1 2014-04-07 Charles Baylis <charles.baylis@linaro.org>
4 * config/arm/arm.h (ASM_OUTPUT_CASE_END): Remove.
5 (LABEL_ALIGN_AFTER_BARRIER): Align barriers which occur after
8 2014-04-07 Richard Biener <rguenther@suse.de>
10 PR tree-optimization/60766
11 * tree-ssa-loop-ivopts.c (cand_value_at): Compute in an
13 (may_eliminate_iv): Convert cand_value_at result to desired
16 2014-04-07 Jason Merrill <jason@redhat.com>
19 * common.opt (-fno-gnu-unique): Add.
20 * config/elfos.h (USE_GNU_UNIQUE_OBJECT): Check it.
22 2014-04-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24 * haifa-sched.c: Fix outdated function reference and minor
25 grammar errors in introductory comment.
27 2014-04-07 Richard Biener <rguenther@suse.de>
30 * tree-ssa-operands.c (maybe_add_call_vops): Also add VDEFs
32 * tree-cfgcleanup.c (fixup_noreturn_call): Do not remove VDEFs.
34 2014-04-06 John David Anglin <danglin@gcc.gnu.org>
37 * config/pa/pa.c (pa_output_function_epilogue): Skip address and code
38 size accounting for thunks.
39 (pa_asm_output_mi_thunk): Use final_start_function() and
40 final_end_function() to output function start and end directives.
42 2014-04-05 Pitchumani Sivanupandi <Pitchumani.S@atmel.com>
44 * config/avr/avr-arch.h (avr_mcu_t): Add dev_attribute field to have device
45 specific ISA/ feature information. Remove short_sp and errata_skip ds.
46 Add avr_device_specific_features enum to have device specific info.
47 * config/avr/avr-c.c (avr_cpu_cpp_builtins): use dev_attribute to check
48 errata_skip. Add __AVR_ISA_RMW__ builtin macro if RMW ISA available.
49 * config/avr/avr-devices.c (avr_mcu_types): Update AVR_MCU macro for
50 updated device specific info.
51 * config/avr/avr-mcus.def: Merge device specific details to
53 * config/avr/avr.c (avr_2word_insn_p): use dev_attribute field to check
55 * config/avr/avr.h (AVR_HAVE_8BIT_SP): same for short sp info.
56 * config/avr/driver-avr.c (avr_device_to_as): Pass -mrmw option to
57 assembler if RMW isa supported by current device.
58 * config/avr/genmultilib.awk: Update as device info structure changed.
59 * doc/invoke.texi: Add info for __AVR_ISA_RMW__ builtin macro
61 2014-04-04 Cong Hou <congh@google.com>
63 PR tree-optimization/60656
64 * tree-vect-stmts.c (supportable_widening_operation):
65 Fix a bug that elements in a vector with vect_used_by_reduction
66 property are incorrectly reordered when the operation on it is not
67 consistant with the one in reduction operation.
69 2014-04-04 John David Anglin <danglin@gcc.gnu.org>
71 PR rtl-optimization/60155
72 * gcse.c (record_set_data): New function.
73 (single_set_gcse): New function.
74 (gcse_emit_move_after): Use single_set_gcse instead of single_set.
75 (hoist_code): Likewise.
76 (get_pressure_class_and_nregs): Likewise.
78 2014-04-04 Eric Botcazou <ebotcazou@adacore.com>
80 * explow.c (probe_stack_range): Emit a final optimization blockage.
82 2014-04-04 Anthony Green <green@moxielogic.com>
84 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2): Fix
87 2014-04-04 Jan Hubicka <hubicka@ucw.cz>
90 * lto-cgraph.c (input_overwrite_node): Check that partitioning
91 flags are set only during streaming.
92 * ipa.c (process_references, walk_polymorphic_call_targets,
93 symtab_remove_unreachable_nodes): Drop bodies of always inline
95 (symtab_remove_unreachable_nodes): Remove always_inline attribute.
97 2014-04-04 Jakub Jelinek <jakub@redhat.com>
98 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
101 * dwarf2out.c (const_ok_for_output_1): Reject expressions
104 2014-04-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
107 * config/arm/cortex-a53.md (cortex_a53_fdivs): Reduce reservation
109 (cortex_a53_fdivd): Likewise.
111 2014-04-04 Martin Jambor <mjambor@suse.cz>
114 * cgraph.h (cgraph_clone_node): New parameter added to declaration.
116 * cgraph.c (clone_of_p): Also return true if thunks match.
117 (verify_edge_corresponds_to_fndecl): Removed extraneous call to
118 cgraph_function_or_thunk_node and an obsolete comment.
119 * cgraphclones.c (build_function_type_skip_args): Moved upwards in the
121 (build_function_decl_skip_args): Likewise.
122 (set_new_clone_decl_and_node_flags): New function.
123 (duplicate_thunk_for_node): Likewise.
124 (redirect_edge_duplicating_thunks): Likewise.
125 (cgraph_clone_node): New parameter args_to_skip, pass it to
126 redirect_edge_duplicating_thunks which is called instead of
127 cgraph_redirect_edge_callee.
128 (cgraph_create_virtual_clone): Pass args_to_skip to cgraph_clone_node,
129 moved setting of a lot of flags to set_new_clone_decl_and_node_flags.
131 2014-04-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
134 * fortran/trans-types.c (gfc_get_function_type): In case of recursion
135 build a variadic function type with empty argument list instead of a
136 stdarg-like function type with incomplete argument list.
138 2014-04-04 Jeff Law <law@redhat.com>
141 * config/arm/predicates.md (const_int_I_operand): New predicate.
142 (const_int_M_operand): Similarly.
143 * config/arm/arm.md (insv_zero): Use const_int_M_operand instead of
145 (insv_t2, extv_reg, extzv_t2): Likewise.
146 (load_multiple_with_writeback): Similarly for const_int_I_operand.
147 (pop_multiple_with_writeback_and_return): Likewise.
148 (vfp_pop_multiple_with_writeback): Likewise
150 2014-04-04 Richard Biener <rguenther@suse.de>
153 * tree-ssanames.c (make_ssa_name_fn): Fix assert.
154 * gimple.c (gimple_set_bb): Avoid ICEing for NULL cfun for
156 * gimplify.h (gimple_add_tmp_var_fn): Declare.
157 * gimplify.c (gimple_add_tmp_var_fn): New function.
158 * gimple-expr.h (create_tmp_reg_fn): Declare.
159 * gimple-expr.c (create_tmp_reg_fn): New function.
160 * gimple-low.c (record_vars_into): Don't change cfun.
161 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix
162 code generation without cfun.
164 2014-04-04 Thomas Schwinge <thomas@codesourcery.com>
167 * Makefile.in (install-driver): Fix shell scripting.
169 2014-04-03 Cong Hou <congh@google.com>
171 PR tree-optimization/60505
172 * tree-vectorizer.h (struct _stmt_vec_info): Add th field as the
173 threshold of number of iterations below which no vectorization will be
175 * tree-vect-loop.c (new_loop_vec_info):
176 Initialize LOOP_VINFO_COST_MODEL_THRESHOLD.
177 * tree-vect-loop.c (vect_analyze_loop_operations):
178 Set LOOP_VINFO_COST_MODEL_THRESHOLD.
179 * tree-vect-loop.c (vect_transform_loop):
180 Use LOOP_VINFO_COST_MODEL_THRESHOLD.
181 * tree-vect-loop.c (vect_analyze_loop_2): Check the maximum number
182 of iterations of the loop and see if we should build the epilogue.
184 2014-04-03 Richard Biener <rguenther@suse.de>
186 * tree-streamer.h (struct streamer_tree_cache_d): Add next_idx
188 (streamer_tree_cache_create): Adjust.
189 * tree-streamer.c (streamer_tree_cache_add_to_node_array): Adjust
190 to allow optional nodes array.
191 (streamer_tree_cache_insert_1): Use next_idx to assign idx.
192 (streamer_tree_cache_append): Likewise.
193 (streamer_tree_cache_create): Create nodes array optionally
194 as specified by parameter.
195 * lto-streamer-out.c (create_output_block): Avoid maintaining
196 the node array in the writer cache.
197 (DFS_write_tree): Remove assertion.
198 (produce_asm_for_decls): Free the out decl state hash table
200 * lto-streamer-in.c (lto_data_in_create): Adjust for
201 streamer_tree_cache_create prototype change.
203 2014-04-03 Richard Biener <rguenther@suse.de>
205 * tree-streamer-out.c (streamer_write_chain): Do not temporarily
206 set TREE_CHAIN to NULL_TREE.
208 2014-04-03 Richard Biener <rguenther@suse.de>
210 PR tree-optimization/60740
211 * graphite-scop-detection.c (stmt_simple_for_scop_p): Iterate
212 over all GIMPLE_COND operands.
214 2014-04-03 Nathan Sidwell <nathan@codesourcery.com>
216 * doc/invoke.texi (Wnon-virtual-dtor): Adjust documentation.
217 (Weffc++): Remove Scott's numbering, merge lists and reference
223 * class.c (accessible_nvdtor_p): New.
224 (check_bases): Don't check base destructor here ...
225 (check_bases_and_members): ... check them here. Trigger on
226 Wnon-virtual-dtor flag.
227 (finish_struct_1): Use accessible_nvdtor_p.
230 * g++.dg/warn/Wnvdtor.C: Add non-polymorphic case.
231 * g++.dg/warn/Wnvdtor-2.C: New.
232 * g++.dg/warn/Wnvdtor-3.C: New.
233 * g++.dg/warn/Wnvdtor-4.C: New.
234 * g++.dg/warn/Weff1.C: Delete.
235 * g++.old-deja/g++.benjamin/15309-1.C: Delete.
236 * g++.old-deja/g++.benjamin/15309-2.C: Delete.
238 2014-04-03 Nick Clifton <nickc@redhat.com>
240 * config/rl78/rl78-expand.md (movqi): Handle (SUBREG (SYMBOL_REF))
243 2014-04-03 Martin Jambor <mjambor@suse.cz>
245 * ipa-cp.c (ipcp_verify_propagated_values): Also dump symtab and
246 mention gcc_unreachable before failing.
247 * ipa.c (symtab_remove_unreachable_nodes): Also print order of
250 2014-04-02 Jan Hubicka <hubicka@ucw.cz>
253 * ipa-devirt.c (get_polymorphic_call_info): Do not ICE on type inconsistent
254 code and instead mark the context inconsistent.
255 (possible_polymorphic_call_targets): For inconsistent contexts
256 return empty complete list.
258 2014-04-02 Anthony Green <green@moxielogic.com>
260 * config/moxie/moxie.md (zero_extendqisi2, zero_extendhisi2)
261 (extendqisi2, extendhisi2): Define.
262 * config/moxie/moxie.h (DEFAULT_SIGNED_CHAR): Change to 0.
263 (WCHAR_TYPE): Change to unsigned int.
265 2014-04-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
267 PR tree-optimization/60733
268 * gimple-ssa-strength-reduction.c (ncd_with_phi): Change required
269 insertion point for PHI candidates to be the end of the feeding
270 block for the PHI argument.
272 2014-04-02 Vladimir Makarov <vmakarov@redhat.com>
274 PR rtl-optimization/60650
275 * lra-constraints.c (process_alt_operands): Decrease reject for
276 earlyclobber matching.
278 2014-04-02 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
280 * config/s390/s390.c (s390_expand_insv): Use GET_MODE_BITSIZE.
282 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
284 * config/spu/spu.c (pad_bb): Do not crash when the last
285 insn is CODE_FOR_blockage.
287 2014-04-02 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
289 * config/spu/spu.md ("insv"): Fail if bitoffset+bitsize
290 lies outside the target mode.
292 2014-04-02 Michael Meissner <meissner@linux.vnet.ibm.com>
295 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): If we have
296 software floating point or no floating point registers, do not
297 allow any type in the FPRs. Eliminate a test for SPE SIMD types
298 in GPRs that occurs after we tested for GPRs that would never be
301 * config/rs6000/rs6000.md (mov<mode>_softfloat32, FMOVE64):
302 Rewrite tests to use TARGET_DOUBLE_FLOAT and TARGET_E500_DOUBLE,
303 since the FMOVE64 type is DFmode/DDmode. If TARGET_E500_DOUBLE,
304 specifically allow DDmode, since that does not use the SPE SIMD
307 2014-04-02 Richard Biener <rguenther@suse.de>
310 * optabs.c (expand_abs_nojump): Honor flag_trapv only for
311 MODE_INTs. Properly use negv_optab.
312 (expand_abs): Likewise.
314 2014-04-02 Richard Biener <rguenther@suse.de>
317 * Makefile.in (install-driver): Guard extra installs with special
320 2014-04-01 Michael Meissner <meissner@linux.vnet.ibm.com>
322 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
325 2014-04-01 Richard Henderson <rth@redhat.com>
328 * config/i386/i386.md (*float<SWI48><MODEF>2_sse): Leave the second
329 alternative enabled before register allocation.
331 2014-04-01 Chung-Lin Tang <cltang@codesourcery.com>
333 * config/nios2/nios2.md (unspec): Remove UNSPEC_TLS, UNSPEC_TLS_LDM.
334 * config/nios2/nios2.c (nios2_function_profiler): Fix addi operand
336 (nios2_large_got_address): Remove unneeded 'sym' parameter.
337 (nios2_got_address): Update nios2_large_got_address call site.
338 (nios2_delegitimize_address): New function.
339 (TARGET_DELEGITIMIZE_ADDRESS): Define to nios2_delegitimize_address.
340 * config/nios2/linux.h (GLIBC_DYNAMIC_LINKER): Define.
341 (LINK_SPEC): Specify dynamic linker using GNU_USER_DYNAMIC_LINKER.
343 2014-04-01 Martin Husemann <martin@duskware.de>
345 * config/mips/netbsd.h (TARGET_OS_CPP_BUILTINS): Define __mips_o32
348 2014-04-01 Richard Sandiford <rdsandiford@googlemail.com>
350 PR rtl-optimization/60604
351 * recog.c (general_operand): Incorporate REG_CANNOT_CHANGE_MODE_P
352 check from register_operand.
353 (register_operand): Redefine in terms of general_operand.
354 (nonmemory_operand): Use register_operand for the non-constant cases.
356 2014-04-01 Richard Biener <rguenther@suse.de>
358 * gimple.h (struct gimple_statement_base): Align subcode to
361 2014-04-01 Sebastian Huber <sebastian.huber@embedded-brains.de>
363 * doc/invoke.texi (mapp-regs): Clarify.
365 2014-03-31 Ulrich Drepper <drepper@gmail.com>
367 * config/i386/avx512fintrin.h (__v32hi): Define type.
369 (_mm512_set1_epi8): Define.
370 (_mm512_set1_epi16): Define.
371 (_mm512_set4_epi32): Define.
372 (_mm512_set4_epi64): Define.
373 (_mm512_set4_pd): Define.
374 (_mm512_set4_ps): Define.
375 (_mm512_setr4_epi64): Define.
376 (_mm512_setr4_epi32): Define.
377 (_mm512_setr4_pd): Define.
378 (_mm512_setr4_ps): Define.
379 (_mm512_setzero_epi32): Define.
381 2014-03-31 Martin Jambor <mjambor@suse.cz>
384 * tree-sra.c (callsite_has_enough_arguments_p): Renamed to
385 callsite_arguments_match_p. Updated all callers. Also check types of
386 corresponding formal parameters and actual arguments.
387 (not_all_callers_have_enough_arguments_p) Renamed to
388 some_callers_have_mismatched_arguments_p.
390 2014-03-31 Yuri Rumyantsev <ysrumyan@gmail.com>
392 * tree-inline.c (copy_loops): Add missed copy of 'safelen'.
394 2014-03-31 Kugan Vivekanandarajah <kuganv@linaro.org>
397 * aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
400 2014-03-30 Uros Bizjak <ubizjak@gmail.com>
402 * config/i386/sse.md (FMAMODE_NOVF512): New mode iterator.
403 (<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>):
405 <sd_mask_codefor>fma_fmadd_<VF_512:mode><sd_maskz_name><round_name>.
406 Use FMAMODE_NOVF512 mode iterator.
407 (<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>): Ditto.
408 (<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>): Ditto.
409 (<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>): Ditto.
410 (<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>):
412 <sd_mask_codefor>fma_fmaddsub_<VF_512:mode><sd_maskz_name><round_name>.
413 Use VF_128_256 mode iterator.
414 (<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>):
417 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
419 * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Clear
420 static chain if needed.
422 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
425 * lra-constraints.c (index_part_to_reg): New.
426 (process_address): Use it.
428 2014-03-27 Jeff Law <law@redhat.com>
429 Jakub Jelinek <jakub@redhat.com>
432 * expr.c (do_tablejump): Use simplify_gen_binary rather than
433 gen_rtx_{PLUS,MULT} to build up the address expression.
435 * i386/i386.c (ix86_legitimize_address): Use copy_addr_to_reg to avoid
436 creating non-canonical RTL.
438 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
441 * ipa-inline.c (want_inline_small_function_p): Short circuit large
442 functions; reorganize to make cheap checks first.
443 (inline_small_functions): Do not estimate growth when dumping;
445 * ipa-inline.h (inline_summary): Add min_size.
446 (growth_likely_positive): New function.
447 * ipa-inline-analysis.c (dump_inline_summary): Add min_size.
448 (set_cond_stmt_execution_predicate): Cleanup.
449 (estimate_edge_size_and_time): Compute min_size.
450 (estimate_calls_size_and_time): Likewise.
451 (estimate_node_size_and_time): Likewise.
452 (inline_update_overall_summary): Update min_size.
453 (do_estimate_edge_time): Likewise.
454 (do_estimate_edge_size): Update.
455 (do_estimate_edge_hints): Update.
456 (growth_likely_positive): New function.
458 2014-03-28 Jakub Jelinek <jakub@redhat.com>
461 * config/i386/i386.c (ix86_copy_addr_to_reg): Call copy_addr_to_reg
462 also if addr has VOIDmode.
464 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
466 * config/arm/aarch-common.c (aarch_crypto_can_dual_issue): New.
467 * config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
469 * config/arm/cortex-a53.md: Add reservations and bypass for crypto
470 instructions as well as AdvancedSIMD loads.
472 2014-03-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
474 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
475 Use crypto_aese type.
476 (aarch64_crypto_aes<aesmc_op>v16qi): Use crypto_aesmc type.
477 * config/arm/arm.md (is_neon_type): Replace crypto_aes with
478 crypto_aese, crypto_aesmc. Move to types.md.
