1 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
4 * lto-cgraph.c (input_overwrite_node): Check that partitioning
5 flags are set only during streaming.
6 * ipa.c (process_references, walk_polymorphic_call_targets,
7 symtab_remove_unreachable_nodes): Drop bodies of always inline
9 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
11 2014-04-04 Jakub Jelinek <jakub@redhat.com>
12 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15 * dwarf2out.c (const_ok_for_output_1): Reject expressions
18 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
23 (cortex_a53_fdivd): Likewise.
25 2014-04-04 Martin Jambor <mjambor@suse.cz>
28 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
30 * cgraph.c (clone_of_p): Also return true if thunks match.
31 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
32 cgraph_function_or_thunk_node and an obsolete comment.
33 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
35 (build_function_decl_skip_args): Likewise.
36 (set_new_clone_decl_and_node_flags): New function.
37 (duplicate_thunk_for_node): Likewise.
38 (redirect_edge_duplicating_thunks): Likewise.
39 (cgraph_clone_node): New parameter args_to_skip, pass it to
40 redirect_edge_duplicating_thunks which is called instead of
41 cgraph_redirect_edge_callee.
42 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
43 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
45 2014-04-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
48 * fortran/trans-types.c (gfc_get_function_type): In case of recursion
49 build a variadic function type with empty argument list instead of a
50 stdarg-like function type with incomplete argument list.
52 2014-04-04 Jeff Law <law@redhat.com>
55 * config/arm/predicates.md (const_int_I_operand): New predicate.
56 (const_int_M_operand): Similarly.
57 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
59 (insv_t2, extv_reg, extzv_t2): Likewise.
60 (load_multiple_with_writeback): Similarly for const_int_I_operand.
61 (pop_multiple_with_writeback_and_return): Likewise.
62 (vfp_pop_multiple_with_writeback): Likewise
64 2014-04-04 Richard Biener <rguenther@suse.de>
67 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
68 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
70 * gimplify.h (gimple_add_tmp_var_fn): Declare.
71 * gimplify.c (gimple_add_tmp_var_fn): New function.
72 * gimple-expr.h (create_tmp_reg_fn): Declare.
73 * gimple-expr.c (create_tmp_reg_fn): New function.
74 * gimple-low.c (record_vars_into): Don't change cfun.
75 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
76 code generation without cfun.
78 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
81 * Makefile.in (install-driver): Fix shell scripting.
83 2014-04-03 Cong Hou <congh@google.com>
85 PR tree-optimization/60505
86 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
87 threshold of number of iterations below which no vectorization will be
89 * tree-vect-loop.c (new_loop_vec_info):
90 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
91 * tree-vect-loop.c (vect_analyze_loop_operations):
92 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
93 * tree-vect-loop.c (vect_transform_loop):
94 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
95 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
96 of iterations of the loop and see if we should build the epilogue.
98 2014-04-03 Richard Biener <rguenther@suse.de>
100 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
102 (streamer_tree_cache_create): Adjust.
103 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
104 to allow optional nodes array.
105 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
106 (streamer_tree_cache_append): Likewise.
107 (streamer_tree_cache_create): Create nodes array optionally
108 as specified by parameter.
109 * lto-streamer-out.c (create_output_block): Avoid maintaining
110 the node array in the writer cache.
111 (DFS_write_tree): Remove assertion.
112 (produce_asm_for_decls): Free the out decl state hash table
114 * lto-streamer-in.c (lto_data_in_create): Adjust for
115 streamer_tree_cache_create prototype change.
117 2014-04-03 Richard Biener <rguenther@suse.de>
119 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
120 set TREE_CHAIN to NULL_TREE.
122 2014-04-03 Richard Biener <rguenther@suse.de>
124 PR tree-optimization/60740
125 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
126 over all GIMPLE_COND operands.
128 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
130 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
131 (Weffc++): Remove Scott's numbering, merge lists and reference
137 * class.c (accessible_nvdtor_p): New.
138 (check_bases): Don't check base destructor here ...
139 (check_bases_and_members): ... check them here. Trigger on
140 Wnon-virtual-dtor flag.
141 (finish_struct_1): Use accessible_nvdtor_p.
144 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
145 * g++.dg/warn/Wnvdtor-2.C: New.
146 * g++.dg/warn/Wnvdtor-3.C: New.
147 * g++.dg/warn/Wnvdtor-4.C: New.
148 * g++.dg/warn/Weff1.C: Delete.
149 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
150 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
152 2014-04-03 Nick Clifton <nickc@redhat.com>
154 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
157 2014-04-03 Martin Jambor <mjambor@suse.cz>
159 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
160 mention gcc_unreachable before failing.
161 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
164 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
167 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
168 code and instead mark the context inconsistent.
169 (possible_polymorphic_call_targets): For inconsistent contexts
170 return empty complete list.
172 2014-04-02 Anthony Green <green@moxielogic.com>
174 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
175 (extendqisi2, extendhisi2): Define.
176 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
177 (WCHAR_TYPE): Change to unsigned int.
179 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
181 PR tree-optimization/60733
182 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
183 insertion point for PHI candidates to be the end of the feeding
184 block for the PHI argument.
186 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
188 PR rtl-optimization/60650
189 * lra-constraints.c (process_alt_operands): Decrease reject for
190 earlyclobber matching.
192 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
194 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
196 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
198 * config/spu/spu.c (pad_bb): Do not crash when the last
199 insn is CODE_FOR_blockage.
201 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
203 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
204 lies outside the target mode.
206 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
209 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
210 software floating point or no floating point registers, do not
211 allow any type in the FPRs. Eliminate a test for SPE SIMD types
212 in GPRs that occurs after we tested for GPRs that would never be
215 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
216 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
217 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
218 specifically allow DDmode, since that does not use the SPE SIMD
221 2014-04-02 Richard Biener <rguenther@suse.de>
224 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
225 MODE_INTs. Properly use negv_optab.
226 (expand_abs): Likewise.
228 2014-04-02 Richard Biener <rguenther@suse.de>
231 * Makefile.in (install-driver): Guard extra installs with special
234 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
236 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
239 2014-04-01 Richard Henderson <rth@redhat.com>
242 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
243 alternative enabled before register allocation.
245 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
247 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
248 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
250 (nios2_large_got_address): Remove unneeded 'sym' parameter.
251 (nios2_got_address): Update nios2_large_got_address call site.
252 (nios2_delegitimize_address): New function.
253 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
254 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
255 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
257 2014-04-01 Martin Husemann <martin@duskware.de>
259 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
262 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
264 PR rtl-optimization/60604
265 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
266 check from register_operand.
267 (register_operand): Redefine in terms of general_operand.
268 (nonmemory_operand): Use register_operand for the non-constant cases.
270 2014-04-01 Richard Biener <rguenther@suse.de>
272 * gimple.h (struct gimple_statement_base): Align subcode to
275 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
277 * doc/invoke.texi (mapp-regs): Clarify.
279 2014-03-31 Ulrich Drepper <drepper@gmail.com>
281 * config/i386/avx512fintrin.h (__v32hi): Define type.
283 (_mm512_set1_epi8): Define.
284 (_mm512_set1_epi16): Define.
285 (_mm512_set4_epi32): Define.
286 (_mm512_set4_epi64): Define.
287 (_mm512_set4_pd): Define.
288 (_mm512_set4_ps): Define.
289 (_mm512_setr4_epi64): Define.
290 (_mm512_setr4_epi32): Define.
291 (_mm512_setr4_pd): Define.
292 (_mm512_setr4_ps): Define.
293 (_mm512_setzero_epi32): Define.
295 2014-03-31 Martin Jambor <mjambor@suse.cz>
298 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
299 callsite_arguments_match_p. Updated all callers. Also check types of
300 corresponding formal parameters and actual arguments.
301 (not_all_callers_have_enough_arguments_p) Renamed to
302 some_callers_have_mismatched_arguments_p.
304 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
306 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
308 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
311 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
314 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
316 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
317 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
319 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
320 Use FMAMODE_NOVF512 mode iterator.
321 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
322 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
323 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
324 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
326 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
327 Use VF_128_256 mode iterator.
328 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
331 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
333 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
334 static chain if needed.
336 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
339 * lra-constraints.c (index_part_to_reg): New.
340 (process_address): Use it.
342 2014-03-27 Jeff Law <law@redhat.com>
343 Jakub Jelinek <jakub@redhat.com>
346 * expr.c (do_tablejump): Use simplify_gen_binary rather than
347 gen_rtx_{PLUS,MULT} to build up the address expression.
349 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
350 creating non-canonical RTL.
352 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
355 * ipa-inline.c (want_inline_small_function_p): Short circuit large
356 functions; reorganize to make cheap checks first.
357 (inline_small_functions): Do not estimate growth when dumping;
359 * ipa-inline.h (inline_summary): Add min_size.
360 (growth_likely_positive): New function.
361 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
362 (set_cond_stmt_execution_predicate): Cleanup.
363 (estimate_edge_size_and_time): Compute min_size.
364 (estimate_calls_size_and_time): Likewise.
365 (estimate_node_size_and_time): Likewise.
366 (inline_update_overall_summary): Update min_size.
367 (do_estimate_edge_time): Likewise.
368 (do_estimate_edge_size): Update.
369 (do_estimate_edge_hints): Update.
370 (growth_likely_positive): New function.
372 2014-03-28 Jakub Jelinek <jakub@redhat.com>
375 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
376 also if addr has VOIDmode.
378 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
380 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
381 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
383 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
384 instructions as well as AdvancedSIMD loads.
386 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
388 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
389 Use crypto_aese type.
390 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
391 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
392 crypto_aese, crypto_aesmc. Move to types.md.
393 * config/arm/types.md (crypto_aes): Split into crypto_aese,
395 * config/arm/iterators.md (crypto_type): Likewise.
397 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
399 * cgraph.c: Include expr.h and tree-dfa.h.
400 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
403 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
406 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
407 regs from checking multi-reg pseudos.
409 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
411 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
413 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
415 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
416 if it would clobber the stack pointer, even temporarily.
418 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
420 * mode-switching.c: Make small adjustments to the top comment.
422 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
424 * config/rs6000/constraints.md (wD constraint): New constraint to
425 match the constant integer to get the top DImode/DFmode out of a
426 vector in a VSX register.
428 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
429 match the constant integer to get the top DImode/DFmode out of a
430 vector in a VSX register.
432 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
435 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
438 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
439 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
441 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
442 Optimize vec_extract of 64-bit values, where the value being
443 extracted is in the top word, where we can use scalar
444 instructions. Add direct move and store support. Combine the big
445 endian/little endian vector select load support into a single insn.
446 (vsx_extract_<mode>_internal1): Likewise.
447 (vsx_extract_<mode>_internal2): Likewise.
448 (vsx_extract_<mode>_load): Likewise.
449 (vsx_extract_<mode>_store): Likewise.
450 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
451 combined into vsx_extract_<mode>_load.
452 (vsx_extract_<mode>_one_le): Likewise.
454 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
455 define the top 64-bit vector element.
457 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
460 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
461 Document vec_vbpermq builtin.
464 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
465 enable use of xxsldwi and xxpermdi builtin functions.
466 (vec_xxpermdi): Likewise.
468 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
469 Document use of vec_xxsldwi and vec_xxpermdi builtins.
471 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
473 PR rtl-optimization/60650
474 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
476 (find_spills_for): New.
477 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
478 Spill all pseudos on the second iteration.
480 2014-03-27 Marek Polacek <polacek@redhat.com>
483 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
486 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
488 * config/s390/s390.c (s390_can_use_return_insn): Check for
489 call-saved FPRs on 31 bit.
491 2014-03-27 Jakub Jelinek <jakub@redhat.com>
494 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
495 if they need regimplification, just drop them instead of
496 calling gimple_regimplify_operands on them.
498 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
501 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
502 (aarch64_frame_pointer_required): Adjust logic.
503 (aarch64_can_eliminate): Adjust logic.
504 (aarch64_override_options_after_change): Adjust logic.
506 2014-03-27 Dehao Chen <dehao@google.com>
508 * ipa-inline.c (early_inliner): Update node's inline info.
510 2014-03-26 Dehao Chen <dehao@google.com>
512 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
513 compiler inserted conditional jumps for NAN float check.
515 2014-03-26 Jakub Jelinek <jakub@redhat.com>
517 * ubsan.h (ubsan_create_data): Change second argument's type
518 to const location_t *.
519 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
521 (ubsan_create_data): Change second argument to const location_t *PLOC.
522 Create Loc field whenever PLOC is non-NULL.
523 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
524 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
528 * real.c (real_to_integer2): Change type of low to UHWI.
530 2014-03-26 Tobias Burnus <burnus@net-b.de>
532 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
533 (CILK_SELF_SPECS): New define.
534 (driver_self_specs): Use it.
536 2014-03-26 Richard Biener <rguenther@suse.de>
538 * tree-pretty-print.c (percent_K_format): Implement special
539 case for LTO and its stripped down BLOCK tree.
541 2014-03-26 Jakub Jelinek <jakub@redhat.com>
544 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
546 * tree-vrp.c (simplify_internal_call_using_ranges): If only
547 one range is range_int_cst_p, but not both, at least optimize
548 addition/subtraction of 0 and multiplication by 0 or 1.
549 * gimple-fold.c (gimple_fold_call): Fold
550 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
551 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
552 INTEGER_CSTs, try to fold at least x * 0 and y - y.
554 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
556 PR rtl-optimization/60452
557 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
558 <case REG>: Return 1 for invalid offsets from the frame pointer.
560 2014-03-26 Marek Polacek <polacek@redhat.com>
563 * doc/extend.texi (C Extensions): Mention variable-length arrays in
566 2014-03-26 Marek Polacek <polacek@redhat.com>
569 * doc/extend.texi (Designated Inits): Describe what happens to omitted
572 2014-03-26 Marek Polacek <polacek@redhat.com>
575 * ira-color.c (update_conflict_hard_regno_costs): Perform the
576 multiplication in unsigned type.
578 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
580 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
582 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
584 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
586 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
589 * cif-code.def (UNREACHABLE) New code.
590 * ipa-inline.c (inline_small_functions): Skip edges to
591 __builtlin_unreachable.
592 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
593 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
594 predicate to __bulitin_unreachable.
595 (set_cond_stmt_execution_predicate): Fix issue when
596 invert_tree_comparison returns ERROR_MARK.
597 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
598 propagate to inline clones.
599 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
601 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
602 * cgraphclones.c (cgraph_clone_node): If call destination is already
603 ureachable, do not redirect it back.
604 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
607 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
609 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
610 Do not modify inline clones.
612 2014-03-25 Jakub Jelinek <jakub@redhat.com>
614 * config/i386/i386.md (general_sext_operand): New mode attr.
615 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
616 don't generate (sign_extend (const_int)).
617 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
618 operands[2]. Use We constraint instead of <i> and
619 <general_sext_operand> predicate instead of <general_operand>.
620 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
621 * config/i386/constraints.md (We): New constraint.
622 * config/i386/predicates.md (x86_64_sext_operand,
623 sext_operand): New predicates.
625 2014-03-25 Martin Jambor <mjambor@suse.cz>
628 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
629 inconsistent devirtualizations to __builtin_unreachable.
631 2014-03-25 Marek Polacek <polacek@redhat.com>
634 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
636 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
638 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
639 order of elements for big-endian.
641 2014-03-25 Richard Biener <rguenther@suse.de>
644 * gimplify-me.c (gimple_regimplify_operands): Update the
647 2014-03-25 Martin Jambor <mjambor@suse.cz>
650 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
651 (lto_output_varpool_node): Likewise.
652 (input_overwrite_node): Likewise.
653 (input_varpool_node): Likewise.
655 2014-03-25 Richard Biener <rguenther@suse.de>
657 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
660 2014-03-25 Jakub Jelinek <jakub@redhat.com>
662 * combine.c (simplify_compare_const): Add MODE argument.
663 Handle mode_width 0 as very large mode_width.
664 (try_combine, simplify_comparison): Adjust callers.
666 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
667 type to avoid signed integer overflow.
668 * explow.c (plus_constant): Likewise.
670 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
672 * doc/generic.texi: Correct typos.
674 2014-03-24 Tobias Burnus <burnus@net-b.de>
676 * doc/invoke.texi (-flto): Expand section about
677 using static libraries with LTO.
679 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
681 PR rtl-optimization/60501
682 * optabs.def (addptr3_optab): New optab.
683 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
684 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
685 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
687 * lra.c (emit_add3_insn): Use the addptr pattern if available.
689 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
691 2014-03-24 Ulrich Drepper <drepper@gmail.com>
693 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
696 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
697 (_mm256_undefined_ps): Define.
698 (_mm256_undefined_pd): Define.
699 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
700 (_mm_undefined_pd): Define.
701 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
702 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
703 (_mm512_undefined_ps): Define.
704 (_mm512_undefined_pd): Define.
705 Use _mm*_undefined_*.
706 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
708 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
710 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
711 (lshr_simd): DI mode added.
712 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
713 (aarch64_ushr_simddi): Likewise.
714 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
715 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
716 (vshrd_n_u64): Likewise.
718 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
720 * Makefile.in (s-macro_list): Depend on cc1.
722 2014-03-23 Teresa Johnson <tejohnson@google.com>
724 * ipa-utils.c (ipa_print_order): Use specified dump file.
726 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
728 PR rtl-optimization/60601
729 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
731 * gcc.c (eval_spec_function): Initialize save_growing_value.
733 2014-03-22 Jakub Jelinek <jakub@redhat.com>
736 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
737 code == MINUS_EXPR, never swap op0 with op1.
739 * toplev.c (init_local_tick): Avoid signed integer multiplication
741 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
742 shift by first operand's bitsize.
744 2014-03-21 Jakub Jelinek <jakub@redhat.com>
747 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
749 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
750 TARGET_ISA_64BIT_P(x).
752 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
754 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
755 pattern for vector nor instead of subtract from splat(-1).
756 (altivec_expand_vec_perm_const_le): Likewise.
758 2014-03-21 Richard Henderson <rth@twiddle.net>
761 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
762 related insns after epilogue_completed.
764 2014-03-21 Martin Jambor <mjambor@suse.cz>
767 * cgraph.h (symtab_node): New flag body_removed.
768 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
769 when removing bodies.
770 * symtab.c (dump_symtab_base): Dump body_removed flag.
771 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
772 had their bodies removed.
774 2014-03-21 Martin Jambor <mjambor@suse.cz>
777 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
780 2014-03-21 Richard Biener <rguenther@suse.de>
782 PR tree-optimization/60577
783 * tree-core.h (struct tree_base): Document nothrow_flag use
784 in VAR_DECL_NONALIASED.
785 * tree.h (VAR_DECL_NONALIASED): New.
786 (may_be_aliased): Adjust.
787 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
789 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
791 * expr.c (expand_expr_real_1): Remove outdated comment.
793 2014-03-20 Jakub Jelinek <jakub@redhat.com>
796 * ira.c (adjust_cleared_regs): Call copy_rtx on
797 *reg_equiv[REGNO (loc)].src_p before passing it to
798 simplify_replace_fn_rtx.
801 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
802 into CONST, put pic register as first operand of PLUS. Use
803 gen_const_mem for both 32-bit and 64-bit PIC got loads.
