtin): Expand new built-ins.
4309 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4310 fix memory access data type.
4311 (*avx512pf_gatherpf<mode>_mask): Ditto.
4312 (*avx512pf_gatherpf<mode>): Ditto.
4313 (avx512pf_scatterpf<mode>): Ditto.
4314 (*avx512pf_scatterpf<mode>_mask): Ditto.
4315 (*avx512pf_scatterpf<mode>): Ditto.
4316 (GATHER_SCATTER_SF_MEM_MODE): New.
4317 (avx512pf_gatherpf<mode>df): Ditto.
4318 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4319 (*avx512pf_scatterpf<mode>df): Ditto.
4321 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4324 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4325 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4328 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4330 * common/config/arm/arm-common.c
4331 (arm_rewrite_mcpu): Handle multiple names.
4333 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4335 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4337 * gimple-builder.h (create_gimple_tmp): Delete.
4339 2014-01-27 Christian Bruel <christian.bruel@st.com>
4341 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4344 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4346 * config/pa/pa.md (call): Generate indirect long calls to non-local
4347 functions when outputing 32-bit code.
4348 (call_value): Likewise except for special call to buggy powf function.
4350 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4351 portable runtime and PIC indirect calls.
4352 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4353 and PIC call sequences. Use ldo instead of blr to set return register
4354 in PIC call sequence.
4356 2014-01-25 Walter Lee <walt@tilera.com>
4358 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4359 avoid clobbering a live register.
4361 2014-01-25 Walter Lee <walt@tilera.com>
4363 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4364 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4365 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4366 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4368 2014-01-25 Walter Lee <walt@tilera.com>
4370 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4371 arguments on even registers.
4372 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4374 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4375 (BIGGEST_ALIGNMENT): Ditto.
4376 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4378 2014-01-25 Walter Lee <walt@tilera.com>
4380 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4381 insns before bundling.
4382 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4384 2014-01-25 Walter Lee <walt@tilera.com>
4386 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4387 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4388 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4390 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4392 * config/mips/constraints.md (kl): Delete.
4393 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4394 define expands, using...
4395 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4396 instructions for MIPS16.
4397 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4398 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4400 2014-01-25 Walter Lee <walt@tilera.com>
4402 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4406 2014-01-25 Walter Lee <walt@tilera.com>
4408 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4409 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4411 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4413 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4416 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4418 * print-rtl.c (in_call_function_usage): New var.
4419 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4420 EXPR_LIST mode as mode and not as reg note name.
4423 * cfgloopmanip.c (copy_loop_info): If
4424 loop->warned_aggressive_loop_optimizations, make sure
4425 the flag is set in target loop too.
4427 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4429 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4431 * builtins.def: Likewise.
4432 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4433 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4434 * ira.c (ira_setup_eliminable_regset): Likewise.
4435 * omp-low.c (gate_expand_omp): Likewise.
4436 (execute_lower_omp): Likewise.
4437 (diagnose_sb_0): Likewise.
4438 (gate_diagnose_omp_blocks): Likewise.
4439 (simd_clone_clauses_extract): Likewise.
4442 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4444 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4445 correction for little endian...
4446 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4449 2014-01-24 Jeff Law <law@redhat.com>
4451 PR tree-optimization/59919
4452 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4453 for non-returning calls.
4455 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4457 * common/config/aarch64/aarch64-common.c
4458 (aarch64_rewrite_mcpu): Handle multiple names.
4459 * config/aarch64/aarch64.h
4460 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4462 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4464 * input.c (add_file_to_cache_tab): Handle the case where fopen
4467 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4470 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4471 from push operand if code of push isn't PRE_DEC.
4473 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4476 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4477 -mquad-memory-atomic. Update -mquad-memory documentation to say
4478 it is only used for non-atomic loads/stores.
4480 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4481 -mquad-memory or -mquad-memory-atomic switches.
4483 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4484 -mquad-memory-atomic to ISA 2.07 support.
4486 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4487 to separate support of normal quad word memory operations (ldq, stq)
4488 from the atomic quad word memory operations.
4490 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4491 support to separate non-atomic quad word operations from atomic
4492 quad word operations. Disable non-atomic quad word operations in
4493 little endian mode so that we don't have to swap words after the
4494 load and before the store.
4495 (quad_load_store_p): Add comment about atomic quad word support.
4496 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4497 options printed with -mdebug=reg.
4499 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4500 -mquad-memory-atomic as the test for whether we have quad word
4501 atomic instructions.
4502 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4503 or -mp8-vector are used, allow byte/half-word atomic operations.
4505 * config/rs6000/sync.md (load_lockedti): Insure that the address
4506 is a proper indexed or indirect address for the lqarx instruction.
4507 On little endian systems, swap the hi/lo registers after the lqarx
4509 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4510 insure the address is valid for the lqarx instruction.
4511 (store_conditionalti): Insure that the address is a proper indexed
4512 or indirect address for the stqcrx. instruction. On little endian
4513 systems, swap the hi/lo registers before doing the stqcrx.
4515 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4516 insure the address is valid for the stqcrx. instruction.
4518 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4519 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4520 type of quad memory support is available.
4522 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4525 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4526 there is a danger of looping.
4528 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4530 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4531 force flag_ira_loop_pressure if set via command line.
4533 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4535 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4536 (ashr_simd): New builtin handling DI mode.
4537 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4538 (aarch64_sshr_simddi): New match pattern.
4539 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4540 (vshrd_n_s64): Likewise.
4541 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4543 2014-01-23 Nick Clifton <nickc@redhat.com>
4545 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4546 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4547 favour of mcu specific scripts.
4548 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4551 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4552 Alex Velenko <Alex.Velenko@arm.com>
4554 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4555 (vaddv_s16): Likewise.
4556 (vaddv_s32): Likewise.
4557 (vaddv_u8): Likewise.
4558 (vaddv_u16): Likewise.
4559 (vaddv_u32): Likewise.
4560 (vaddvq_s8): Likewise.
4561 (vaddvq_s16): Likewise.