479 * config/arm/types.md (crypto_aes): Split into crypto_aese,
481 * config/arm/iterators.md (crypto_type): Likewise.
483 2014-03-28 Jan Hubicka <hubicka@ucw.cz>
485 * cgraph.c: Include expr.h and tree-dfa.h.
486 (cgraph_redirect_edge_call_stmt_to_callee): If call in noreturn;
489 2014-03-28 Vladimir Makarov <vmakarov@redhat.com>
492 * lra-assigns.c (find_hard_regno_for): Remove unavailable hard
493 regs from checking multi-reg pseudos.
495 2014-03-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
497 * config/arm/t-aprofile (MULTILIB_MATCHES): Correct A12 rule.
499 2014-03-28 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
501 * config/rs6000/rs6000.c (fusion_gpr_load_p): Refuse optimization
502 if it would clobber the stack pointer, even temporarily.
504 2014-03-28 Eric Botcazou <ebotcazou@adacore.com>
506 * mode-switching.c: Make small adjustments to the top comment.
508 2014-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
510 * config/rs6000/constraints.md (wD constraint): New constraint to
511 match the constant integer to get the top DImode/DFmode out of a
512 vector in a VSX register.
514 * config/rs6000/predicates.md (vsx_scalar_64bit): New predicate to
515 match the constant integer to get the top DImode/DFmode out of a
516 vector in a VSX register.
518 * config/rs6000/rs6000-builtins.def (VBPERMQ): Add vbpermq builtin
521 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
524 * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
525 -mdebug=reg, print value of VECTOR_ELEMENT_SCALAR_64BIT.
527 * config/rs6000/vsx.md (vsx_extract_<mode>, V2DI/V2DF modes):
528 Optimize vec_extract of 64-bit values, where the value being
529 extracted is in the top word, where we can use scalar
530 instructions. Add direct move and store support. Combine the big
531 endian/little endian vector select load support into a single insn.
532 (vsx_extract_<mode>_internal1): Likewise.
533 (vsx_extract_<mode>_internal2): Likewise.
534 (vsx_extract_<mode>_load): Likewise.
535 (vsx_extract_<mode>_store): Likewise.
536 (vsx_extract_<mode>_zero): Delete, big and little endian insns are
537 combined into vsx_extract_<mode>_load.
538 (vsx_extract_<mode>_one_le): Likewise.
540 * config/rs6000/rs6000.h (VECTOR_ELEMENT_SCALAR_64BIT): Macro to
541 define the top 64-bit vector element.
543 * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wD
546 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
547 Document vec_vbpermq builtin.
550 * config/rs6000/altivec.h (vec_xxsldwi): Add missing define to
551 enable use of xxsldwi and xxpermdi builtin functions.
552 (vec_xxpermdi): Likewise.
554 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
555 Document use of vec_xxsldwi and vec_xxpermdi builtins.
557 2014-03-27 Vladimir Makarov <vmakarov@redhat.com>
559 PR rtl-optimization/60650
560 * lra-assign.c (find_hard_regno_for, spill_for): Add parameter
562 (find_spills_for): New.
563 (assign_by_spills): Pass the new parameter to find_hard_regno_for.
564 Spill all pseudos on the second iteration.
566 2014-03-27 Marek Polacek <polacek@redhat.com>
569 * doc/extend.texi (ffs Builtins): Change unsigned types to signed
572 2014-03-27 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
574 * config/s390/s390.c (s390_can_use_return_insn): Check for
575 call-saved FPRs on 31 bit.
577 2014-03-27 Jakub Jelinek <jakub@redhat.com>
580 * omp-low.c (lower_omp_1): For gimple_clobber_p stmts,
581 if they need regimplification, just drop them instead of
582 calling gimple_regimplify_operands on them.
584 2014-03-27 Marcus Shawcroft <marcus.shawcroft@arm.com>
587 * config/aarch64/aarch64.c (faked_omit_frame_pointer): Remove.
588 (aarch64_frame_pointer_required): Adjust logic.
589 (aarch64_can_eliminate): Adjust logic.
590 (aarch64_override_options_after_change): Adjust logic.
592 2014-03-27 Dehao Chen <dehao@google.com>
594 * ipa-inline.c (early_inliner): Update node's inline info.
596 2014-03-26 Dehao Chen <dehao@google.com>
598 * dojump.c (do_compare_rtx_and_jump): Sets correct probability for
599 compiler inserted conditional jumps for NAN float check.
601 2014-03-26 Jakub Jelinek <jakub@redhat.com>
603 * ubsan.h (ubsan_create_data): Change second argument's type
604 to const location_t *.
605 * ubsan.c (ubsan_source_location): If xloc.file is NULL, set it to
607 (ubsan_create_data): Change second argument to const location_t *PLOC.
608 Create Loc field whenever PLOC is non-NULL.
609 (ubsan_instrument_unreachable, ubsan_expand_null_ifn,
610 ubsan_build_overflow_builtin, instrument_bool_enum_load): Adjust
614 * real.c (real_to_integer2): Change type of low to UHWI.
616 2014-03-26 Tobias Burnus <burnus@net-b.de>
618 * gcc.c (LINK_COMMAND_SPEC): Use libcilkrts.spec for -fcilkplus.
619 (CILK_SELF_SPECS): New define.
620 (driver_self_specs): Use it.
622 2014-03-26 Richard Biener <rguenther@suse.de>
624 * tree-pretty-print.c (percent_K_format): Implement special
625 case for LTO and its stripped down BLOCK tree.
627 2014-03-26 Jakub Jelinek <jakub@redhat.com>
630 * ubsan.c (instrument_si_overflow): Instrument ABS_EXPR.
632 * tree-vrp.c (simplify_internal_call_using_ranges): If only
633 one range is range_int_cst_p, but not both, at least optimize
634 addition/subtraction of 0 and multiplication by 0 or 1.
635 * gimple-fold.c (gimple_fold_call): Fold
636 IFN_UBSAN_CHECK_{ADD,SUB,MUL}.
637 (gimple_fold_stmt_to_constant_1): If both op0 and op1 aren't
638 INTEGER_CSTs, try to fold at least x * 0 and y - y.
640 2014-03-26 Eric Botcazou <ebotcazou@adacore.com>
642 PR rtl-optimization/60452
643 * rtlanal.c (rtx_addr_can_trap_p_1): Fix head comment.
644 <case REG>: Return 1 for invalid offsets from the frame pointer.
646 2014-03-26 Marek Polacek <polacek@redhat.com>
649 * doc/extend.texi (C Extensions): Mention variable-length arrays in
652 2014-03-26 Marek Polacek <polacek@redhat.com>
655 * doc/extend.texi (Designated Inits): Describe what happens to omitted
658 2014-03-26 Marek Polacek <polacek@redhat.com>
661 * ira-color.c (update_conflict_hard_regno_costs): Perform the
662 multiplication in unsigned type.
664 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
666 * doc/install.texi: Document nds32le-*-elf and nds32be-*-elf.
668 2014-03-26 Chung-Ju Wu <jasonwucj@gmail.com>
670 * doc/contrib.texi: Add myself as Andes nds32 port contributor.
672 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
675 * cif-code.def (UNREACHABLE) New code.
676 * ipa-inline.c (inline_small_functions): Skip edges to
677 __builtlin_unreachable.
678 (estimate_edge_growth): Allow edges to __builtlin_unreachable.
679 * ipa-inline-analysis.c (edge_set_predicate): Redirect edges with false
680 predicate to __bulitin_unreachable.
681 (set_cond_stmt_execution_predicate): Fix issue when
682 invert_tree_comparison returns ERROR_MARK.
683 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow): Do not
684 propagate to inline clones.
685 * cgraph.c (verify_edge_corresponds_to_fndecl): Allow redirection
687 * ipa-cp.c (create_specialized_node): Be ready for new node to appear.
688 * cgraphclones.c (cgraph_clone_node): If call destination is already
689 ureachable, do not redirect it back.
690 * tree-inline.c (fold_marked_statements): Hanlde calls becoming
693 2014-03-25 Jan Hubicka <hubicka@ucw.cz>
695 * ipa-pure-const.c (propagate_pure_const, propagate_nothrow):
696 Do not modify inline clones.
698 2014-03-25 Jakub Jelinek <jakub@redhat.com>
700 * config/i386/i386.md (general_sext_operand): New mode attr.
701 (addv<mode>4, subv<mode>4, mulv<mode>4): If operands[2] is CONST_INT,
702 don't generate (sign_extend (const_int)).
703 (*addv<mode>4, *subv<mode>4, *mulv<mode>4): Disallow CONST_INT_P
704 operands[2]. Use We constraint instead of <i> and
705 <general_sext_operand> predicate instead of <general_operand>.
706 (*addv<mode>4_1, *subv<mode>4_1, *mulv<mode>4_1): New insns.
707 * config/i386/constraints.md (We): New constraint.
708 * config/i386/predicates.md (x86_64_sext_operand,
709 sext_operand): New predicates.
711 2014-03-25 Martin Jambor <mjambor@suse.cz>
714 * ipa-cp.c (ipa_get_indirect_edge_target_1): Redirect type
715 inconsistent devirtualizations to __builtin_unreachable.
717 2014-03-25 Marek Polacek <polacek@redhat.com>
720 * doc/extend.texi (Example of asm with clobbered asm reg): Fix typo.
722 2014-03-25 Alan Lawrence <alan.lawrence@arm.com>
724 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Reverse
725 order of elements for big-endian.
727 2014-03-25 Richard Biener <rguenther@suse.de>
730 * gimplify-me.c (gimple_regimplify_operands): Update the
733 2014-03-25 Martin Jambor <mjambor@suse.cz>
736 * lto-cgraph.c (lto_output_node): Stream body_removed flag.
737 (lto_output_varpool_node): Likewise.
738 (input_overwrite_node): Likewise.
739 (input_varpool_node): Likewise.
741 2014-03-25 Richard Biener <rguenther@suse.de>
743 * lto-wrapper.c (merge_and_complain): Handle OPT_fPIE like OPT_fpie.
746 2014-03-25 Jakub Jelinek <jakub@redhat.com>
748 * combine.c (simplify_compare_const): Add MODE argument.
749 Handle mode_width 0 as very large mode_width.
750 (try_combine, simplify_comparison): Adjust callers.
752 * cselib.c (cselib_hash_rtx): Perform addition in unsigned
753 type to avoid signed integer overflow.
754 * explow.c (plus_constant): Likewise.
756 2014-03-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
758 * doc/generic.texi: Correct typos.
760 2014-03-24 Tobias Burnus <burnus@net-b.de>
762 * doc/invoke.texi (-flto): Expand section about
763 using static libraries with LTO.
765 2014-03-24 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
767 PR rtl-optimization/60501
768 * optabs.def (addptr3_optab): New optab.
769 * optabs.c (gen_addptr3_insn, have_addptr3_insn): New function.
770 * doc/md.texi ("addptrm3"): Document new RTL standard expander.
771 * expr.h (gen_addptr3_insn, have_addptr3_insn): Add prototypes.
773 * lra.c (emit_add3_insn): Use the addptr pattern if available.
775 * config/s390/s390.md ("addptrdi3", "addptrsi3"): New expanders.
777 2014-03-24 Ulrich Drepper <drepper@gmail.com>
779 * config/i386/avx512fintrin.h: Define _mm512_set1_ps and
782 * config/i386/avxintrin.h (_mm256_undefined_si256): Define.
783 (_mm256_undefined_ps): Define.
784 (_mm256_undefined_pd): Define.
785 * config/i386/emmintrin.h (_mm_undefined_si128): Define.
786 (_mm_undefined_pd): Define.
787 * config/i386/xmmintrin.h (_mm_undefined_ps): Define.
788 * config/i386/avx512fintrin.h (_mm512_undefined_si512): Define.
789 (_mm512_undefined_ps): Define.
790 (_mm512_undefined_pd): Define.
791 Use _mm*_undefined_*.
792 * config/i386/avx2intrin.h: Use _mm*_undefined_*.
794 2014-03-24 Alex Velenko <Alex.Velenko@arm.com>
796 * config/aarch64/aarch64-simd-builtins.def (lshr): DI mode excluded.
797 (lshr_simd): DI mode added.
798 * config/aarch64/aarch64-simd.md (aarch64_lshr_simddi): New pattern.
799 (aarch64_ushr_simddi): Likewise.
800 * config/aarch64/aarch64.md (UNSPEC_USHR64): New unspec.
801 * config/aarch64/arm_neon.h (vshr_n_u64): Intrinsic fixed.
802 (vshrd_n_u64): Likewise.
804 2014-03-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
806 * Makefile.in (s-macro_list): Depend on cc1.
808 2014-03-23 Teresa Johnson <tejohnson@google.com>
810 * ipa-utils.c (ipa_print_order): Use specified dump file.
812 2014-03-23 Eric Botcazou <ebotcazou@adacore.com>
814 PR rtl-optimization/60601
815 * bb-reorder.c (fix_up_fall_thru_edges): Test EDGE_FALLTHRU everywhere.
817 * gcc.c (eval_spec_function): Initialize save_growing_value.
819 2014-03-22 Jakub Jelinek <jakub@redhat.com>
822 * internal-fn.c (ubsan_expand_si_overflow_addsub_check): For
823 code == MINUS_EXPR, never swap op0 with op1.
825 * toplev.c (init_local_tick): Avoid signed integer multiplication
827 * genautomata.c (reserv_sets_hash_value): Fix rotate idiom, avoid
828 shift by first operand's bitsize.
830 2014-03-21 Jakub Jelinek <jakub@redhat.com>
833 * config/i386/i386.h (TARGET_64BIT_P): If not TARGET_BI_ARCH,
835 * config/i386/darwin.h (TARGET_64BIT_P): Redefine to
836 TARGET_ISA_64BIT_P(x).
838 2014-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
840 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Generate a
841 pattern for vector nor instead of subtract from splat(-1).
842 (altivec_expand_vec_perm_const_le): Likewise.
844 2014-03-21 Richard Henderson <rth@twiddle.net>
847 * ifcvt.c (dead_or_predicable): Return FALSE if there are any frame
848 related insns after epilogue_completed.
850 2014-03-21 Martin Jambor <mjambor@suse.cz>
853 * cgraph.h (symtab_node): New flag body_removed.
854 * ipa.c (symtab_remove_unreachable_nodes): Set body_removed flag
855 when removing bodies.
856 * symtab.c (dump_symtab_base): Dump body_removed flag.
857 * cgraph.c (verify_edge_corresponds_to_fndecl): Skip nodes which
858 had their bodies removed.
860 2014-03-21 Martin Jambor <mjambor@suse.cz>
863 * ipa.c (symtab_remove_unreachable_nodes): Clear thunk flag of nodes
866 2014-03-21 Richard Biener <rguenther@suse.de>
868 PR tree-optimization/60577
869 * tree-core.h (struct tree_base): Document nothrow_flag use
870 in VAR_DECL_NONALIASED.
871 * tree.h (VAR_DECL_NONALIASED): New.
872 (may_be_aliased): Adjust.
873 * coverage.c (build_var): Set VAR_DECL_NONALIASED.
875 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
877 * expr.c (expand_expr_real_1): Remove outdated comment.
879 2014-03-20 Jakub Jelinek <jakub@redhat.com>
882 * ira.c (adjust_cleared_regs): Call copy_rtx on
883 *reg_equiv[REGNO (loc)].src_p before passing it to
884 simplify_replace_fn_rtx.
887 * config/i386/i386.c (x86_output_mi_thunk): Surround UNSPEC_GOT
888 into CONST, put pic register as first operand of PLUS. Use
889 gen_const_mem for both 32-bit and 64-bit PIC got loads.
891 2014-03-20 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
893 * config/aarch64/aarch64.c (MEMORY_MOVE_COST): Delete.
895 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
897 * config/sparc/sparc.c (sparc_do_work_around_errata): Implement work
898 around for store forwarding issue in the FPU on the UT699.
899 * config/sparc/sparc.md (in_branch_delay): Return false for single FP
900 loads and operations if -mfix-ut699 is specified.
901 (divtf3_hq): Tweak attribute.
902 (sqrttf2_hq): Likewise.
904 2014-03-20 Eric Botcazou <ebotcazou@adacore.com>
906 * calls.c (store_one_arg): Remove incorrect const qualification on the
907 type of the temporary.
908 * cfgexpand.c (expand_return): Likewise.
909 * expr.c (expand_constructor): Likewise.
910 (expand_expr_real_1): Likewise.
912 2014-03-20 Zhenqiang Chen <zhenqiang.chen@linaro.org>
914 * config/arm/arm.c (arm_dwarf_register_span): Update the element number
917 2014-03-19 Kaz Kojima <kkojima@gcc.gnu.org>
920 * config/sh/sh.md (udivsi3_i1): Clobber R1 register.
922 2014-03-19 James Greenhalgh <james.greenhalgh@arm.com>
924 * config/arm/aarch-common-protos.h
925 (alu_cost_table): Fix spelling of "extend".
926 * config/arm/arm.c (arm_new_rtx_costs): Fix spelling of "extend".
928 2014-03-19 Richard Biener <rguenther@suse.de>
931 * tree-core.h (tree_type_common): Re-order pointer members
932 to reduce recursion depth during GC walks.
934 2014-03-19 Marek Polacek <polacek@redhat.com>
937 * ubsan.c (ubsan_type_descriptor): Check that DECL_NAME is nonnull
940 2014-03-19 Richard Biener <rguenther@suse.de>
943 * lto-streamer-in.c (input_function): In WPA stage do not drop
946 2014-03-19 Jakub Jelinek <jakub@redhat.com>
948 PR tree-optimization/60559
949 * vectorizable_mask_load_store): Replace scalar MASK_LOAD
950 with build_zero_cst assignment.
952 2014-03-18 Kai Tietz <ktietz@redhat.com>
954 PR rtl-optimization/56356
955 * sdbout.c (sdbout_parms): Verify that parms'
956 incoming argument is valid.
957 (sdbout_reg_parms): Likewise.
959 2014-03-18 Richard Henderson <rth@redhat.com>
962 * config/i386/i386.md (*float<SWI48x><MODEF>2_i387): Move down to
963 be shadowed by *float<SWI48><MODEF>2_sse. Test X87_ENABLE_FLOAT.
964 (*float<SWI48><MODEF>2_sse): Check X87_ENABLE_FLOAT for alternative 0.
966 2014-03-18 Basile Starynkevitch <basile@starynkevitch.net>
968 * plugin.def: Improve comment for PLUGIN_INCLUDE_FILE.
969 * doc/plugins.texi (Plugin callbacks): Mention PLUGIN_INCLUDE_FILE.