805 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
807 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
809 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
811 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
812 around for store forwarding issue in the FPU on the UT699.
813 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
814 loads and operations if -mfix-ut699 is specified.
815 (divtf3_hq): Tweak attribute.
816 (sqrttf2_hq): Likewise.
818 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
820 * calls.c (store_one_arg): Remove incorrect const qualification on the
821 type of the temporary.
822 * cfgexpand.c (expand_return): Likewise.
823 * expr.c (expand_constructor): Likewise.
824 (expand_expr_real_1): Likewise.
826 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
828 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
831 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
834 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
836 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
838 * config/arm/aarch-common-protos.h
839 (alu_cost_table): Fix spelling of "extend".
840 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
842 2014-03-19 Richard Biener <rguenther@suse.de>
845 * tree-core.h (tree_type_common): Re-order pointer members
846 to reduce recursion depth during GC walks.
848 2014-03-19 Marek Polacek <polacek@redhat.com>
851 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
854 2014-03-19 Richard Biener <rguenther@suse.de>
857 * lto-streamer-in.c (input_function): In WPA stage do not drop
860 2014-03-19 Jakub Jelinek <jakub@redhat.com>
862 PR tree-optimization/60559
863 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
864 with build_zero_cst assignment.
866 2014-03-18 Kai Tietz <ktietz@redhat.com>
868 PR rtl-optimization/56356
869 * sdbout.c (sdbout_parms): Verify that parms'
870 incoming argument is valid.
871 (sdbout_reg_parms): Likewise.
873 2014-03-18 Richard Henderson <rth@redhat.com>
876 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
877 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
878 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
880 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
882 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
883 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
884 Italicize plugin event names in description. Explain that
885 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
886 Remind that no GCC functions should be called after PLUGIN_FINISH.
887 Explain what pragmas with expansion are.
889 2014-03-18 Martin Liska <mliska@suse.cz>
891 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
892 gimple call statement is update.
893 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
894 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
896 2014-03-18 Jakub Jelinek <jakub@redhat.com>
899 * ubsan.c (ubsan_instrument_unreachable): Call
900 initialize_sanitizer_builtins.
901 (ubsan_pass): Likewise.
904 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
905 varpool_finalize_decl instead of rest_of_decl_compilation.
907 2014-03-18 Richard Biener <rguenther@suse.de>
909 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
910 by using bitmap_and_compl instead of bitmap_and_compl_into.
911 (df_rd_transfer_function): Likewise.
913 2014-03-18 Richard Biener <rguenther@suse.de>
915 * doc/lto.texi (fresolution): Fix typo.
917 2014-03-18 Richard Biener <rguenther@suse.de>
919 * doc/invoke.texi (flto): Update for changes in 4.9.
921 2014-03-18 Richard Biener <rguenther@suse.de>
923 * doc/loop.texi: Remove section on the removed lambda framework.
924 Update loop docs with recent changes in preserving loop structure.
926 2014-03-18 Richard Biener <rguenther@suse.de>
928 * doc/lto.texi (-fresolution): Document.
930 2014-03-18 Richard Biener <rguenther@suse.de>
932 * doc/contrib.texi: Adjust my name.
934 2014-03-18 Jakub Jelinek <jakub@redhat.com>
937 * internal-fn.c: Include diagnostic-core.h.
938 (expand_BUILTIN_EXPECT): New function.
939 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
940 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
941 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
942 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
944 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
945 Revert 3 argument __builtin_expect code.
946 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
947 * gimple-fold.c (gimple_fold_call): Likewise.
948 * tree.h (fold_builtin_expect): New prototype.
949 * builtins.c (build_builtin_expect_predicate): Add predictor
950 argument, if non-NULL, create 3 argument __builtin_expect.
951 (fold_builtin_expect): No longer static. Add ARG2 argument,
952 pass it through to build_builtin_expect_predicate.
953 (fold_builtin_2): Adjust caller.
954 (fold_builtin_3): Handle BUILT_IN_EXPECT.
955 * internal-fn.def (BUILTIN_EXPECT): New.
957 2014-03-18 Tobias Burnus <burnus@net-b.de>
960 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
961 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
962 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
964 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
967 * predict.c (combine_predictions_for_bb): Fix up formatting.
968 (expr_expected_value_1, expr_expected_value): Add predictor argument,
969 fill what it points to if non-NULL.
970 (tree_predict_by_opcode): Adjust caller, use the predictor.
971 * predict.def (PRED_COMPARE_AND_SWAP): Add.
973 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
975 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
976 proper constant for the store mode.
978 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
980 * symtab.c (change_decl_assembler_name): Fix transparent alias
983 2014-03-16 Renlin Li <Renlin.Li@arm.com>
985 * config/aarch64/aarch64.c: Correct the comments about the
986 aarch64 stack layout.
988 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
990 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
991 check for GF_OMP_FOR_KIND_FOR.
993 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
995 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
996 ymm and zmm register names.
998 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1001 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1002 note creation for the 2010-08-31 changes.
1004 2014-03-17 Marek Polacek <polacek@redhat.com>
1007 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1008 as -fno-tree-loop-vectorize.
1009 (expand_omp_simd): Likewise.
1011 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1013 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1014 (eligible_for_call_delay): New prototype.
1015 * config/sparc/sparc.c (tls_call_delay): Rename into...
1016 (eligible_for_call_delay): ...this. Return false if the instruction
1017 cannot be put in the delay slot of a branch.
1018 (eligible_for_restore_insn): Simplify.
1019 (eligible_for_return_delay): Return false if the instruction cannot be
1020 put in the delay slot of a branch and simplify.
1021 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1022 put in the delay slot of a branch.
1023 * config/sparc/sparc.md (fix_ut699): New attribute.
1024 (tls_call_delay): Delete.
1025 (in_call_delay): Reimplement.
1026 (eligible_for_sibcall_delay): Rename into...
1027 (in_sibcall_delay): ...this.
1028 (eligible_for_return_delay): Rename into...
1029 (in_return_delay): ...this.
1030 (in_branch_delay): Reimplement.
1031 (in_uncond_branch_delay): Delete.
1032 (in_annul_branch_delay): Delete.
1034 2014-03-14 Richard Henderson <rth@redhat.com>
1037 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1038 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1039 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1040 (floathi splitters): Remove.
1041 (float<SWI48x>xf2): New pattern.
1042 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1043 code that tried to handle DImode for 32-bit, but which was excluded
1044 by the pattern's condition. Drop allocation of stack temporary.
1045 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1046 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1047 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1048 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1049 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1050 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1051 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1052 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1053 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1054 (*float<SWI48x><X87MODEF>2_i387): Remove.
1055 (all float _with_temp splitters): Remove.
1056 (*float<SWI48x><MODEF>2_i387): New pattern.
1057 (*float<SWI48><MODEF>2_sse): New pattern.
1058 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1059 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1061 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1062 Marek Polacek <polacek@redhat.com>
1065 * common.opt (dump_base_name_prefixed): New Variable.
1066 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1067 if x_dump_base_name_prefixed is already set, set it at the end.
1069 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1071 PR rtl-optimization/60508
1072 * lra-constraints.c (get_reload_reg): Add new parameter
1074 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1075 Pass the new parameter values.
1077 2014-03-14 Richard Biener <rguenther@suse.de>
1079 * common.opt: Revert unintented changes from r205065.
1082 2014-03-14 Richard Biener <rguenther@suse.de>
1085 * cfghooks.c (split_block): Properly adjust all loops the
1086 block was a latch of.
1088 2014-03-14 Martin Jambor <mjambor@suse.cz>
1091 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1094 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1097 * config/avr/avr.c (avr_set_current_function): Pass function name
1098 through default_strip_name_encoding before sanity checking instead
1099 of skipping the first char of the assembler name.
1101 2014-03-13 Richard Henderson <rth@redhat.com>
1104 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1105 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1106 * config/i386/i386-protos.h: Likewise.
1107 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1108 in the expander instead of a splitter.
1109 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1110 any possibility of requiring a memory.
1111 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1112 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1113 (fp branch splitters): Update for ix86_split_fp_branch.
1114 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1115 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1116 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1117 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1118 (*fop_<MODEF>_3_i387): Likewise.
1119 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1120 (splitters for the fop_* register patterns): Remove.
1121 (fscalexf4_i387): Rename from *fscalexf4_i387.
1122 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1124 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1126 PR tree-optimization/59779
1127 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1128 type for bitsize and maxsize instead of HOST_WIDE_INT.
1130 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1132 PR rtl-optimization/57320
1133 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1134 the CFG after thread_prologue_and_epilogue_insns.
1136 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1138 PR rtl-optimization/57189
1139 * lra-constraints.c (process_alt_operands): Disfavor spilling
1142 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1144 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1146 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1148 PR tree-optimization/59025
1150 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1151 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1153 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1156 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1157 calls of avr_out_plus_1.
1159 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1161 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1162 BB's single pred and update the father loop's latch info later.
1164 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1166 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1170 (VEC_base): Likewise.
1171 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1172 registers, we need to swap double words in little endian mode.
1174 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1175 to be a container mode for 128-bit integer operations added in ISA
1176 2.07. Unlike TImode and PTImode, the preferred register set is
1177 the Altivec/VMX registers for the 128-bit operations.
1179 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1181 (rs6000_split_128bit_ok_p): Likewise.
1183 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1184 macros for creating ISA 2.07 normal and overloaded builtin
1185 functions with 3 arguments.
1186 (BU_P8V_OVERLOAD_3): Likewise.
1187 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1188 for use as overloaded functions.
1189 (VPERM_1TI_UNS): Likewise.
1190 (VSEL_1TI): Likewise.
1191 (VSEL_1TI_UNS): Likewise.
1192 (ST_INTERNAL_1ti): Likewise.
1193 (LD_INTERNAL_1ti): Likewise.
1194 (XXSEL_1TI): Likewise.
1195 (XXSEL_1TI_UNS): Likewise.
1196 (VPERM_1TI): Likewise.
1197 (VPERM_1TI_UNS): Likewise.
1198 (XXPERMDI_1TI): Likewise.
1199 (SET_1TI): Likewise.
1200 (LXVD2X_V1TI): Likewise.
1201 (STXVD2X_V1TI): Likewise.
1202 (VEC_INIT_V1TI): Likewise.
1203 (VEC_SET_V1TI): Likewise.
1204 (VEC_EXT_V1TI): Likewise.
1205 (EQV_V1TI): Likewise.
1206 (NAND_V1TI): Likewise.
1207 (ORC_V1TI): Likewise.
1208 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1209 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1211 (VADDUQM): Likewise.
1212 (VSUBCUQ): Likewise.
1213 (VADDEUQM): Likewise.
1214 (VADDECUQ): Likewise.
1215 (VSUBEUQM): Likewise.
1216 (VSUBECUQ): Likewise.
1218 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1219 __int128_t and __uint128_t types.
1220 (__uint128_type): Likewise.
1221 (altivec_categorize_keyword): Add support for vector __int128_t,
1222 vector __uint128_t, vector __int128, and vector unsigned __int128
1223 as a container type for TImode operations that need to be done in
1224 VSX/Altivec registers.
1225 (rs6000_macro_to_expand): Likewise.
1226 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1227 to support 128-bit integer instructions vaddcuq, vadduqm,
1228 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1229 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1231 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1232 for V1TImode, and set up preferences to use VSX/Altivec registers.
1233 Setup VSX reload handlers.
1234 (rs6000_debug_reg_global): Likewise.
1235 (rs6000_init_hard_regno_mode_ok): Likewise.
1236 (rs6000_preferred_simd_mode): Likewise.
1237 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1238 (easy_altivec_constant): Likewise.
1239 (output_vec_const_move): Likewise.
1240 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1242 (rs6000_expand_vector_extract): Likewise.
1243 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1245 (rs6000_const_vec): Add support for V1TImode.
1246 (rs6000_emit_le_vsx_load): Swap double words when loading or
1247 storing TImode/V1TImode.
1248 (rs6000_emit_le_vsx_store): Likewise.
1249 (rs6000_emit_le_vsx_move): Likewise.
1250 (rs6000_emit_move): Add support for V1TImode.
1251 (altivec_expand_ld_builtin): Likewise.
1252 (altivec_expand_st_builtin): Likewise.
1253 (altivec_expand_vec_init_builtin): Likewise.
1254 (altivec_expand_builtin): Likewise.
1255 (rs6000_init_builtins): Add support for V1TImode type. Add
1256 support for ISA 2.07 128-bit integer builtins. Define type names
1257 for the VSX/Altivec vector types.
1258 (altivec_init_builtins): Add support for overloaded vector
1259 functions with V1TImode type.
1260 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1261 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1263 (rs6000_split_128bit_ok_p): Likewise.
1264 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1265 __int128_t and vector __uint128_t.
1267 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1268 and mode attributes.
1275 (VS_scalar): Likewise.
1276 (VS_double): Likewise.
1277 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1279 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1280 we support the ISA 2.07 128-bit integer arithmetic instructions.
1281 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1282 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1283 and TImode types for use with the builtin functions.
1284 (V1TI_type_node): Likewise.
1285 (unsigned_V1TI_type_node): Likewise.
1286 (intTI_type_internal_node): Likewise.
1287 (uintTI_type_internal_node): Likewise.
1289 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1290 128-bit builtin functions.
1291 (UNSPEC_VADDEUQM): Likewise.
1292 (UNSPEC_VADDECUQ): Likewise.
1293 (UNSPEC_VSUBCUQ): Likewise.
1294 (UNSPEC_VSUBEUQM): Likewise.
1295 (UNSPEC_VSUBECUQ): Likewise.
1296 (VM): Add V1TImode to vector mode iterators.
1298 (VI_unit): Likewise.
1299 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1300 (altivec_vaddcuq): Likewise.
1301 (altivec_vsubuqm): Likewise.
1302 (altivec_vsubcuq): Likewise.
1303 (altivec_vaddeuqm): Likewise.
1304 (altivec_vaddecuq): Likewise.
1305 (altivec_vsubeuqm): Likewise.
1306 (altivec_vsubecuq): Likewise.
1308 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1310 (BOOL_128): Likewise.
1311 (BOOL_REGS_OUTPUT): Likewise.
1312 (BOOL_REGS_OP1): Likewise.
1313 (BOOL_REGS_OP2): Likewise.
1314 (BOOL_REGS_UNARY): Likewise.
1315 (BOOL_REGS_AND_CR0): Likewise.
1317 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1318 128-bit integer builtin support.
1319 (vec_vadduqm): Likewise.
1320 (vec_vaddecuq): Likewise.
1321 (vec_vaddeuqm): Likewise.
1322 (vec_vsubecuq): Likewise.
1323 (vec_vsubeuqm): Likewise.
1324 (vec_vsubcuq): Likewise.
1325 (vec_vsubuqm): Likewise.
1327 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1328 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1329 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1330 128-bit integer add/subtract to ISA 2.07.
1332 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1334 * config/arc/arc.c (arc_predicate_delay_insns):
1335 Fix third argument passed to conditionalize_nonjump.
1337 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1339 * config/aarch64/aarch64-builtins.c
1340 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1341 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1342 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1343 instead of __builtin_lfloor.
1344 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1346 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1348 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1349 (tree_ssa_ifcombine_bb_1): New function.
1350 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1351 is an empty forwarder block to then_bb or vice versa and then_bb
1352 and else_bb are effectively swapped.
1354 2014-03-12 Christian Bruel <christian.bruel@st.com>
1357 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1358 REG_CFA_DEF_CFA note.
1359 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1360 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1362 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1364 PR tree-optimization/60454
1365 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1367 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1369 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1370 Do not define target_cpu_default2 to generic.
1371 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1372 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1373 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1375 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1376 Marc Glisse <marc.glisse@inria.fr>
1378 PR tree-optimization/60502
1379 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1380 instead of build_low_bits_mask.
1382 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1385 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1386 if there are multiple uses, but op doesn't live on E edge.
1387 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1388 clobber stmts before __builtin_unreachable.
1390 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1392 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1393 hard_frame_pointer_rtx.
1394 * cse.c (cse_insn): Remove volatile check.
1395 * cselib.c (cselib_process_insn): Likewise.
1396 * dse.c (scan_insn): Likewise.
1398 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1400 * config/arc/arc.c (conditionalize_nonjump): New function,
1402 (arc_ifcvt): ... this.
1403 (arc_predicate_delay_insns): Use it.
1405 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1407 * config/arc/predicates.md (extend_operand): During/after reload,
1408 allow const_int_operand.
1409 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1410 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1411 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1413 (umulsi3_highpart_i): Likewise.
1415 2014-03-11 Richard Biener <rguenther@suse.de>
1417 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1418 Add asserts to guard possible wrong-code bugs.
1420 2014-03-11 Richard Biener <rguenther@suse.de>
1422 PR tree-optimization/60429
1423 PR tree-optimization/60485
1424 * tree-ssa-structalias.c (set_union_with_increment): Properly
1425 take into account all fields that overlap the shifted vars.
1426 (do_sd_constraint): Likewise.
1427 (do_ds_constraint): Likewise.
1428 (get_constraint_for_ptr_offset): Likewise.
1430 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1432 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1433 (nios2_compute_frame_layout):
1434 Add calculation of cfun->machine->fp_save_offset.
1435 (nios2_expand_prologue): Correct setting of frame pointer register
1437 (nios2_expand_epilogue): Update recovery of stack pointer from
1438 frame pointer accordingly.
1439 (nios2_initial_elimination_offset): Update calculation of offset
1440 for eliminating to HARD_FRAME_POINTER_REGNUM.
1442 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1445 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1446 cgraph_get_create_node on VAR_DECLs.
1448 2014-03-10 Richard Biener <rguenther@suse.de>
1451 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1453 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1455 * config/vms/vms.opt (vms_float_format): New variable.
1457 2014-03-08 Tobias Burnus <burnus@net-b.de>
1459 * doc/invoke.texi (-fcilkplus): Update implementation status.
1461 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1462 Richard Biener <rguenther@suse.de>
1464 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1465 consistently accross all TUs.
1466 (run_gcc): Enable -fshort-double automatically at link at link-time
1467 and disallow override.
1469 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1472 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1473 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1474 if they can't be used.
1476 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1478 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1479 for Solaris 11/x86 ld.
1480 * configure: Regenerate.
1482 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1484 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1485 (LIB_TLS_SPEC): Save as ld_tls_libs.
1486 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1487 (HAVE_AS_IX86_TLSLDM): New test.
1488 * configure, config.in: Regenerate.
1489 * config/i386/i386.c (legitimize_tls_address): Fall back to
1490 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1491 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1492 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1493 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1495 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1497 * common.opt (fira-loop-pressure): Mark as optimization.
1499 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1501 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1502 an OpenMP mappable type.
1504 2014-03-06 Matthias Klose <doko@ubuntu.com>
1506 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1507 MULTILIB_OSDIRNAMES is not defined.
1509 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1510 Meador Inge <meadori@codesourcery.com>
1513 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1514 (arm_legitimize_address): Call legitimize_tls_address for any
1515 arm_tls_referenced_p expression, handle constant addend. Call it
1516 before testing for !TARGET_ARM.
1517 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1519 2014-03-06 Richard Biener <rguenther@suse.de>
1525 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1527 * tree-streamer.c (record_common_node): Assert we don't record
1528 nodes with type double.