4562 (vaddvq_s32): Likewise.
4563 (vaddvq_s64): Likewise.
4564 (vaddvq_u8): Likewise.
4565 (vaddvq_u16): Likewise.
4566 (vaddvq_u32): Likewise.
4567 (vaddvq_u64): Likewise.
4568 (vaddv_f32): Likewise.
4569 (vaddvq_f32): Likewise.
4570 (vaddvq_f64): Likewise.
4571 (vmaxv_f32): Likewise.
4572 (vmaxv_s8): Likewise.
4573 (vmaxv_s16): Likewise.
4574 (vmaxv_s32): Likewise.
4575 (vmaxv_u8): Likewise.
4576 (vmaxv_u16): Likewise.
4577 (vmaxv_u32): Likewise.
4578 (vmaxvq_f32): Likewise.
4579 (vmaxvq_f64): Likewise.
4580 (vmaxvq_s8): Likewise.
4581 (vmaxvq_s16): Likewise.
4582 (vmaxvq_s32): Likewise.
4583 (vmaxvq_u8): Likewise.
4584 (vmaxvq_u16): Likewise.
4585 (vmaxvq_u32): Likewise.
4586 (vmaxnmv_f32): Likewise.
4587 (vmaxnmvq_f32): Likewise.
4588 (vmaxnmvq_f64): Likewise.
4589 (vminv_f32): Likewise.
4590 (vminv_s8): Likewise.
4591 (vminv_s16): Likewise.
4592 (vminv_s32): Likewise.
4593 (vminv_u8): Likewise.
4594 (vminv_u16): Likewise.
4595 (vminv_u32): Likewise.
4596 (vminvq_f32): Likewise.
4597 (vminvq_f64): Likewise.
4598 (vminvq_s8): Likewise.
4599 (vminvq_s16): Likewise.
4600 (vminvq_s32): Likewise.
4601 (vminvq_u8): Likewise.
4602 (vminvq_u16): Likewise.
4603 (vminvq_u32): Likewise.
4604 (vminnmv_f32): Likewise.
4605 (vminnmvq_f32): Likewise.
4606 (vminnmvq_f64): Likewise.
4608 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4610 * config/aarch64/aarch64-simd.md
4611 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4612 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4613 (*aarch64_mul3_elt<mode>): Likewise.
4614 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4615 (*aarch64_mul3_elt_to_64v2df): Likewise.
4616 (*aarch64_mla_elt<mode>): Likewise.
4617 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4618 (*aarch64_mls_elt<mode>): Likewise.
4619 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4620 (*aarch64_fma4_elt<mode>): Likewise.
4621 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4622 (*aarch64_fma4_elt_to_64v2df): Likewise.
4623 (*aarch64_fnma4_elt<mode>): Likewise.
4624 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4625 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4626 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4627 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4628 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4629 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4630 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4631 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4632 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4634 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4636 * config/aarch64/aarch64-simd.md
4637 (aarch64_be_checked_get_lane<mode>): New define_expand.
4638 * config/aarch64/aarch64-simd-builtins.def
4639 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4640 New builtin definition.
4641 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4642 Use new safe be builtin.
4644 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4646 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4648 (aarch64_be_st1<mode>): Likewise.
4649 (aarch_ld1<VALL:mode>): Define_expand modified.
4650 (aarch_st1<VALL:mode>): Likewise.
4651 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4652 (UNSPEC_ST1): Likewise.
4654 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4656 * config/microblaze/microblaze.md: Add trap insn and attribute
4658 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4660 PR preprocessor/58580
4661 * input.h (location_get_source_line): Take an additional line_size
4663 (void diagnostics_file_cache_fini): Declare new function.
4664 * input.c (struct fcache): New type.
4665 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4666 New static constants.
4667 (diagnostic_file_cache_init, total_lines_num)
4668 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4669 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4670 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4671 (get_next_line, read_next_line, goto_next_line, read_line_num):
4672 New static function definitions.
4673 (diagnostic_file_cache_fini): New function.
4674 (location_get_source_line): Take an additional output line_len
4675 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4677 * diagnostic.c (diagnostic_finish): Call
4678 diagnostic_file_cache_fini.
4679 (adjust_line): Take an additional input parameter for the length
4680 of the line, rather than calculating it with strlen.
4681 (diagnostic_show_locus): Adjust the use of
4682 location_get_source_line and adjust_line with respect to their new
4683 signature. While displaying a line now, do not stop at the first
4684 null byte. Rather, display the zero byte as a space and keep
4685 going until we reach the size of the line.
4686 * Makefile.in: Add vec.o to OBJS-libcommon
4688 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4689 Ilya Tocar <ilya.tocar@intel.com>
4691 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4692 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4693 (__builtin_ia32_kmov16): Ditto.
4694 * config/i386/i386.md (UNSPEC_KMOV): New.
4697 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4699 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4700 (_mm512_storeu_si512): Ditto.
4702 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4705 * rtl.h (get_referenced_operands): Declare.
4706 * recog.c (get_referenced_operands): New function.
4707 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4708 operands have been referenced when recording LO_SUM references.
4710 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4712 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4714 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4716 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4717 Enable for generic and recent AMD targets.
4719 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4721 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4722 ARG_SIZE note when adjustment was eliminated.
4724 2014-01-22 Jeff Law <law@redhat.com>
4726 PR tree-optimization/59597
4727 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4728 in file. Accept new argument REGISTERING and use it to modify
4729 dump output appropriately.
4730 (register_jump_thread): Corresponding changes.
4731 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4732 thread paths involving joiner blocks. Add code to dump cancelled
4733 jump threading paths.
4735 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4737 PR rtl-optimization/59477
4738 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4739 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4741 2014-01-22 Tom Tromey <tromey@redhat.com>
4743 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4745 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4747 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4749 PR rtl-optimization/59896
4750 * lra-constraints.c (process_alt_operands): Check unused note for
4751 matched operands of insn with no output reloads.
4753 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4755 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4756 (mips_move_from_gpr_cost): Likewise.