970 Italicize plugin event names in description. Explain that
971 PLUGIN_PRAGMAS has no sense for lto1. Explain PLUGIN_INCLUDE_FILE.
972 Remind that no GCC functions should be called after PLUGIN_FINISH.
973 Explain what pragmas with expansion are.
975 2014-03-18 Martin Liska <mliska@suse.cz>
977 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Added case when
978 gimple call statement is update.
979 * gimple-fold.c (gimple_fold_call): Changed order for GIMPLE_ASSIGN and
980 GIMPLE_CALL, where gsi iterator still points to GIMPLE CALL.
982 2014-03-18 Jakub Jelinek <jakub@redhat.com>
985 * ubsan.c (ubsan_instrument_unreachable): Call
986 initialize_sanitizer_builtins.
987 (ubsan_pass): Likewise.
990 * ubsan.c (ubsan_type_descriptor, ubsan_create_data): Call
991 varpool_finalize_decl instead of rest_of_decl_compilation.
993 2014-03-18 Richard Biener <rguenther@suse.de>
995 * df-problems.c (df_rd_confluence_n): Avoid bitmap_copy
996 by using bitmap_and_compl instead of bitmap_and_compl_into.
997 (df_rd_transfer_function): Likewise.
999 2014-03-18 Richard Biener <rguenther@suse.de>
1001 * doc/lto.texi (fresolution): Fix typo.
1003 2014-03-18 Richard Biener <rguenther@suse.de>
1005 * doc/invoke.texi (flto): Update for changes in 4.9.
1007 2014-03-18 Richard Biener <rguenther@suse.de>
1009 * doc/loop.texi: Remove section on the removed lambda framework.
1010 Update loop docs with recent changes in preserving loop structure.
1012 2014-03-18 Richard Biener <rguenther@suse.de>
1014 * doc/lto.texi (-fresolution): Document.
1016 2014-03-18 Richard Biener <rguenther@suse.de>
1018 * doc/contrib.texi: Adjust my name.
1020 2014-03-18 Jakub Jelinek <jakub@redhat.com>
1023 * internal-fn.c: Include diagnostic-core.h.
1024 (expand_BUILTIN_EXPECT): New function.
1025 * gimplify.c (gimplify_call_expr): Use false instead of FALSE.
1026 (gimplify_modify_expr): Gimplify 3 argument __builtin_expect into
1027 IFN_BUILTIN_EXPECT call instead of __builtin_expect builtin call.
1028 * ipa-inline-analysis.c (find_foldable_builtin_expect): Handle
1030 * predict.c (expr_expected_value_1): Handle IFN_BUILTIN_EXPECT.
1031 Revert 3 argument __builtin_expect code.
1032 (strip_predict_hints): Handle IFN_BUILTIN_EXPECT.
1033 * gimple-fold.c (gimple_fold_call): Likewise.
1034 * tree.h (fold_builtin_expect): New prototype.
1035 * builtins.c (build_builtin_expect_predicate): Add predictor
1036 argument, if non-NULL, create 3 argument __builtin_expect.
1037 (fold_builtin_expect): No longer static. Add ARG2 argument,
1038 pass it through to build_builtin_expect_predicate.
1039 (fold_builtin_2): Adjust caller.
1040 (fold_builtin_3): Handle BUILT_IN_EXPECT.
1041 * internal-fn.def (BUILTIN_EXPECT): New.
1043 2014-03-18 Tobias Burnus <burnus@net-b.de>
1046 * predict.def (PRED_FORTRAN_OVERFLOW, PRED_FORTRAN_FAIL_ALLOC,
1047 PRED_FORTRAN_FAIL_IO, PRED_FORTRAN_WARN_ONCE, PRED_FORTRAN_SIZE_ZERO,
1048 PRED_FORTRAN_INVALID_BOUND, PRED_FORTRAN_ABSENT_DUMMY): Add.
1050 2014-03-18 Jan Hubicka <hubicka@ucw.cz>
1053 * predict.c (combine_predictions_for_bb): Fix up formatting.
1054 (expr_expected_value_1, expr_expected_value): Add predictor argument,
1055 fill what it points to if non-NULL.
1056 (tree_predict_by_opcode): Adjust caller, use the predictor.
1057 * predict.def (PRED_COMPARE_AND_SWAP): Add.
1059 2014-03-18 Eric Botcazou <ebotcazou@adacore.com>
1061 * config/sparc/sparc.c (sparc_do_work_around_errata): Speed up and use
1062 proper constant for the store mode.
1064 2014-03-18 Ilya Enkovich <ilya.enkovich@intel.com>
1066 * symtab.c (change_decl_assembler_name): Fix transparent alias
1069 2014-03-16 Renlin Li <Renlin.Li@arm.com>
1071 * config/aarch64/aarch64.c: Correct the comments about the
1072 aarch64 stack layout.
1074 2014-03-18 Thomas Schwinge <thomas@codesourcery.com>
1076 * omp-low.c (lower_rec_input_clauses) <build_omp_barrier>: Restore
1077 check for GF_OMP_FOR_KIND_FOR.
1079 2013-03-18 Kirill Yukhin <kirill.yukhin@intel.com>
1081 * config/i386/i386.h (ADDITIONAL_REGISTER_NAMES): Add
1082 ymm and zmm register names.
1084 2014-03-17 Jakub Jelinek <jakub@redhat.com>
1087 * config/i386/i386.c (ix86_expand_epilogue): Adjust REG_CFA_ADJUST_CFA
1088 note creation for the 2010-08-31 changes.
1090 2014-03-17 Marek Polacek <polacek@redhat.com>
1093 * omp-low.c (omp_max_vf): Treat -fno-tree-loop-optimize the same
1094 as -fno-tree-loop-vectorize.
1095 (expand_omp_simd): Likewise.
1097 2014-03-15 Eric Botcazou <ebotcazou@adacore.com>
1099 * config/sparc/sparc-protos.h (tls_call_delay): Delete.
1100 (eligible_for_call_delay): New prototype.
1101 * config/sparc/sparc.c (tls_call_delay): Rename into...
1102 (eligible_for_call_delay): ...this. Return false if the instruction
1103 cannot be put in the delay slot of a branch.
1104 (eligible_for_restore_insn): Simplify.
1105 (eligible_for_return_delay): Return false if the instruction cannot be
1106 put in the delay slot of a branch and simplify.
1107 (eligible_for_sibcall_delay): Return false if the instruction cannot be
1108 put in the delay slot of a branch.
1109 * config/sparc/sparc.md (fix_ut699): New attribute.
1110 (tls_call_delay): Delete.
1111 (in_call_delay): Reimplement.
1112 (eligible_for_sibcall_delay): Rename into...
1113 (in_sibcall_delay): ...this.
1114 (eligible_for_return_delay): Rename into...
1115 (in_return_delay): ...this.
1116 (in_branch_delay): Reimplement.
1117 (in_uncond_branch_delay): Delete.
1118 (in_annul_branch_delay): Delete.
1120 2014-03-14 Richard Henderson <rth@redhat.com>
1123 * config/i386/i386.md (floathi<X87MODEF>2): Delete expander; rename
1124 define_insn from *floathi<X87MODEF>2_i387; allow nonimmediate_operand.
1125 (*floathi<X87MODEF>2_i387_with_temp): Remove.
1126 (floathi splitters): Remove.
1127 (float<SWI48x>xf2): New pattern.
1128 (float<SWI48><MODEF>2): Rename from float<SWI48x><X87MODEF>2. Drop
1129 code that tried to handle DImode for 32-bit, but which was excluded
1130 by the pattern's condition. Drop allocation of stack temporary.
1131 (*floatsi<MODEF>2_vector_mixed_with_temp): Remove.
1132 (*float<SWI48><MODEF>2_mixed_with_temp): Remove.
1133 (*float<SWI48><MODEF>2_mixed_interunit): Remove.
1134 (*float<SWI48><MODEF>2_mixed_nointerunit): Remove.
1135 (*floatsi<MODEF>2_vector_sse_with_temp): Remove.
1136 (*float<SWI48><MODEF>2_sse_with_temp): Remove.
1137 (*float<SWI48><MODEF>2_sse_interunit): Remove.
1138 (*float<SWI48><MODEF>2_sse_nointerunit): Remove.
1139 (*float<SWI48x><X87MODEF>2_i387_with_temp): Remove.
1140 (*float<SWI48x><X87MODEF>2_i387): Remove.
1141 (all float _with_temp splitters): Remove.
1142 (*float<SWI48x><MODEF>2_i387): New pattern.
1143 (*float<SWI48><MODEF>2_sse): New pattern.
1144 (float TARGET_USE_VECTOR_CONVERTS splitters): Merge them.
1145 (float TARGET_SSE_PARTIAL_REG_DEPENDENCY splitters): Merge them.
1147 2014-03-14 Jakub Jelinek <jakub@redhat.com>
1148 Marek Polacek <polacek@redhat.com>
1151 * common.opt (dump_base_name_prefixed): New Variable.
1152 * opts.c (finish_options): Don't prepend directory to x_dump_base_name
1153 if x_dump_base_name_prefixed is already set, set it at the end.
1155 2014-03-14 Vladimir Makarov <vmakarov@redhat.com>
1157 PR rtl-optimization/60508
1158 * lra-constraints.c (get_reload_reg): Add new parameter
1160 (process_addr_reg, simplify_operand_subreg, curr_insn_transform):
1161 Pass the new parameter values.
1163 2014-03-14 Richard Biener <rguenther@suse.de>
1165 * common.opt: Revert unintented changes from r205065.
1168 2014-03-14 Richard Biener <rguenther@suse.de>
1171 * cfghooks.c (split_block): Properly adjust all loops the
1172 block was a latch of.
1174 2014-03-14 Martin Jambor <mjambor@suse.cz>
1177 * ipa-prop.c (ipa_modify_call_arguments): Fix iteration condition
1180 2014-03-14 Georg-Johann Lay <avr@gjlay.de>
1183 * config/avr/avr.c (avr_set_current_function): Pass function name
1184 through default_strip_name_encoding before sanity checking instead
1185 of skipping the first char of the assembler name.
1187 2014-03-13 Richard Henderson <rth@redhat.com>
1190 * config/i386/i386.c (ix86_split_fp_branch): Remove pushed argument.
1191 (ix86_force_to_memory, ix86_free_from_memory): Remove.
1192 * config/i386/i386-protos.h: Likewise.
1193 * config/i386/i386.md (floathi<X87MODEF>2): Use assign_386_stack_local
1194 in the expander instead of a splitter.
1195 (float<SWI48x><X87MODEF>2): Use assign_386_stack_local if there is
1196 any possibility of requiring a memory.
1197 (*floatsi<MODEF>2_vector_mixed): Remove, and the splitters.
1198 (*floatsi<MODEF>2_vector_sse): Remove, and the splitters.
1199 (fp branch splitters): Update for ix86_split_fp_branch.
1200 (*jcc<X87MODEF>_<SWI24>_i387): Remove r/f alternative.
1201 (*jcc<X87MODEF>_<SWI24>_r_i387): Likewise.
1202 (splitter for jcc<X87MODEF>_<SWI24>_i387 r/f): Remove.
1203 (*fop_<MODEF>_2_i387): Remove f/r alternative.
1204 (*fop_<MODEF>_3_i387): Likewise.
1205 (*fop_xf_2_i387, *fop_xf_3_i387): Likewise.
1206 (splitters for the fop_* register patterns): Remove.
1207 (fscalexf4_i387): Rename from *fscalexf4_i387.
1208 (ldexpxf3): Use gen_floatsixf2 and gen_fscalexf4_i387.
1210 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1212 PR tree-optimization/59779
1213 * tree-dfa.c (get_ref_base_and_extent): Use double_int
1214 type for bitsize and maxsize instead of HOST_WIDE_INT.
1216 2014-03-13 Steven Bosscher <steven@gcc.gnu.org>
1218 PR rtl-optimization/57320
1219 * function.c (rest_of_handle_thread_prologue_and_epilogue): Cleanup
1220 the CFG after thread_prologue_and_epilogue_insns.
1222 2014-03-13 Vladimir Makarov <vmakarov@redhat.com>
1224 PR rtl-optimization/57189
1225 * lra-constraints.c (process_alt_operands): Disfavor spilling
1228 2014-03-13 Cesar Philippidis <cesar@codesourcery.com>
1230 * lto-wrapper.c (maybe_unlink_file): Suppress diagnostic messages.
1232 2014-03-13 Jakub Jelinek <jakub@redhat.com>
1234 PR tree-optimization/59025
1236 * tree-ssa-reassoc.c (sort_by_operand_rank): For SSA_NAMEs with the
1237 same rank, sort by bb_rank and gimple_uid of SSA_NAME_DEF_STMT first.
1239 2014-03-13 Georg-Johann Lay <avr@gjlay.de>
1242 * config/avr/avr.c (avr_out_plus): Swap cc_plus and cc_minus in
1243 calls of avr_out_plus_1.
1245 2014-03-13 Bin Cheng <bin.cheng@arm.com>
1247 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Record
1248 BB's single pred and update the father loop's latch info later.
1250 2014-03-12 Michael Meissner <meissner@linux.vnet.ibm.com>
1252 * config/rs6000/vector.md (VEC_L): Add V1TI mode to vector types.
1256 (VEC_base): Likewise.
1257 (mov<MODE>, VEC_M modes): If we are loading TImode into VSX
1258 registers, we need to swap double words in little endian mode.
1260 * config/rs6000/rs6000-modes.def (V1TImode): Add new vector mode
1261 to be a container mode for 128-bit integer operations added in ISA
1262 2.07. Unlike TImode and PTImode, the preferred register set is
1263 the Altivec/VMX registers for the 128-bit operations.
1265 * config/rs6000/rs6000-protos.h (rs6000_move_128bit_ok_p): Add
1267 (rs6000_split_128bit_ok_p): Likewise.
1269 * config/rs6000/rs6000-builtin.def (BU_P8V_AV_3): Add new support
1270 macros for creating ISA 2.07 normal and overloaded builtin
1271 functions with 3 arguments.
1272 (BU_P8V_OVERLOAD_3): Likewise.
1273 (VPERM_1T): Add support for V1TImode in 128-bit vector operations
1274 for use as overloaded functions.
1275 (VPERM_1TI_UNS): Likewise.
1276 (VSEL_1TI): Likewise.
1277 (VSEL_1TI_UNS): Likewise.
1278 (ST_INTERNAL_1ti): Likewise.
1279 (LD_INTERNAL_1ti): Likewise.
1280 (XXSEL_1TI): Likewise.
1281 (XXSEL_1TI_UNS): Likewise.
1282 (VPERM_1TI): Likewise.
1283 (VPERM_1TI_UNS): Likewise.
1284 (XXPERMDI_1TI): Likewise.
1285 (SET_1TI): Likewise.
1286 (LXVD2X_V1TI): Likewise.
1287 (STXVD2X_V1TI): Likewise.
1288 (VEC_INIT_V1TI): Likewise.
1289 (VEC_SET_V1TI): Likewise.
1290 (VEC_EXT_V1TI): Likewise.
1291 (EQV_V1TI): Likewise.
1292 (NAND_V1TI): Likewise.
1293 (ORC_V1TI): Likewise.
1294 (VADDCUQ): Add support for 128-bit integer arithmetic instructions
1295 added in ISA 2.07. Add both normal 'altivec' builtins, and the
1297 (VADDUQM): Likewise.
1298 (VSUBCUQ): Likewise.
1299 (VADDEUQM): Likewise.
1300 (VADDECUQ): Likewise.
1301 (VSUBEUQM): Likewise.
1302 (VSUBECUQ): Likewise.
1304 * config/rs6000/rs6000-c.c (__int128_type): New static to hold
1305 __int128_t and __uint128_t types.
1306 (__uint128_type): Likewise.
1307 (altivec_categorize_keyword): Add support for vector __int128_t,
1308 vector __uint128_t, vector __int128, and vector unsigned __int128
1309 as a container type for TImode operations that need to be done in
1310 VSX/Altivec registers.
1311 (rs6000_macro_to_expand): Likewise.
1312 (altivec_overloaded_builtins): Add ISA 2.07 overloaded functions
1313 to support 128-bit integer instructions vaddcuq, vadduqm,
1314 vaddecuq, vaddeuqm, vsubcuq, vsubuqm, vsubecuq, vsubeuqm.
1315 (altivec_resolve_overloaded_builtin): Add support for V1TImode.
1317 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Add support
1318 for V1TImode, and set up preferences to use VSX/Altivec registers.
1319 Setup VSX reload handlers.
1320 (rs6000_debug_reg_global): Likewise.
1321 (rs6000_init_hard_regno_mode_ok): Likewise.
1322 (rs6000_preferred_simd_mode): Likewise.
1323 (vspltis_constant): Do not allow V1TImode as easy altivec constants.
1324 (easy_altivec_constant): Likewise.
1325 (output_vec_const_move): Likewise.
1326 (rs6000_expand_vector_set): Convert V1TImode set and extract to
1328 (rs6000_expand_vector_extract): Likewise.
1329 (reg_offset_addressing_ok_p): Setup V1TImode to use VSX reg+reg
1331 (rs6000_const_vec): Add support for V1TImode.
1332 (rs6000_emit_le_vsx_load): Swap double words when loading or
1333 storing TImode/V1TImode.
1334 (rs6000_emit_le_vsx_store): Likewise.
1335 (rs6000_emit_le_vsx_move): Likewise.
1336 (rs6000_emit_move): Add support for V1TImode.
1337 (altivec_expand_ld_builtin): Likewise.
1338 (altivec_expand_st_builtin): Likewise.
1339 (altivec_expand_vec_init_builtin): Likewise.
1340 (altivec_expand_builtin): Likewise.
1341 (rs6000_init_builtins): Add support for V1TImode type. Add
1342 support for ISA 2.07 128-bit integer builtins. Define type names
1343 for the VSX/Altivec vector types.
1344 (altivec_init_builtins): Add support for overloaded vector
1345 functions with V1TImode type.
1346 (rs6000_preferred_reload_class): Prefer Altivec registers for V1TImode.
1347 (rs6000_move_128bit_ok_p): Move 128-bit move/split validation to
1349 (rs6000_split_128bit_ok_p): Likewise.
1350 (rs6000_handle_altivec_attribute): Create V1TImode from vector
1351 __int128_t and vector __uint128_t.
1353 * config/rs6000/vsx.md (VSX_L): Add V1TImode to vector iterators
1354 and mode attributes.
1361 (VS_scalar): Likewise.