1529 (preload_common_node): Skip type double, complex double and double
1530 pointer since it is now frontend dependent due to fshort-double option.
1532 2014-03-06 Richard Biener <rguenther@suse.de>
1534 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1535 or -fno-lto is specified and the linker has full plugin support.
1536 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1537 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1538 * lto-wrapper.c (merge_and_complain): Merge compile-time
1539 optimization levels.
1540 (run_gcc): And pass it through to the link options.
1542 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1546 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1548 * cselib.c (remove_useless_values): Skip to avoid quadratic
1549 behavior if the condition moved from...
1550 (cselib_process_insn): ... here holds.
1552 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1555 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1556 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1559 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1560 (TM_H): Add x86-tune.def.
1562 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1564 * config/aarch64/aarch64.c (generic_tunings):
1565 Use cortexa57_extra_costs.
1567 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1570 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1571 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1572 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1575 2014-03-04 Heiher <r@hev.cc>
1577 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1578 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1580 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1582 * config/i386/predicates.md (const2356_operand): Change to ...
1583 (const2367_operand): ... this.
1584 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1586 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1587 (*avx512pf_scatterpf<mode>sf): Ditto.
1588 (avx512pf_scatterpf<mode>df): Ditto.
1589 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1590 (*avx512pf_scatterpf<mode>df): Ditto.
1591 * config/i386/i386.c (ix86_expand_builtin): Update
1592 incorrect hint operand error message.
1594 2014-03-04 Richard Biener <rguenther@suse.de>
1596 * lto-section-in.c (lto_get_section_data): Fix const cast.
1598 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1600 * tree-streamer.c (record_common_node): Assert we don't record
1601 nodes with type double.
1602 (preload_common_node): Skip type double, complex double and double
1603 pointer since it is now frontend dependent due to fshort-double option.
1605 2014-03-04 Richard Biener <rguenther@suse.de>
1608 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1609 (lto_input_toplevel_asms): Likewise.
1610 * lto-section-in.c (lto_get_section_data): Instead do it here
1613 2014-03-04 Richard Biener <rguenther@suse.de>
1615 PR tree-optimization/60382
1616 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1617 dead PHIs a reduction.
1619 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1621 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1623 (_mm_prefetch): Move out of GCC target("sse") pragma.
1624 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1625 GCC target("prfchw") pragma.
1626 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1628 * config/i386/i386.c (ix86_option_override_internal): Enable
1629 -mprfchw with -mprefetchwt1.
1631 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1633 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1636 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1638 * opts.h (CL_PCH_IGNORE): Define.
1639 * targhooks.c (option_affects_pch_p):
1640 Return false for options that have CL_PCH_IGNORE set.
1641 * opt-functions.awk: Process PchIgnore.
1642 * doc/options.texi: Document PchIgnore.
1644 * config/arc/arc.opt (misize): Add PchIgnore property.
1646 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1648 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1649 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1650 constraint on constants to permit them being loaded into
1651 GENERAL_REGS or BASE_REGS.
1653 2014-03-03 Nick Clifton <nickc@redhat.com>
1655 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1656 anti-cacnonical alternatives.
1657 (negandhi3_real): New pattern.
1658 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1660 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1662 * config/avr/avr-mcus.def: Remove atxmega16x1.
1663 * config/avr/avr-tables.opt: Regenerate.
1664 * config/avr/t-multilib: Regenerate.
1665 * doc/avr-mmcu.texi: Regenerate.
1667 2014-03-03 Tobias Grosser <tobias@grosser.es>
1668 Mircea Namolaru <mircea.namolaru@inria.fr>
1670 PR tree-optimization/58028
1671 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1674 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1676 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1677 not handled by recognizers.
1679 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1682 * function.c (expand_function_end): Don't emit
1683 clobber_return_register sequence if clobber_after is a BARRIER.
1684 * cfgexpand.c (construct_exit_block): Append instructions before
1685 return_label to prev_bb.
1687 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1689 * config/rs6000/constraints.md: Document reserved use of "wc".
1691 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1694 * ipa.c (function_and_variable_visibility): When dissolving comdat
1695 group, also set all symbols to local.
1697 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1702 2013-12-14 Jan Hubicka <jh@suse.cz>
1704 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1706 2014-03-02 Jon Beniston <jon@beniston.com>
1712 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1713 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1714 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1715 (simple_return, *simple_return): New patterns
1716 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1717 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1719 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1721 * dwarf2out.c (gen_subprogram_die): Tidy.
1723 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1726 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1727 (*mov_t_msb_neg_negc): ... this new insn.
1729 2014-02-28 Jason Merrill <jason@redhat.com>
1732 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1735 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1738 * dwarf2out.c (decltype_auto_die): New static.
1739 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1740 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1741 (is_cxx_auto): Likewise.
1743 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1745 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1746 we are not using general regs only.
1748 2014-02-28 Richard Biener <rguenther@suse.de>
1751 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1752 previous fix and only allow to remove trivial pre-headers
1753 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1754 (remove_forwarder_block): Properly update the latch of a loop.
1756 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1759 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1760 (cselib_preserved_hash_table): New.
1761 (preserve_constants_and_equivs): Move preserved vals to it.
1762 (cselib_find_slot): Look it up first.
1763 (cselib_init): Initialize it.
1764 (cselib_finish): Release it.
1765 (dump_cselib_table): Dump it.
1767 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1770 * cselib.c (remove_useless_values): Skip to avoid quadratic
1771 behavior if the condition moved from...
1772 (cselib_process_insn): ... here holds.
1774 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1777 * var-tracking.c (vt_initialize): Apply the same condition to
1778 preserve the CFA base value.
1780 2014-02-28 Joey Ye <joey.ye@arm.com>
1783 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1784 if reload in progress or completed.
1786 2014-02-28 Tobias Burnus <burnus@net-b.de>
1789 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1792 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1794 * doc/tm.texi.in (Condition Code Status): Update documention for
1795 relative locations of cc0-setter and cc0-user.
1797 2014-02-27 Jeff Law <law@redhat.com>
1799 PR rtl-optimization/52714
1800 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1801 into two independent simple sets, if I3 is a jump, ensure the
1802 pattern we place into I3 is a (set (pc) ...).
1804 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1805 Jeff Law <law@redhat.com>
1807 PR rtl-optimization/49847
1808 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1809 are in different blocks.
1810 * doc/tm.texi (Condition Code Status): Update documention for
1811 relative locations of cc0-setter and cc0-user.
1813 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1816 * lra.c (lra_emit_add): Check SUBREG too.
1818 2014-02-27 Andreas Schwab <schwab@suse.de>
1820 * config/m68k/m68k.c (m68k_option_override): Disable
1821 -flive-range-shrinkage for classic m68k.
1822 (m68k_override_options_after_change): Likewise.
1824 2014-02-27 Marek Polacek <polacek@redhat.com>
1827 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1828 -Wmaybe-uninitialized.
1830 2014-02-27 Alan Modra <amodra@gmail.com>
1833 * reload1.c (emit_input_reload_insns): When reload_override_in,
1834 set old to rl->in_reg when rl->in_reg is a subreg.
1836 2014-02-26 Richard Biener <rguenther@suse.de>
1839 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1841 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1843 * common/config/i386/predicates.md (const1256_operand): Remove.
1844 (const2356_operand): New.
1845 (const_1_to_2_operand): Remove.
1846 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1847 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1848 (*avx512pf_gatherpf<mode>sf): Ditto.
1849 (avx512pf_gatherpf<mode>df): Ditto.
1850 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1851 (*avx512pf_gatherpf<mode>df): Ditto.
1852 (avx512pf_scatterpf<mode>sf): Ditto.
1853 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1854 (*avx512pf_scatterpf<mode>sf): Ditto.
1855 (avx512pf_scatterpf<mode>df): Ditto.
1856 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1857 (*avx512pf_scatterpf<mode>df): Ditto.
1858 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1860 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1862 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1863 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1864 (_mm512_mask_testn_epi64_mask): Move to ...
1865 * config/i386/avx512cdintrin.h: Here.
1866 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1867 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1868 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1869 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1870 TARGET_AVX512F from TARGET_AVX512CD.
1872 2014-02-26 Richard Biener <rguenther@suse.de>
1875 * ipa.c (walk_polymorphic_call_targets): Properly guard
1876 call to inline_update_overall_summary.
1878 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1881 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1882 and latches only if requested. Fix latch if it is removed.
1883 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1884 LOOPS_HAVE_PREHEADERS.
1886 2014-02-25 Andrew Pinski <apinski@cavium.com>
1888 * builtins.c (expand_builtin_thread_pointer): Create a new target
1889 when the target is NULL.
1891 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1893 PR rtl-optimization/60317
1894 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1895 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1896 * lra-assigns.c: Include params.h.
1897 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1898 other reload pseudos considerations.
1900 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1902 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1903 to use canonical form for nor<mode>3.
1905 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1908 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1911 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1913 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
1914 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
1915 (ix86_handle_option): Handle OPT_mprefetchwt1.
1916 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
1917 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
1919 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1920 OPTION_MASK_ISA_PREFETCHWT1.
1921 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
1922 (PTA_PREFETCHWT1): New.
1923 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
1924 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
1925 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
1926 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
1927 (*prefetch_avx512pf_<mode>_: Change into ...
1928 (*prefetch_prefetchwt1_<mode>: This.
1929 * config/i386/i386.opt (mprefetchwt1): New.
1930 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
1931 (_mm_prefetch): Handle intent to write.
1932 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
1934 2014-02-25 Richard Biener <rguenther@suse.de>
1937 * emit-rtl.c (mem_attrs_htab): Remove.
1938 (mem_attrs_htab_hash): Likewise.
1939 (mem_attrs_htab_eq): Likewise.
1940 (set_mem_attrs): Always allocate new mem-attrs when something changed.
1941 (init_emit_once): Do not allocate mem_attrs_htab.
1943 2014-02-25 Richard Biener <rguenther@suse.de>
1946 * lto-opts.c (lto_write_options): Output non-explicit conservative
1947 -fwrapv, -fno-trapv and -fno-strict-overflow.
1948 * lto-wrapper.c (merge_and_complain): Handle merging those options.
1949 (run_gcc): And pass them through.
1951 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1953 * sel-sched.c (calculate_new_fences): New parameter ptime.
1954 Calculate it as a maximum over all fence cycles.
1955 (sel_sched_region_2): Adjust the call to calculate_new_fences.
1956 Print the final schedule timing when sched_verbose.
1958 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
1960 PR rtl-optimization/60292
1961 * sel-sched.c (fill_vec_av_set): Do not reset target availability
1962 bit fot the fence instruction.
1964 2014-02-24 Alangi Derick <alangiderick@gmail.com>
1966 * calls.h: Fix typo in comment.
1968 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
1970 * config/pa/pa.c (pa_output_move_double): Don't valididate when
1971 adjusting offsetable addresses.
1973 2014-02-24 Guozhi Wei <carrot@google.com>
1975 * sparseset.h (sparseset_pop): Fix the wrong index.
1977 2014-02-24 Walter Lee <walt@tilera.com>
1979 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
1980 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
1982 * common/config/tilegx/tilegx-common.c
1983 (TARGET_DEFAULT_TARGET_FLAGS): Define.
1984 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
1986 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
1987 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
1988 (tilegx_gimplify_va_arg_expr): Handle big endian.
1989 (tilegx_expand_unaligned_load): Ditto.
1990 (tilegx_expand_unaligned_store): Ditto.
1991 (TARGET_RETURN_IN_MSB): New.
1992 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
1993 (TARGET_ENDIAN_DEFAULT): New.
1994 (TARGET_BIG_ENDIAN): Handle big endian.
1995 (BYTES_BIG_ENDIAN): Ditto.
1996 (WORDS_BIG_ENDIAN): Ditto.
1997 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2000 * config/tilegx/tilegx.md (extv): Handle big endian.
2002 (insn_st<n>): Ditto.
2003 (insn_st<n>_add<bitsuffix>): Ditto.
2004 (insn_stnt<n>): Ditto.
2005 (insn_stnt<n>_add<bitsuffix>):Ditto.
2006 (vec_interleave_highv8qi): Handle big endian.
2007 (vec_interleave_highv8qi_be): New.
2008 (vec_interleave_highv8qi_le): New.
2009 (insn_v1int_h): Handle big endian.
2010 (vec_interleave_lowv8qi): Handle big endian.
2011 (vec_interleave_lowv8qi_be): New.
2012 (vec_interleave_lowv8qi_le): New.
2013 (insn_v1int_l): Handle big endian.
2014 (vec_interleave_highv4hi): Handle big endian.
2015 (vec_interleave_highv4hi_be): New.
2016 (vec_interleave_highv4hi_le): New.
2017 (insn_v2int_h): Handle big endian.
2018 (vec_interleave_lowv4hi): Handle big endian.
2019 (vec_interleave_lowv4hi_be): New.
2020 (vec_interleave_lowv4hi_le): New.
2021 (insn_v2int_l): Handle big endian.
2022 (vec_interleave_highv2si): Handle big endian.
2023 (vec_interleave_highv2si_be): New.
2024 (vec_interleave_highv2si_le): New.
2025 (insn_v4int_h): Handle big endian.
2026 (vec_interleave_lowv2si): Handle big endian.
2027 (vec_interleave_lowv2si_be): New.
2028 (vec_interleave_lowv2si_le): New.
2029 (insn_v4int_l): Handle big endian.
2030 * config/tilegx/tilegx.opt (mbig-endian): New option.
2031 (mlittle-endian): New option.
2032 * doc/install.texi: Document tilegxbe-linux.
2033 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2035 2014-02-24 Martin Jambor <mjambor@suse.cz>
2038 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2039 there are no parameter descriptors.
2041 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2043 PR rtl-optimization/60268
2044 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2045 initialization to ...
2046 (sched_rgn_init): ... here.
2047 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2049 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2051 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2054 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2056 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2059 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2061 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2062 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2064 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2066 * config/microblaze/predicates.md: Add cmp_op predicate.
2067 * config/microblaze/microblaze.md: Add branch_compare instruction
2068 which uses cmp_op predicate and emits cmp insn before branch.
2069 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2070 to microblaze_expand_conditional_branch and consolidate logic.
2071 (microblaze_expand_conditional_branch): emit branch_compare
2072 insn instead of handling cmp op separate from branch insn.
2074 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2076 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2079 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2081 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2082 define_insn with define_expand and new define_insn
2083 *altivec_lve<VI_char>x_internal.
2084 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2085 and new define_insn *altivec_stve<VI_char>x_internal.
2086 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2088 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2090 (altivec_expand_stvex_be): New function.
2092 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2094 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2095 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2096 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2097 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2099 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2102 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2103 instead of emit_move_insn.
2105 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2107 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2109 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2110 gen_altivec_vsumsws.
2112 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2114 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2115 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2116 (altivec_lvxl_<mode>): New define_expand incorporating
2117 -maltivec=be semantics where needed.
2118 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2119 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2120 semantics where needed.
2121 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2122 (altivec_stvx_<mode>): New define_expand incorporating
2123 -maltivec=be semantics where needed.
2124 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2125 VM2 iterator instead of V4SI.
2126 (altivec_stvxl_<mode>): New define_expand incorporating
2127 -maltivec=be semantics where needed.
2128 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2129 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2130 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2131 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2132 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2133 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2134 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2135 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2136 ALTIVEC_BUILTIN_STVXL.
2137 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2138 (altivec_expand_stvx_be): Likewise.
2139 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2140 (altivec_expand_lvx_be): Likewise.
2141 (altivec_expand_stvx_be): Likewise.
2142 (altivec_expand_builtin): Add cases for
2143 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2144 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2145 (altivec_init_builtins): Add definitions for
2146 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2147 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2149 2014-02-21 Catherine Moore <clm@codesourcery.com>
2151 * doc/invoke.texi (mvirt, mno-virt): Document.
2152 * config/mips/mips.opt (mvirt): New option.
2153 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2155 2014-02-21 Richard Biener <rguenther@suse.de>
2157 PR tree-optimization/60276
2158 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2159 (STMT_VINFO_MIN_NEG_DIST): New macro.
2160 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2161 STMT_VINFO_MIN_NEG_DIST.
2162 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2163 made for negative dependence distances still hold.
2165 2014-02-21 Richard Biener <rguenther@suse.de>
2168 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2169 DECL_INITIAL for globals not in the current function context.
2171 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2173 PR tree-optimization/56490
2174 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2175 * tree-ssa-uninit.c: Include params.h.
2176 (compute_control_dep_chain): Add num_calls argument, return false
2177 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2178 num_calls to recursive call.
2179 (find_predicates): Change dep_chain into normal array,
2180 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2181 variable and adjust compute_control_dep_chain caller.
2182 (find_def_preds): Likewise.
2184 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2186 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2187 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2189 2014-02-21 Nick Clifton <nickc@redhat.com>
2191 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2192 (pushhi1): Likewise.
2193 (popqi1): Add mode to pre_dec.
2196 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2198 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2199 mode for mask of V8SFmode permutation.
2201 2014-02-20 Richard Henderson <rth@redhat.com>
2204 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2205 a new pseudo for OLDVAL.
2207 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2210 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2211 gen_reg_rtx if d->testing_p.
2212 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2213 if d->testing_p and we will certainly return true.
2214 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2217 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2219 * emit-rtl.c (gen_reg_rtx): Assert that
2220 crtl->emit.regno_pointer_align_length is non-zero.
2222 2014-02-20 Richard Henderson <rth@redhat.com>
2225 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2226 on failure the store back into EXPECT.
2228 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2229 Sandra Loosemore <sandra@codesourcery.com>
2231 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2232 * config/nios2/nios2.c (nios2_function_profiler): Add
2233 -fPIC (flag_pic == 2) support.
2234 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2235 (nios2_large_offset_p): New function.
2236 (nios2_unspec_reloc_p): Move up position, update to use
2237 nios2_large_offset_p.
2238 (nios2_unspec_address): Remove function.
2239 (nios2_unspec_offset): New function.
2240 (nios2_large_got_address): New function.
2241 (nios2_got_address): Add large offset support.
2242 (nios2_legitimize_tls_address): Update usage of removed and new
2244 (nios2_symbol_binds_local_p): New function.
2245 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2246 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2247 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2248 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2249 processing for (const (unspec ...)).
2250 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2252 2014-02-20 Richard Biener <rguenther@suse.de>
2254 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2255 doing the substitution.
2256 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2258 2014-02-20 Martin Jambor <mjambor@suse.cz>
2261 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2262 info when checking whether lattices are bottom.
2264 2014-02-20 Richard Biener <rguenther@suse.de>
2267 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2270 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2273 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2274 parameter specifying the scaling.
2275 (inline_call): Update.
2276 (want_inline_recursively): Guard division by zero.
2277 (recursive_inlining): Update.
2278 * ipa-inline.h (clone_inlined_nodes): Update.
2280 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2283 * config/i386/i386.c (classify_argument): Pass structures of size
2284 64 bytes or less in register.
2286 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2287 Kirill Yukhin <kirill.yukhin@intel.com>
2289 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2290 (_mm_rcp28_round_ss): Ditto.
2291 (_mm_rsqrt28_round_sd): Ditto.