4758 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4760 PR rtl-optimization/59858
4761 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4762 ira_class_hard_regs_num.
4763 (process_alt_operands): Increase reject for dying matched operand.
4765 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4768 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4769 smaller than size, perform several stores or loads and stores
4770 at dst + count - size to store or copy all of size bytes, rather
4771 than just last modesize bytes.
4773 2014-01-20 DJ Delorie <dj@redhat.com>
4775 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4776 that CLOBBERs are REGs before propogating their values.
4778 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4781 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4782 (cgraph_inline_failed_type): New function.
4783 * cgraph.h (DEFCIFCODE): Add type.
4784 (cgraph_inline_failed_type_t): New enum.
4785 (cgraph_inline_failed_type): New prototype.
4786 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4787 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4788 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4789 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4790 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4791 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4792 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4793 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4794 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4795 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4796 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4797 OPTIMIZATION_MISMATCH.
4798 * tree-inline.c (expand_call_inline): Emit errors during
4799 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4801 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4804 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4805 mode attribute in insn output.
4807 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4809 * output.h (output_constant): Delete.
4810 * varasm.c (output_constant): Make private.
4812 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4814 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4816 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4819 * tree.h (fold_builtin_strcat): New prototype.
4820 * builtins.c (fold_builtin_strcat): No longer static. Add len
4821 argument, if non-NULL, don't call c_strlen. Optimize
4822 directly into __builtin_memcpy instead of __builtin_strcpy.
4823 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4824 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4826 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4828 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4829 for SImode_address_operand operands, having only a REG argument.
4831 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4833 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4834 loader name using mbig-endian.
4835 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4837 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4839 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4843 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4845 * config/aarch64/aarch64-protos.h
4846 (aarch64_cannot_change_mode_class_ptr): Declare.
4847 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4848 aarch64_cannot_change_mode_class_ptr): New.
4849 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4850 backend hook aarch64_cannot_change_mode_class.
4852 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4854 * common/config/aarch64/aarch64-common.c
4855 (aarch64_handle_option): Don't handle any option order logic here.
4856 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4857 selected_cpu, warn on architecture version mismatch.
4858 (aarch64_override_options): Fix parsing order for option strings.
4860 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4861 Iain Sandoe <iain@codesourcery.com>
4864 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4865 warning. Amend comment to reflect current functionality.
4867 2014-01-20 Richard Biener <rguenther@suse.de>
4870 * builtins.c (fold_builtin_strcat): Remove case better handled
4871 by tree-ssa-strlen.c.
4873 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4875 * config/aarch64/aarch64.opt
4876 (mcpu, march, mtune): Make case-insensitive.
4878 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4881 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4882 if operands[1] is a REG or ZERO_EXTEND of a REG.
4884 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4886 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4888 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4890 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4891 long non-pic millicode calls.
4893 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4895 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4897 2014-01-19 Kito Cheng <kito@0xlab.org>
4899 * builtins.c (expand_movstr): Check movstr expand done or fail.
4901 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4902 H.J. Lu <hongjiu.lu@intel.com>
4905 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4906 to DImode for zero-extended addresses.
4908 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4910 PR rtl-optimization/57763
4911 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4912 on the new indirect jump_insn and increment LABEL_NUSES (label).
4914 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4918 * config.gcc (x86_archs): New variable.
4919 (x86_64_archs): Likewise.
4920 (x86_cpus): Likewise.
4921 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4922 --with-arch/--with-cpu= options.
4923 Support --with-arch=/--with-cpu={nehalem,westmere,
4924 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4926 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4928 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4929 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4931 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4933 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4935 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4938 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4939 clear cpp_get_options (parse_in)->warn_unused_macros for
4940 ix86_target_macros_internal with cpp_define.
4942 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4944 * jump.c (delete_related_insns): Keep (use (insn))s.
4945 * reorg.c (redundant_insn): Check for barriers too.
4947 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4949 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4951 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4953 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4954 call to $$dyncall when TARGET_LONG_CALLS is true.
4956 2014-01-17 Jeff Law <law@redhat.com>
4958 * ree.c (combine_set_extension): Temporarily disable test for
4959 changing number of hard registers.
4961 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4964 * ipa-inline-analysis.c (inline_free_summary):
4965 Do not free summary of aliases.
4967 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4970 * gimplify.c (gimplify_expr): Use create_tmp_var
4971 instead of create_tmp_var_raw. If cond doesn't have
4972 integral type, don't add the IFN_ANNOTATE builtin at all.
4974 2014-01-17 Martin Jambor <mjambor@suse.cz>
4977 * ipa-cp.c (prev_edge_clone): New variable.
4978 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4979 Also resize prev_edge_clone vector.
4980 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4981 (ipcp_edge_removal_hook): New function.
4982 (ipcp_driver): Register ipcp_edge_removal_hook.
4984 2014-01-17 Andrew Pinski <apinski@cavium.com>
4985 Steve Ellcey <sellcey@mips.com>
4988 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4991 2014-01-17 Jeff Law <law@redhat.com>
4994 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4995 so that pass_ccp runs first.
4997 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4999 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
5000 (ix86_adjust_cost): Use !TARGET_XXX.
5001 (do_reorder_for_imul): Likewise.
5002 (swap_top_of_ready_list): Likewise.
5003 (ix86_sched_reorder): Likewise.
5005 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
5007 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5008 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5009 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5010 (intel_memset): New. Duplicate slm_memset.
5011 (intel_cost): New. Duplicate slm_cost.
5012 (m_INTEL): New macro.
5013 (processor_target_table): Add "intel".
5014 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5015 with PROCESSOR_INTEL for "intel".
5016 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5017 PROCESSOR_SILVERMONT.
5018 (ix86_issue_rate): Likewise.
5019 (ix86_adjust_cost): Likewise.
5020 (ia32_multipass_dfa_lookahead): Likewise.
5021 (swap_top_of_ready_list): Likewise.
5022 (ix86_sched_reorder): Likewise.
5023 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5024 instead of TARGET_OPT_AGU.