1362 (VS_double): Likewise.
1363 (vsx_set_v1ti): New builtin function to create V1TImode from TImode.
1365 * config/rs6000/rs6000.h (TARGET_VADDUQM): New macro to say whether
1366 we support the ISA 2.07 128-bit integer arithmetic instructions.
1367 (ALTIVEC_OR_VSX_VECTOR_MODE): Add V1TImode.
1368 (enum rs6000_builtin_type_index): Add fields to hold V1TImode
1369 and TImode types for use with the builtin functions.
1370 (V1TI_type_node): Likewise.
1371 (unsigned_V1TI_type_node): Likewise.
1372 (intTI_type_internal_node): Likewise.
1373 (uintTI_type_internal_node): Likewise.
1375 * config/rs6000/altivec.md (UNSPEC_VADDCUQ): New unspecs for ISA 2.07
1376 128-bit builtin functions.
1377 (UNSPEC_VADDEUQM): Likewise.
1378 (UNSPEC_VADDECUQ): Likewise.
1379 (UNSPEC_VSUBCUQ): Likewise.
1380 (UNSPEC_VSUBEUQM): Likewise.
1381 (UNSPEC_VSUBECUQ): Likewise.
1382 (VM): Add V1TImode to vector mode iterators.
1384 (VI_unit): Likewise.
1385 (altivec_vadduqm): Add ISA 2.07 128-bit binary builtins.
1386 (altivec_vaddcuq): Likewise.
1387 (altivec_vsubuqm): Likewise.
1388 (altivec_vsubcuq): Likewise.
1389 (altivec_vaddeuqm): Likewise.
1390 (altivec_vaddecuq): Likewise.
1391 (altivec_vsubeuqm): Likewise.
1392 (altivec_vsubecuq): Likewise.
1394 * config/rs6000/rs6000.md (FMOVE128_GPR): Add V1TImode to vector
1396 (BOOL_128): Likewise.
1397 (BOOL_REGS_OUTPUT): Likewise.
1398 (BOOL_REGS_OP1): Likewise.
1399 (BOOL_REGS_OP2): Likewise.
1400 (BOOL_REGS_UNARY): Likewise.
1401 (BOOL_REGS_AND_CR0): Likewise.
1403 * config/rs6000/altivec.h (vec_vaddcuq): Add support for ISA 2.07
1404 128-bit integer builtin support.
1405 (vec_vadduqm): Likewise.
1406 (vec_vaddecuq): Likewise.
1407 (vec_vaddeuqm): Likewise.
1408 (vec_vsubecuq): Likewise.
1409 (vec_vsubeuqm): Likewise.
1410 (vec_vsubcuq): Likewise.
1411 (vec_vsubuqm): Likewise.
1413 * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
1414 Document vec_vaddcuq, vec_vadduqm, vec_vaddecuq, vec_vaddeuqm,
1415 vec_subecuq, vec_subeuqm, vec_vsubcuq, vec_vsubeqm builtins adding
1416 128-bit integer add/subtract to ISA 2.07.
1418 2014-03-12 Joern Rennecke <joern.rennecke@embecosm.com>
1420 * config/arc/arc.c (arc_predicate_delay_insns):
1421 Fix third argument passed to conditionalize_nonjump.
1423 2014-03-12 Yufeng Zhang <yufeng.zhang@arm.com>
1425 * config/aarch64/aarch64-builtins.c
1426 (aarch64_builtin_vectorized_function): Add BUILT_IN_LFLOORF,
1427 BUILT_IN_LLFLOOR, BUILT_IN_LCEILF and BUILT_IN_LLCEIL.
1428 * config/aarch64/arm_neon.h (vcvtaq_u64_f64): Call __builtin_llfloor
1429 instead of __builtin_lfloor.
1430 (vcvtnq_u64_f64): Call __builtin_llceil instead of __builtin_lceil.
1432 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1434 * tree-ssa-ifcombine.c (forwarder_block_to): New function.
1435 (tree_ssa_ifcombine_bb_1): New function.
1436 (tree_ssa_ifcombine_bb): Use it. Handle also cases where else_bb
1437 is an empty forwarder block to then_bb or vice versa and then_bb
1438 and else_bb are effectively swapped.
1440 2014-03-12 Christian Bruel <christian.bruel@st.com>
1443 * config/arm/arm.c (arm_emit_vfp_multi_reg_pop): Emit a
1444 REG_CFA_DEF_CFA note.
1445 (arm_expand_epilogue_apcs_frame): call arm_add_cfa_adjust_cfa_note.
1446 (arm_unwind_emit): Allow REG_CFA_DEF_CFA.
1448 2014-03-12 Thomas Preud'homme <thomas.preudhomme@arm.com>
1450 PR tree-optimization/60454
1451 * tree-ssa-math-opts.c (find_bswap_1): Fix bswap detection.
1453 2014-03-12 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1455 * config.gcc (aarch64*-*-*): Use ISA flags from aarch64-arches.def.
1456 Do not define target_cpu_default2 to generic.
1457 * config/aarch64/aarch64.h (TARGET_CPU_DEFAULT): Use generic cpu.
1458 * config/aarch64/aarch64.c (aarch64_override_options): Update comment.
1459 * config/aarch64/aarch64-arches.def (armv8-a): Use generic cpu.
1461 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1462 Marc Glisse <marc.glisse@inria.fr>
1464 PR tree-optimization/60502
1465 * tree-ssa-reassoc.c (eliminate_not_pairs): Use build_all_ones_cst
1466 instead of build_low_bits_mask.
1468 2014-03-12 Jakub Jelinek <jakub@redhat.com>
1471 * tree-vrp.c (register_edge_assert_for_1): Don't add assert
1472 if there are multiple uses, but op doesn't live on E edge.
1473 * tree-cfg.c (assert_unreachable_fallthru_edge_p): Also ignore
1474 clobber stmts before __builtin_unreachable.
1476 2014-03-11 Richard Sandiford <rdsandiford@googlemail.com>
1478 * builtins.c (expand_builtin_setjmp_receiver): Use and clobber
1479 hard_frame_pointer_rtx.
1480 * cse.c (cse_insn): Remove volatile check.
1481 * cselib.c (cselib_process_insn): Likewise.
1482 * dse.c (scan_insn): Likewise.
1484 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1486 * config/arc/arc.c (conditionalize_nonjump): New function,
1488 (arc_ifcvt): ... this.
1489 (arc_predicate_delay_insns): Use it.
1491 2014-03-11 Joern Rennecke <joern.rennecke@embecosm.com>
1493 * config/arc/predicates.md (extend_operand): During/after reload,
1494 allow const_int_operand.
1495 * config/arc/arc.md (mulsidi3_700): Use extend_operand predicate.
1496 (umulsidi3_700): Likewise. Change operand 2 constraint back to "cL".
1497 (mulsi3_highpart): Change operand 2 constraint alternatives 2 and 3
1499 (umulsi3_highpart_i): Likewise.
1501 2014-03-11 Richard Biener <rguenther@suse.de>
1503 * tree-ssa-structalias.c (get_constraint_for_ptr_offset):
1504 Add asserts to guard possible wrong-code bugs.
1506 2014-03-11 Richard Biener <rguenther@suse.de>
1508 PR tree-optimization/60429
1509 PR tree-optimization/60485
1510 * tree-ssa-structalias.c (set_union_with_increment): Properly
1511 take into account all fields that overlap the shifted vars.
1512 (do_sd_constraint): Likewise.
1513 (do_ds_constraint): Likewise.
1514 (get_constraint_for_ptr_offset): Likewise.
1516 2014-03-11 Chung-Lin Tang <cltang@codesourcery.com>
1518 * config/nios2/nios2.c (machine_function): Add fp_save_offset field.
1519 (nios2_compute_frame_layout):
1520 Add calculation of cfun->machine->fp_save_offset.
1521 (nios2_expand_prologue): Correct setting of frame pointer register
1523 (nios2_expand_epilogue): Update recovery of stack pointer from
1524 frame pointer accordingly.
1525 (nios2_initial_elimination_offset): Update calculation of offset
1526 for eliminating to HARD_FRAME_POINTER_REGNUM.
1528 2014-03-10 Jakub Jelinek <jakub@redhat.com>
1531 * ipa.c (symtab_remove_unreachable_nodes): Don't call
1532 cgraph_get_create_node on VAR_DECLs.
1534 2014-03-10 Richard Biener <rguenther@suse.de>
1537 * tree.c (signed_or_unsigned_type_for): Handle OFFSET_TYPEs.
1539 2014-03-08 Douglas B Rupp <rupp@gnat.com>
1541 * config/vms/vms.opt (vms_float_format): New variable.
1543 2014-03-08 Tobias Burnus <burnus@net-b.de>
1545 * doc/invoke.texi (-fcilkplus): Update implementation status.
1547 2014-03-08 Paulo Matos <paulo@matos-sorge.com>
1548 Richard Biener <rguenther@suse.de>
1550 * lto-wrapper.c (merge_and_complain): Ensure -fshort-double is used
1551 consistently accross all TUs.
1552 (run_gcc): Enable -fshort-double automatically at link at link-time
1553 and disallow override.
1555 2014-03-08 Richard Sandiford <rdsandiford@googlemail.com>
1558 * config/mips/mips.c (mips_option_override): Promote -mpaired-single
1559 warning to an error. Disable TARGET_PAIRED_SINGLE and TARGET_MIPS3D
1560 if they can't be used.
1562 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1564 * configure.ac (HAVE_AS_IX86_TLSLDMPLT): Improve test
1565 for Solaris 11/x86 ld.
1566 * configure: Regenerate.
1568 2014-03-07 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1570 * configure.ac (TLS_SECTION_ASM_FLAG): Save as tls_section_flag.
1571 (LIB_TLS_SPEC): Save as ld_tls_libs.
1572 (HAVE_AS_IX86_TLSLDMPLT): Define as 1/0.
1573 (HAVE_AS_IX86_TLSLDM): New test.
1574 * configure, config.in: Regenerate.
1575 * config/i386/i386.c (legitimize_tls_address): Fall back to
1576 TLS_MODEL_GLOBAL_DYNAMIC on 32-bit Solaris/x86 if tool chain
1577 cannot support TLS_MODEL_LOCAL_DYNAMIC.
1578 * config/i386/i386.md (*tls_local_dynamic_base_32_gnu): Use if
1579 instead of #ifdef in HAVE_AS_IX86_TLSLDMPLT test.
1581 2014-03-07 Paulo Matos <paulo@matos-sorge.com>
1583 * common.opt (fira-loop-pressure): Mark as optimization.
1585 2014-03-07 Thomas Schwinge <thomas@codesourcery.com>
1587 * langhooks.c (lhd_omp_mappable_type): The error_mark_node is not
1588 an OpenMP mappable type.
1590 2014-03-06 Matthias Klose <doko@ubuntu.com>
1592 * Makefile.in (s-mlib): Only pass MULTIARCH_DIRNAME if
1593 MULTILIB_OSDIRNAMES is not defined.
1595 2014-03-06 Jakub Jelinek <jakub@redhat.com>
1596 Meador Inge <meadori@codesourcery.com>
1599 * config/arm/arm.c (arm_tls_symbol_p): Remove.
1600 (arm_legitimize_address): Call legitimize_tls_address for any
1601 arm_tls_referenced_p expression, handle constant addend. Call it
1602 before testing for !TARGET_ARM.
1603 (thumb_legitimize_address): Don't handle arm_tls_symbol_p here.
1605 2014-03-06 Richard Biener <rguenther@suse.de>
1611 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1613 * tree-streamer.c (record_common_node): Assert we don't record
1614 nodes with type double.
1615 (preload_common_node): Skip type double, complex double and double
1616 pointer since it is now frontend dependent due to fshort-double option.
1618 2014-03-06 Richard Biener <rguenther@suse.de>
1620 * gcc.c (PLUGIN_COND): Always enable unless -fno-use-linker-plugin
1621 or -fno-lto is specified and the linker has full plugin support.
1622 * collect2.c (lto_mode): Default to LTO_MODE_WHOPR if LTO is enabled.
1623 (main): Remove -flto processing, adjust lto_mode using use_plugin late.
1624 * lto-wrapper.c (merge_and_complain): Merge compile-time
1625 optimization levels.
1626 (run_gcc): And pass it through to the link options.
1628 2014-03-06 Alexandre Oliva <aoliva@redhat.com>
1632 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1634 * cselib.c (remove_useless_values): Skip to avoid quadratic
1635 behavior if the condition moved from...
1636 (cselib_process_insn): ... here holds.
1638 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1641 * Makefile.in (PLUGIN_HEADERS): Add tree-phinodes.h, stor-layout.h,
1642 ssa-iterators.h, $(RESOURCE_H) and tree-cfgcleanup.h.
1645 * config/i386/t-i386 (OPTIONS_H_EXTRA): Add stringop.def.
1646 (TM_H): Add x86-tune.def.
1648 2014-03-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1650 * config/aarch64/aarch64.c (generic_tunings):
1651 Use cortexa57_extra_costs.
1653 2014-03-05 Jakub Jelinek <jakub@redhat.com>
1656 * cfgexpand.c (expand_used_vars): Do not assume all SSA_NAMEs
1657 of PARM/RESULT_DECLs must be coalesced with optimize && in_lto_p.
1658 * tree-ssa-coalesce.c (coalesce_ssa_name): Use MUST_COALESCE_COST - 1
1661 2014-03-04 Heiher <r@hev.cc>
1663 * config/mips/mips-cpus.def (loongson3a): Mark as a MIPS64r2 processor.
1664 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Adjust accordingly.
1666 2014-03-04 Uros Bizjak <ubizjak@gmail.com>
1668 * config/i386/predicates.md (const2356_operand): Change to ...
1669 (const2367_operand): ... this.
1670 * config/i386/sse.md (avx512pf_scatterpf<mode>sf): Use
1672 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1673 (*avx512pf_scatterpf<mode>sf): Ditto.
1674 (avx512pf_scatterpf<mode>df): Ditto.
1675 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1676 (*avx512pf_scatterpf<mode>df): Ditto.
1677 * config/i386/i386.c (ix86_expand_builtin): Update
1678 incorrect hint operand error message.
1680 2014-03-04 Richard Biener <rguenther@suse.de>
1682 * lto-section-in.c (lto_get_section_data): Fix const cast.
1684 2014-03-04 Paulo Matos <paulo@matos-sorge.com>
1686 * tree-streamer.c (record_common_node): Assert we don't record
1687 nodes with type double.
1688 (preload_common_node): Skip type double, complex double and double
1689 pointer since it is now frontend dependent due to fshort-double option.
1691 2014-03-04 Richard Biener <rguenther@suse.de>
1694 * lto-streamer-in.c (lto_read_body): Remove LTO bytecode version check.
1695 (lto_input_toplevel_asms): Likewise.
1696 * lto-section-in.c (lto_get_section_data): Instead do it here
1699 2014-03-04 Richard Biener <rguenther@suse.de>
1701 PR tree-optimization/60382
1702 * tree-vect-loop.c (vect_is_simple_reduction_1): Do not consider
1703 dead PHIs a reduction.
1705 2014-03-03 Uros Bizjak <ubizjak@gmail.com>
1707 * config/i386/xmmintrin.h (enum _mm_hint) <_MM_HINT_ET0>: Correct
1709 (_mm_prefetch): Move out of GCC target("sse") pragma.
1710 * config/i386/prfchwintrin.h (_m_prefetchw): Move out of
1711 GCC target("prfchw") pragma.
1712 * config/i386/i386.md (prefetch): Emit prefetchwt1 only
1714 * config/i386/i386.c (ix86_option_override_internal): Enable
1715 -mprfchw with -mprefetchwt1.
1717 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1719 * config/arc/arc.md (casesi_load) <length attribute alternative 0>:
1722 2014-03-03 Joern Rennecke <joern.rennecke@embecosm.com>
1724 * opts.h (CL_PCH_IGNORE): Define.
1725 * targhooks.c (option_affects_pch_p):
1726 Return false for options that have CL_PCH_IGNORE set.
1727 * opt-functions.awk: Process PchIgnore.
1728 * doc/options.texi: Document PchIgnore.
1730 * config/arc/arc.opt (misize): Add PchIgnore property.
1732 2014-03-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1734 * config/rs6000/rs6000.c (rs6000_preferred_reload_class): Disallow
1735 reload of PLUS rtx's outside of GENERAL_REGS or BASE_REGS; relax
1736 constraint on constants to permit them being loaded into
1737 GENERAL_REGS or BASE_REGS.
1739 2014-03-03 Nick Clifton <nickc@redhat.com>
1741 * config/rl78/rl78-real.md (cbranchsi4_real_signed): Add
1742 anti-cacnonical alternatives.
1743 (negandhi3_real): New pattern.
1744 * config/rl78/rl78-virt.md (negandhi3_virt): New pattern.
1746 2014-03-03 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
1748 * config/avr/avr-mcus.def: Remove atxmega16x1.
1749 * config/avr/avr-tables.opt: Regenerate.
1750 * config/avr/t-multilib: Regenerate.
1751 * doc/avr-mmcu.texi: Regenerate.
1753 2014-03-03 Tobias Grosser <tobias@grosser.es>
1754 Mircea Namolaru <mircea.namolaru@inria.fr>
1756 PR tree-optimization/58028
1757 * graphite-clast-to-gimple.c (set_cloog_options): Don't remove
1760 2014-03-03 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1762 * config/arm/neon.md (*movmisalign<mode>): Legitimize addresses
1763 not handled by recognizers.
1765 2014-03-03 Jakub Jelinek <jakub@redhat.com>
1768 * function.c (expand_function_end): Don't emit
1769 clobber_return_register sequence if clobber_after is a BARRIER.
1770 * cfgexpand.c (construct_exit_block): Append instructions before
1771 return_label to prev_bb.
1773 2014-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1775 * config/rs6000/constraints.md: Document reserved use of "wc".
1777 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1780 * ipa.c (function_and_variable_visibility): When dissolving comdat
1781 group, also set all symbols to local.
1783 2014-03-02 Jan Hubicka <hubicka@ucw.cz>
1788 2013-12-14 Jan Hubicka <jh@suse.cz>
1790 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Skip clobbers.
1792 2014-03-02 Jon Beniston <jon@beniston.com>
1798 * config/lm32/lm32.c (lm32_legitimate_constant_p): Remove, as incorrect.
1799 (TARGET_LEGITIMATE_CONSTANT_P): Undefine, as not needed.
1800 * config/lm32/lm32.md (movsi_insn): Add 32-bit immediate support.