2292 (_mm_rsqrt28_round_ss): Ditto.
2293 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2294 (_mm_rcp14_round_ss): Ditto.
2295 (_mm_rsqrt14_round_sd): Ditto.
2296 (_mm_rsqrt14_round_ss): Ditto.
2297 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2298 the first input operand, get rid of match_dup.
2299 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2301 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2303 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2304 operand as the first input operand, set type attribute.
2305 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2307 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2308 operand as the first input operand, set type attribute.
2310 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2312 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2315 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2318 * config/i386/i386.c (construct_container): Remove TFmode check
2319 for X86_64_INTEGER_CLASS.
2321 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2324 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2325 only when -Wpsabi is enabled.
2327 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2330 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2331 passing arrays in registers are the same as for structs, so remove the
2332 special case for them.
2334 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2336 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2337 destination type, extract only the valid bits if the source type is not
2338 integral and has a different mode.
2340 2014-02-19 Richard Biener <rguenther@suse.de>
2343 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2346 2014-02-19 Richard Biener <rguenther@suse.de>
2349 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2350 (ipa_modify_call_arguments): Emit an argument load explicitely and
2351 preserve virtual SSA form there and for the replacement call.
2352 Do not update SSA form nor free dominance info.
2354 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2356 * ipa.c (function_and_variable_visibility): Also clear WEAK
2357 flag when disolving COMDAT_GROUP.
2359 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2361 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2362 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2364 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2365 do more sanity checks.
2366 (detect_type_change): Return true when giving up early.
2367 (compute_complex_assign_jump_func): Fix type parameter of
2368 ipa_set_ancestor_jf.
2369 (compute_complex_ancestor_jump_func): Likewise.
2370 (update_jump_functions_after_inlining): Fix updating of
2372 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2374 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2376 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2377 inline clones when edge disappears.
2379 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2382 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2383 Split 64-bit moves into 2 patterns. Do not allow the use of
2384 direct move for TDmode in little endian, since the decimal value
2385 has little endian bytes within a word, but the 64-bit pieces are
2386 ordered in a big endian fashion, and normal subreg's of TDmode are
2388 (mov<mode>_64bit_dm): Likewise.
2389 (movtd_64bit_nodm): Likewise.
2391 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2393 PR tree-optimization/60174
2394 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2395 statement of an SSA_NAME that occurs in an abnormal PHI node.
2397 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2400 * final.c (SEEN_BB): Remove.
2401 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2402 (final_scan_insn): Don't force_source_line on second
2403 NOTE_INSN_BASIC_BLOCK.
2405 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2408 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2409 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2410 (type_natural_mode): Warn ABI change when %zmm register is not
2411 available for AVX512F vector value passing.
2413 2014-02-18 Kai Tietz <ktietz@redhat.com>
2416 * config/i386/i386.c (ix86_expand_prologue): Use value in
2417 rax register as displacement when restoring %r10 or %rax.
2418 Fix wrong offset when restoring both registers.
2420 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2422 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2423 assertion with conditional return.
2425 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2426 Uros Bizjak <ubizjak@gmail.com>
2429 * config/i386/driver-i386.c (host_detect_local_cpu): If
2430 YMM state is not saved by the OS, also clear has_f16c. Move
2431 CPUID 0x80000001 handling before YMM state saving checking.
2433 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2435 PR rtl-optimization/58960
2436 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2437 factored out from ...
2438 (sched_init): ... here.
2439 (free_global_sched_pressure_data): New, factored out from ...
2440 (sched_finish): ... here.
2441 * sched-int.h (free_global_sched_pressure_data): Declare.
2442 * sched-rgn.c (nr_regions_initial): New static global.
2443 (haifa_find_rgns): Initialize it.
2444 (schedule_region): Disable sched-pressure for the newly
2447 2014-02-17 Richard Biener <rguenther@suse.de>
2449 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2450 release SSA defs of pattern stmts.
2452 2014-02-17 Richard Biener <rguenther@suse.de>
2454 * tree-inline.c (expand_call_inline): Release the virtual
2455 operand defined by the call we are about to inline.
2457 2014-02-17 Richard Biener <rguenther@suse.de>
2459 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2461 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2462 Ilya Tocar <ilya.tocar@intel.com>
2464 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2465 arguments order in builtin.
2466 (_mm512_permutexvar_epi64): Ditto.
2467 (_mm512_mask_permutexvar_epi64): Ditto
2468 (_mm512_maskz_permutexvar_epi32): Ditto
2469 (_mm512_permutexvar_epi32): Ditto
2470 (_mm512_mask_permutexvar_epi32): Ditto
2472 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2474 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2475 (p8_vmrgow): Likewise.
2477 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2479 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2482 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2485 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2486 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2487 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2488 using direct move instructions on ISA 2.07. Also adjust
2489 instruction length for 64-bit.
2490 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2491 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2493 2014-02-15 Alan Modra <amodra@gmail.com>
2497 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2498 find_replacement on parts of insn rtl that might be reloaded.
2500 2014-02-15 Richard Biener <rguenther@suse.de>
2502 PR tree-optimization/60183
2503 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2504 (tree_ssa_phiprop): Calculate and free post-dominators.
2506 2014-02-14 Jeff Law <law@redhat.com>
2508 PR rtl-optimization/60131
2509 * ree.c (get_extended_src_reg): New function.
2510 (combine_reaching_defs): Use it rather than assuming location of REG.
2511 (find_and_remove_re): Verify first operand of extension is
2512 a REG before adding the insns to the copy list.
2514 2014-02-14 Roland McGrath <mcgrathr@google.com>
2516 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2517 * configure: Regenerated.
2518 * config.in: Regenerated.
2519 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2520 instead of ASM_SHORT.
2522 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2523 Richard Earnshaw <rearnsha@arm.com>
2525 PR rtl-optimization/59535
2526 * lra-constraints.c (process_alt_operands): Encourage alternative
2527 when unassigned pseudo class is superset of the alternative class.
2528 (inherit_reload_reg): Don't inherit when optimizing for code size.
2529 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2530 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2531 modes not less than 4 for Thumb1.
2533 2014-02-14 Kyle McMartin <kyle@redhat.com>
2536 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2538 2014-02-14 Richard Biener <rguenther@suse.de>
2540 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2541 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2542 Do not strip INDIRECT_REFs.
2544 2014-02-14 Richard Biener <rguenther@suse.de>
2547 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2548 DECL_FUNCTION_SPECIFIC_TARGET.
2549 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2550 * tree-streamer-out.c (pack_ts_target_option): Remove.
2551 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2552 (write_ts_function_decl_tree_pointers): Do not stream
2553 DECL_FUNCTION_SPECIFIC_TARGET.
2554 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2555 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2556 (lto_input_ts_function_decl_tree_pointers): Do not stream
2557 DECL_FUNCTION_SPECIFIC_TARGET.
2559 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2561 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2562 (get_initial_def_for_induction, vectorizable_induction): Ignore
2563 debug stmts when looking for exit_phi.
2564 (vectorizable_live_operation): Fix up condition.
2566 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2568 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2569 nreverse() because it changes the content of original tree list.
2571 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2573 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2574 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2576 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2578 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2579 GNU coding standards.
2581 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2584 * dwarf2out.c (gen_subprogram_die): Don't call
2585 add_calling_convention_attribute if subr_die is old_die.
2587 2014-02-13 Sharad Singhai <singhai@google.com>
2589 * doc/optinfo.texi: Fix order of nodes.
2591 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2593 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2594 operands[2], not operands[3].
2596 2014-02-13 Richard Biener <rguenther@suse.de>
2599 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2600 mention the possibility of an in-tree build.
2601 (CLooG): Update recommended version to 0.18.1, mention the
2602 possibility of an in-tree build and clarify that the ISL
2603 bundled with CLooG does not work.
2605 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2608 * expr.c (compress_float_constant): If x is a hard register,
2609 extend into a pseudo and then move to x.
2611 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2613 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2614 caused by bad second argument to warning_at() with -mhotpatch and
2615 nested functions (e.g. with gfortran).
2617 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2619 * opts.c (option_name): Remove "enabled by default" rider.
2621 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2623 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2625 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2626 Uros Bizjak <ubizjak@gmail.com>
2629 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2630 * configure: Regenerated.
2632 2014-02-12 Richard Biener <rguenther@suse.de>
2634 * vec.c (vec_prefix::calculate_allocation): Move as
2635 inline variant to vec.h.
2636 (vec_prefix::calculate_allocation_1): New out-of-line version.
2637 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2638 (vec_prefix::m_has_auto_buf): Rename to ...
2639 (vec_prefix::m_using_auto_storage): ... this.
2640 (vec_prefix::calculate_allocation): Inline the easy cases
2641 and dispatch to calculate_allocation_1 which doesn't need the
2643 (va_heap::reserve): Use gcc_checking_assert.
2644 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2645 m_using_auto_storage.
2646 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2648 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2649 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2650 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2652 2014-02-12 Richard Biener <rguenther@suse.de>
2654 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2655 when we found a dependence.
2657 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2659 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2661 (maybe_fold_stmt): ... into this new function.
2662 * omp-low.c (lower_omp): Update comment.
2664 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2667 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2670 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2672 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2673 identifiers in comments.
2674 (cortexa53_extra_costs): Likewise.
2675 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2676 (cortexa7_extra_costs): Likewise.
2677 (cortexa12_extra_costs): Likewise.
2678 (cortexa15_extra_costs): Likewise.
2679 (v7m_extra_costs): Likewise.
2681 2014-02-12 Richard Biener <rguenther@suse.de>
2684 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2685 of posix_memalign being successful.
2686 (lower_stmt): Restrict lowering of posix_memalign to when
2687 -ftree-bit-ccp is enabled.
2689 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2691 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2693 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2695 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2697 PR rtl-optimization/60116
2698 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2699 other_insn once the combination has been validated.
2701 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2704 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2706 * ipa-devirt.c: Include demangle.h
2707 (odr_violation_reported): New static variable.
2708 (add_type_duplicate): Update odr_violations.
2709 (maybe_record_node): Add completep parameter; update it.
2710 (record_target_from_binfo): Add COMPLETEP parameter;
2711 update it as needed.
2712 (possible_polymorphic_call_targets_1): Likewise.
2713 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2714 rename FINAL to COMPLETE.
2715 (record_targets_from_bases): Sanity check we found the binfo;
2716 fix COMPLETEP updating.
2717 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2718 parameter, fix computing of COMPLETEP.
2719 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2720 at LTO time do demangling.
2721 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2722 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2724 (gimple_get_virt_method_for_binfo): Likewise.
2725 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2726 gimple_get_virt_method_for_vtable): Update prototypes.
2728 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2731 * genautomata.c (add_presence_absence): Fix typo with
2732 {final_}presence_list.
2734 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2737 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2738 for VSX/Altivec vectors that land in GPR registers.
2740 2014-02-11 Richard Henderson <rth@redhat.com>
2741 Jakub Jelinek <jakub@redhat.com>
2744 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2745 around drhs if type conversion to lacc->type is not useless.
2747 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2749 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2751 (cortex-a57.cortex-a53): Likewise.
2752 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2754 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2756 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2759 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2761 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2762 add_options_for_arm_vfp3.
2764 2014-02-11 Jeff Law <law@redhat.com>
2767 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2768 object with an undesirable mode.
2770 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2773 * config/i386/sol2-9.h: New file.
2774 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2775 *-*-solaris2.9*): Use it.
2777 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2779 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2780 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2782 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2784 * config/microblaze/microblaze.c: Extend mcpu version format
2786 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2788 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2790 2014-02-10 Richard Henderson <rth@redhat.com>
2793 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2794 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2795 ms-abi vs -mno-accumulate-outgoing-args.
2796 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2797 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2800 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2803 * cfgexpand.c (expand_asm_operands): Attach source location to
2804 ASM_INPUT rtx objects.
2805 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2807 2014-02-10 Nick Clifton <nickc@redhat.com>
2809 * config/mn10300/mn10300.c (popcount): New function.
2810 (mn10300_expand_prologue): Include saved registers in stack usage
2813 2014-02-10 Jeff Law <law@redhat.com>
2816 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2817 when changing the SET_DEST of a prior insn to avoid an input reload.
2819 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2821 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2822 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2823 -mcall-openbsd, or -mcall-linux.
2824 (CC1_ENDIAN_BIG_SPEC): Remove.
2825 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2826 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2827 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2828 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2829 and %cc1_endian_default.
2830 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2832 2014-02-10 Richard Biener <rguenther@suse.de>
2834 PR tree-optimization/60115
2835 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2836 MEM_REF handling. Properly verify that the accesses are not
2837 out of the objects bound.
2839 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2841 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2844 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2846 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2847 proper constants and fix formatting.
2848 (possible_polymorphic_call_targets): Fix formatting.
2850 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2851 Ilya Tocar <ilya.tocar@intel.com>
2853 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2854 (_mm512_loadu_epi32): Renamed into...
2855 (_mm512_loadu_si512): This.
2856 (_mm512_storeu_epi32): Renamed into...
2857 (_mm512_storeu_si512): This.
2858 (_mm512_maskz_ceil_ps): Removed.
2859 (_mm512_maskz_ceil_pd): Ditto.
2860 (_mm512_maskz_floor_ps): Ditto.
2861 (_mm512_maskz_floor_pd): Ditto.
2862 (_mm512_floor_round_ps): Ditto.
2863 (_mm512_floor_round_pd): Ditto.
2864 (_mm512_ceil_round_ps): Ditto.
2865 (_mm512_ceil_round_pd): Ditto.
2866 (_mm512_mask_floor_round_ps): Ditto.
2867 (_mm512_mask_floor_round_pd): Ditto.
2868 (_mm512_mask_ceil_round_ps): Ditto.
2869 (_mm512_mask_ceil_round_pd): Ditto.
2870 (_mm512_maskz_floor_round_ps): Ditto.
2871 (_mm512_maskz_floor_round_pd): Ditto.
2872 (_mm512_maskz_ceil_round_ps): Ditto.
2873 (_mm512_maskz_ceil_round_pd): Ditto.
2874 (_mm512_expand_pd): Ditto.
2875 (_mm512_expand_ps): Ditto.
2876 * config/i386/i386.c (ix86_builtins): Remove
2877 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2878 (bdesc_args): Ditto.
2879 * config/i386/predicates.md (const1256_operand): New.
2880 (const_1_to_2_operand): Ditto.
2881 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2882 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2883 (*avx512pf_gatherpf<mode>sf): Ditto.
2884 (avx512pf_gatherpf<mode>df): Ditto.
2885 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2886 (*avx512pf_gatherpf<mode>df): Ditto.
2887 (avx512pf_scatterpf<mode>sf): Ditto.
2888 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2889 (*avx512pf_scatterpf<mode>sf): Ditto.
2890 (avx512pf_scatterpf<mode>df): Ditto.
2891 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2892 (*avx512pf_scatterpf<mode>df): Ditto.
2893 (avx512f_expand<mode>): Removed.
2894 (<shift_insn><mode>3<mask_name>): Change predicate type.
2896 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2898 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2899 not at the end of datarefs vector use ordered_remove to avoid
2900 reordering datarefs vector.
2903 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2904 mark local addressable non-static vars as GOVD_PRIVATE
2905 instead of GOVD_LOCAL.
2906 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2907 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2911 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2912 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2913 assume_aligned or alloc_align attributes.
2914 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
2915 arguments. Handle also assume_aligned and alloc_align attributes.
2916 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
2917 calls to functions with assume_aligned or alloc_align attributes.
2918 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
2920 2014-02-08 Terry Guo <terry.guo@arm.com>
2922 * doc/invoke.texi: Document ARM -march=armv7e-m.
2924 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2926 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
2927 flag on __cilkrts_rethrow builtin.
2930 * ipa-cp.c (determine_versionability): Fail at -O0
2931 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
2932 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
2935 2014-02-04 Jakub Jelinek <jakub@redhat.com>
2938 * tree-inline.c (copy_forbidden): Fail for
2939 __attribute__((optimize (0))) functions.
2941 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2943 * varpool.c: Include pointer-set.h.
2944 (varpool_remove_unreferenced_decls): Variables in other partitions
2945 will not be output; be however careful to not lose information
2948 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
2950 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
2951 lookup in the vtable constructor.
2953 2014-02-07 Jeff Law <law@redhat.com>
2956 * config/m68k/m68k.md (ashldi_extsi): Turn into a
2957 define_insn_and_split.
2959 * ipa-inline.c (inline_small_functions): Fix typos.
2961 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2963 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
2964 (s390_can_use_return_insn): Declare.
2965 * config/s390/s390.h (EPILOGUE_USES): Define.
2966 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
2968 (s390_chunkify_start): Handle return JUMP_LABELs.
2969 (s390_early_mach): Emit a main_pool instruction on the entry edge.
2970 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
2971 (s390_can_use_return_insn): New functions.
2972 (s390_fix_long_loop_prediction): Handle conditional returns.
2973 (TARGET_SET_UP_BY_PROLOGUE): Define.
2974 * config/s390/s390.md (ANY_RETURN): New code iterator.
2975 (*creturn, *csimple_return, return, simple_return): New patterns.
2977 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2979 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
2980 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
2981 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
2982 REG_CFA_RESTORE list when deciding not to restore a register.
2984 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2986 * config/s390/s390.c: Include tree-pass.h and context.h.
2987 (s390_early_mach): New function, split out from...
2988 (s390_emit_prologue): ...here.
2989 (pass_data_s390_early_mach): New pass structure.
2990 (pass_s390_early_mach): New class.
2991 (s390_option_override): Create and register early_mach pass.
2992 Move to end of file.
2994 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2996 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
2997 to match for the exit block.
2999 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3001 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3002 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3003 Reject misaligned operands.
3005 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3007 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3009 2014-02-07 Richard Biener <rguenther@suse.de>
3012 * gimple-low.c (lower_builtin_posix_memalign): New function.
3013 (lower_stmt): Call it to lower posix_memalign in a way
3014 to make alignment info accessible.
3016 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3019 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3020 __builtin_setjmp_receiver.
3022 2014-02-07 Richard Biener <rguenther@suse.de>
3025 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3026 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3027 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3028 Handle BUILT_IN_POSIX_MEMALIGN.
3029 (find_func_clobbers): Likewise.
3030 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3031 (call_may_clobber_ref_p_1): Likewise.
3033 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3036 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3039 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3042 * lto-cgraph.c (lto_output_node): Use
3043 symtab_get_symbol_partitioning_class.
3044 (lto_output_varpool_node): likewise.
3045 (symtab_get_symbol_partitioning_class): Move here from
3047 * cgraph.h (symbol_partitioning_class): Likewise.
3048 (symtab_get_symbol_partitioning_class): Declare.
3050 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3052 * ggc.h (ggc_internal_cleared_alloc): New macro.
3053 * vec.h (vec_safe_copy): Handle memory stats.
3054 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3055 * target-globals.c (save_target_globals): Likewise.
3057 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3060 * expr.c (emit_move_resolve_push): Export; be bit more selective
3061 on when to clear alias set.
3062 * expr.h (emit_move_resolve_push): Declare.
3063 * function.h (struct function): Add tail_call_marked.
3064 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3065 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3066 * config/i386/i386.md (TImode move expander): De not call
3068 (FP push expanders): Preserve memory attributes.