5025 * config/i386/i386.h (TARGET_INTEL): New.
5026 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5027 (processor_type): Add PROCESSOR_INTEL.
5028 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5029 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5031 2014-01-17 Marek Polacek <polacek@redhat.com>
5034 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5037 2014-01-17 Richard Biener <rguenther@suse.de>
5039 PR tree-optimization/46590
5040 * opts.c (default_options_table): Add entries for
5041 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5042 all enabled at -O1 but not for -Og.
5043 * common.opt (fbranch-count-reg): Remove Init(1).
5044 (fmove-loop-invariants): Likewise.
5045 (ftree-pta): Likewise.
5047 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5049 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5050 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5051 decls to at least the GCC 4.8 used alignments.
5054 * tree-nested.c (convert_nonlocal_reference_stmt,
5055 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5056 of GIMPLE_BIND stmts, adjust associated decls.
5058 2014-01-17 Richard Biener <rguenther@suse.de>
5060 PR tree-optimization/46590
5061 * vec.h (vec<>::bseach): New member function implementing
5062 binary search according to C89 bsearch.
5063 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5064 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5065 bitmap pointer again. Make accesses_in_loop a flat array.
5066 (mem_ref_obstack): New global.
5067 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5068 (mark_ref_stored): Likewise.
5069 (ref_indep_loop_p_2): Likewise.
5070 (set_ref_stored_in_loop): New helper function.
5071 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5072 (memref_free): Adjust.
5073 (record_mem_ref_loc): Simplify.
5074 (gather_mem_refs_stmt): Adjust.
5075 (sort_locs_in_loop_postorder_cmp): New function.
5076 (analyze_memory_references): Sort accesses_in_loop after
5077 loop postorder number.
5078 (find_ref_loc_in_loop_cmp): New function.
5079 (for_all_locs_in_loop): Find relevant cluster of locs in
5080 accesses_in_loop and iterate without recursion.
5081 (execute_sm): Avoid uninit warning.
5082 (struct ref_always_accessed): Simplify.
5083 (ref_always_accessed::operator ()): Likewise.
5084 (ref_always_accessed_p): Likewise.
5085 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5086 loop postorder numbers here.
5087 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5090 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5093 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5094 on decls for which assemble_alias has been called.
5096 2014-01-17 Nick Clifton <nickc@redhat.com>
5098 * config/msp430/msp430.opt: (mcpu): New option.
5099 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5100 (msp430_option_override): Parse target_cpu. If the MCU name
5101 matches a generic string, clear target_mcu.
5102 (msp430_attr): Allow numeric interrupt values up to 63.
5103 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5104 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5106 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5108 * config/msp430/msp430.md (popm): Use %J rather than %I.
5109 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5110 (addhi_cy_i): Use immediate_operand for operand 2.
5111 * doc/invoke.texi: Document -mcpu option.
5113 2014-01-17 Richard Biener <rguenther@suse.de>
5115 PR rtl-optimization/38518
5116 * df.h (df_analyze_loop): Declare.
5117 * df-core.c: Include cfgloop.h.
5118 (df_analyze_1): Split out main part of df_analyze.
5119 (df_analyze): Adjust.
5120 (loop_inverted_post_order_compute): New function.
5121 (loop_post_order_compute): Likewise.
5122 (df_analyze_loop): New function avoiding whole-function
5124 * loop-invariant.c (find_defs): Use df_analyze_loop.
5125 (find_invariants): Adjust.
5126 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5128 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5130 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5131 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5133 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5135 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5136 traversal when removing references.
5138 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5141 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5143 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5146 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5147 pushing a reload for an autoinc when we had previously reloaded an
5148 inner part of the address.
5150 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5152 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5154 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5155 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5156 when not giving up or versioning for alias only because of
5158 (vect_analyze_data_ref_dependences): Set to true.
5159 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5161 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5162 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5166 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5169 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5170 operand 0 predicate for gathers, use a new pseudo as subtarget.
5172 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5175 * lra-constraints.c (process_alt_operands): Add printing debug info.
5176 Check absence of input/output reloads for matched operands too.
5178 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5180 PR rtl-optimization/59835
5181 * ira.c (ira_init_register_move_cost): Increase cost for
5184 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5186 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5188 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5191 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5192 non-register objects. Use gen_(high/low)part more consistently.
5195 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5198 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5199 endian support, remove tests for WORDS_BIG_ENDIAN.
5200 (p8_mfvsrd_3_<mode>): Likewise.
5201 (reload_gpr_from_vsx<mode>): Likewise.
5202 (reload_gpr_from_vsxsf): Likewise.
5203 (p8_mfvsrd_4_disf): Likewise.
5205 2014-01-16 Richard Biener <rguenther@suse.de>
5207 PR rtl-optimization/46590
5208 * lcm.c (compute_antinout_edge): Use postorder iteration.
5209 (compute_laterin): Use inverted postorder iteration.
5211 2014-01-16 Nick Clifton <nickc@redhat.com>
5214 * varasm.c (output_constant): Return the number of bytes actually
5216 (output_constructor_array_range): Update the field size with the
5217 number of bytes emitted by output_constant.
5218 (output_constructor_regular_field): Likewise. Also do not
5219 complain if the total number of bytes emitted is now greater
5220 than the expected fieldpos.
5221 * output.h (output_constant): Update prototype and descriptive comment.
5223 2014-01-16 Marek Polacek <polacek@redhat.com>
5226 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5227 it is error_mark_node.
5229 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5231 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5232 VALID_AVX256_REG_OR_OI_MODE.
5234 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5236 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5237 current procedure should be profiled.
5239 2014-01-15 Andrew Pinski <apinski@cavium.com>
5241 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5242 of moving from/to the STACK_REG register class.
5244 2014-01-15 Richard Henderson <rth@redhat.com>
5247 * reginfo.c (global_regs_decl): Globalize.
5248 * rtl.h (global_regs_decl): Declare.
5249 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5250 reserved via global_regs.