1801 (simple_return, *simple_return): New patterns
1802 * config/lm32/predicates.md (movsi_rhs_operand): Remove as obsolete.
1803 * configure.ac (force_sjlj_exceptions): Force sjlj exceptions for lm32.
1805 2014-03-01 Paolo Carlini <paolo.carlini@oracle.com>
1807 * dwarf2out.c (gen_subprogram_die): Tidy.
1809 2014-03-01 Oleg Endo <olegendo@gcc.gnu.org>
1812 * config/sh/sh.md (*mov_t_msb_neg): Split into ...
1813 (*mov_t_msb_neg_negc): ... this new insn.
1815 2014-02-28 Jason Merrill <jason@redhat.com>
1818 * ipa-devirt.c (ipa_devirt): Don't choose an implicitly-declared
1821 2014-02-28 Paolo Carlini <paolo.carlini@oracle.com>
1824 * dwarf2out.c (decltype_auto_die): New static.
1825 (gen_subprogram_die): Handle 'decltype(auto)' like 'auto'.
1826 (gen_type_die_with_usage): Handle 'decltype(auto)'.
1827 (is_cxx_auto): Likewise.
1829 2014-02-28 Ian Bolton <ian.bolton@arm.com>
1831 * config/aarch64/aarch64.h: Define __ARM_NEON by default if
1832 we are not using general regs only.
1834 2014-02-28 Richard Biener <rguenther@suse.de>
1837 * tree-cfgcleanup.c (tree_forwarder_block_p): Restrict
1838 previous fix and only allow to remove trivial pre-headers
1839 and latches. Also honor LOOPS_MAY_HAVE_MULTIPLE_LATCHES.
1840 (remove_forwarder_block): Properly update the latch of a loop.
1842 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1845 * cselib.c (cselib_hasher::equal): Special-case VALUE lookup.
1846 (cselib_preserved_hash_table): New.
1847 (preserve_constants_and_equivs): Move preserved vals to it.
1848 (cselib_find_slot): Look it up first.
1849 (cselib_init): Initialize it.
1850 (cselib_finish): Release it.
1851 (dump_cselib_table): Dump it.
1853 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1856 * cselib.c (remove_useless_values): Skip to avoid quadratic
1857 behavior if the condition moved from...
1858 (cselib_process_insn): ... here holds.
1860 2014-02-28 Alexandre Oliva <aoliva@redhat.com>
1863 * var-tracking.c (vt_initialize): Apply the same condition to
1864 preserve the CFA base value.
1866 2014-02-28 Joey Ye <joey.ye@arm.com>
1869 * config/arm/arm.c (thumb_far_jump_used_p): Don't change
1870 if reload in progress or completed.
1872 2014-02-28 Tobias Burnus <burnus@net-b.de>
1875 * tree-pretty-print.c (dump_generic_node, print_declaration): Handle
1878 2014-02-27 H.J. Lu <hongjiu.lu@intel.com>
1880 * doc/tm.texi.in (Condition Code Status): Update documention for
1881 relative locations of cc0-setter and cc0-user.
1883 2014-02-27 Jeff Law <law@redhat.com>
1885 PR rtl-optimization/52714
1886 * combine.c (try_combine): When splitting an unrecognized PARALLEL
1887 into two independent simple sets, if I3 is a jump, ensure the
1888 pattern we place into I3 is a (set (pc) ...).
1890 2014-02-27 Mikael Pettersson <mikpe@it.uu.se>
1891 Jeff Law <law@redhat.com>
1893 PR rtl-optimization/49847
1894 * cse.c (fold_rtx) Handle case where cc0 setter and cc0 user
1895 are in different blocks.
1896 * doc/tm.texi (Condition Code Status): Update documention for
1897 relative locations of cc0-setter and cc0-user.
1899 2014-02-27 Vladimir Makarov <vmakarov@redhat.com>
1902 * lra.c (lra_emit_add): Check SUBREG too.
1904 2014-02-27 Andreas Schwab <schwab@suse.de>
1906 * config/m68k/m68k.c (m68k_option_override): Disable
1907 -flive-range-shrinkage for classic m68k.
1908 (m68k_override_options_after_change): Likewise.
1910 2014-02-27 Marek Polacek <polacek@redhat.com>
1913 * tree-ssa-uninit.c (gate_warn_uninitialized): Run the pass even for
1914 -Wmaybe-uninitialized.
1916 2014-02-27 Alan Modra <amodra@gmail.com>
1919 * reload1.c (emit_input_reload_insns): When reload_override_in,
1920 set old to rl->in_reg when rl->in_reg is a subreg.
1922 2014-02-26 Richard Biener <rguenther@suse.de>
1925 * lra-assigns.c (spill_for): Avoid mixed-sign comparison.
1927 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1929 * common/config/i386/predicates.md (const1256_operand): Remove.
1930 (const2356_operand): New.
1931 (const_1_to_2_operand): Remove.
1932 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
1933 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1934 (*avx512pf_gatherpf<mode>sf): Ditto.
1935 (avx512pf_gatherpf<mode>df): Ditto.
1936 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1937 (*avx512pf_gatherpf<mode>df): Ditto.
1938 (avx512pf_scatterpf<mode>sf): Ditto.
1939 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1940 (*avx512pf_scatterpf<mode>sf): Ditto.
1941 (avx512pf_scatterpf<mode>df): Ditto.
1942 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1943 (*avx512pf_scatterpf<mode>df): Ditto.
1944 * common/config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET0.
1946 2014-02-26 Ilya Tocar <ilya.tocar@intel.com>
1948 * config/i386/avx512fintrin.h (_mm512_testn_epi32_mask),
1949 (_mm512_mask_testn_epi32_mask), (_mm512_testn_epi64_mask),
1950 (_mm512_mask_testn_epi64_mask): Move to ...
1951 * config/i386/avx512cdintrin.h: Here.
1952 * config/i386/i386.c (bdesc_args): Change MASK_ISA for testnm.
1953 * config/i386/sse.md (avx512f_vmscalef<mode><round_name>): Remove %.
1954 (avx512f_scalef<mode><mask_name><round_name>): Ditto.
1955 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Change conditon to
1956 TARGET_AVX512F from TARGET_AVX512CD.
1958 2014-02-26 Richard Biener <rguenther@suse.de>
1961 * ipa.c (walk_polymorphic_call_targets): Properly guard
1962 call to inline_update_overall_summary.
1964 2014-02-26 Bin Cheng <bin.cheng@arm.com>
1967 * tree-cfgcleanup.c (tree_forwarder_block_p): Protect loop preheaders
1968 and latches only if requested. Fix latch if it is removed.
1969 * tree-ssa-dom.c (tree_ssa_dominator_optimize): Set
1970 LOOPS_HAVE_PREHEADERS.
1972 2014-02-25 Andrew Pinski <apinski@cavium.com>
1974 * builtins.c (expand_builtin_thread_pointer): Create a new target
1975 when the target is NULL.
1977 2014-02-25 Vladimir Makarov <vmakarov@redhat.com>
1979 PR rtl-optimization/60317
1980 * params.def (PARAM_LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1981 * params.h (LRA_MAX_CONSIDERED_RELOAD_PSEUDOS): New.
1982 * lra-assigns.c: Include params.h.
1983 (spill_for): Use LRA_MAX_CONSIDERED_RELOAD_PSEUDOS as guard for
1984 other reload pseudos considerations.
1986 2014-02-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1988 * config/rs6000/vector.md (*vector_unordered<mode>): Change split
1989 to use canonical form for nor<mode>3.
1991 2014-02-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1994 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Allow 128 to 64-bit
1997 2014-02-25 Ilya Tocar <ilya.tocar@intel.com>
1999 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PREFETCHWT1_SET),
2000 (OPTION_MASK_ISA_PREFETCHWT1_UNSET): New.
2001 (ix86_handle_option): Handle OPT_mprefetchwt1.
2002 * config/i386/cpuid.h (bit_PREFETCHWT1): New.
2003 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
2005 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2006 OPTION_MASK_ISA_PREFETCHWT1.
2007 * config/i386/i386.c (ix86_target_string): Handle mprefetchwt1.
2008 (PTA_PREFETCHWT1): New.
2009 (ix86_option_override_internal): Handle PTA_PREFETCHWT1.
2010 (ix86_valid_target_attribute_inner_p): Handle OPT_mprefetchwt1.
2011 * config/i386/i386.h (TARGET_PREFETCHWT1, TARGET_PREFETCHWT1_P): New.
2012 * config/i386/i386.md (prefetch): Check TARGET_PREFETCHWT1
2013 (*prefetch_avx512pf_<mode>_: Change into ...
2014 (*prefetch_prefetchwt1_<mode>: This.
2015 * config/i386/i386.opt (mprefetchwt1): New.
2016 * config/i386/xmmintrin.h (_mm_hint): Add _MM_HINT_ET1.
2017 (_mm_prefetch): Handle intent to write.
2018 * doc/invoke.texi (mprefetchwt1), (mno-prefetchwt1): Doccument.
2020 2014-02-25 Richard Biener <rguenther@suse.de>
2023 * emit-rtl.c (mem_attrs_htab): Remove.
2024 (mem_attrs_htab_hash): Likewise.
2025 (mem_attrs_htab_eq): Likewise.
2026 (set_mem_attrs): Always allocate new mem-attrs when something changed.
2027 (init_emit_once): Do not allocate mem_attrs_htab.
2029 2014-02-25 Richard Biener <rguenther@suse.de>
2032 * lto-opts.c (lto_write_options): Output non-explicit conservative
2033 -fwrapv, -fno-trapv and -fno-strict-overflow.
2034 * lto-wrapper.c (merge_and_complain): Handle merging those options.
2035 (run_gcc): And pass them through.
2037 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2039 * sel-sched.c (calculate_new_fences): New parameter ptime.
2040 Calculate it as a maximum over all fence cycles.
2041 (sel_sched_region_2): Adjust the call to calculate_new_fences.
2042 Print the final schedule timing when sched_verbose.
2044 2014-02-25 Andrey Belevantsev <abel@ispras.ru>
2046 PR rtl-optimization/60292
2047 * sel-sched.c (fill_vec_av_set): Do not reset target availability
2048 bit fot the fence instruction.
2050 2014-02-24 Alangi Derick <alangiderick@gmail.com>
2052 * calls.h: Fix typo in comment.
2054 2014-02-24 John David Anglin <danglin@gcc.gnu.org>
2056 * config/pa/pa.c (pa_output_move_double): Don't valididate when
2057 adjusting offsetable addresses.
2059 2014-02-24 Guozhi Wei <carrot@google.com>
2061 * sparseset.h (sparseset_pop): Fix the wrong index.
2063 2014-02-24 Walter Lee <walt@tilera.com>
2065 * config.gcc (tilepro-*-*): Change to tilepro*-*-*.
2066 (tilegx-*-linux*): Change to tilegx*-*-linux*; Support tilegxbe
2068 * common/config/tilegx/tilegx-common.c
2069 (TARGET_DEFAULT_TARGET_FLAGS): Define.
2070 * config/tilegx/linux.h (ASM_SPEC): Add endian_spec.
2072 * config/tilegx/sync.md (atomic_test_and_set): Handle big endian.
2073 * config/tilegx/tilegx.c (tilegx_return_in_msb): New.
2074 (tilegx_gimplify_va_arg_expr): Handle big endian.
2075 (tilegx_expand_unaligned_load): Ditto.
2076 (tilegx_expand_unaligned_store): Ditto.
2077 (TARGET_RETURN_IN_MSB): New.
2078 * config/tilegx/tilegx.h (TARGET_DEFAULT): New.
2079 (TARGET_ENDIAN_DEFAULT): New.
2080 (TARGET_BIG_ENDIAN): Handle big endian.
2081 (BYTES_BIG_ENDIAN): Ditto.
2082 (WORDS_BIG_ENDIAN): Ditto.
2083 (FLOAT_WORDS_BIG_ENDIAN): Ditto.
2086 * config/tilegx/tilegx.md (extv): Handle big endian.
2088 (insn_st<n>): Ditto.
2089 (insn_st<n>_add<bitsuffix>): Ditto.
2090 (insn_stnt<n>): Ditto.
2091 (insn_stnt<n>_add<bitsuffix>):Ditto.
2092 (vec_interleave_highv8qi): Handle big endian.
2093 (vec_interleave_highv8qi_be): New.
2094 (vec_interleave_highv8qi_le): New.
2095 (insn_v1int_h): Handle big endian.
2096 (vec_interleave_lowv8qi): Handle big endian.
2097 (vec_interleave_lowv8qi_be): New.
2098 (vec_interleave_lowv8qi_le): New.
2099 (insn_v1int_l): Handle big endian.
2100 (vec_interleave_highv4hi): Handle big endian.
2101 (vec_interleave_highv4hi_be): New.
2102 (vec_interleave_highv4hi_le): New.
2103 (insn_v2int_h): Handle big endian.
2104 (vec_interleave_lowv4hi): Handle big endian.
2105 (vec_interleave_lowv4hi_be): New.
2106 (vec_interleave_lowv4hi_le): New.
2107 (insn_v2int_l): Handle big endian.
2108 (vec_interleave_highv2si): Handle big endian.
2109 (vec_interleave_highv2si_be): New.
2110 (vec_interleave_highv2si_le): New.
2111 (insn_v4int_h): Handle big endian.
2112 (vec_interleave_lowv2si): Handle big endian.
2113 (vec_interleave_lowv2si_be): New.
2114 (vec_interleave_lowv2si_le): New.
2115 (insn_v4int_l): Handle big endian.
2116 * config/tilegx/tilegx.opt (mbig-endian): New option.
2117 (mlittle-endian): New option.
2118 * doc/install.texi: Document tilegxbe-linux.
2119 * doc/invoke.texi: Document -mbig-endian and -mlittle-endian.
2121 2014-02-24 Martin Jambor <mjambor@suse.cz>
2124 * ipa-cp.c (propagate_constants_accross_call): Bail out early if
2125 there are no parameter descriptors.
2127 2014-02-24 Andrey Belevantsev <abel@ispras.ru>
2129 PR rtl-optimization/60268
2130 * sched-rgn.c (haifa_find_rgns): Move the nr_regions_initial variable
2131 initialization to ...
2132 (sched_rgn_init): ... here.
2133 (schedule_region): Check for SCHED_PRESSURE_NONE earlier.
2135 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2137 * config/microblaze/microblaze.md: Correct ashrsi_reg / lshrsi_reg
2140 2014-02-23 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2142 * config/microblaze/microblaze.h: Remove SECONDARY_MEMORY_NEEDED
2145 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2147 * /config/microblaze/microblaze.c: Add microblaze_asm_output_mi_thunk,
2148 define TARGET_ASM_OUTPUT_MI_THUNK and TARGET_ASM_CAN_OUTPUT_MI_THUNK.
2150 2014-02-23 David Holsgrove <david.holsgrove@xilinx.com>
2152 * config/microblaze/predicates.md: Add cmp_op predicate.
2153 * config/microblaze/microblaze.md: Add branch_compare instruction
2154 which uses cmp_op predicate and emits cmp insn before branch.
2155 * config/microblaze/microblaze.c (microblaze_emit_compare): Rename
2156 to microblaze_expand_conditional_branch and consolidate logic.
2157 (microblaze_expand_conditional_branch): emit branch_compare
2158 insn instead of handling cmp op separate from branch insn.
2160 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2162 * config/rs6000/rs6000.c (rs6000_emit_le_vsx_move): Relax assert
2165 2014-02-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2167 * config/rs6000/altivec.md (altivec_lve<VI_char>x): Replace
2168 define_insn with define_expand and new define_insn
2169 *altivec_lve<VI_char>x_internal.
2170 (altivec_stve<VI_char>x): Replace define_insn with define_expand
2171 and new define_insn *altivec_stve<VI_char>x_internal.
2172 * config/rs6000/rs6000-protos.h (altivec_expand_stvex_be): New
2174 * config/rs6000/rs6000.c (altivec_expand_lvx_be): Document use by
2176 (altivec_expand_stvex_be): New function.
2178 2014-02-22 Joern Rennecke <joern.rennecke@embecosm.com>
2180 * config/avr/avr.c (avr_can_eliminate): Allow elimination from
2181 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM if !frame_pointer_needed.
2182 * config/avr/avr.c (ELIMINABLE_REGS): Add elimination from
2183 ARG_POINTER_REGNUM to STACK_POINTER_REGNUM.
2185 2014-02-21 Vladimir Makarov <vmakarov@redhat.com>
2188 * lra-constraints.c (inherit_reload_reg): Use lra_emit_move
2189 instead of emit_move_insn.
2191 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2193 * config/rs6000/altivec.md (altivec_vsumsws): Replace second
2195 (reduc_uplus_v16qi): Use gen_altivec_vsumsws_direct instead of
2196 gen_altivec_vsumsws.
2198 2014-02-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2200 * config/rs6000/altivec.md (altivec_lvxl): Rename as
2201 *altivec_lvxl_<mode>_internal and use VM2 iterator instead of V4SI.
2202 (altivec_lvxl_<mode>): New define_expand incorporating
2203 -maltivec=be semantics where needed.
2204 (altivec_lvx): Rename as *altivec_lvx_<mode>_internal.
2205 (altivec_lvx_<mode>): New define_expand incorporating -maltivec=be
2206 semantics where needed.
2207 (altivec_stvx): Rename as *altivec_stvx_<mode>_internal.
2208 (altivec_stvx_<mode>): New define_expand incorporating
2209 -maltivec=be semantics where needed.
2210 (altivec_stvxl): Rename as *altivec_stvxl_<mode>_internal and use
2211 VM2 iterator instead of V4SI.
2212 (altivec_stvxl_<mode>): New define_expand incorporating
2213 -maltivec=be semantics where needed.
2214 * config/rs6000/rs6000-builtin.def: Add new built-in definitions
2215 LVXL_V2DF, LVXL_V2DI, LVXL_V4SF, LVXL_V4SI, LVXL_V8HI, LVXL_V16QI,
2216 LVX_V2DF, LVX_V2DI, LVX_V4SF, LVX_V4SI, LVX_V8HI, LVX_V16QI, STVX_V2DF,
2217 STVX_V2DI, STVX_V4SF, STVX_V4SI, STVX_V8HI, STVX_V16QI, STVXL_V2DF,
2218 STVXL_V2DI, STVXL_V4SF, STVXL_V4SI, STVXL_V8HI, STVXL_V16QI.
2219 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Replace
2220 ALTIVEC_BUILTIN_LVX with ALTIVEC_BUILTIN_LVX_<MODE> throughout;
2221 similarly for ALTIVEC_BUILTIN_LVXL, ALTIVEC_BUILTIN_STVX, and
2222 ALTIVEC_BUILTIN_STVXL.
2223 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be): New prototype.