3069 * config/i386/sse.md (push<mode>1): Remove.
3070 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3071 (ix86_expand_push): Remove.
3072 * config/i386/mmx.md (push<mode>1): Remove.
3074 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3076 PR rtl-optimization/60030
3077 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3078 lopart with paradoxical subreg before shifting it up by hprec.
3080 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3082 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3083 Remove extra newline at end of file.
3084 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3085 (arm_issue_rate): Handle cortexa57.
3086 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3087 (cortex-a57.cortex-a53): Likewise.
3089 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3092 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3093 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3095 (arm_expand_prologue): Adjust all callers.
3096 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3097 info, registers also at the lowest numbered registers side. Use
3098 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3102 * var-tracking.c (adjust_mems): Before adding a SET to
3103 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3105 2014-02-06 Alan Modra <amodra@gmail.com>
3108 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3109 change SDmode to DDmode when lra_in_progress.
3111 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3114 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3115 free_data_ref on the dr first, and before goto again also set dr
3116 to the next dr. For simd_lane_access, free old datarefs[i] before
3117 overwriting it. For get_vectype_for_scalar_type failure, don't
3118 free_data_ref if simd_lane_access.
3120 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3123 * tree.h (opts_for_fn): New inline function.
3124 (opt_for_fn): Define.
3125 * config/i386/i386.c (ix86_function_regparm): Use
3126 opt_for_fn (decl, optimize) instead of optimize.
3128 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3130 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3131 for SYMBOL_REF in large memory model.
3133 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3135 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3137 (cortex-a57): Likewise.
3138 (cortex-a57.cortex-a53): Likewise.
3140 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3141 Kugan Vivekanandarajah <kuganv@linaro.org>
3143 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3145 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3147 2014-02-06 Richard Biener <rguenther@suse.de>
3149 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3150 set_loop_copy and initialize_original_copy_tables.
3152 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3154 * config/aarch64/aarch64-simd.md
3155 (aarch64_ashr_simddi): Change QI to SI.
3157 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3158 Jakub Jelinek <jakub@redhat.com>
3161 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3164 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3166 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3167 CODE_FOR_altivec_vpku[hw]um to
3168 CODE_FOR_altivec_vpku[hw]um_direct.
3169 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3170 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3171 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3172 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3174 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3176 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3177 generation for -maltivec=be.
3178 (altivec_vsumsws): Simplify redundant test.
3180 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3182 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3183 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3184 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3185 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3186 gen_altivec_vpkuwum.
3187 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3189 (altivec_vpks<VI_char>ss): Likewise.
3190 (altivec_vpks<VI_char>us): Likewise.
3191 (altivec_vpku<VI_char>us): Likewise.
3192 (altivec_vpku<VI_char>um): Likewise.
3193 (altivec_vpku<VI_char>um_direct): New (copy of
3194 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3196 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3197 target is little endian and -maltivec=be is not specified.
3198 (*altivec_vupkhs<VU_char>_direct): New (copy of
3199 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3200 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3201 target is little endian and -maltivec=be is not specified.
3202 (*altivec_vupkls<VU_char>_direct): New (copy of
3203 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3204 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3205 little endian and -maltivec=be is not specified.
3206 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3207 little endian and -maltivec=be is not specified.
3209 2014-02-05 Richard Henderson <rth@redhat.com>
3212 * combine-stack-adj.c: Revert r206943.
3213 * sched-int.h (struct deps_desc): Add last_args_size.
3214 * sched-deps.c (init_deps): Initialize it.
3215 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3216 contain REG_ARGS_SIZE notes.
3218 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3220 * lto-cgraph.c (asm_nodes_output): Make global.
3221 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3222 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3223 (driver_handle_option): Handle OPT_fwpa.
3225 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3228 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3229 a comment typo and formatting issue. If odr_hash hasn't been
3230 created, return vNULL and set *completep to false.
3233 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3234 bb with no successors.
3236 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3239 * doc/invoke.texi (-march): Clarify documentation for ARM.
3243 2014-02-05 Richard Biener <rguenther@suse.de>
3245 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3246 when not vectorizing because of too many alias checks.
3247 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3248 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3250 2014-02-05 Nick Clifton <nickc@redhat.com>
3252 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3253 accept extended registers in any mode when compiling for the MN10300.
3255 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3257 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3258 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3259 sanitization attributes.
3260 (can_inline_edge_p): Likewise.
3261 (sanitize_attrs_match_for_inline_p): New function.
3263 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3265 * ipa-prop.c (detect_type_change): Shor circuit testing of
3266 type changes on THIS pointer.
3268 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3271 * config/pa/pa.c (legitimize_tls_address): Return original address
3272 if not passed a SYMBOL_REF rtx.
3273 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3275 (pa_emit_move_sequence): Simplify TLS source operands.
3276 (pa_legitimate_constant_p): Reject all TLS constants.
3277 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3278 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3280 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3282 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3283 groups when we know they are controlled by LTO.
3284 * varasm.c (default_binds_local_p_1): If object is in other partition,
3285 it will be resolved locally.
3287 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3289 * config/host-linux.c (linux_gt_pch_use_address): Don't
3290 use SSIZE_MAX because it is not always defined.
3292 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3295 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3296 threshold for pseudo splitting.
3297 (update_ebb_live_info): Process call argument hard registers and
3298 hard registers from insn definition too.
3299 (max_small_class_regs_num): New constant.
3300 (inherit_in_ebb): Update live hard regs through EBBs. Update
3301 reloads_num only for small register classes. Don't split for
3304 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3307 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3310 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3312 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3315 2014-02-04 Marek Polacek <polacek@redhat.com>
3317 * gdbinit.in (pel): Define.
3319 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3321 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3324 2014-02-04 Richard Biener <rguenther@suse.de>
3327 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3328 in function context local.
3329 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3330 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3331 similar to LTO_imported_decl_ref.
3333 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3335 PR tree-optimization/60002
3336 * cgraphclones.c (build_function_decl_skip_args): Clear
3339 PR tree-optimization/60023
3340 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3341 false to gsi_replace.
3342 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3343 has been in some EH region and vec_stmt could throw, add
3344 vec_stmt into the same EH region.
3345 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3346 has no lhs, ignore it.
3347 * internal-fn.c (expand_MASK_LOAD): Likewise.
3350 * tree-inline.c (copy_forbidden): Fail for
3351 __attribute__((optimize (0))) functions.
3354 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3355 is set, copy one less argument.
3356 (expand_simd_clones): Don't subtract clone_info->inbranch
3357 from simd_clone_struct_alloc argument.
3359 PR rtl-optimization/57915
3360 * recog.c (simplify_while_replacing): If all unary/binary/relational
3361 operation arguments are constant, attempt to simplify those.
3364 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3365 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3367 2014-02-04 Richard Biener <rguenther@suse.de>
3369 PR tree-optimization/60012
3370 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3371 TBAA disambiguation to all DDRs.
3373 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3376 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3377 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3379 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3382 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3384 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3386 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3387 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3389 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3392 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3393 to figure out targets of polymorphic calls with known decl.
3394 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3395 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3396 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3397 (get_polymorphic_call_info): ... here.
3398 (get_polymorphic_call_info_from_invariant): New function.
3400 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3402 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3403 lookup via vtable pointer; check for type consistency
3404 and turn inconsitent facts into UNREACHABLE.
3405 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3406 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3407 type inconsistent querries; return UNREACHABLE instead.
3409 2014-02-03 Richard Henderson <rth@twiddle.net>
3412 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3413 already processed this node.
3414 (normalize_one_pred_1): Pass along mark_set.
3415 (normalize_one_pred): Create and destroy a pointer_set_t.
3416 (normalize_one_pred_chain): Likewise.
3418 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3420 PR gcov-profile/58602
3421 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3423 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3426 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3427 -fno-devirtualize; try to devirtualize by the knowledge of
3428 virtual table pointer given by aggregate propagation.
3429 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3430 (ipa_print_node_jump_functions): Dump also offset that
3431 is relevant for polymorphic calls.
3432 (determine_known_aggregate_parts): Add arg_type parameter; use it
3433 instead of determining the type from pointer type.
3434 (ipa_compute_jump_functions_for_edge): Update call of
3435 determine_known_aggregate_parts.
3436 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3437 (gimple_get_virt_method_for_binfo): ... here; simplify using
3438 vtable_pointer_value_to_vtable.
3439 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3440 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3441 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3442 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3444 (vtable_pointer_value_to_binfo): ... here.
3445 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3447 2014-02-03 Teresa Johnson <tejohnson@google.com>
3449 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3450 redef of outer loop index variable.
3452 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3456 * doc/extend.texi (Function Attributes): Typo.
3458 2014-02-03 Cong Hou <congh@google.com>
3460 PR tree-optimization/60000
3461 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3462 if the vectorized statement is a store. A store statement can only
3463 appear at the end of pattern statements.
3465 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3467 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3468 (ix86_option_override_internal): Default long double to 64-bit for
3469 32-bit Bionic and to 128-bit for 64-bit Bionic.
3471 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3472 TARGET_LONG_DOUBLE_128 is true.
3473 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3475 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3476 (mlong-double-64): Negate -mlong-double-128.
3477 (mlong-double-128): New option.
3479 * config/i386/i386-c.c (ix86_target_macros): Define
3480 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3482 * doc/invoke.texi: Document -mlong-double-128.
3484 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3486 PR rtl-optimization/60024
3487 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3489 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3491 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3493 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3495 PR rtl-optimization/57662
3496 * sel-sched.c (code_motion_path_driver): Do not mark already not
3497 existing blocks in the visiting bitmap.
3499 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3501 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3502 on the insn being emitted.
3504 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3505 Will Deacon <will.deacon@arm.com>
3507 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3509 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3511 * config/arm/arm-tables.opt: Regenerate.
3513 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3515 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3516 for vector types other than V16QImode.
3517 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3518 define_expand, and call altivec_expand_vec_perm_le when producing
3519 code with little endian element order.
3520 (*altivec_vperm_<mode>_internal): New insn having previous
3521 behavior of altivec_vperm_<mode>.
3522 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3523 altivec_expand_vec_perm_le when producing code with little endian
3525 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3526 behavior of altivec_vperm_<mode>_uns.
3528 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3530 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3531 (altivec_vsumsws): Add handling for -maltivec=be with a little
3533 (altivec_vsumsws_direct): New.
3534 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3535 gen_altivec_vsumsws.
3537 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3539 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3540 vtable_pointer_value_to_binfo): New functions.
3541 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3542 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3544 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3546 * config/nios2/nios2.md (load_got_register): Initialize GOT
3547 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3548 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3550 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3552 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3553 preserverd by passthrough, do not propagate the type.
3555 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3557 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3558 (mips_atomic_assign_expand_fenv): New function.
3559 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3561 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3563 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3564 (__builtin_mips_set_fcsr): Likewise.
3565 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3566 MIPS_USI_FTYPE_VOID.
3567 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3568 (mips16_expand_set_fcsr): Likewise.
3569 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3570 (mips16_set_fcsr_stub): Likewise.
3571 (mips16_get_fcsr_one_only_stub): New class.
3572 (mips16_set_fcsr_one_only_stub): Likewise.
3573 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3574 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3575 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3576 (hard_float): New availability predicate.
3577 (mips_builtins): Add get_fcsr and set_fcsr.
3578 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3579 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3580 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3581 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3582 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3585 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3587 * config/mips/mips.c (mips_one_only_stub): New class.
3588 (mips_need_mips16_rdhwr_p): Replace with...
3589 (mips16_rdhwr_stub): ...this new variable.
3590 (mips16_stub_call_address): New function.
3591 (mips16_rdhwr_one_only_stub): New class.
3592 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3593 (mips_output_mips16_rdhwr): Delete.
3594 (mips_finish_stub): New function.
3595 (mips_code_end): Use it to handle rdhwr stubs.
3597 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3600 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3601 when calculating size of integer atomic types.
3603 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3605 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3607 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3609 PR tree-optimization/60003
3610 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3611 * profile.c (branch_prob): Use gimple_call_builtin_p
3612 to check for BUILT_IN_SETJMP_RECEIVER.
3613 * tree-inline.c (copy_bb): Call notice_special_calls.
3615 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3618 * lra-constraints.c (process_alt_operands): Update reload_sum only
3621 2014-01-31 Richard Henderson <rth@redhat.com>
3624 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3625 until after else_eh is processed.
3627 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3629 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3630 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3631 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3632 in smmintrin.h, remove them.
3633 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3634 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3635 * config/i386/i386.md (ROUND_SAE): Fix value.
3636 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3637 (const48_operand): New.
3638 * config/i386/subst.md (round), (round_expand): Use
3639 const_4_or_8_to_11_operand.
3640 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3642 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3644 * config/i386/constraints.md (Yk): Swap meaning with k.
3645 * config/i386/i386.md (movhi_internal): Change Yk to k.
3646 (movqi_internal): Ditto.
3647 (*k<logic><mode>): Ditto.
3650 (kandn<mode>): Ditto.
3651 (*<code>hi_1): Ditto.
3652 (*<code>qi_1): Ditto.
3653 (kxnor<mode>): Ditto.
3654 (kortestzhi): Ditto.
3655 (kortestchi): Ditto.
3657 (*one_cmplhi2_1): Ditto.
3658 (*one_cmplqi2_1): Ditto.
3659 * config/i386/sse.md (): Change k to Yk.
3660 (avx512f_load<mode>_mask): Ditto.
3661 (avx512f_blendm<mode>): Ditto.
3662 (avx512f_store<mode>_mask): Ditto.
3663 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3664 (avx512f_storedqu<mode>_mask): Ditto.
3665 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3667 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3668 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3669 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3670 (avx512f_maskcmp<mode>3): Ditto.
3671 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3672 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3673 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3674 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3675 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3676 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3677 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3678 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3679 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3680 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3681 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3682 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3683 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3684 (vec_extract_lo_<mode>_maskm): Ditto.
3685 (vec_extract_hi_<mode>_maskm): Ditto.
3686 (avx512f_vternlog<mode>_mask): Ditto.
3687 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3688 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3689 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3690 (avx512f_<code>v8div16qi2_mask): Ditto.
3691 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3692 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3693 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3694 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3695 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3696 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3697 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3698 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3699 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3700 (avx512cd_maskb_vec_dupv8di): Ditto.
3701 (avx512cd_maskw_vec_dupv16si): Ditto.
3702 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3703 (avx512f_vpermi2var<mode>3_mask): Ditto.
3704 (avx512f_vpermi2var<mode>3_mask): Ditto.
3705 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3706 (*avx512f_gathersi<mode>): Ditto.
3707 (*avx512f_gathersi<mode>_2): Ditto.
3708 (*avx512f_gatherdi<mode>): Ditto.
3709 (*avx512f_gatherdi<mode>_2): Ditto.
3710 (*avx512f_scattersi<mode>): Ditto.
3711 (*avx512f_scatterdi<mode>): Ditto.
3712 (avx512f_compress<mode>_mask): Ditto.
3713 (avx512f_compressstore<mode>_mask): Ditto.
3714 (avx512f_expand<mode>_mask): Ditto.
3715 * config/i386/subst.md (mask): Change k to Yk.
3716 (mask_scalar_merge): Ditto.
3719 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3721 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3723 2014-01-31 Richard Biener <rguenther@suse.de>
3726 * builtins.c (fold_builtin_memory_op): Make sure to not
3727 use a floating-point mode or a boolean or enumeral type for
3730 2014-01-30 DJ Delorie <dj@redhat.com>
3732 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3733 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3734 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3735 whenever main() has an epilogue.
3737 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3739 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3740 unused variable "field".
3741 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3742 (vsx_mergeh_<mode>): Likewise.
3743 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3744 (altivec_vmrghh): Likewise.
3745 (altivec_vmrghw): Likewise.
3746 (altivec_vmrglb): Likewise.
3747 (altivec_vmrglh): Likewise.
3748 (altivec_vmrglw): Likewise.
3749 (altivec_vspltb): Add missing uses.
3750 (altivec_vsplth): Likewise.
3751 (altivec_vspltw): Likewise.
3752 (altivec_vspltsf): Likewise.
3754 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3757 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3758 frame related instructions.
3760 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3762 PR rtl-optimization/59959
3763 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3764 any reload of register whose subreg is invalid.
3766 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3768 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3769 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3770 Add missing return type - void.
3772 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3774 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3775 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3776 remove element index adjustment for endian (now handled in vsx.md
3778 (altivec_expand_vec_perm_const): Use
3779 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3780 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3781 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3782 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3783 define_expand and a new define_insn *altivec_vspltb_internal;
3784 adjust for -maltivec=be on a little endian target.
3785 (altivec_vspltb_direct): New.
3786 (altivec_vsplth): Divide into a define_expand and a new
3787 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3788 little endian target.
3789 (altivec_vsplth_direct): New.
3790 (altivec_vspltw): Divide into a define_expand and a new
3791 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3792 little endian target.
3793 (altivec_vspltw_direct): New.
3794 (altivec_vspltsf): Divide into a define_expand and a new
3795 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3796 a little endian target.
3798 2014-01-30 Richard Biener <rguenther@suse.de>
3800 PR tree-optimization/59993
3801 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3802 can propagate form the earlier stmt and avoid the transform
3803 when the intermediate result is needed.
3805 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3807 * README.Portability: Fix typo.
3809 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3811 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3812 comparison_operator with ordered_comparison_operator.
3814 2014-01-30 Nick Clifton <nickc@redhat.com>
3816 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3817 Rename to mn10300_store_multiple_regs.
3818 * config/mn10300/mn10300.c: Likewise.
3819 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3820 store_multiple_regs.
3821 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3822 Call mn10300_store_multiple_regs.
3824 2014-01-30 Nick Clifton <nickc@redhat.com>
3825 DJ Delorie <dj@redhat.com>
3827 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3828 %fp 2 to keep registers after it properly word-aligned.
3829 (rl78_alloc_physical_registers_umul): Handle the case where both
3830 input operands are the same.
3832 2014-01-30 Richard Biener <rguenther@suse.de>
3834 PR tree-optimization/59903
3835 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3838 2014-01-30 Jason Merrill <jason@redhat.com>
3841 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3844 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3846 2014-01-30 Richard Biener <rguenther@suse.de>
3848 PR tree-optimization/59951
3849 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3851 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3854 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3855 SFmode to DFmode case.
3857 2014-01-29 DJ Delorie <dj@redhat.com>
3859 * config/msp430/msp430.opt (-minrt): New.
3860 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3862 (ENDFILE_SPEC): Likewise.
3864 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3866 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3867 (estimate_function_body_sizes): Use it.
3869 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3872 * dwarf2out.c (is_cxx_auto): New.
3873 (is_base_type): Use it.
3874 (gen_type_die_with_usage): Likewise.
3876 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3878 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3879 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3880 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3881 -maltivec=be with LE targets.
3882 (vsx_mergeh_<mode>): Likewise.
3883 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3884 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3885 (altivec_vmrghb): Replace with define_expand and new
3886 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3887 (altivec_vmrghb_direct): New define_insn.
3888 (altivec_vmrghh): Replace with define_expand and new
3889 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3890 (altivec_vmrghh_direct): New define_insn.