5252 2014-01-15 Teresa Johnson <tejohnson@google.com>
5254 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5256 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5258 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5259 and vmulosh rather than call gen_vec_widen_smult_*.
5260 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5261 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5262 (vec_widen_smult_even_v16qi): Likewise.
5263 (vec_widen_umult_even_v8hi): Likewise.
5264 (vec_widen_smult_even_v8hi): Likewise.
5265 (vec_widen_umult_odd_v16qi): Likewise.
5266 (vec_widen_smult_odd_v16qi): Likewise.
5267 (vec_widen_umult_odd_v8hi): Likewise.
5268 (vec_widen_smult_odd_v8hi): Likewise.
5269 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5270 vmuloub rather than call gen_vec_widen_umult_*.
5271 (vec_widen_umult_lo_v16qi): Likewise.
5272 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5273 vmulosb rather than call gen_vec_widen_smult_*.
5274 (vec_widen_smult_lo_v16qi): Likewise.
5275 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5276 rather than call gen_vec_widen_umult_*.
5277 (vec_widen_umult_lo_v8hi): Likewise.
5278 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5279 rather than call gen_vec_widen_smult_*.
5280 (vec_widen_smult_lo_v8hi): Likewise.
5282 2014-01-15 Jeff Law <law@redhat.com>
5284 PR tree-optimization/59747
5285 * ree.c (find_and_remove_re): Properly handle case where a second
5286 eliminated extension requires widening a copy created for elimination
5287 of a prior extension.
5288 (combine_set_extension): Ensure that the number of hard regs needed
5289 for a destination register does not change when we widen it.
5291 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5293 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5294 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5295 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5296 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5297 (avr-*-rtems*): Likewise.
5298 (bfin*-rtems*): Likewise.
5299 (moxie-*-rtems*): Likewise.
5300 (h8300-*-rtems*): Likewise.
5301 (i[34567]86-*-rtems*): Likewise.
5302 (lm32-*-rtems*): Likewise.
5303 (m32r-*-rtems*): Likewise.
5304 (m68k-*-rtems*): Likewise.
5305 (microblaze*-*-rtems*): Likewise.
5306 (mips*-*-rtems*): Likewise.
5307 (powerpc-*-rtems*): Likewise.
5308 (sh-*-rtems*): Likewise.
5309 (sparc-*-rtems*): Likewise.
5310 (sparc64-*-rtems*): Likewise.
5311 (v850-*-rtems*): Likewise.
5312 (m32c-*-rtems*): Likewise.
5314 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5316 PR rtl-optimization/59511
5317 * ira.c (ira_init_register_move_cost): Use memory costs for some
5318 cases of register move cost calculations.
5319 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5320 instead of BB frequency.
5321 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5322 * lra-assigns.c (find_hard_regno_for): Ditto.
5324 2014-01-15 Richard Biener <rguenther@suse.de>
5326 PR tree-optimization/59822
5327 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5328 (vectorizable_load): Use it to hoist defs of uses of invariant
5329 loads out of the loop.
5331 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5332 Kugan Vivekanandarajah <kuganv@linaro.org>
5335 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5338 2014-01-15 Richard Biener <rguenther@suse.de>
5340 PR rtl-optimization/59802
5341 * lcm.c (compute_available): Use inverted postorder to seed
5342 the initial worklist.
5344 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5347 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5348 ADDR_REGS for invalid symrefs in non-PIC code.
5350 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5353 * builtins.c (determine_block_size): Initialize *probable_max_size
5354 even if len_rtx is CONST_INT.
5356 2014-01-14 Andrew Pinski <apinski@cavium.com>
5358 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5359 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5360 (cortexa53_tunings): Likewise.
5361 (aarch64_sched_issue_rate): New function.
5362 (TARGET_SCHED_ISSUE_RATE): Define.
5364 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5366 * ira-costs.c (find_costs_and_classes): Add missed
5367 ira_init_register_move_cost_if_necessary.
5369 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5372 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5374 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5377 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5378 to indicate if type is used for function return value. Warn ABI
5379 change if the vector mode isn't available for function return value.
5380 (ix86_function_arg_advance): Pass false to type_natural_mode.
5381 (ix86_function_arg): Likewise.
5382 (ix86_gimplify_va_arg): Likewise.
5383 (function_arg_32): Don't warn ABI change.
5384 (ix86_function_value): Pass true to type_natural_mode.
5385 (ix86_return_in_memory): Likewise.
5386 (ix86_struct_value_rtx): Removed.
5387 (TARGET_STRUCT_VALUE_RTX): Likewise.
5389 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5391 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5392 converting a conditional jump into a conditional return.
5394 2014-01-14 Richard Biener <rguenther@suse.de>
5396 PR tree-optimization/58921
5397 PR tree-optimization/59006
5398 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5399 hoisting invariant stmts.
5400 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5401 invariant loads on the preheader edge if possible.
5403 2014-01-14 Joey Ye <joey.ye@arm.com>
5405 * doc/plugin.texi (Building GCC plugins): Update to C++.
5407 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5409 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5410 (_mm_rcp28_round_ss): Ditto.
5411 (_mm_rsqrt28_round_sd): Ditto.
5412 (_mm_rsqrt28_round_ss): Ditto.
5413 (_mm_rcp28_sd): Ditto.
5414 (_mm_rcp28_ss): Ditto.
5415 (_mm_rsqrt28_sd): Ditto.
5416 (_mm_rsqrt28_ss): Ditto.
5417 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5418 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5419 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5420 (IX86_BUILTIN_RCP28SD): Ditto.
5421 (IX86_BUILTIN_RCP28SS): Ditto.
5422 (IX86_BUILTIN_RSQRT28SD): Ditto.
5423 (IX86_BUILTIN_RSQRT28SS): Ditto.
5424 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5425 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5426 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5427 (ix86_expand_special_args_builtin): Expand new FTYPE.
5428 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5429 (srcp14<mode>): Make insn unary.
5430 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5431 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5432 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5433 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5434 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5435 Fix rounding: make it SAE only.