2224 (altivec_expand_stvx_be): Likewise.
2225 * config/rs6000/rs6000.c (swap_selector_for_mode): New function.
2226 (altivec_expand_lvx_be): Likewise.
2227 (altivec_expand_stvx_be): Likewise.
2228 (altivec_expand_builtin): Add cases for
2229 ALTIVEC_BUILTIN_STVX_<MODE>, ALTIVEC_BUILTIN_STVXL_<MODE>,
2230 ALTIVEC_BUILTIN_LVXL_<MODE>, and ALTIVEC_BUILTIN_LVX_<MODE>.
2231 (altivec_init_builtins): Add definitions for
2232 __builtin_altivec_lvxl_<mode>, __builtin_altivec_lvx_<mode>,
2233 __builtin_altivec_stvx_<mode>, and __builtin_altivec_stvxl_<mode>.
2235 2014-02-21 Catherine Moore <clm@codesourcery.com>
2237 * doc/invoke.texi (mvirt, mno-virt): Document.
2238 * config/mips/mips.opt (mvirt): New option.
2239 * config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.
2241 2014-02-21 Richard Biener <rguenther@suse.de>
2243 PR tree-optimization/60276
2244 * tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
2245 (STMT_VINFO_MIN_NEG_DIST): New macro.
2246 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
2247 STMT_VINFO_MIN_NEG_DIST.
2248 * tree-vect-stmts.c (vectorizable_load): Verify if assumptions
2249 made for negative dependence distances still hold.
2251 2014-02-21 Richard Biener <rguenther@suse.de>
2254 * tree-ssa-live.c (mark_all_vars_used_1): Do not walk
2255 DECL_INITIAL for globals not in the current function context.
2257 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2259 PR tree-optimization/56490
2260 * params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
2261 * tree-ssa-uninit.c: Include params.h.
2262 (compute_control_dep_chain): Add num_calls argument, return false
2263 if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
2264 num_calls to recursive call.
2265 (find_predicates): Change dep_chain into normal array,
2266 cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
2267 variable and adjust compute_control_dep_chain caller.
2268 (find_def_preds): Likewise.
2270 2014-02-21 Thomas Schwinge <thomas@codesourcery.com>
2272 * gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
2273 <case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.
2275 2014-02-21 Nick Clifton <nickc@redhat.com>
2277 * config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
2278 (pushhi1): Likewise.
2279 (popqi1): Add mode to pre_dec.
2282 2014-02-21 Jakub Jelinek <jakub@redhat.com>
2284 * config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
2285 mode for mask of V8SFmode permutation.
2287 2014-02-20 Richard Henderson <rth@redhat.com>
2290 * builtins.c (expand_builtin_atomic_compare_exchange): Always make
2291 a new pseudo for OLDVAL.
2293 2014-02-20 Jakub Jelinek <jakub@redhat.com>
2296 * config/i386/i386.c (expand_vec_perm_interleave2): Don't call
2297 gen_reg_rtx if d->testing_p.
2298 (expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
2299 if d->testing_p and we will certainly return true.
2300 (expand_vec_perm_even_odd_1): Likewise. Don't call gen_reg_rtx
2303 2014-02-20 Uros Bizjak <ubizjak@gmail.com>
2305 * emit-rtl.c (gen_reg_rtx): Assert that
2306 crtl->emit.regno_pointer_align_length is non-zero.
2308 2014-02-20 Richard Henderson <rth@redhat.com>
2311 * builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
2312 on failure the store back into EXPECT.
2314 2014-02-20 Chung-Lin Tang <cltang@codesourcery.com>
2315 Sandra Loosemore <sandra@codesourcery.com>
2317 * config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
2318 * config/nios2/nios2.c (nios2_function_profiler): Add
2319 -fPIC (flag_pic == 2) support.
2320 (nios2_handle_custom_fpu_cfg): Fix warning parameter.
2321 (nios2_large_offset_p): New function.
2322 (nios2_unspec_reloc_p): Move up position, update to use
2323 nios2_large_offset_p.
2324 (nios2_unspec_address): Remove function.
2325 (nios2_unspec_offset): New function.
2326 (nios2_large_got_address): New function.
2327 (nios2_got_address): Add large offset support.
2328 (nios2_legitimize_tls_address): Update usage of removed and new
2330 (nios2_symbol_binds_local_p): New function.
2331 (nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
2332 (nios2_legitimize_address): Update to use nios2_large_offset_p.
2333 (nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
2334 (nios2_print_operand): Merge H/L processing, add hiadj/lo
2335 processing for (const (unspec ...)).
2336 (nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.
2338 2014-02-20 Richard Biener <rguenther@suse.de>
2340 * tree-cfg.c (replace_uses_by): Mark altered BBs before
2341 doing the substitution.
2342 (verify_gimple_assign_single): Also verify bare MEM_REFs on the lhs.
2344 2014-02-20 Martin Jambor <mjambor@suse.cz>
2347 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
2348 info when checking whether lattices are bottom.
2350 2014-02-20 Richard Biener <rguenther@suse.de>
2353 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
2356 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
2359 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale
2360 parameter specifying the scaling.
2361 (inline_call): Update.
2362 (want_inline_recursively): Guard division by zero.
2363 (recursive_inlining): Update.
2364 * ipa-inline.h (clone_inlined_nodes): Update.
2366 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2369 * config/i386/i386.c (classify_argument): Pass structures of size
2370 64 bytes or less in register.
2372 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
2373 Kirill Yukhin <kirill.yukhin@intel.com>
2375 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
2376 (_mm_rcp28_round_ss): Ditto.
2377 (_mm_rsqrt28_round_sd): Ditto.
2378 (_mm_rsqrt28_round_ss): Ditto.
2379 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
2380 (_mm_rcp14_round_ss): Ditto.
2381 (_mm_rsqrt14_round_sd): Ditto.
2382 (_mm_rsqrt14_round_ss): Ditto.
2383 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
2384 the first input operand, get rid of match_dup.
2385 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
2387 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2389 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
2390 operand as the first input operand, set type attribute.
2391 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2393 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
2394 operand as the first input operand, set type attribute.
2396 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2398 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
2401 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
2404 * config/i386/i386.c (construct_container): Remove TFmode check
2405 for X86_64_INTEGER_CLASS.
2407 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
2410 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
2411 only when -Wpsabi is enabled.
2413 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
2416 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
2417 passing arrays in registers are the same as for structs, so remove the
2418 special case for them.
2420 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
2422 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
2423 destination type, extract only the valid bits if the source type is not
2424 integral and has a different mode.
2426 2014-02-19 Richard Biener <rguenther@suse.de>
2429 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
2432 2014-02-19 Richard Biener <rguenther@suse.de>
2435 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
2436 (ipa_modify_call_arguments): Emit an argument load explicitely and
2437 preserve virtual SSA form there and for the replacement call.
2438 Do not update SSA form nor free dominance info.
2440 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2442 * ipa.c (function_and_variable_visibility): Also clear WEAK
2443 flag when disolving COMDAT_GROUP.
2445 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2447 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
2448 * ipa-prop.c (ipa_set_jf_known_type): Return early when
2450 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
2451 do more sanity checks.
2452 (detect_type_change): Return true when giving up early.
2453 (compute_complex_assign_jump_func): Fix type parameter of
2454 ipa_set_ancestor_jf.
2455 (compute_complex_ancestor_jump_func): Likewise.
2456 (update_jump_functions_after_inlining): Fix updating of
2458 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
2460 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
2462 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
2463 inline clones when edge disappears.
2465 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
2468 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
2469 Split 64-bit moves into 2 patterns. Do not allow the use of
2470 direct move for TDmode in little endian, since the decimal value
2471 has little endian bytes within a word, but the 64-bit pieces are
2472 ordered in a big endian fashion, and normal subreg's of TDmode are
2474 (mov<mode>_64bit_dm): Likewise.
2475 (movtd_64bit_nodm): Likewise.
2477 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2479 PR tree-optimization/60174
2480 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
2481 statement of an SSA_NAME that occurs in an abnormal PHI node.
2483 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2486 * final.c (SEEN_BB): Remove.
2487 (SEEN_NOTE, SEEN_EMITTED): Renumber.
2488 (final_scan_insn): Don't force_source_line on second
2489 NOTE_INSN_BASIC_BLOCK.
2491 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
2494 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
2495 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
2496 (type_natural_mode): Warn ABI change when %zmm register is not
2497 available for AVX512F vector value passing.
2499 2014-02-18 Kai Tietz <ktietz@redhat.com>
2502 * config/i386/i386.c (ix86_expand_prologue): Use value in
2503 rax register as displacement when restoring %r10 or %rax.
2504 Fix wrong offset when restoring both registers.
2506 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
2508 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
2509 assertion with conditional return.
2511 2014-02-18 Jakub Jelinek <jakub@redhat.com>
2512 Uros Bizjak <ubizjak@gmail.com>
2515 * config/i386/driver-i386.c (host_detect_local_cpu): If
2516 YMM state is not saved by the OS, also clear has_f16c. Move
2517 CPUID 0x80000001 handling before YMM state saving checking.
2519 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
2521 PR rtl-optimization/58960
2522 * haifa-sched.c (alloc_global_sched_pressure_data): New,
2523 factored out from ...
2524 (sched_init): ... here.
2525 (free_global_sched_pressure_data): New, factored out from ...
2526 (sched_finish): ... here.
2527 * sched-int.h (free_global_sched_pressure_data): Declare.
2528 * sched-rgn.c (nr_regions_initial): New static global.
2529 (haifa_find_rgns): Initialize it.
2530 (schedule_region): Disable sched-pressure for the newly
2533 2014-02-17 Richard Biener <rguenther@suse.de>
2535 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
2536 release SSA defs of pattern stmts.
2538 2014-02-17 Richard Biener <rguenther@suse.de>
2540 * tree-inline.c (expand_call_inline): Release the virtual
2541 operand defined by the call we are about to inline.
2543 2014-02-17 Richard Biener <rguenther@suse.de>
2545 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
2547 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
2548 Ilya Tocar <ilya.tocar@intel.com>
2550 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
2551 arguments order in builtin.
2552 (_mm512_permutexvar_epi64): Ditto.
2553 (_mm512_mask_permutexvar_epi64): Ditto
2554 (_mm512_maskz_permutexvar_epi32): Ditto
2555 (_mm512_permutexvar_epi32): Ditto
2556 (_mm512_mask_permutexvar_epi32): Ditto
2558 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2560 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
2561 (p8_vmrgow): Likewise.
2563 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2565 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
2568 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2571 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
2572 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
2573 into 64-bit and 32-bit moves. On 64-bit moves, add support for
2574 using direct move instructions on ISA 2.07. Also adjust
2575 instruction length for 64-bit.
2576 (mov<mode>_64bit, TFmode/TDmode): Likewise.
2577 (mov<mode>_32bit, TFmode/TDmode): Likewise.
2579 2014-02-15 Alan Modra <amodra@gmail.com>
2583 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
2584 find_replacement on parts of insn rtl that might be reloaded.
2586 2014-02-15 Richard Biener <rguenther@suse.de>
2588 PR tree-optimization/60183
2589 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
2590 (tree_ssa_phiprop): Calculate and free post-dominators.
2592 2014-02-14 Jeff Law <law@redhat.com>
2594 PR rtl-optimization/60131
2595 * ree.c (get_extended_src_reg): New function.
2596 (combine_reaching_defs): Use it rather than assuming location of REG.
2597 (find_and_remove_re): Verify first operand of extension is
2598 a REG before adding the insns to the copy list.
2600 2014-02-14 Roland McGrath <mcgrathr@google.com>
2602 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
2603 * configure: Regenerated.
2604 * config.in: Regenerated.
2605 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
2606 instead of ASM_SHORT.
2608 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
2609 Richard Earnshaw <rearnsha@arm.com>
2611 PR rtl-optimization/59535
2612 * lra-constraints.c (process_alt_operands): Encourage alternative
2613 when unassigned pseudo class is superset of the alternative class.
2614 (inherit_reload_reg): Don't inherit when optimizing for code size.
2615 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
2616 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
2617 modes not less than 4 for Thumb1.
2619 2014-02-14 Kyle McMartin <kyle@redhat.com>
2622 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
2624 2014-02-14 Richard Biener <rguenther@suse.de>
2626 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
2627 (get_frame_arg): Drop the assert with langhook types_compatible_p.
2628 Do not strip INDIRECT_REFs.
2630 2014-02-14 Richard Biener <rguenther@suse.de>
2633 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
2634 DECL_FUNCTION_SPECIFIC_TARGET.
2635 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
2636 * tree-streamer-out.c (pack_ts_target_option): Remove.
2637 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
2638 (write_ts_function_decl_tree_pointers): Do not stream
2639 DECL_FUNCTION_SPECIFIC_TARGET.
2640 * tree-streamer-in.c (unpack_ts_target_option): Remove.
2641 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
2642 (lto_input_ts_function_decl_tree_pointers): Do not stream
2643 DECL_FUNCTION_SPECIFIC_TARGET.
2645 2014-02-14 Jakub Jelinek <jakub@redhat.com>
2647 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
2648 (get_initial_def_for_induction, vectorizable_induction): Ignore
2649 debug stmts when looking for exit_phi.
2650 (vectorizable_live_operation): Fix up condition.
2652 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2654 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
2655 nreverse() because it changes the content of original tree list.
2657 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2659 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
2660 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
2662 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
2664 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
2665 GNU coding standards.
2667 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2670 * dwarf2out.c (gen_subprogram_die): Don't call
2671 add_calling_convention_attribute if subr_die is old_die.
2673 2014-02-13 Sharad Singhai <singhai@google.com>
2675 * doc/optinfo.texi: Fix order of nodes.
2677 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
2679 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
2680 operands[2], not operands[3].
2682 2014-02-13 Richard Biener <rguenther@suse.de>
2685 * doc/install.texi (ISL): Update recommended version to 0.12.2,
2686 mention the possibility of an in-tree build.
2687 (CLooG): Update recommended version to 0.18.1, mention the
2688 possibility of an in-tree build and clarify that the ISL
2689 bundled with CLooG does not work.
2691 2014-02-13 Jakub Jelinek <jakub@redhat.com>
2694 * expr.c (compress_float_constant): If x is a hard register,
2695 extend into a pseudo and then move to x.
2697 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
2699 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
2700 caused by bad second argument to warning_at() with -mhotpatch and
2701 nested functions (e.g. with gfortran).
2703 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
2705 * opts.c (option_name): Remove "enabled by default" rider.
2707 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
2709 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
2711 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
2712 Uros Bizjak <ubizjak@gmail.com>
2715 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
2716 * configure: Regenerated.
2718 2014-02-12 Richard Biener <rguenther@suse.de>
2720 * vec.c (vec_prefix::calculate_allocation): Move as
2721 inline variant to vec.h.
2722 (vec_prefix::calculate_allocation_1): New out-of-line version.
2723 * vec.h (vec_prefix::calculate_allocation_1): Declare.
2724 (vec_prefix::m_has_auto_buf): Rename to ...
2725 (vec_prefix::m_using_auto_storage): ... this.
2726 (vec_prefix::calculate_allocation): Inline the easy cases
2727 and dispatch to calculate_allocation_1 which doesn't need the
2729 (va_heap::reserve): Use gcc_checking_assert.
2730 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
2731 m_using_auto_storage.
2732 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
2734 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
2735 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
2736 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
2738 2014-02-12 Richard Biener <rguenther@suse.de>
2740 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
2741 when we found a dependence.
2743 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
2745 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
2747 (maybe_fold_stmt): ... into this new function.
2748 * omp-low.c (lower_omp): Update comment.
2750 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
2753 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
2756 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
2758 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
2759 identifiers in comments.
2760 (cortexa53_extra_costs): Likewise.
2761 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
2762 (cortexa7_extra_costs): Likewise.
2763 (cortexa12_extra_costs): Likewise.
2764 (cortexa15_extra_costs): Likewise.
2765 (v7m_extra_costs): Likewise.
2767 2014-02-12 Richard Biener <rguenther@suse.de>
2770 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
2771 of posix_memalign being successful.
2772 (lower_stmt): Restrict lowering of posix_memalign to when
2773 -ftree-bit-ccp is enabled.
2775 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
2777 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
2779 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
2781 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
2783 PR rtl-optimization/60116
2784 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
2785 other_insn once the combination has been validated.
2787 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
2790 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
2792 * ipa-devirt.c: Include demangle.h
2793 (odr_violation_reported): New static variable.
2794 (add_type_duplicate): Update odr_violations.
2795 (maybe_record_node): Add completep parameter; update it.
2796 (record_target_from_binfo): Add COMPLETEP parameter;
2797 update it as needed.
2798 (possible_polymorphic_call_targets_1): Likewise.
2799 (struct polymorphic_call_target_d): Add nonconstruction_targets;
2800 rename FINAL to COMPLETE.
2801 (record_targets_from_bases): Sanity check we found the binfo;
2802 fix COMPLETEP updating.
2803 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
2804 parameter, fix computing of COMPLETEP.
2805 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
2806 at LTO time do demangling.
2807 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
2808 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
2810 (gimple_get_virt_method_for_binfo): Likewise.
2811 * gimple-fold.h (gimple_get_virt_method_for_binfo,
2812 gimple_get_virt_method_for_vtable): Update prototypes.
2814 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
2817 * genautomata.c (add_presence_absence): Fix typo with
2818 {final_}presence_list.
2820 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
2823 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
2824 for VSX/Altivec vectors that land in GPR registers.
2826 2014-02-11 Richard Henderson <rth@redhat.com>
2827 Jakub Jelinek <jakub@redhat.com>
2830 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
2831 around drhs if type conversion to lacc->type is not useless.
2833 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2835 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
2837 (cortex-a57.cortex-a53): Likewise.
2838 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
2840 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2842 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
2845 2014-02-11 Renlin Li <Renlin.Li@arm.com>
2847 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
2848 add_options_for_arm_vfp3.
2850 2014-02-11 Jeff Law <law@redhat.com>
2853 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
2854 object with an undesirable mode.
2856 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2859 * config/i386/sol2-9.h: New file.
2860 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
2861 *-*-solaris2.9*): Use it.
2863 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2865 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
2866 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
2868 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
2870 * config/microblaze/microblaze.c: Extend mcpu version format
2872 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
2874 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
2876 2014-02-10 Richard Henderson <rth@redhat.com>
2879 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
2880 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
2881 ms-abi vs -mno-accumulate-outgoing-args.