3891 (altivec_vmrghw): Replace with define_expand and new
3892 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3893 (altivec_vmrghw_direct): New define_insn.
3894 (*altivec_vmrghsf): Adjust for endianness.
3895 (altivec_vmrglb): Replace with define_expand and new
3896 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3897 (altivec_vmrglb_direct): New define_insn.
3898 (altivec_vmrglh): Replace with define_expand and new
3899 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3900 (altivec_vmrglh_direct): New define_insn.
3901 (altivec_vmrglw): Replace with define_expand and new
3902 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3903 (altivec_vmrglw_direct): New define_insn.
3904 (*altivec_vmrglsf): Adjust for endianness.
3905 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3906 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3907 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3908 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3909 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3910 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3911 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3912 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3914 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
3916 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
3917 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
3920 2014-01-29 Richard Biener <rguenther@suse.de>
3922 PR tree-optimization/58742
3923 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
3924 associate_pointerplus_align.
3925 (associate_pointerplus_diff): New function.
3926 (associate_pointerplus): Likewise. Call associate_pointerplus_align
3927 and associate_pointerplus_diff.
3929 2014-01-29 Richard Biener <rguenther@suse.de>
3931 * lto-streamer.h (LTO_major_version): Bump to 3.
3932 (LTO_minor_version): Reset to 0.
3934 2014-01-29 Renlin Li <Renlin.Li@arm.com>
3936 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
3937 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
3938 (arm_file_start): Generate correct asm header for armv7ve.
3939 * config/arm/bpabi.h: Add multilib support for armv7ve.
3940 * config/arm/driver-arm.c: Change the architectures of cortex-a7
3941 and cortex-a15 to armv7ve.
3942 * config/arm/t-aprofile: Add multilib support for armv7ve.
3943 * doc/invoke.texi: Document -march=armv7ve.
3945 2014-01-29 Richard Biener <rguenther@suse.de>
3947 PR tree-optimization/58742
3948 * tree-ssa-forwprop.c (associate_plusminus): Return true
3949 if we changed sth, defer EH cleanup to ...
3950 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
3951 (simplify_mult): New function.
3953 2014-01-29 Jakub Jelinek <jakub@redhat.com>
3956 PR tree-optimization/59920
3957 * tree.c (build_common_builtin_nodes): Remove
3958 __builtin_setjmp_dispatcher initialization.
3959 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
3960 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
3961 instead of gsi_after_labels + manually skipping debug stmts.
3962 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
3963 ignore bbs with IFN_ABNORMAL_DISPATCHER.
3964 * tree-inline.c (copy_edges_for_bb): Remove
3965 can_make_abnormal_goto argument, instead add abnormal_goto_dest
3966 argument. Ignore computed_goto_p stmts. Don't call
3967 make_abnormal_goto_edges. If a call might need abnormal edges
3968 for non-local gotos, see if it already has an edge to
3969 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
3970 with true argument, don't do anything then, otherwise add
3971 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
3972 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
3974 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
3975 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
3976 (lower_stmt): Don't set data->calls_builtin_setjmp.
3977 (lower_builtin_setjmp): Adjust comment.
3978 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
3979 * tree-cfg.c (found_computed_goto): Remove.
3980 (factor_computed_gotos): Remove.
3981 (make_goto_expr_edges): Return bool, true for computed gotos.
3982 Don't call make_abnormal_goto_edges.
3983 (build_gimple_cfg): Don't set found_computed_goto, don't call
3984 factor_computed_gotos.
3985 (computed_goto_p): No longer static.
3986 (make_blocks): Don't set found_computed_goto.
3987 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
3988 (make_edges): If make_goto_expr_edges returns true, push bb
3989 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
3990 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
3991 vector. Record mapping between bbs and OpenMP regions if there
3992 are any, adjust make_gimple_omp_edges caller. Call
3993 handle_abnormal_edges.
3994 (make_abnormal_goto_edges): Remove.
3995 * tree-cfg.h (make_abnormal_goto_edges): Remove.
3996 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
3997 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
3998 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
3999 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4000 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4001 filling *region also set *region_idx to (*region)->entry->index.
4004 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4005 For REGs set ORIGINAL_REGNO.
4007 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4009 * doc/md.texi: Mention that a target shouldn't implement
4010 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4013 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4015 PR tree-optimization/59594
4016 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4017 a copy of the datarefs vector rather than the vector itself.
4019 2014-01-28 Jason Merrill <jason@redhat.com>
4022 * dwarf2out.c (auto_die): New static.
4023 (gen_type_die_with_usage): Handle C++1y 'auto'.
4024 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4027 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4030 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4031 (SPEC_X32): Likewise.
4032 (SPEC_64): Likewise.
4033 * config/i386/i386.c (ix86_option_override_internal): Turn off
4034 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4036 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4037 * config/i386/i386.h (TARGET_16BIT): New macro.
4038 (TARGET_16BIT_P): Likewise.
4039 * config/i386/i386.opt: Add m16.
4040 * doc/invoke.texi: Document -m16.
4042 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4044 PR preprocessor/59935
4045 * input.c (location_get_source_line): Bail out on when line number
4046 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4048 2014-01-28 Richard Biener <rguenther@suse.de>
4050 PR tree-optimization/58742
4051 * tree-ssa-forwprop.c (associate_plusminus): Handle
4052 pointer subtraction of the form (T)(P + A) - (T)P.
4054 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4056 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4059 2014-01-28 Richard Biener <rguenther@suse.de>
4062 2014-01-28 Richard Biener <rguenther@suse.de>
4064 PR rtl-optimization/45364
4065 PR rtl-optimization/59890
4066 * var-tracking.c (local_get_addr_clear_given_value): Handle
4067 already cleared slot.
4068 (val_reset): Handle not allocated local_get_addr_cache.
4069 (vt_find_locations): Use post-order on the inverted CFG.
4071 2014-01-28 Richard Biener <rguenther@suse.de>
4073 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4075 2014-01-28 Richard Biener <rguenther@suse.de>
4077 PR rtl-optimization/45364
4078 PR rtl-optimization/59890
4079 * var-tracking.c (local_get_addr_clear_given_value): Handle
4080 already cleared slot.
4081 (val_reset): Handle not allocated local_get_addr_cache.
4082 (vt_find_locations): Use post-order on the inverted CFG.
4084 2014-01-28 Alan Modra <amodra@gmail.com>
4086 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4087 * configure.ac <recursive call for build != host>: Define
4088 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4089 and LD_FOR_BUILD too.
4090 * configure: Regenerate.
4092 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4094 * config/i386/i386.c (get_builtin_code_for_version): Separate
4095 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4096 Broadwell from Haswell.
4098 2014-01-27 Steve Ellcey <sellcey@mips.com>
4100 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4101 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4102 * config/mips/mips.c (mips_option_override): Change setting
4104 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4105 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4106 Change from Mask to Var.
4108 2014-01-27 Jeff Law <law@redhat.com>
4110 * ipa-inline.c (inline_small_functions): Fix typo.
4112 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4114 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4115 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4116 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4117 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4118 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4119 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4120 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4121 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4122 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4123 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4124 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4125 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4126 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4127 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4128 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4129 (_mm512_storeu_epi64): Ditto.
4130 (_mm512_cmpge_epi32_mask): Ditto.
4131 (_mm512_cmpge_epu32_mask): Ditto.
4132 (_mm512_cmpge_epi64_mask): Ditto.
4133 (_mm512_cmpge_epu64_mask): Ditto.
4134 (_mm512_cmple_epi32_mask): Ditto.
4135 (_mm512_cmple_epu32_mask): Ditto.
4136 (_mm512_cmple_epi64_mask): Ditto.
4137 (_mm512_cmple_epu64_mask): Ditto.
4138 (_mm512_cmplt_epi32_mask): Ditto.
4139 (_mm512_cmplt_epu32_mask): Ditto.
4140 (_mm512_cmplt_epi64_mask): Ditto.
4141 (_mm512_cmplt_epu64_mask): Ditto.
4142 (_mm512_cmpneq_epi32_mask): Ditto.
4143 (_mm512_cmpneq_epu32_mask): Ditto.
4144 (_mm512_cmpneq_epi64_mask): Ditto.
4145 (_mm512_cmpneq_epu64_mask): Ditto.
4146 (_mm512_expand_pd): Ditto.
4147 (_mm512_expand_ps): Ditto.
4148 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4149 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4150 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4151 * config/i386/i386.c (ix86_builtins): Add
4152 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4153 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4154 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4155 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4156 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4157 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4158 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4159 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4160 IX86_BUILTIN_PMOVUSQW512_MEM.
4161 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4162 __builtin_ia32_pmovsqd512mem_mask,
4163 __builtin_ia32_pmovqd512mem_mask,
4164 __builtin_ia32_pmovusqw512mem_mask,
4165 __builtin_ia32_pmovsqw512mem_mask,
4166 __builtin_ia32_pmovqw512mem_mask,
4167 __builtin_ia32_pmovusdw512mem_mask,
4168 __builtin_ia32_pmovsdw512mem_mask,
4169 __builtin_ia32_pmovdw512mem_mask,
4170 __builtin_ia32_pmovqb512mem_mask,
4171 __builtin_ia32_pmovusqb512mem_mask,
4172 __builtin_ia32_pmovsqb512mem_mask,
4173 __builtin_ia32_pmovusdb512mem_mask,
4174 __builtin_ia32_pmovsdb512mem_mask,
4175 __builtin_ia32_pmovdb512mem_mask.
4176 (bdesc_args): Add __builtin_ia32_expanddf512,
4177 __builtin_ia32_expandsf512.
4178 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4179 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4180 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4181 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4182 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4183 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4184 (avx512f_<code>v8div16qi2_mask_store): This.
4185 (avx512f_expand<mode>): New.
4187 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4189 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4191 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4192 (_mm512_prefetch_i32scatter_pd): Ditto.
4193 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4194 (_mm512_prefetch_i64scatter_pd): Ditto.
4195 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4196 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4197 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4198 (_mm512_prefetch_i32scatter_ps): Ditto.
4199 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4200 (_mm512_prefetch_i64scatter_ps): Ditto.
4201 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4202 * config/i386/i386-builtin-types.def: Define
4203 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4204 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4205 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4206 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4207 IX86_BUILTIN_SCATTERPFQPD.
4208 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4209 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4210 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4211 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4212 __builtin_ia32_scatterpfqps.
4213 (ix86_expand_builtin): Expand new built-ins.
4214 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4215 fix memory access data type.
4216 (*avx512pf_gatherpf<mode>_mask): Ditto.
4217 (*avx512pf_gatherpf<mode>): Ditto.
4218 (avx512pf_scatterpf<mode>): Ditto.
4219 (*avx512pf_scatterpf<mode>_mask): Ditto.
4220 (*avx512pf_scatterpf<mode>): Ditto.
4221 (GATHER_SCATTER_SF_MEM_MODE): New.
4222 (avx512pf_gatherpf<mode>df): Ditto.
4223 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4224 (*avx512pf_scatterpf<mode>df): Ditto.
4226 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4229 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4230 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4233 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4235 * common/config/arm/arm-common.c
4236 (arm_rewrite_mcpu): Handle multiple names.
4238 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4240 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4242 * gimple-builder.h (create_gimple_tmp): Delete.
4244 2014-01-27 Christian Bruel <christian.bruel@st.com>
4246 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4249 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4251 * config/pa/pa.md (call): Generate indirect long calls to non-local
4252 functions when outputing 32-bit code.
4253 (call_value): Likewise except for special call to buggy powf function.
4255 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4256 portable runtime and PIC indirect calls.
4257 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4258 and PIC call sequences. Use ldo instead of blr to set return register
4259 in PIC call sequence.
4261 2014-01-25 Walter Lee <walt@tilera.com>
4263 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4264 avoid clobbering a live register.
4266 2014-01-25 Walter Lee <walt@tilera.com>
4268 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4269 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4270 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4271 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4273 2014-01-25 Walter Lee <walt@tilera.com>
4275 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4276 arguments on even registers.
4277 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4279 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4280 (BIGGEST_ALIGNMENT): Ditto.
4281 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4283 2014-01-25 Walter Lee <walt@tilera.com>
4285 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4286 insns before bundling.
4287 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4289 2014-01-25 Walter Lee <walt@tilera.com>
4291 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4292 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4293 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4295 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4297 * config/mips/constraints.md (kl): Delete.
4298 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4299 define expands, using...
4300 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4301 instructions for MIPS16.
4302 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4303 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4305 2014-01-25 Walter Lee <walt@tilera.com>
4307 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4311 2014-01-25 Walter Lee <walt@tilera.com>
4313 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4314 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4316 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4318 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4321 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4323 * print-rtl.c (in_call_function_usage): New var.
4324 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4325 EXPR_LIST mode as mode and not as reg note name.
4328 * cfgloopmanip.c (copy_loop_info): If
4329 loop->warned_aggressive_loop_optimizations, make sure
4330 the flag is set in target loop too.
4332 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4334 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4336 * builtins.def: Likewise.
4337 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4338 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4339 * ira.c (ira_setup_eliminable_regset): Likewise.
4340 * omp-low.c (gate_expand_omp): Likewise.
4341 (execute_lower_omp): Likewise.
4342 (diagnose_sb_0): Likewise.
4343 (gate_diagnose_omp_blocks): Likewise.
4344 (simd_clone_clauses_extract): Likewise.
4347 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4349 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4350 correction for little endian...
4351 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4354 2014-01-24 Jeff Law <law@redhat.com>
4356 PR tree-optimization/59919
4357 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4358 for non-returning calls.
4360 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4362 * common/config/aarch64/aarch64-common.c
4363 (aarch64_rewrite_mcpu): Handle multiple names.
4364 * config/aarch64/aarch64.h
4365 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4367 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4369 * input.c (add_file_to_cache_tab): Handle the case where fopen
4372 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4375 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4376 from push operand if code of push isn't PRE_DEC.
4378 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4381 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4382 -mquad-memory-atomic. Update -mquad-memory documentation to say
4383 it is only used for non-atomic loads/stores.
4385 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4386 -mquad-memory or -mquad-memory-atomic switches.
4388 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4389 -mquad-memory-atomic to ISA 2.07 support.
4391 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4392 to separate support of normal quad word memory operations (ldq, stq)
4393 from the atomic quad word memory operations.
4395 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4396 support to separate non-atomic quad word operations from atomic
4397 quad word operations. Disable non-atomic quad word operations in
4398 little endian mode so that we don't have to swap words after the
4399 load and before the store.
4400 (quad_load_store_p): Add comment about atomic quad word support.
4401 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4402 options printed with -mdebug=reg.
4404 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4405 -mquad-memory-atomic as the test for whether we have quad word
4406 atomic instructions.
4407 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4408 or -mp8-vector are used, allow byte/half-word atomic operations.
4410 * config/rs6000/sync.md (load_lockedti): Insure that the address
4411 is a proper indexed or indirect address for the lqarx instruction.
4412 On little endian systems, swap the hi/lo registers after the lqarx
4414 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4415 insure the address is valid for the lqarx instruction.
4416 (store_conditionalti): Insure that the address is a proper indexed
4417 or indirect address for the stqcrx. instruction. On little endian
4418 systems, swap the hi/lo registers before doing the stqcrx.
4420 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4421 insure the address is valid for the stqcrx. instruction.
4423 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4424 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4425 type of quad memory support is available.
4427 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4430 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4431 there is a danger of looping.
4433 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4435 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4436 force flag_ira_loop_pressure if set via command line.
4438 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4440 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4441 (ashr_simd): New builtin handling DI mode.
4442 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4443 (aarch64_sshr_simddi): New match pattern.
4444 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4445 (vshrd_n_s64): Likewise.
4446 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4448 2014-01-23 Nick Clifton <nickc@redhat.com>
4450 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4451 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4452 favour of mcu specific scripts.
4453 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4456 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4457 Alex Velenko <Alex.Velenko@arm.com>
4459 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4460 (vaddv_s16): Likewise.
4461 (vaddv_s32): Likewise.
4462 (vaddv_u8): Likewise.
4463 (vaddv_u16): Likewise.
4464 (vaddv_u32): Likewise.
4465 (vaddvq_s8): Likewise.
4466 (vaddvq_s16): Likewise.
4467 (vaddvq_s32): Likewise.
4468 (vaddvq_s64): Likewise.
4469 (vaddvq_u8): Likewise.
4470 (vaddvq_u16): Likewise.
4471 (vaddvq_u32): Likewise.
4472 (vaddvq_u64): Likewise.
4473 (vaddv_f32): Likewise.
4474 (vaddvq_f32): Likewise.
4475 (vaddvq_f64): Likewise.
4476 (vmaxv_f32): Likewise.
4477 (vmaxv_s8): Likewise.
4478 (vmaxv_s16): Likewise.
4479 (vmaxv_s32): Likewise.
4480 (vmaxv_u8): Likewise.
4481 (vmaxv_u16): Likewise.
4482 (vmaxv_u32): Likewise.
4483 (vmaxvq_f32): Likewise.
4484 (vmaxvq_f64): Likewise.
4485 (vmaxvq_s8): Likewise.
4486 (vmaxvq_s16): Likewise.
4487 (vmaxvq_s32): Likewise.
4488 (vmaxvq_u8): Likewise.
4489 (vmaxvq_u16): Likewise.
4490 (vmaxvq_u32): Likewise.
4491 (vmaxnmv_f32): Likewise.
4492 (vmaxnmvq_f32): Likewise.
4493 (vmaxnmvq_f64): Likewise.
4494 (vminv_f32): Likewise.
4495 (vminv_s8): Likewise.
4496 (vminv_s16): Likewise.
4497 (vminv_s32): Likewise.
4498 (vminv_u8): Likewise.
4499 (vminv_u16): Likewise.
4500 (vminv_u32): Likewise.
4501 (vminvq_f32): Likewise.
4502 (vminvq_f64): Likewise.
4503 (vminvq_s8): Likewise.
4504 (vminvq_s16): Likewise.
4505 (vminvq_s32): Likewise.
4506 (vminvq_u8): Likewise.
4507 (vminvq_u16): Likewise.
4508 (vminvq_u32): Likewise.
4509 (vminnmv_f32): Likewise.
4510 (vminnmvq_f32): Likewise.
4511 (vminnmvq_f64): Likewise.
4513 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4515 * config/aarch64/aarch64-simd.md
4516 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4517 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4518 (*aarch64_mul3_elt<mode>): Likewise.
4519 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4520 (*aarch64_mul3_elt_to_64v2df): Likewise.
4521 (*aarch64_mla_elt<mode>): Likewise.
4522 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4523 (*aarch64_mls_elt<mode>): Likewise.
4524 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4525 (*aarch64_fma4_elt<mode>): Likewise.
4526 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4527 (*aarch64_fma4_elt_to_64v2df): Likewise.
4528 (*aarch64_fnma4_elt<mode>): Likewise.
4529 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4530 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4531 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4532 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4533 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4534 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4535 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4536 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4537 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4539 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4541 * config/aarch64/aarch64-simd.md
4542 (aarch64_be_checked_get_lane<mode>): New define_expand.
4543 * config/aarch64/aarch64-simd-builtins.def
4544 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4545 New builtin definition.
4546 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4547 Use new safe be builtin.
4549 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4551 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4553 (aarch64_be_st1<mode>): Likewise.