5436 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5438 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5440 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5441 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5442 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5443 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5444 (round_saeonly_mask_scalar_operand4): Ditto.
5445 (round_saeonly_mask_scalar_op3): Ditto.
5446 (round_saeonly_mask_scalar_op4): Ditto.
5448 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5450 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5451 Implement -maltivec=be for vec_insert and vec_extract.
5453 2014-01-10 DJ Delorie <dj@redhat.com>
5455 * config/msp430/msp430.md (call_internal): Don't allow memory
5456 references with SP as the base register.
5457 (call_value_internal): Likewise.
5458 * config/msp430/constraints.md (Yc): New. For memory references
5459 that don't use SP as a base register.
5461 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5462 "an integer without a # prefix"
5463 * config/msp430/msp430.md (epilogue_helper): Use it.
5465 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5468 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5469 AVX512F gather builtins.
5470 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5471 on gather decls with INTEGER_TYPE masktype.
5472 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5473 directly into the builtin rather than hoisting it before loop.
5475 PR tree-optimization/59387
5476 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5477 (scev_const_prop): If folded_casts and type has undefined overflow,
5478 use force_gimple_operand instead of force_gimple_operand_gsi and
5479 for each added stmt if it is assign with
5480 arith_code_with_undefined_signed_overflow, call
5481 rewrite_to_defined_overflow.
5482 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5483 gimple-fold.h instead.
5484 (arith_code_with_undefined_signed_overflow,
5485 rewrite_to_defined_overflow): Moved to ...
5486 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5487 rewrite_to_defined_overflow): ... here. No longer static.
5488 Include gimplify-me.h.
5489 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5490 rewrite_to_defined_overflow): New prototypes.
5492 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5494 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5496 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5498 * builtins.c (get_object_alignment_2): Minor tweak.
5499 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5501 2014-01-13 Christian Bruel <christian.bruel@st.com>
5503 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5504 optimized non constant lengths.
5506 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5509 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5510 load as __atomic_load_N if possible.
5512 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5514 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5516 (rs6000_expand_builtin): Adjust call.
5518 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5521 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5522 * config/rs6000/rs6000.c: Include target-globals.h.
5523 (rs6000_set_current_function): Instead of doing target_reinit
5524 unconditionally, use save_target_globals_default_opts and
5525 restore_target_globals.
5527 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5529 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5530 (rs6000_expand_builtin): Handle mffs and mtfsf.
5531 (rs6000_init_builtins): Define mffs and mtfsf.
5532 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5533 (rs6000_mffs): New pattern.
5534 (rs6000_mtfsf): New pattern.
5536 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5538 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5539 Start narrowing with START. Apply candidate-use pair
5540 and check overall cost in narrowing.
5541 (iv_ca_prune): Pass new argument.
5543 2014-01-10 Jeff Law <law@redhat.com>
5546 * ree.c (combine_reaching_defs): Ensure the defining statement
5547 occurs before the extension when optimizing extensions with
5548 different source and destination hard registers.
5550 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5553 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5554 vtables into the type inheritance graph.
5556 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5558 PR rtl-optimization/59754
5559 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5560 modes in the REGNO != REGNO case.
5562 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5564 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5566 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5568 PR tree-optimization/59745
5569 * tree-predcom.c (tree_predictive_commoning_loop): Call
5570 free_affine_expand_cache if giving up because components is NULL.
5572 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5573 GC in payload of target_globals struct instead of allocating them on
5574 the heap and the larger structs separately using GC.
5575 * target-globals.h (struct target_globals): Make regs, hard_regs,
5576 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5577 of GTY((skip)) and change type to void *.
5578 (reset_target_globals): Cast loads from those fields to corresponding
5581 2014-01-10 Steve Ellcey <sellcey@mips.com>
5584 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5585 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5586 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5588 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5591 * aarch64-modes.def (CC_Zmode): New flags mode.
5592 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5593 represents an equality.
5594 (aarch64_get_condition_code): Handle CC_Zmode.
5595 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5597 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5599 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5600 extraction in good case.
5602 2014-01-10 Richard Biener <rguenther@suse.de>
5604 PR tree-optimization/59374
5605 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5606 checking after SLP discovery. Mark stmts not participating
5607 in any SLP instance properly.
5609 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5611 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5612 when handling a SET rtx.
5614 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5616 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5617 (cortex-a57): Likewise.
5618 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5620 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5622 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5623 non-iwmmxt builtins.
5625 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5629 * ipa-devirt.c record_target_from_binfo): Take as argument
5630 stack of binfos and lookup matching one for virtual inheritance.
5631 (possible_polymorphic_call_targets_1): Update.
5633 2014-01-10 Huacai Chen <chenhc@lemote.com>
5635 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5636 kernel strings for Loongson-2E/2F/3A.
5638 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5641 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5642 is_gimple_call before calling gimple_call_internal_p.
5644 2014-01-09 Steve Ellcey <sellcey@mips.com>
5646 * Makefile.in (TREE_FLOW_H): Remove.
5647 (TREE_SSA_H): Add file names from tree-flow.h.
5648 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5649 * tree.h: Remove tree-flow.h reference.
5650 * hash-table.h: Remove tree-flow.h reference.
5651 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5652 reference with tree-ssa-loop.h.
5654 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5656 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5657 default element-order behavior for -maltivec.
5658 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5659 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5660 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5661 when targeting big endian, at least for now.
5662 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5664 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5667 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5668 var satisfies use_register_for_decl, just take into account type
5669 alignment, rather than decl alignment.
5671 PR tree-optimization/59622
5672 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5673 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5674 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5675 Don't devirtualize for inplace at all. For targets.length () == 1,
5676 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5678 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5680 * config/i386/i386.md (cpu): Remove the unused btver1.
5682 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5684 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5686 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5689 * tree-core.h (struct target_globals): New forward declaration.
5690 (struct tree_target_option): Add globals field.
5691 * tree.h (TREE_TARGET_GLOBALS): Define.
5692 (prepare_target_option_nodes_for_pch): New prototype.