2882 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
2883 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
2886 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2889 * cfgexpand.c (expand_asm_operands): Attach source location to
2890 ASM_INPUT rtx objects.
2891 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
2893 2014-02-10 Nick Clifton <nickc@redhat.com>
2895 * config/mn10300/mn10300.c (popcount): New function.
2896 (mn10300_expand_prologue): Include saved registers in stack usage
2899 2014-02-10 Jeff Law <law@redhat.com>
2902 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
2903 when changing the SET_DEST of a prior insn to avoid an input reload.
2905 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
2907 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
2908 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
2909 -mcall-openbsd, or -mcall-linux.
2910 (CC1_ENDIAN_BIG_SPEC): Remove.
2911 (CC1_ENDIAN_LITTLE_SPEC): Remove.
2912 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2913 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
2914 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
2915 and %cc1_endian_default.
2916 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
2918 2014-02-10 Richard Biener <rguenther@suse.de>
2920 PR tree-optimization/60115
2921 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
2922 MEM_REF handling. Properly verify that the accesses are not
2923 out of the objects bound.
2925 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2927 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
2930 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
2932 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
2933 proper constants and fix formatting.
2934 (possible_polymorphic_call_targets): Fix formatting.
2936 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
2937 Ilya Tocar <ilya.tocar@intel.com>
2939 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
2940 (_mm512_loadu_epi32): Renamed into...
2941 (_mm512_loadu_si512): This.
2942 (_mm512_storeu_epi32): Renamed into...
2943 (_mm512_storeu_si512): This.
2944 (_mm512_maskz_ceil_ps): Removed.
2945 (_mm512_maskz_ceil_pd): Ditto.
2946 (_mm512_maskz_floor_ps): Ditto.
2947 (_mm512_maskz_floor_pd): Ditto.
2948 (_mm512_floor_round_ps): Ditto.
2949 (_mm512_floor_round_pd): Ditto.
2950 (_mm512_ceil_round_ps): Ditto.
2951 (_mm512_ceil_round_pd): Ditto.
2952 (_mm512_mask_floor_round_ps): Ditto.
2953 (_mm512_mask_floor_round_pd): Ditto.
2954 (_mm512_mask_ceil_round_ps): Ditto.
2955 (_mm512_mask_ceil_round_pd): Ditto.
2956 (_mm512_maskz_floor_round_ps): Ditto.
2957 (_mm512_maskz_floor_round_pd): Ditto.
2958 (_mm512_maskz_ceil_round_ps): Ditto.
2959 (_mm512_maskz_ceil_round_pd): Ditto.
2960 (_mm512_expand_pd): Ditto.
2961 (_mm512_expand_ps): Ditto.
2962 * config/i386/i386.c (ix86_builtins): Remove
2963 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
2964 (bdesc_args): Ditto.
2965 * config/i386/predicates.md (const1256_operand): New.
2966 (const_1_to_2_operand): Ditto.
2967 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
2968 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
2969 (*avx512pf_gatherpf<mode>sf): Ditto.
2970 (avx512pf_gatherpf<mode>df): Ditto.
2971 (*avx512pf_gatherpf<mode>df_mask): Ditto.
2972 (*avx512pf_gatherpf<mode>df): Ditto.
2973 (avx512pf_scatterpf<mode>sf): Ditto.
2974 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
2975 (*avx512pf_scatterpf<mode>sf): Ditto.
2976 (avx512pf_scatterpf<mode>df): Ditto.
2977 (*avx512pf_scatterpf<mode>df_mask): Ditto.
2978 (*avx512pf_scatterpf<mode>df): Ditto.
2979 (avx512f_expand<mode>): Removed.
2980 (<shift_insn><mode>3<mask_name>): Change predicate type.
2982 2014-02-08 Jakub Jelinek <jakub@redhat.com>
2984 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
2985 not at the end of datarefs vector use ordered_remove to avoid
2986 reordering datarefs vector.
2989 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
2990 mark local addressable non-static vars as GOVD_PRIVATE
2991 instead of GOVD_LOCAL.
2992 * omp-low.c (lower_omp_for): Move gimple_bind_vars
2993 and BLOCK_VARS of gimple_bind_block to new_stmt rather
2997 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
2998 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
2999 assume_aligned or alloc_align attributes.
3000 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
3001 arguments. Handle also assume_aligned and alloc_align attributes.
3002 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
3003 calls to functions with assume_aligned or alloc_align attributes.
3004 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
3006 2014-02-08 Terry Guo <terry.guo@arm.com>
3008 * doc/invoke.texi: Document ARM -march=armv7e-m.
3010 2014-02-08 Jakub Jelinek <jakub@redhat.com>
3012 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
3013 flag on __cilkrts_rethrow builtin.
3016 * ipa-cp.c (determine_versionability): Fail at -O0
3017 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
3018 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
3021 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3024 * tree-inline.c (copy_forbidden): Fail for
3025 __attribute__((optimize (0))) functions.
3027 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3029 * varpool.c: Include pointer-set.h.
3030 (varpool_remove_unreferenced_decls): Variables in other partitions
3031 will not be output; be however careful to not lose information
3034 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
3036 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
3037 lookup in the vtable constructor.
3039 2014-02-07 Jeff Law <law@redhat.com>
3042 * config/m68k/m68k.md (ashldi_extsi): Turn into a
3043 define_insn_and_split.
3045 * ipa-inline.c (inline_small_functions): Fix typos.
3047 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3049 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
3050 (s390_can_use_return_insn): Declare.
3051 * config/s390/s390.h (EPILOGUE_USES): Define.
3052 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
3054 (s390_chunkify_start): Handle return JUMP_LABELs.
3055 (s390_early_mach): Emit a main_pool instruction on the entry edge.
3056 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
3057 (s390_can_use_return_insn): New functions.
3058 (s390_fix_long_loop_prediction): Handle conditional returns.
3059 (TARGET_SET_UP_BY_PROLOGUE): Define.
3060 * config/s390/s390.md (ANY_RETURN): New code iterator.
3061 (*creturn, *csimple_return, return, simple_return): New patterns.
3063 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3065 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
3066 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
3067 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
3068 REG_CFA_RESTORE list when deciding not to restore a register.
3070 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3072 * config/s390/s390.c: Include tree-pass.h and context.h.
3073 (s390_early_mach): New function, split out from...
3074 (s390_emit_prologue): ...here.
3075 (pass_data_s390_early_mach): New pass structure.
3076 (pass_s390_early_mach): New class.
3077 (s390_option_override): Create and register early_mach pass.
3078 Move to end of file.
3080 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3082 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
3083 to match for the exit block.
3085 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3087 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
3088 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
3089 Reject misaligned operands.
3091 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3093 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
3095 2014-02-07 Richard Biener <rguenther@suse.de>
3098 * gimple-low.c (lower_builtin_posix_memalign): New function.
3099 (lower_stmt): Call it to lower posix_memalign in a way
3100 to make alignment info accessible.
3102 2014-02-07 Jakub Jelinek <jakub@redhat.com>
3105 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
3106 __builtin_setjmp_receiver.
3108 2014-02-07 Richard Biener <rguenther@suse.de>
3111 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
3112 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
3113 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
3114 Handle BUILT_IN_POSIX_MEMALIGN.
3115 (find_func_clobbers): Likewise.
3116 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3117 (call_may_clobber_ref_p_1): Likewise.
3119 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3122 * ipa-devirt.c (record_target_from_binfo): Remove overactive
3125 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3128 * lto-cgraph.c (lto_output_node): Use
3129 symtab_get_symbol_partitioning_class.
3130 (lto_output_varpool_node): likewise.
3131 (symtab_get_symbol_partitioning_class): Move here from
3133 * cgraph.h (symbol_partitioning_class): Likewise.
3134 (symtab_get_symbol_partitioning_class): Declare.
3136 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3138 * ggc.h (ggc_internal_cleared_alloc): New macro.
3139 * vec.h (vec_safe_copy): Handle memory stats.
3140 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
3141 * target-globals.c (save_target_globals): Likewise.
3143 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
3146 * expr.c (emit_move_resolve_push): Export; be bit more selective
3147 on when to clear alias set.
3148 * expr.h (emit_move_resolve_push): Declare.
3149 * function.h (struct function): Add tail_call_marked.
3150 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
3151 * config/i386/i386-protos.h (ix86_expand_push): Remove.
3152 * config/i386/i386.md (TImode move expander): De not call
3154 (FP push expanders): Preserve memory attributes.
3155 * config/i386/sse.md (push<mode>1): Remove.
3156 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
3157 (ix86_expand_push): Remove.
3158 * config/i386/mmx.md (push<mode>1): Remove.
3160 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3162 PR rtl-optimization/60030
3163 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
3164 lopart with paradoxical subreg before shifting it up by hprec.
3166 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3168 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
3169 Remove extra newline at end of file.
3170 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
3171 (arm_issue_rate): Handle cortexa57.
3172 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
3173 (cortex-a57.cortex-a53): Likewise.
3175 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3178 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
3179 don't record in REG_FRAME_RELATED_EXPR registers not set in that
3181 (arm_expand_prologue): Adjust all callers.
3182 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
3183 info, registers also at the lowest numbered registers side. Use
3184 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
3188 * var-tracking.c (adjust_mems): Before adding a SET to
3189 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
3191 2014-02-06 Alan Modra <amodra@gmail.com>
3194 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
3195 change SDmode to DDmode when lra_in_progress.
3197 2014-02-06 Jakub Jelinek <jakub@redhat.com>
3200 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
3201 free_data_ref on the dr first, and before goto again also set dr
3202 to the next dr. For simd_lane_access, free old datarefs[i] before
3203 overwriting it. For get_vectype_for_scalar_type failure, don't
3204 free_data_ref if simd_lane_access.
3206 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
3209 * tree.h (opts_for_fn): New inline function.
3210 (opt_for_fn): Define.
3211 * config/i386/i386.c (ix86_function_regparm): Use
3212 opt_for_fn (decl, optimize) instead of optimize.
3214 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
3216 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
3217 for SYMBOL_REF in large memory model.
3219 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3221 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
3223 (cortex-a57): Likewise.
3224 (cortex-a57.cortex-a53): Likewise.
3226 2014-02-06 Yury Gribov <y.gribov@samsung.com>
3227 Kugan Vivekanandarajah <kuganv@linaro.org>
3229 * config/arm/arm.c (arm_vector_alignment_reachable): Check
3231 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
3233 2014-02-06 Richard Biener <rguenther@suse.de>
3235 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
3236 set_loop_copy and initialize_original_copy_tables.
3238 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
3240 * config/aarch64/aarch64-simd.md
3241 (aarch64_ashr_simddi): Change QI to SI.
3243 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3244 Jakub Jelinek <jakub@redhat.com>
3247 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
3250 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3252 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
3253 CODE_FOR_altivec_vpku[hw]um to
3254 CODE_FOR_altivec_vpku[hw]um_direct.
3255 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
3256 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
3257 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
3258 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
3260 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3262 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
3263 generation for -maltivec=be.
3264 (altivec_vsumsws): Simplify redundant test.
3266 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3268 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
3269 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
3270 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
3271 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
3272 gen_altivec_vpkuwum.
3273 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
3275 (altivec_vpks<VI_char>ss): Likewise.
3276 (altivec_vpks<VI_char>us): Likewise.
3277 (altivec_vpku<VI_char>us): Likewise.
3278 (altivec_vpku<VI_char>um): Likewise.
3279 (altivec_vpku<VI_char>um_direct): New (copy of
3280 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
3282 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
3283 target is little endian and -maltivec=be is not specified.
3284 (*altivec_vupkhs<VU_char>_direct): New (copy of
3285 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
3286 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
3287 target is little endian and -maltivec=be is not specified.
3288 (*altivec_vupkls<VU_char>_direct): New (copy of
3289 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
3290 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
3291 little endian and -maltivec=be is not specified.
3292 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
3293 little endian and -maltivec=be is not specified.
3295 2014-02-05 Richard Henderson <rth@redhat.com>
3298 * combine-stack-adj.c: Revert r206943.
3299 * sched-int.h (struct deps_desc): Add last_args_size.
3300 * sched-deps.c (init_deps): Initialize it.
3301 (sched_analyze_insn): Add OUTPUT dependencies between insns that
3302 contain REG_ARGS_SIZE notes.
3304 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
3306 * lto-cgraph.c (asm_nodes_output): Make global.
3307 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
3308 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
3309 (driver_handle_option): Handle OPT_fwpa.
3311 2014-02-05 Jakub Jelinek <jakub@redhat.com>
3314 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
3315 a comment typo and formatting issue. If odr_hash hasn't been
3316 created, return vNULL and set *completep to false.
3319 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
3320 bb with no successors.
3322 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
3325 * doc/invoke.texi (-march): Clarify documentation for ARM.
3329 2014-02-05 Richard Biener <rguenther@suse.de>
3331 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
3332 when not vectorizing because of too many alias checks.
3333 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
3334 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
3336 2014-02-05 Nick Clifton <nickc@redhat.com>
3338 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
3339 accept extended registers in any mode when compiling for the MN10300.
3341 2014-02-05 Yury Gribov <y.gribov@samsung.com>
3343 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
3344 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
3345 sanitization attributes.
3346 (can_inline_edge_p): Likewise.
3347 (sanitize_attrs_match_for_inline_p): New function.
3349 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3351 * ipa-prop.c (detect_type_change): Shor circuit testing of
3352 type changes on THIS pointer.
3354 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
3357 * config/pa/pa.c (legitimize_tls_address): Return original address
3358 if not passed a SYMBOL_REF rtx.
3359 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
3361 (pa_emit_move_sequence): Simplify TLS source operands.
3362 (pa_legitimate_constant_p): Reject all TLS constants.
3363 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
3364 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
3366 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3368 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
3369 groups when we know they are controlled by LTO.
3370 * varasm.c (default_binds_local_p_1): If object is in other partition,
3371 it will be resolved locally.
3373 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3375 * config/host-linux.c (linux_gt_pch_use_address): Don't
3376 use SSIZE_MAX because it is not always defined.
3378 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
3381 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
3382 threshold for pseudo splitting.
3383 (update_ebb_live_info): Process call argument hard registers and
3384 hard registers from insn definition too.
3385 (max_small_class_regs_num): New constant.
3386 (inherit_in_ebb): Update live hard regs through EBBs. Update
3387 reloads_num only for small register classes. Don't split for
3390 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
3393 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
3396 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
3398 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
3401 2014-02-04 Marek Polacek <polacek@redhat.com>
3403 * gdbinit.in (pel): Define.
3405 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3407 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
3410 2014-02-04 Richard Biener <rguenther@suse.de>
3413 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
3414 in function context local.
3415 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
3416 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
3417 similar to LTO_imported_decl_ref.
3419 2014-02-04 Jakub Jelinek <jakub@redhat.com>
3421 PR tree-optimization/60002
3422 * cgraphclones.c (build_function_decl_skip_args): Clear
3425 PR tree-optimization/60023
3426 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
3427 false to gsi_replace.
3428 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
3429 has been in some EH region and vec_stmt could throw, add
3430 vec_stmt into the same EH region.
3431 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
3432 has no lhs, ignore it.
3433 * internal-fn.c (expand_MASK_LOAD): Likewise.
3436 * tree-inline.c (copy_forbidden): Fail for
3437 __attribute__((optimize (0))) functions.
3440 * omp-low.c (simd_clone_struct_copy): If from->inbranch
3441 is set, copy one less argument.
3442 (expand_simd_clones): Don't subtract clone_info->inbranch
3443 from simd_clone_struct_alloc argument.
3445 PR rtl-optimization/57915
3446 * recog.c (simplify_while_replacing): If all unary/binary/relational
3447 operation arguments are constant, attempt to simplify those.
3450 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
3451 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
3453 2014-02-04 Richard Biener <rguenther@suse.de>
3455 PR tree-optimization/60012
3456 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
3457 TBAA disambiguation to all DDRs.
3459 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3462 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
3463 (LINK_SPEC): Use it for -shared, -shared-libgcc.
3465 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3468 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
3470 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3472 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
3473 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
3475 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3478 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
3479 to figure out targets of polymorphic calls with known decl.
3480 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3481 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
3482 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
3483 (get_polymorphic_call_info): ... here.
3484 (get_polymorphic_call_info_from_invariant): New function.
3486 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3488 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
3489 lookup via vtable pointer; check for type consistency
3490 and turn inconsitent facts into UNREACHABLE.
3491 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3492 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
3493 type inconsistent querries; return UNREACHABLE instead.
3495 2014-02-03 Richard Henderson <rth@twiddle.net>
3498 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
3499 already processed this node.
3500 (normalize_one_pred_1): Pass along mark_set.
3501 (normalize_one_pred): Create and destroy a pointer_set_t.
3502 (normalize_one_pred_chain): Likewise.
3504 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
3506 PR gcov-profile/58602
3507 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
3509 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
3512 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
3513 -fno-devirtualize; try to devirtualize by the knowledge of
3514 virtual table pointer given by aggregate propagation.
3515 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
3516 (ipa_print_node_jump_functions): Dump also offset that
3517 is relevant for polymorphic calls.
3518 (determine_known_aggregate_parts): Add arg_type parameter; use it
3519 instead of determining the type from pointer type.
3520 (ipa_compute_jump_functions_for_edge): Update call of
3521 determine_known_aggregate_parts.
3522 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
3523 (gimple_get_virt_method_for_binfo): ... here; simplify using
3524 vtable_pointer_value_to_vtable.
3525 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
3526 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
3527 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
3528 (vtable_pointer_value_to_vtable): Break out from ...; handle also
3530 (vtable_pointer_value_to_binfo): ... here.
3531 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
3533 2014-02-03 Teresa Johnson <tejohnson@google.com>
3535 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
3536 redef of outer loop index variable.
3538 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
3542 * doc/extend.texi (Function Attributes): Typo.
3544 2014-02-03 Cong Hou <congh@google.com>
3546 PR tree-optimization/60000
3547 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
3548 if the vectorized statement is a store. A store statement can only
3549 appear at the end of pattern statements.
3551 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3553 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
3554 (ix86_option_override_internal): Default long double to 64-bit for
3555 32-bit Bionic and to 128-bit for 64-bit Bionic.
3557 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
3558 TARGET_LONG_DOUBLE_128 is true.
3559 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
3561 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
3562 (mlong-double-64): Negate -mlong-double-128.
3563 (mlong-double-128): New option.
3565 * config/i386/i386-c.c (ix86_target_macros): Define
3566 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
3568 * doc/invoke.texi: Document -mlong-double-128.