4554 (aarch_ld1<VALL:mode>): Define_expand modified.
4555 (aarch_st1<VALL:mode>): Likewise.
4556 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4557 (UNSPEC_ST1): Likewise.
4559 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4561 * config/microblaze/microblaze.md: Add trap insn and attribute
4563 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4565 PR preprocessor/58580
4566 * input.h (location_get_source_line): Take an additional line_size
4568 (void diagnostics_file_cache_fini): Declare new function.
4569 * input.c (struct fcache): New type.
4570 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4571 New static constants.
4572 (diagnostic_file_cache_init, total_lines_num)
4573 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4574 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4575 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4576 (get_next_line, read_next_line, goto_next_line, read_line_num):
4577 New static function definitions.
4578 (diagnostic_file_cache_fini): New function.
4579 (location_get_source_line): Take an additional output line_len
4580 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4582 * diagnostic.c (diagnostic_finish): Call
4583 diagnostic_file_cache_fini.
4584 (adjust_line): Take an additional input parameter for the length
4585 of the line, rather than calculating it with strlen.
4586 (diagnostic_show_locus): Adjust the use of
4587 location_get_source_line and adjust_line with respect to their new
4588 signature. While displaying a line now, do not stop at the first
4589 null byte. Rather, display the zero byte as a space and keep
4590 going until we reach the size of the line.
4591 * Makefile.in: Add vec.o to OBJS-libcommon
4593 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4594 Ilya Tocar <ilya.tocar@intel.com>
4596 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4597 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4598 (__builtin_ia32_kmov16): Ditto.
4599 * config/i386/i386.md (UNSPEC_KMOV): New.
4602 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4604 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4605 (_mm512_storeu_si512): Ditto.
4607 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4610 * rtl.h (get_referenced_operands): Declare.
4611 * recog.c (get_referenced_operands): New function.
4612 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4613 operands have been referenced when recording LO_SUM references.
4615 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4617 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4619 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4621 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4622 Enable for generic and recent AMD targets.
4624 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4626 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4627 ARG_SIZE note when adjustment was eliminated.
4629 2014-01-22 Jeff Law <law@redhat.com>
4631 PR tree-optimization/59597
4632 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4633 in file. Accept new argument REGISTERING and use it to modify
4634 dump output appropriately.
4635 (register_jump_thread): Corresponding changes.
4636 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4637 thread paths involving joiner blocks. Add code to dump cancelled
4638 jump threading paths.
4640 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4642 PR rtl-optimization/59477
4643 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4644 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4646 2014-01-22 Tom Tromey <tromey@redhat.com>
4648 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4650 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4652 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4654 PR rtl-optimization/59896
4655 * lra-constraints.c (process_alt_operands): Check unused note for
4656 matched operands of insn with no output reloads.
4658 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4660 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4661 (mips_move_from_gpr_cost): Likewise.
4663 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4665 PR rtl-optimization/59858
4666 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4667 ira_class_hard_regs_num.
4668 (process_alt_operands): Increase reject for dying matched operand.
4670 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4673 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4674 smaller than size, perform several stores or loads and stores
4675 at dst + count - size to store or copy all of size bytes, rather
4676 than just last modesize bytes.
4678 2014-01-20 DJ Delorie <dj@redhat.com>
4680 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4681 that CLOBBERs are REGs before propogating their values.
4683 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4686 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4687 (cgraph_inline_failed_type): New function.
4688 * cgraph.h (DEFCIFCODE): Add type.
4689 (cgraph_inline_failed_type_t): New enum.
4690 (cgraph_inline_failed_type): New prototype.
4691 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4692 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4693 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4694 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4695 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4696 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4697 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4698 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4699 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4700 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4701 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4702 OPTIMIZATION_MISMATCH.
4703 * tree-inline.c (expand_call_inline): Emit errors during
4704 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4706 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4709 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4710 mode attribute in insn output.
4712 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4714 * output.h (output_constant): Delete.
4715 * varasm.c (output_constant): Make private.
4717 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4719 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4721 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4724 * tree.h (fold_builtin_strcat): New prototype.
4725 * builtins.c (fold_builtin_strcat): No longer static. Add len
4726 argument, if non-NULL, don't call c_strlen. Optimize
4727 directly into __builtin_memcpy instead of __builtin_strcpy.
4728 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4729 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4731 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4733 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4734 for SImode_address_operand operands, having only a REG argument.
4736 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4738 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4739 loader name using mbig-endian.
4740 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4742 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4744 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4748 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4750 * config/aarch64/aarch64-protos.h
4751 (aarch64_cannot_change_mode_class_ptr): Declare.
4752 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4753 aarch64_cannot_change_mode_class_ptr): New.
4754 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4755 backend hook aarch64_cannot_change_mode_class.
4757 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4759 * common/config/aarch64/aarch64-common.c
4760 (aarch64_handle_option): Don't handle any option order logic here.
4761 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4762 selected_cpu, warn on architecture version mismatch.
4763 (aarch64_override_options): Fix parsing order for option strings.
4765 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4766 Iain Sandoe <iain@codesourcery.com>
4769 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4770 warning. Amend comment to reflect current functionality.
4772 2014-01-20 Richard Biener <rguenther@suse.de>
4775 * builtins.c (fold_builtin_strcat): Remove case better handled
4776 by tree-ssa-strlen.c.
4778 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4780 * config/aarch64/aarch64.opt
4781 (mcpu, march, mtune): Make case-insensitive.
4783 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4786 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4787 if operands[1] is a REG or ZERO_EXTEND of a REG.
4789 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4791 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4793 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4795 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4796 long non-pic millicode calls.
4798 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4800 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4802 2014-01-19 Kito Cheng <kito@0xlab.org>
4804 * builtins.c (expand_movstr): Check movstr expand done or fail.
4806 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4807 H.J. Lu <hongjiu.lu@intel.com>
4810 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4811 to DImode for zero-extended addresses.
4813 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4815 PR rtl-optimization/57763
4816 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4817 on the new indirect jump_insn and increment LABEL_NUSES (label).
4819 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4823 * config.gcc (x86_archs): New variable.
4824 (x86_64_archs): Likewise.
4825 (x86_cpus): Likewise.
4826 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4827 --with-arch/--with-cpu= options.
4828 Support --with-arch=/--with-cpu={nehalem,westmere,
4829 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4831 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4833 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4834 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4836 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4838 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4840 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4843 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4844 clear cpp_get_options (parse_in)->warn_unused_macros for
4845 ix86_target_macros_internal with cpp_define.
4847 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4849 * jump.c (delete_related_insns): Keep (use (insn))s.
4850 * reorg.c (redundant_insn): Check for barriers too.
4852 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4854 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4856 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4858 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4859 call to $$dyncall when TARGET_LONG_CALLS is true.
4861 2014-01-17 Jeff Law <law@redhat.com>
4863 * ree.c (combine_set_extension): Temporarily disable test for
4864 changing number of hard registers.
4866 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4869 * ipa-inline-analysis.c (inline_free_summary):
4870 Do not free summary of aliases.
4872 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4875 * gimplify.c (gimplify_expr): Use create_tmp_var
4876 instead of create_tmp_var_raw. If cond doesn't have
4877 integral type, don't add the IFN_ANNOTATE builtin at all.
4879 2014-01-17 Martin Jambor <mjambor@suse.cz>
4882 * ipa-cp.c (prev_edge_clone): New variable.
4883 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4884 Also resize prev_edge_clone vector.
4885 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4886 (ipcp_edge_removal_hook): New function.
4887 (ipcp_driver): Register ipcp_edge_removal_hook.
4889 2014-01-17 Andrew Pinski <apinski@cavium.com>
4890 Steve Ellcey <sellcey@mips.com>
4893 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4896 2014-01-17 Jeff Law <law@redhat.com>
4899 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4900 so that pass_ccp runs first.
4902 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4904 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4905 (ix86_adjust_cost): Use !TARGET_XXX.
4906 (do_reorder_for_imul): Likewise.
4907 (swap_top_of_ready_list): Likewise.
4908 (ix86_sched_reorder): Likewise.
4910 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4912 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4913 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
4914 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
4915 (intel_memset): New. Duplicate slm_memset.
4916 (intel_cost): New. Duplicate slm_cost.
4917 (m_INTEL): New macro.
4918 (processor_target_table): Add "intel".
4919 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
4920 with PROCESSOR_INTEL for "intel".
4921 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
4922 PROCESSOR_SILVERMONT.
4923 (ix86_issue_rate): Likewise.
4924 (ix86_adjust_cost): Likewise.
4925 (ia32_multipass_dfa_lookahead): Likewise.
4926 (swap_top_of_ready_list): Likewise.
4927 (ix86_sched_reorder): Likewise.
4928 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
4929 instead of TARGET_OPT_AGU.
4930 * config/i386/i386.h (TARGET_INTEL): New.
4931 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
4932 (processor_type): Add PROCESSOR_INTEL.
4933 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
4934 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
4936 2014-01-17 Marek Polacek <polacek@redhat.com>
4939 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
4942 2014-01-17 Richard Biener <rguenther@suse.de>
4944 PR tree-optimization/46590
4945 * opts.c (default_options_table): Add entries for
4946 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
4947 all enabled at -O1 but not for -Og.
4948 * common.opt (fbranch-count-reg): Remove Init(1).
4949 (fmove-loop-invariants): Likewise.
4950 (ftree-pta): Likewise.
4952 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4954 * config/i386/i386.c (ix86_data_alignment): For compatibility with
4955 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
4956 decls to at least the GCC 4.8 used alignments.
4959 * tree-nested.c (convert_nonlocal_reference_stmt,
4960 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
4961 of GIMPLE_BIND stmts, adjust associated decls.
4963 2014-01-17 Richard Biener <rguenther@suse.de>
4965 PR tree-optimization/46590
4966 * vec.h (vec<>::bseach): New member function implementing
4967 binary search according to C89 bsearch.
4968 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
4969 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
4970 bitmap pointer again. Make accesses_in_loop a flat array.
4971 (mem_ref_obstack): New global.
4972 (outermost_indep_loop): Adjust for mem_ref->stored changes.
4973 (mark_ref_stored): Likewise.
4974 (ref_indep_loop_p_2): Likewise.
4975 (set_ref_stored_in_loop): New helper function.
4976 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
4977 (memref_free): Adjust.
4978 (record_mem_ref_loc): Simplify.
4979 (gather_mem_refs_stmt): Adjust.
4980 (sort_locs_in_loop_postorder_cmp): New function.
4981 (analyze_memory_references): Sort accesses_in_loop after
4982 loop postorder number.
4983 (find_ref_loc_in_loop_cmp): New function.
4984 (for_all_locs_in_loop): Find relevant cluster of locs in
4985 accesses_in_loop and iterate without recursion.
4986 (execute_sm): Avoid uninit warning.
4987 (struct ref_always_accessed): Simplify.
4988 (ref_always_accessed::operator ()): Likewise.
4989 (ref_always_accessed_p): Likewise.
4990 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
4991 loop postorder numbers here.
4992 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
4995 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4998 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
4999 on decls for which assemble_alias has been called.
5001 2014-01-17 Nick Clifton <nickc@redhat.com>
5003 * config/msp430/msp430.opt: (mcpu): New option.
5004 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5005 (msp430_option_override): Parse target_cpu. If the MCU name
5006 matches a generic string, clear target_mcu.
5007 (msp430_attr): Allow numeric interrupt values up to 63.
5008 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5009 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5011 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5013 * config/msp430/msp430.md (popm): Use %J rather than %I.
5014 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5015 (addhi_cy_i): Use immediate_operand for operand 2.
5016 * doc/invoke.texi: Document -mcpu option.
5018 2014-01-17 Richard Biener <rguenther@suse.de>
5020 PR rtl-optimization/38518
5021 * df.h (df_analyze_loop): Declare.
5022 * df-core.c: Include cfgloop.h.
5023 (df_analyze_1): Split out main part of df_analyze.
5024 (df_analyze): Adjust.
5025 (loop_inverted_post_order_compute): New function.
5026 (loop_post_order_compute): Likewise.
5027 (df_analyze_loop): New function avoiding whole-function
5029 * loop-invariant.c (find_defs): Use df_analyze_loop.
5030 (find_invariants): Adjust.
5031 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5033 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5035 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5036 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5038 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5040 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5041 traversal when removing references.
5043 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5046 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5048 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5051 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5052 pushing a reload for an autoinc when we had previously reloaded an
5053 inner part of the address.
5055 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5057 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5059 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5060 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5061 when not giving up or versioning for alias only because of
5063 (vect_analyze_data_ref_dependences): Set to true.
5064 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5066 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5067 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5071 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5074 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5075 operand 0 predicate for gathers, use a new pseudo as subtarget.
5077 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5080 * lra-constraints.c (process_alt_operands): Add printing debug info.
5081 Check absence of input/output reloads for matched operands too.
5083 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5085 PR rtl-optimization/59835
5086 * ira.c (ira_init_register_move_cost): Increase cost for
5089 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5091 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5093 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5096 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5097 non-register objects. Use gen_(high/low)part more consistently.
5100 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5103 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5104 endian support, remove tests for WORDS_BIG_ENDIAN.
5105 (p8_mfvsrd_3_<mode>): Likewise.
5106 (reload_gpr_from_vsx<mode>): Likewise.
5107 (reload_gpr_from_vsxsf): Likewise.
5108 (p8_mfvsrd_4_disf): Likewise.
5110 2014-01-16 Richard Biener <rguenther@suse.de>
5112 PR rtl-optimization/46590
5113 * lcm.c (compute_antinout_edge): Use postorder iteration.
5114 (compute_laterin): Use inverted postorder iteration.
5116 2014-01-16 Nick Clifton <nickc@redhat.com>
5119 * varasm.c (output_constant): Return the number of bytes actually
5121 (output_constructor_array_range): Update the field size with the
5122 number of bytes emitted by output_constant.
5123 (output_constructor_regular_field): Likewise. Also do not
5124 complain if the total number of bytes emitted is now greater
5125 than the expected fieldpos.
5126 * output.h (output_constant): Update prototype and descriptive comment.
5128 2014-01-16 Marek Polacek <polacek@redhat.com>
5131 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5132 it is error_mark_node.
5134 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5136 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5137 VALID_AVX256_REG_OR_OI_MODE.
5139 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5141 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5142 current procedure should be profiled.
5144 2014-01-15 Andrew Pinski <apinski@cavium.com>
5146 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5147 of moving from/to the STACK_REG register class.
5149 2014-01-15 Richard Henderson <rth@redhat.com>
5152 * reginfo.c (global_regs_decl): Globalize.
5153 * rtl.h (global_regs_decl): Declare.
5154 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5155 reserved via global_regs.
5157 2014-01-15 Teresa Johnson <tejohnson@google.com>
5159 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5161 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5163 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5164 and vmulosh rather than call gen_vec_widen_smult_*.
5165 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5166 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5167 (vec_widen_smult_even_v16qi): Likewise.
5168 (vec_widen_umult_even_v8hi): Likewise.
5169 (vec_widen_smult_even_v8hi): Likewise.
5170 (vec_widen_umult_odd_v16qi): Likewise.
5171 (vec_widen_smult_odd_v16qi): Likewise.
5172 (vec_widen_umult_odd_v8hi): Likewise.
5173 (vec_widen_smult_odd_v8hi): Likewise.
5174 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5175 vmuloub rather than call gen_vec_widen_umult_*.
5176 (vec_widen_umult_lo_v16qi): Likewise.
5177 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5178 vmulosb rather than call gen_vec_widen_smult_*.
5179 (vec_widen_smult_lo_v16qi): Likewise.
5180 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5181 rather than call gen_vec_widen_umult_*.
5182 (vec_widen_umult_lo_v8hi): Likewise.
5183 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5184 rather than call gen_vec_widen_smult_*.
5185 (vec_widen_smult_lo_v8hi): Likewise.
5187 2014-01-15 Jeff Law <law@redhat.com>
5189 PR tree-optimization/59747
5190 * ree.c (find_and_remove_re): Properly handle case where a second
5191 eliminated extension requires widening a copy created for elimination
5192 of a prior extension.
5193 (combine_set_extension): Ensure that the number of hard regs needed
5194 for a destination register does not change when we widen it.
5196 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5198 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5199 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5200 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5201 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5202 (avr-*-rtems*): Likewise.
5203 (bfin*-rtems*): Likewise.
5204 (moxie-*-rtems*): Likewise.
5205 (h8300-*-rtems*): Likewise.
5206 (i[34567]86-*-rtems*): Likewise.
5207 (lm32-*-rtems*): Likewise.
5208 (m32r-*-rtems*): Likewise.
5209 (m68k-*-rtems*): Likewise.
5210 (microblaze*-*-rtems*): Likewise.
5211 (mips*-*-rtems*): Likewise.
5212 (powerpc-*-rtems*): Likewise.
5213 (sh-*-rtems*): Likewise.
5214 (sparc-*-rtems*): Likewise.
5215 (sparc64-*-rtems*): Likewise.
5216 (v850-*-rtems*): Likewise.
5217 (m32c-*-rtems*): Likewise.
5219 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5221 PR rtl-optimization/59511
5222 * ira.c (ira_init_register_move_cost): Use memory costs for some
5223 cases of register move cost calculations.
5224 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5225 instead of BB frequency.
5226 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5227 * lra-assigns.c (find_hard_regno_for): Ditto.
5229 2014-01-15 Richard Biener <rguenther@suse.de>
5231 PR tree-optimization/59822
5232 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5233 (vectorizable_load): Use it to hoist defs of uses of invariant
5234 loads out of the loop.
5236 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5237 Kugan Vivekanandarajah <kuganv@linaro.org>
5240 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5243 2014-01-15 Richard Biener <rguenther@suse.de>
5245 PR rtl-optimization/59802
5246 * lcm.c (compute_available): Use inverted postorder to seed
5247 the initial worklist.
5249 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5252 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5253 ADDR_REGS for invalid symrefs in non-PIC code.
5255 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5258 * builtins.c (determine_block_size): Initialize *probable_max_size
5259 even if len_rtx is CONST_INT.
5261 2014-01-14 Andrew Pinski <apinski@cavium.com>
5263 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5264 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5265 (cortexa53_tunings): Likewise.
5266 (aarch64_sched_issue_rate): New function.
5267 (TARGET_SCHED_ISSUE_RATE): Define.
5269 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5271 * ira-costs.c (find_costs_and_classes): Add missed
5272 ira_init_register_move_cost_if_necessary.
5274 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5277 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5279 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5282 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5283 to indicate if type is used for function return value. Warn ABI
5284 change if the vector mode isn't available for function return value.
5285 (ix86_function_arg_advance): Pass false to type_natural_mode.
5286 (ix86_function_arg): Likewise.
5287 (ix86_gimplify_va_arg): Likewise.
5288 (function_arg_32): Don't warn ABI change.
5289 (ix86_function_value): Pass true to type_natural_mode.
5290 (ix86_return_in_memory): Likewise.
5291 (ix86_struct_value_rtx): Removed.
5292 (TARGET_STRUCT_VALUE_RTX): Likewise.
5294 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5296 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5297 converting a conditional jump into a conditional return.
5299 2014-01-14 Richard Biener <rguenther@suse.de>
5301 PR tree-optimization/58921
5302 PR tree-optimization/59006
5303 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5304 hoisting invariant stmts.