5693 * target-globals.h (struct target_globals): Define even if
5695 * tree.c (prepare_target_option_node_for_pch,
5696 prepare_target_option_nodes_for_pch): New functions.
5697 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5698 * config/i386/i386.c: Include target-globals.h.
5699 (ix86_set_current_function): Instead of doing target_reinit
5700 unconditionally, use save_target_globals_default_opts and
5701 restore_target_globals.
5703 2014-01-09 Richard Biener <rguenther@suse.de>
5705 PR tree-optimization/59715
5706 * tree-cfg.h (split_critical_edges): Declare.
5707 * tree-cfg.c (split_critical_edges): Export.
5708 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5710 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5712 * cfgexpand.c (expand_stack_vars): Optionally disable
5713 asan stack protection.
5714 (expand_used_vars): Likewise.
5715 (partition_stack_vars): Likewise.
5716 * asan.c (asan_emit_stack_protection): Optionally disable
5717 after return stack usage.
5718 (instrument_derefs): Optionally disable memory access instrumentation.
5719 (instrument_builtin_call): Likewise.
5720 (instrument_strlen_call): Likewise.
5721 (asan_protect_global): Optionally disable global variables protection.
5722 * doc/invoke.texi: Added doc for new options.
5723 * params.def: Added new options.
5724 * params.h: Likewise.
5726 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5728 PR rtl-optimization/59724
5729 * ifcvt.c (cond_exec_process_if_block): Don't call
5730 flow_find_head_matching_sequence with 0 longest_match.
5731 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5732 non-active insns if !stop_after.
5733 (try_head_merge_bb): Revert 2014-01-07 changes.
5735 2014-01-08 Jeff Law <law@redhat.com>
5737 * ree.c (get_sub_rtx): New function, extracted from...
5738 (merge_def_and_ext): Here.
5739 (combine_reaching_defs): Use get_sub_rtx.
5741 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5743 * cgraph.h (varpool_variable_node): Do not choke on null node.
5745 2014-01-08 Catherine Moore <clm@codesourcery.com>
5747 * config/mips/mips.md (simple_return): Attempt to use JRC
5749 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5751 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5753 PR rtl-optimization/59137
5754 * reorg.c (steal_delay_list_from_target): Call update_block for
5756 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5758 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5761 two duplicate entries.
5763 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5766 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5768 * config/mips/mips.c (mips_truncated_op_cost): New function.
5769 (mips_rtx_costs): Adjust test for BADDU.
5770 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5772 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5774 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5775 (*baddu_si): ...this new pattern.
5777 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5780 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5782 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5785 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5787 (expand_expr, expand_normal): Adjust.
5788 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5789 inner_reference_p. Use inner_reference_p to expand inner references.
5790 (store_expr): Adjust.
5791 * cfgexpand.c (expand_call_stmt): Adjust.
5793 2014-01-08 Rong Xu <xur@google.com>
5795 * gcov-io.c (gcov_var): Move from gcov-io.h.
5796 (gcov_position): Ditto.
5797 (gcov_is_error): Ditto.
5798 (gcov_rewrite): Ditto.
5799 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5800 only part to libgcc/libgcov.h.
5802 2014-01-08 Marek Polacek <polacek@redhat.com>
5805 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5807 2014-01-08 Marek Polacek <polacek@redhat.com>
5810 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5812 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5814 PR rtl-optimization/59649
5815 * stor-layout.c (get_mode_bounds): For BImode return
5816 0 and STORE_FLAG_VALUE.
5818 2014-01-08 Richard Biener <rguenther@suse.de>
5821 * gimple.h (is_gimple_builtin_call): Remove.
5822 (gimple_builtin_call_types_compatible_p): New.
5823 (gimple_call_builtin_p): New overload.
5824 * gimple.c (is_gimple_builtin_call): Remove.
5825 (validate_call): Rename to ...
5826 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5828 (validate_type): New static function.
5829 (gimple_call_builtin_p): New overload and adjust.
5830 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5831 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5832 (gimple_fold_stmt_to_constant_1): Likewise.
5833 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5835 2014-01-08 Richard Biener <rguenther@suse.de>
5838 * gimplify.c (gimplify_expr): Gimplify register-register type
5839 VIEW_CONVERT_EXPRs to separate stmts.
5841 2014-01-07 Jeff Law <law@redhat.com>
5844 * ree.c (combine_set_extension): Handle case where source
5845 and destination registers in an extension insn are different.
5846 (combine_reaching_defs): Allow source and destination registers
5847 in extension to be different under limited circumstances.
5848 (add_removable_extension): Remove restriction that the
5849 source and destination registers in the extension are the same.
5850 (find_and_remove_re): Emit a copy from the extension's
5851 destination to its source after the defining insn if
5852 the source and destination registers are different.
5855 * ifcvt.c (merge_if_block): If we are merging a block with more than
5856 one successor with a block with no successors, remove any BARRIER
5857 after the second block.
5859 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5861 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5863 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5866 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5867 for 14-bit register offsets when INT14_OK_STRICT is false.
5869 2014-01-07 Roland Stigge <stigge@antcom.de>
5870 Michael Meissner <meissner@linux.vnet.ibm.com>
5873 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5874 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5876 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5878 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5881 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5883 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5884 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5885 rtx is const0_rtx or not.
5887 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5890 * target-globals.c (save_target_globals): Remove this_fn_optab
5892 * toplev.c: Include optabs.h.
5893 (target_reinit): Temporarily restore the global options if another
5894 set of options are in force.
5896 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5898 PR rtl-optimization/58668
5899 * cfgcleanup.c (flow_find_cross_jump): Don't count
5900 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5901 to determine what is counted.
5902 (flow_find_head_matching_sequence): Use active_insn_p to determine
5904 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5906 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5907 determine what is counted.
5909 PR tree-optimization/59643
5910 * tree-predcom.c (split_data_refs_to_components): If one dr is
5911 read and one write, determine_offset fails and the write isn't
5912 in the bad component, just put the read into the bad component.