5305 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5306 invariant loads on the preheader edge if possible.
5308 2014-01-14 Joey Ye <joey.ye@arm.com>
5310 * doc/plugin.texi (Building GCC plugins): Update to C++.
5312 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5314 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5315 (_mm_rcp28_round_ss): Ditto.
5316 (_mm_rsqrt28_round_sd): Ditto.
5317 (_mm_rsqrt28_round_ss): Ditto.
5318 (_mm_rcp28_sd): Ditto.
5319 (_mm_rcp28_ss): Ditto.
5320 (_mm_rsqrt28_sd): Ditto.
5321 (_mm_rsqrt28_ss): Ditto.
5322 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5323 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5324 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5325 (IX86_BUILTIN_RCP28SD): Ditto.
5326 (IX86_BUILTIN_RCP28SS): Ditto.
5327 (IX86_BUILTIN_RSQRT28SD): Ditto.
5328 (IX86_BUILTIN_RSQRT28SS): Ditto.
5329 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5330 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5331 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5332 (ix86_expand_special_args_builtin): Expand new FTYPE.
5333 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5334 (srcp14<mode>): Make insn unary.
5335 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5336 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5337 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5338 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5339 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5340 Fix rounding: make it SAE only.
5341 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5343 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5345 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5346 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5347 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5348 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5349 (round_saeonly_mask_scalar_operand4): Ditto.
5350 (round_saeonly_mask_scalar_op3): Ditto.
5351 (round_saeonly_mask_scalar_op4): Ditto.
5353 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5355 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5356 Implement -maltivec=be for vec_insert and vec_extract.
5358 2014-01-10 DJ Delorie <dj@redhat.com>
5360 * config/msp430/msp430.md (call_internal): Don't allow memory
5361 references with SP as the base register.
5362 (call_value_internal): Likewise.
5363 * config/msp430/constraints.md (Yc): New. For memory references
5364 that don't use SP as a base register.
5366 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5367 "an integer without a # prefix"
5368 * config/msp430/msp430.md (epilogue_helper): Use it.
5370 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5373 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5374 AVX512F gather builtins.
5375 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5376 on gather decls with INTEGER_TYPE masktype.
5377 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5378 directly into the builtin rather than hoisting it before loop.
5380 PR tree-optimization/59387
5381 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5382 (scev_const_prop): If folded_casts and type has undefined overflow,
5383 use force_gimple_operand instead of force_gimple_operand_gsi and
5384 for each added stmt if it is assign with
5385 arith_code_with_undefined_signed_overflow, call
5386 rewrite_to_defined_overflow.
5387 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5388 gimple-fold.h instead.
5389 (arith_code_with_undefined_signed_overflow,
5390 rewrite_to_defined_overflow): Moved to ...
5391 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5392 rewrite_to_defined_overflow): ... here. No longer static.
5393 Include gimplify-me.h.
5394 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5395 rewrite_to_defined_overflow): New prototypes.
5397 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5399 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5401 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5403 * builtins.c (get_object_alignment_2): Minor tweak.
5404 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5406 2014-01-13 Christian Bruel <christian.bruel@st.com>
5408 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5409 optimized non constant lengths.
5411 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5414 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5415 load as __atomic_load_N if possible.
5417 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5419 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5421 (rs6000_expand_builtin): Adjust call.
5423 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5426 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5427 * config/rs6000/rs6000.c: Include target-globals.h.
5428 (rs6000_set_current_function): Instead of doing target_reinit
5429 unconditionally, use save_target_globals_default_opts and
5430 restore_target_globals.
5432 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5434 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5435 (rs6000_expand_builtin): Handle mffs and mtfsf.
5436 (rs6000_init_builtins): Define mffs and mtfsf.
5437 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5438 (rs6000_mffs): New pattern.
5439 (rs6000_mtfsf): New pattern.
5441 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5443 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5444 Start narrowing with START. Apply candidate-use pair
5445 and check overall cost in narrowing.
5446 (iv_ca_prune): Pass new argument.
5448 2014-01-10 Jeff Law <law@redhat.com>
5451 * ree.c (combine_reaching_defs): Ensure the defining statement
5452 occurs before the extension when optimizing extensions with
5453 different source and destination hard registers.
5455 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5458 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5459 vtables into the type inheritance graph.
5461 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5463 PR rtl-optimization/59754
5464 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5465 modes in the REGNO != REGNO case.
5467 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5469 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5471 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5473 PR tree-optimization/59745
5474 * tree-predcom.c (tree_predictive_commoning_loop): Call
5475 free_affine_expand_cache if giving up because components is NULL.
5477 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5478 GC in payload of target_globals struct instead of allocating them on
5479 the heap and the larger structs separately using GC.
5480 * target-globals.h (struct target_globals): Make regs, hard_regs,
5481 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5482 of GTY((skip)) and change type to void *.
5483 (reset_target_globals): Cast loads from those fields to corresponding
5486 2014-01-10 Steve Ellcey <sellcey@mips.com>
5489 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5490 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5491 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5493 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5496 * aarch64-modes.def (CC_Zmode): New flags mode.
5497 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5498 represents an equality.
5499 (aarch64_get_condition_code): Handle CC_Zmode.
5500 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5502 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5504 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5505 extraction in good case.
5507 2014-01-10 Richard Biener <rguenther@suse.de>
5509 PR tree-optimization/59374
5510 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5511 checking after SLP discovery. Mark stmts not participating
5512 in any SLP instance properly.
5514 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5516 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5517 when handling a SET rtx.
5519 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5521 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5522 (cortex-a57): Likewise.
5523 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5525 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5527 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5528 non-iwmmxt builtins.
5530 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5534 * ipa-devirt.c record_target_from_binfo): Take as argument
5535 stack of binfos and lookup matching one for virtual inheritance.
5536 (possible_polymorphic_call_targets_1): Update.
5538 2014-01-10 Huacai Chen <chenhc@lemote.com>
5540 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5541 kernel strings for Loongson-2E/2F/3A.
5543 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5546 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5547 is_gimple_call before calling gimple_call_internal_p.
5549 2014-01-09 Steve Ellcey <sellcey@mips.com>
5551 * Makefile.in (TREE_FLOW_H): Remove.
5552 (TREE_SSA_H): Add file names from tree-flow.h.
5553 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5554 * tree.h: Remove tree-flow.h reference.
5555 * hash-table.h: Remove tree-flow.h reference.
5556 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5557 reference with tree-ssa-loop.h.
5559 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5561 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5562 default element-order behavior for -maltivec.
5563 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5564 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5565 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5566 when targeting big endian, at least for now.
5567 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5569 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5572 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5573 var satisfies use_register_for_decl, just take into account type
5574 alignment, rather than decl alignment.
5576 PR tree-optimization/59622
5577 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5578 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5579 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5580 Don't devirtualize for inplace at all. For targets.length () == 1,
5581 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5583 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5585 * config/i386/i386.md (cpu): Remove the unused btver1.
5587 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5589 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5591 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5594 * tree-core.h (struct target_globals): New forward declaration.
5595 (struct tree_target_option): Add globals field.
5596 * tree.h (TREE_TARGET_GLOBALS): Define.
5597 (prepare_target_option_nodes_for_pch): New prototype.
5598 * target-globals.h (struct target_globals): Define even if
5600 * tree.c (prepare_target_option_node_for_pch,
5601 prepare_target_option_nodes_for_pch): New functions.
5602 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5603 * config/i386/i386.c: Include target-globals.h.
5604 (ix86_set_current_function): Instead of doing target_reinit
5605 unconditionally, use save_target_globals_default_opts and
5606 restore_target_globals.
5608 2014-01-09 Richard Biener <rguenther@suse.de>
5610 PR tree-optimization/59715
5611 * tree-cfg.h (split_critical_edges): Declare.
5612 * tree-cfg.c (split_critical_edges): Export.
5613 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5615 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5617 * cfgexpand.c (expand_stack_vars): Optionally disable
5618 asan stack protection.
5619 (expand_used_vars): Likewise.
5620 (partition_stack_vars): Likewise.
5621 * asan.c (asan_emit_stack_protection): Optionally disable
5622 after return stack usage.
5623 (instrument_derefs): Optionally disable memory access instrumentation.
5624 (instrument_builtin_call): Likewise.
5625 (instrument_strlen_call): Likewise.
5626 (asan_protect_global): Optionally disable global variables protection.
5627 * doc/invoke.texi: Added doc for new options.
5628 * params.def: Added new options.
5629 * params.h: Likewise.
5631 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5633 PR rtl-optimization/59724
5634 * ifcvt.c (cond_exec_process_if_block): Don't call
5635 flow_find_head_matching_sequence with 0 longest_match.
5636 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5637 non-active insns if !stop_after.
5638 (try_head_merge_bb): Revert 2014-01-07 changes.
5640 2014-01-08 Jeff Law <law@redhat.com>
5642 * ree.c (get_sub_rtx): New function, extracted from...
5643 (merge_def_and_ext): Here.
5644 (combine_reaching_defs): Use get_sub_rtx.
5646 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5648 * cgraph.h (varpool_variable_node): Do not choke on null node.
5650 2014-01-08 Catherine Moore <clm@codesourcery.com>
5652 * config/mips/mips.md (simple_return): Attempt to use JRC
5654 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5656 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5658 PR rtl-optimization/59137
5659 * reorg.c (steal_delay_list_from_target): Call update_block for
5661 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5663 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5665 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5666 two duplicate entries.
5668 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5671 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5673 * config/mips/mips.c (mips_truncated_op_cost): New function.
5674 (mips_rtx_costs): Adjust test for BADDU.
5675 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5677 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5679 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5680 (*baddu_si): ...this new pattern.
5682 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5685 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5687 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5690 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5692 (expand_expr, expand_normal): Adjust.
5693 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5694 inner_reference_p. Use inner_reference_p to expand inner references.
5695 (store_expr): Adjust.
5696 * cfgexpand.c (expand_call_stmt): Adjust.
5698 2014-01-08 Rong Xu <xur@google.com>
5700 * gcov-io.c (gcov_var): Move from gcov-io.h.
5701 (gcov_position): Ditto.
5702 (gcov_is_error): Ditto.
5703 (gcov_rewrite): Ditto.
5704 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5705 only part to libgcc/libgcov.h.
5707 2014-01-08 Marek Polacek <polacek@redhat.com>
5710 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5712 2014-01-08 Marek Polacek <polacek@redhat.com>
5715 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5717 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5719 PR rtl-optimization/59649
5720 * stor-layout.c (get_mode_bounds): For BImode return
5721 0 and STORE_FLAG_VALUE.
5723 2014-01-08 Richard Biener <rguenther@suse.de>
5726 * gimple.h (is_gimple_builtin_call): Remove.
5727 (gimple_builtin_call_types_compatible_p): New.
5728 (gimple_call_builtin_p): New overload.
5729 * gimple.c (is_gimple_builtin_call): Remove.
5730 (validate_call): Rename to ...
5731 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5733 (validate_type): New static function.
5734 (gimple_call_builtin_p): New overload and adjust.
5735 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5736 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5737 (gimple_fold_stmt_to_constant_1): Likewise.
5738 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5740 2014-01-08 Richard Biener <rguenther@suse.de>
5743 * gimplify.c (gimplify_expr): Gimplify register-register type
5744 VIEW_CONVERT_EXPRs to separate stmts.
5746 2014-01-07 Jeff Law <law@redhat.com>
5749 * ree.c (combine_set_extension): Handle case where source
5750 and destination registers in an extension insn are different.
5751 (combine_reaching_defs): Allow source and destination registers
5752 in extension to be different under limited circumstances.
5753 (add_removable_extension): Remove restriction that the
5754 source and destination registers in the extension are the same.
5755 (find_and_remove_re): Emit a copy from the extension's
5756 destination to its source after the defining insn if
5757 the source and destination registers are different.
5760 * ifcvt.c (merge_if_block): If we are merging a block with more than
5761 one successor with a block with no successors, remove any BARRIER
5762 after the second block.
5764 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5766 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5768 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5771 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5772 for 14-bit register offsets when INT14_OK_STRICT is false.
5774 2014-01-07 Roland Stigge <stigge@antcom.de>
5775 Michael Meissner <meissner@linux.vnet.ibm.com>
5778 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5779 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5781 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5783 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5786 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5788 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5789 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5790 rtx is const0_rtx or not.
5792 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5795 * target-globals.c (save_target_globals): Remove this_fn_optab
5797 * toplev.c: Include optabs.h.
5798 (target_reinit): Temporarily restore the global options if another
5799 set of options are in force.
5801 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5803 PR rtl-optimization/58668
5804 * cfgcleanup.c (flow_find_cross_jump): Don't count
5805 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5806 to determine what is counted.
5807 (flow_find_head_matching_sequence): Use active_insn_p to determine
5809 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5811 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5812 determine what is counted.
5814 PR tree-optimization/59643
5815 * tree-predcom.c (split_data_refs_to_components): If one dr is
5816 read and one write, determine_offset fails and the write isn't
5817 in the bad component, just put the read into the bad component.
5819 2014-01-07 Mike Stump <mikestump@comcast.net>
5820 Jakub Jelinek <jakub@redhat.com>
5823 * tree-core.h (struct tree_optimization_option): Change optabs
5824 type from unsigned char * to void *.
5825 * optabs.c (init_tree_optimization_optabs): Adjust
5826 TREE_OPTIMIZATION_OPTABS initialization.
5828 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5831 * config/i386/i386.h (struct machine_function): Add
5832 no_drap_save_restore field.
5833 * config/i386/i386.c (ix86_save_reg): Use
5834 !cfun->machine->no_drap_save_restore instead of
5835 crtl->stack_realign_needed.
5836 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5837 this function clears frame_pointer_needed. Set
5838 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5839 and DRAP reg is needed.
5841 2014-01-06 Marek Polacek <polacek@redhat.com>
5844 * doc/implement-c.texi: Mention that other integer types are
5845 permitted as bit-field types in strictly conforming mode.
5847 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5849 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5852 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5854 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5856 2014-01-06 Martin Jambor <mjambor@suse.cz>
5859 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5861 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5863 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5867 * var-tracking.c (add_stores): Preserve the value of the source even if
5868 we don't record the store.
5870 2014-01-06 Terry Guo <terry.guo@arm.com>
5872 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5874 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5877 * config/darwin.c (darwin_function_section): Adjust return values to
5878 correspond to optimisation changes made in r206070.
5880 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5882 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5883 from prefetch_block tune setting.
5884 (nocona_cost): Correct size of prefetch block to 64.
5886 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5888 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5889 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5890 used to save the static chain register in the computation of the offset
5891 from which the FP registers need to be restored.
5893 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5895 PR tree-optimization/59519
5896 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5897 ICE if get_current_def (current_new_name) is already non-NULL, as long
5898 as it is a phi result of some other phi in *new_exit_bb that has
5901 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5902 or vmovdqu* for misaligned_operand.
5903 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5904 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5905 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5906 aligned_mem for AVX512F masked aligned load and store builtins and for
5909 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5911 PR tree-optimization/59651
5912 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5913 Address range for negative step should be added by TYPE_SIZE_UNIT.
5915 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
5917 * config/m68k/m68k.c (handle_move_double): Handle pushes with
5918 overlapping registers also for registers other than the stack pointer.
5920 2014-01-03 Marek Polacek <polacek@redhat.com>
5923 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
5926 2014-01-03 Jakub Jelinek <jakub@redhat.com>
5929 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
5932 * config/i386/i386.md (MODE_SIZE): New mode attribute.
5933 (push splitter): Use <P:MODE_SIZE> instead of
5934 GET_MODE_SIZE (<P:MODE>mode).
5935 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
5936 (mov -1, reg peephole2): Likewise.
5937 * config/i386/sse.md (*mov<mode>_internal,
5938 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
5939 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
5940 *<code><mode>3, *andnot<mode>3<mask_name>,
5941 <mask_codefor><code><mode>3<mask_name>): Likewise.
5942 * config/i386/subst.md (mask_mode512bit_condition,
5943 sd_mask_mode512bit_condition): Likewise.
5945 2014-01-02 Xinliang David Li <davidxl@google.com>
5947 PR tree-optimization/59303
5948 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
5949 (dump_predicates): Better output format.
5950 (pred_equal_p): New function.
5951 (is_neq_relop_p): Ditto.
5952 (is_neq_zero_form_p): Ditto.
5953 (pred_expr_equal_p): Ditto.
5954 (pred_neg_p): Ditto.
5955 (simplify_pred): Ditto.
5956 (simplify_preds_2): Ditto.
5957 (simplify_preds_3): Ditto.
5958 (simplify_preds_4): Ditto.
5959 (simplify_preds): Ditto.
5961 (push_to_worklist): Ditto.
5962 (get_pred_info_from_cmp): Ditto.
5963 (is_degenerated_phi): Ditto.
5964 (normalize_one_pred_1): Ditto.
5965 (normalize_one_pred): Ditto.
5966 (normalize_one_pred_chain): Ditto.
5967 (normalize_preds): Ditto.
5968 (normalize_cond_1): Remove function.
5969 (normalize_cond): Ditto.
5970 (is_gcond_subset_of): Ditto.
5971 (is_subset_of_any): Ditto.
5972 (is_or_set_subset_of): Ditto.
5973 (is_and_set_subset_of): Ditto.
5974 (is_norm_cond_subset_of): Ditto.
5975 (pred_chain_length_cmp): Ditto.
5976 (convert_control_dep_chain_into_preds): Type change.
5977 (find_predicates): Ditto.
5978 (find_def_preds): Ditto.
5979 (destroy_predicates_vecs): Ditto.
5980 (find_matching_predicates_in_rest_chains): Ditto.
5981 (use_pred_not_overlap_with_undef_path_pred): Ditto.
5982 (is_pred_expr_subset): Ditto.
5983 (is_pred_chain_subset_of): Ditto.
5984 (is_included_in): Ditto.
5985 (is_superset_of): Ditto.
5987 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5989 Update copyright years.
5991 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
5993 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
5994 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
5995 config/arc/arc.md, config/arc/arc.opt,
5996 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
5997 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
5998 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
5999 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6000 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6001 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6002 vtable-verify.c, vtable-verify.h: Use the standard form for the
6005 2014-01-02 Tobias Burnus <burnus@net-b.de>
6007 * gcc.c (process_command): Update copyright notice dates.
6008 * gcov-dump.c: Ditto.
6010 * doc/cpp.texi: Bump @copying's copyright year.
6011 * doc/cppinternals.texi: Ditto.
6012 * doc/gcc.texi: Ditto.
6013 * doc/gccint.texi: Ditto.
6014 * doc/gcov.texi: Ditto.
6015 * doc/install.texi: Ditto.
6016 * doc/invoke.texi: Ditto.
6018 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6020 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6022 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6024 * config/i386/sse.md (*mov<mode>_internal): Guard
6025 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6027 PR rtl-optimization/59647
6028 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6029 new_rtx into UNSIGNED_FLOAT rtxes.
6031 Copyright (C) 2014 Free Software Foundation, Inc.
6033 Copying and distribution of this file, with or without modification,
6034 are permitted in any medium without royalty provided the copyright
6035 notice and this notice are preserved.