5914 2014-01-07 Mike Stump <mikestump@comcast.net>
5915 Jakub Jelinek <jakub@redhat.com>
5918 * tree-core.h (struct tree_optimization_option): Change optabs
5919 type from unsigned char * to void *.
5920 * optabs.c (init_tree_optimization_optabs): Adjust
5921 TREE_OPTIMIZATION_OPTABS initialization.
5923 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5926 * config/i386/i386.h (struct machine_function): Add
5927 no_drap_save_restore field.
5928 * config/i386/i386.c (ix86_save_reg): Use
5929 !cfun->machine->no_drap_save_restore instead of
5930 crtl->stack_realign_needed.
5931 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5932 this function clears frame_pointer_needed. Set
5933 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5934 and DRAP reg is needed.
5936 2014-01-06 Marek Polacek <polacek@redhat.com>
5939 * doc/implement-c.texi: Mention that other integer types are
5940 permitted as bit-field types in strictly conforming mode.
5942 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5944 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5947 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5949 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5951 2014-01-06 Martin Jambor <mjambor@suse.cz>
5954 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5956 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5958 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5962 * var-tracking.c (add_stores): Preserve the value of the source even if
5963 we don't record the store.
5965 2014-01-06 Terry Guo <terry.guo@arm.com>
5967 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5969 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5972 * config/darwin.c (darwin_function_section): Adjust return values to
5973 correspond to optimisation changes made in r206070.
5975 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5977 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5978 from prefetch_block tune setting.
5979 (nocona_cost): Correct size of prefetch block to 64.
5981 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5983 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5984 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5985 used to save the static chain register in the computation of the offset
5986 from which the FP registers need to be restored.
5988 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5990 PR tree-optimization/59519
5991 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5992 ICE if get_current_def (current_new_name) is already non-NULL, as long
5993 as it is a phi result of some other phi in *new_exit_bb that has
5996 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5997 or vmovdqu* for misaligned_operand.
5998 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5999 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
6000 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
6001 aligned_mem for AVX512F masked aligned load and store builtins and for
6004 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
6006 PR tree-optimization/59651
6007 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
6008 Address range for negative step should be added by TYPE_SIZE_UNIT.
6010 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6012 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6013 overlapping registers also for registers other than the stack pointer.
6015 2014-01-03 Marek Polacek <polacek@redhat.com>
6018 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6021 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6024 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6027 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6028 (push splitter): Use <P:MODE_SIZE> instead of
6029 GET_MODE_SIZE (<P:MODE>mode).
6030 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6031 (mov -1, reg peephole2): Likewise.
6032 * config/i386/sse.md (*mov<mode>_internal,
6033 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6034 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6035 *<code><mode>3, *andnot<mode>3<mask_name>,
6036 <mask_codefor><code><mode>3<mask_name>): Likewise.
6037 * config/i386/subst.md (mask_mode512bit_condition,
6038 sd_mask_mode512bit_condition): Likewise.
6040 2014-01-02 Xinliang David Li <davidxl@google.com>
6042 PR tree-optimization/59303
6043 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6044 (dump_predicates): Better output format.
6045 (pred_equal_p): New function.
6046 (is_neq_relop_p): Ditto.
6047 (is_neq_zero_form_p): Ditto.
6048 (pred_expr_equal_p): Ditto.
6049 (pred_neg_p): Ditto.
6050 (simplify_pred): Ditto.
6051 (simplify_preds_2): Ditto.
6052 (simplify_preds_3): Ditto.
6053 (simplify_preds_4): Ditto.
6054 (simplify_preds): Ditto.
6056 (push_to_worklist): Ditto.
6057 (get_pred_info_from_cmp): Ditto.
6058 (is_degenerated_phi): Ditto.
6059 (normalize_one_pred_1): Ditto.
6060 (normalize_one_pred): Ditto.
6061 (normalize_one_pred_chain): Ditto.
6062 (normalize_preds): Ditto.
6063 (normalize_cond_1): Remove function.
6064 (normalize_cond): Ditto.
6065 (is_gcond_subset_of): Ditto.
6066 (is_subset_of_any): Ditto.
6067 (is_or_set_subset_of): Ditto.
6068 (is_and_set_subset_of): Ditto.
6069 (is_norm_cond_subset_of): Ditto.
6070 (pred_chain_length_cmp): Ditto.
6071 (convert_control_dep_chain_into_preds): Type change.
6072 (find_predicates): Ditto.
6073 (find_def_preds): Ditto.
6074 (destroy_predicates_vecs): Ditto.
6075 (find_matching_predicates_in_rest_chains): Ditto.
6076 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6077 (is_pred_expr_subset): Ditto.
6078 (is_pred_chain_subset_of): Ditto.
6079 (is_included_in): Ditto.
6080 (is_superset_of): Ditto.
6082 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6084 Update copyright years.
6086 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6088 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6089 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6090 config/arc/arc.md, config/arc/arc.opt,
6091 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6092 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6093 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6094 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6095 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6096 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6097 vtable-verify.c, vtable-verify.h: Use the standard form for the
6100 2014-01-02 Tobias Burnus <burnus@net-b.de>
6102 * gcc.c (process_command): Update copyright notice dates.
6103 * gcov-dump.c: Ditto.
6105 * doc/cpp.texi: Bump @copying's copyright year.
6106 * doc/cppinternals.texi: Ditto.
6107 * doc/gcc.texi: Ditto.
6108 * doc/gccint.texi: Ditto.
6109 * doc/gcov.texi: Ditto.
6110 * doc/install.texi: Ditto.
6111 * doc/invoke.texi: Ditto.
6113 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6115 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6117 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6119 * config/i386/sse.md (*mov<mode>_internal): Guard
6120 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6122 PR rtl-optimization/59647
6123 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6124 new_rtx into UNSIGNED_FLOAT rtxes.
6126 Copyright (C) 2014 Free Software Foundation, Inc.
6128 Copying and distribution of this file, with or without modification,
6129 are permitted in any medium without royalty provided the copyright
6130 notice and this notice are preserved.