3570 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
3572 PR rtl-optimization/60024
3573 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
3575 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
3577 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
3579 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3581 PR rtl-optimization/57662
3582 * sel-sched.c (code_motion_path_driver): Do not mark already not
3583 existing blocks in the visiting bitmap.
3585 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
3587 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
3588 on the insn being emitted.
3590 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
3591 Will Deacon <will.deacon@arm.com>
3593 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
3595 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3597 * config/arm/arm-tables.opt: Regenerate.
3599 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3601 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
3602 for vector types other than V16QImode.
3603 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
3604 define_expand, and call altivec_expand_vec_perm_le when producing
3605 code with little endian element order.
3606 (*altivec_vperm_<mode>_internal): New insn having previous
3607 behavior of altivec_vperm_<mode>.
3608 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
3609 altivec_expand_vec_perm_le when producing code with little endian
3611 (*altivec_vperm_<mode>_uns_internal): New insn having previous
3612 behavior of altivec_vperm_<mode>_uns.
3614 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3616 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
3617 (altivec_vsumsws): Add handling for -maltivec=be with a little
3619 (altivec_vsumsws_direct): New.
3620 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
3621 gen_altivec_vsumsws.
3623 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3625 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
3626 vtable_pointer_value_to_binfo): New functions.
3627 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
3628 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
3630 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
3632 * config/nios2/nios2.md (load_got_register): Initialize GOT
3633 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
3634 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
3636 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
3638 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
3639 preserverd by passthrough, do not propagate the type.
3641 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3643 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
3644 (mips_atomic_assign_expand_fenv): New function.
3645 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
3647 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3649 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
3650 (__builtin_mips_set_fcsr): Likewise.
3651 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
3652 MIPS_USI_FTYPE_VOID.
3653 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
3654 (mips16_expand_set_fcsr): Likewise.
3655 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
3656 (mips16_set_fcsr_stub): Likewise.
3657 (mips16_get_fcsr_one_only_stub): New class.
3658 (mips16_set_fcsr_one_only_stub): Likewise.
3659 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
3660 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
3661 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
3662 (hard_float): New availability predicate.
3663 (mips_builtins): Add get_fcsr and set_fcsr.
3664 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
3665 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
3666 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
3667 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
3668 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
3671 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
3673 * config/mips/mips.c (mips_one_only_stub): New class.
3674 (mips_need_mips16_rdhwr_p): Replace with...
3675 (mips16_rdhwr_stub): ...this new variable.
3676 (mips16_stub_call_address): New function.
3677 (mips16_rdhwr_one_only_stub): New class.
3678 (mips_expand_thread_pointer): Use mips16_stub_call_address.
3679 (mips_output_mips16_rdhwr): Delete.
3680 (mips_finish_stub): New function.
3681 (mips_code_end): Use it to handle rdhwr stubs.
3683 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
3686 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
3687 when calculating size of integer atomic types.
3689 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
3691 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
3693 2014-02-01 Jakub Jelinek <jakub@redhat.com>
3695 PR tree-optimization/60003
3696 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
3697 * profile.c (branch_prob): Use gimple_call_builtin_p
3698 to check for BUILT_IN_SETJMP_RECEIVER.
3699 * tree-inline.c (copy_bb): Call notice_special_calls.
3701 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
3704 * lra-constraints.c (process_alt_operands): Update reload_sum only
3707 2014-01-31 Richard Henderson <rth@redhat.com>
3710 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
3711 until after else_eh is processed.
3713 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3715 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
3716 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
3717 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
3718 in smmintrin.h, remove them.
3719 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
3720 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
3721 * config/i386/i386.md (ROUND_SAE): Fix value.
3722 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
3723 (const48_operand): New.
3724 * config/i386/subst.md (round), (round_expand): Use
3725 const_4_or_8_to_11_operand.
3726 (round_saeonly), (round_saeonly_expand): Use const48_operand.
3728 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
3730 * config/i386/constraints.md (Yk): Swap meaning with k.
3731 * config/i386/i386.md (movhi_internal): Change Yk to k.
3732 (movqi_internal): Ditto.
3733 (*k<logic><mode>): Ditto.
3736 (kandn<mode>): Ditto.
3737 (*<code>hi_1): Ditto.
3738 (*<code>qi_1): Ditto.
3739 (kxnor<mode>): Ditto.
3740 (kortestzhi): Ditto.
3741 (kortestchi): Ditto.
3743 (*one_cmplhi2_1): Ditto.
3744 (*one_cmplqi2_1): Ditto.
3745 * config/i386/sse.md (): Change k to Yk.
3746 (avx512f_load<mode>_mask): Ditto.
3747 (avx512f_blendm<mode>): Ditto.
3748 (avx512f_store<mode>_mask): Ditto.
3749 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
3750 (avx512f_storedqu<mode>_mask): Ditto.
3751 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
3753 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
3754 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
3755 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
3756 (avx512f_maskcmp<mode>3): Ditto.
3757 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
3758 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
3759 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
3760 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
3761 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
3762 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
3763 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
3764 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
3765 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
3766 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
3767 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
3768 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
3769 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
3770 (vec_extract_lo_<mode>_maskm): Ditto.
3771 (vec_extract_hi_<mode>_maskm): Ditto.
3772 (avx512f_vternlog<mode>_mask): Ditto.
3773 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
3774 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
3775 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
3776 (avx512f_<code>v8div16qi2_mask): Ditto.
3777 (avx512f_<code>v8div16qi2_mask_store): Ditto.
3778 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
3779 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
3780 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
3781 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
3782 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
3783 (*avx512pf_gatherpf<mode>df_mask): Ditto.
3784 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
3785 (*avx512pf_scatterpf<mode>df_mask): Ditto.
3786 (avx512cd_maskb_vec_dupv8di): Ditto.
3787 (avx512cd_maskw_vec_dupv16si): Ditto.
3788 (avx512f_vpermi2var<mode>3_maskz): Ditto.
3789 (avx512f_vpermi2var<mode>3_mask): Ditto.
3790 (avx512f_vpermi2var<mode>3_mask): Ditto.
3791 (avx512f_vpermt2var<mode>3_maskz): Ditto.
3792 (*avx512f_gathersi<mode>): Ditto.
3793 (*avx512f_gathersi<mode>_2): Ditto.
3794 (*avx512f_gatherdi<mode>): Ditto.
3795 (*avx512f_gatherdi<mode>_2): Ditto.
3796 (*avx512f_scattersi<mode>): Ditto.
3797 (*avx512f_scatterdi<mode>): Ditto.
3798 (avx512f_compress<mode>_mask): Ditto.
3799 (avx512f_compressstore<mode>_mask): Ditto.
3800 (avx512f_expand<mode>_mask): Ditto.
3801 * config/i386/subst.md (mask): Change k to Yk.
3802 (mask_scalar_merge): Ditto.
3805 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
3807 * doc/extend.texi (Vector Extensions): Document ?: in C++.
3809 2014-01-31 Richard Biener <rguenther@suse.de>
3812 * builtins.c (fold_builtin_memory_op): Make sure to not
3813 use a floating-point mode or a boolean or enumeral type for
3816 2014-01-30 DJ Delorie <dj@redhat.com>
3818 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
3819 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
3820 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
3821 whenever main() has an epilogue.
3823 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3825 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
3826 unused variable "field".
3827 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
3828 (vsx_mergeh_<mode>): Likewise.
3829 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
3830 (altivec_vmrghh): Likewise.
3831 (altivec_vmrghw): Likewise.
3832 (altivec_vmrglb): Likewise.
3833 (altivec_vmrglh): Likewise.
3834 (altivec_vmrglw): Likewise.
3835 (altivec_vspltb): Add missing uses.
3836 (altivec_vsplth): Likewise.
3837 (altivec_vspltw): Likewise.
3838 (altivec_vspltsf): Likewise.
3840 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3843 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
3844 frame related instructions.
3846 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
3848 PR rtl-optimization/59959
3849 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
3850 any reload of register whose subreg is invalid.
3852 2014-01-30 Jakub Jelinek <jakub@redhat.com>
3854 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
3855 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
3856 Add missing return type - void.
3858 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3860 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
3861 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
3862 remove element index adjustment for endian (now handled in vsx.md
3864 (altivec_expand_vec_perm_const): Use
3865 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
3866 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
3867 (vsx_xxspltw_<mode>): Adjust element index for little endian.
3868 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
3869 define_expand and a new define_insn *altivec_vspltb_internal;
3870 adjust for -maltivec=be on a little endian target.
3871 (altivec_vspltb_direct): New.
3872 (altivec_vsplth): Divide into a define_expand and a new
3873 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
3874 little endian target.
3875 (altivec_vsplth_direct): New.
3876 (altivec_vspltw): Divide into a define_expand and a new
3877 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
3878 little endian target.
3879 (altivec_vspltw_direct): New.
3880 (altivec_vspltsf): Divide into a define_expand and a new
3881 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
3882 a little endian target.
3884 2014-01-30 Richard Biener <rguenther@suse.de>
3886 PR tree-optimization/59993
3887 * tree-ssa-forwprop.c (associate_pointerplus): Check we
3888 can propagate form the earlier stmt and avoid the transform
3889 when the intermediate result is needed.
3891 2014-01-30 Alangi Derick <alangiderick@gmail.com>
3893 * README.Portability: Fix typo.
3895 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
3897 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
3898 comparison_operator with ordered_comparison_operator.
3900 2014-01-30 Nick Clifton <nickc@redhat.com>
3902 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
3903 Rename to mn10300_store_multiple_regs.
3904 * config/mn10300/mn10300.c: Likewise.
3905 * config/mn10300/mn10300.md (store_movm): Fix typo: call
3906 store_multiple_regs.
3907 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
3908 Call mn10300_store_multiple_regs.
3910 2014-01-30 Nick Clifton <nickc@redhat.com>
3911 DJ Delorie <dj@redhat.com>
3913 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
3914 %fp 2 to keep registers after it properly word-aligned.
3915 (rl78_alloc_physical_registers_umul): Handle the case where both
3916 input operands are the same.
3918 2014-01-30 Richard Biener <rguenther@suse.de>
3920 PR tree-optimization/59903
3921 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
3924 2014-01-30 Jason Merrill <jason@redhat.com>
3927 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
3930 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
3932 2014-01-30 Richard Biener <rguenther@suse.de>
3934 PR tree-optimization/59951
3935 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
3937 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
3940 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
3941 SFmode to DFmode case.
3943 2014-01-29 DJ Delorie <dj@redhat.com>
3945 * config/msp430/msp430.opt (-minrt): New.
3946 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
3948 (ENDFILE_SPEC): Likewise.
3950 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
3952 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
3953 (estimate_function_body_sizes): Use it.
3955 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
3958 * dwarf2out.c (is_cxx_auto): New.
3959 (is_base_type): Use it.
3960 (gen_type_die_with_usage): Likewise.
3962 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3964 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
3965 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
3966 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
3967 -maltivec=be with LE targets.
3968 (vsx_mergeh_<mode>): Likewise.
3969 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
3970 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
3971 (altivec_vmrghb): Replace with define_expand and new
3972 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
3973 (altivec_vmrghb_direct): New define_insn.
3974 (altivec_vmrghh): Replace with define_expand and new
3975 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
3976 (altivec_vmrghh_direct): New define_insn.
3977 (altivec_vmrghw): Replace with define_expand and new
3978 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
3979 (altivec_vmrghw_direct): New define_insn.
3980 (*altivec_vmrghsf): Adjust for endianness.
3981 (altivec_vmrglb): Replace with define_expand and new
3982 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
3983 (altivec_vmrglb_direct): New define_insn.
3984 (altivec_vmrglh): Replace with define_expand and new
3985 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
3986 (altivec_vmrglh_direct): New define_insn.
3987 (altivec_vmrglw): Replace with define_expand and new
3988 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
3989 (altivec_vmrglw_direct): New define_insn.
3990 (*altivec_vmrglsf): Adjust for endianness.
3991 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3992 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3993 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
3994 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
3995 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3996 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
3997 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
3998 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
4000 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
4002 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
4003 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
4006 2014-01-29 Richard Biener <rguenther@suse.de>
4008 PR tree-optimization/58742
4009 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
4010 associate_pointerplus_align.
4011 (associate_pointerplus_diff): New function.
4012 (associate_pointerplus): Likewise. Call associate_pointerplus_align
4013 and associate_pointerplus_diff.
4015 2014-01-29 Richard Biener <rguenther@suse.de>
4017 * lto-streamer.h (LTO_major_version): Bump to 3.
4018 (LTO_minor_version): Reset to 0.
4020 2014-01-29 Renlin Li <Renlin.Li@arm.com>
4022 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
4023 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
4024 (arm_file_start): Generate correct asm header for armv7ve.
4025 * config/arm/bpabi.h: Add multilib support for armv7ve.
4026 * config/arm/driver-arm.c: Change the architectures of cortex-a7
4027 and cortex-a15 to armv7ve.
4028 * config/arm/t-aprofile: Add multilib support for armv7ve.
4029 * doc/invoke.texi: Document -march=armv7ve.
4031 2014-01-29 Richard Biener <rguenther@suse.de>
4033 PR tree-optimization/58742
4034 * tree-ssa-forwprop.c (associate_plusminus): Return true
4035 if we changed sth, defer EH cleanup to ...
4036 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
4037 (simplify_mult): New function.
4039 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4042 PR tree-optimization/59920
4043 * tree.c (build_common_builtin_nodes): Remove
4044 __builtin_setjmp_dispatcher initialization.
4045 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
4046 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
4047 instead of gsi_after_labels + manually skipping debug stmts.
4048 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
4049 ignore bbs with IFN_ABNORMAL_DISPATCHER.
4050 * tree-inline.c (copy_edges_for_bb): Remove
4051 can_make_abnormal_goto argument, instead add abnormal_goto_dest
4052 argument. Ignore computed_goto_p stmts. Don't call
4053 make_abnormal_goto_edges. If a call might need abnormal edges
4054 for non-local gotos, see if it already has an edge to
4055 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
4056 with true argument, don't do anything then, otherwise add
4057 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
4058 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
4060 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
4061 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
4062 (lower_stmt): Don't set data->calls_builtin_setjmp.
4063 (lower_builtin_setjmp): Adjust comment.
4064 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
4065 * tree-cfg.c (found_computed_goto): Remove.
4066 (factor_computed_gotos): Remove.
4067 (make_goto_expr_edges): Return bool, true for computed gotos.
4068 Don't call make_abnormal_goto_edges.
4069 (build_gimple_cfg): Don't set found_computed_goto, don't call
4070 factor_computed_gotos.
4071 (computed_goto_p): No longer static.
4072 (make_blocks): Don't set found_computed_goto.
4073 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
4074 (make_edges): If make_goto_expr_edges returns true, push bb
4075 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
4076 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
4077 vector. Record mapping between bbs and OpenMP regions if there
4078 are any, adjust make_gimple_omp_edges caller. Call
4079 handle_abnormal_edges.
4080 (make_abnormal_goto_edges): Remove.
4081 * tree-cfg.h (make_abnormal_goto_edges): Remove.
4082 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
4083 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
4084 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
4085 * internal-fn.def (ABNORMAL_DISPATCHER): New.
4086 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
4087 filling *region also set *region_idx to (*region)->entry->index.
4090 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
4091 For REGs set ORIGINAL_REGNO.
4093 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
4095 * doc/md.texi: Mention that a target shouldn't implement
4096 vec_widen_(s|u)mul_even/odd pair if it is less efficient
4099 2014-01-29 Jakub Jelinek <jakub@redhat.com>
4101 PR tree-optimization/59594
4102 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
4103 a copy of the datarefs vector rather than the vector itself.
4105 2014-01-28 Jason Merrill <jason@redhat.com>
4108 * dwarf2out.c (auto_die): New static.
4109 (gen_type_die_with_usage): Handle C++1y 'auto'.
4110 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
4113 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
4116 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
4117 (SPEC_X32): Likewise.
4118 (SPEC_64): Likewise.
4119 * config/i386/i386.c (ix86_option_override_internal): Turn off
4120 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
4122 (x86_file_start): Output .code16gcc for TARGET_16BIT.
4123 * config/i386/i386.h (TARGET_16BIT): New macro.
4124 (TARGET_16BIT_P): Likewise.
4125 * config/i386/i386.opt: Add m16.
4126 * doc/invoke.texi: Document -m16.
4128 2014-01-28 Jakub Jelinek <jakub@redhat.com>
4130 PR preprocessor/59935
4131 * input.c (location_get_source_line): Bail out on when line number
4132 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
4134 2014-01-28 Richard Biener <rguenther@suse.de>
4136 PR tree-optimization/58742
4137 * tree-ssa-forwprop.c (associate_plusminus): Handle
4138 pointer subtraction of the form (T)(P + A) - (T)P.
4140 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4142 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
4145 2014-01-28 Richard Biener <rguenther@suse.de>
4148 2014-01-28 Richard Biener <rguenther@suse.de>
4150 PR rtl-optimization/45364
4151 PR rtl-optimization/59890
4152 * var-tracking.c (local_get_addr_clear_given_value): Handle
4153 already cleared slot.
4154 (val_reset): Handle not allocated local_get_addr_cache.
4155 (vt_find_locations): Use post-order on the inverted CFG.
4157 2014-01-28 Richard Biener <rguenther@suse.de>
4159 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
4161 2014-01-28 Richard Biener <rguenther@suse.de>
4163 PR rtl-optimization/45364
4164 PR rtl-optimization/59890
4165 * var-tracking.c (local_get_addr_clear_given_value): Handle
4166 already cleared slot.
4167 (val_reset): Handle not allocated local_get_addr_cache.
4168 (vt_find_locations): Use post-order on the inverted CFG.
4170 2014-01-28 Alan Modra <amodra@gmail.com>
4172 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
4173 * configure.ac <recursive call for build != host>: Define
4174 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
4175 and LD_FOR_BUILD too.
4176 * configure: Regenerate.
4178 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
4180 * config/i386/i386.c (get_builtin_code_for_version): Separate
4181 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
4182 Broadwell from Haswell.
4184 2014-01-27 Steve Ellcey <sellcey@mips.com>
4186 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
4187 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
4188 * config/mips/mips.c (mips_option_override): Change setting
4190 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
4191 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
4192 Change from Mask to Var.
4194 2014-01-27 Jeff Law <law@redhat.com>
4196 * ipa-inline.c (inline_small_functions): Fix typo.
4198 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
4200 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
4201 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
4202 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
4203 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
4204 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
4205 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
4206 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
4207 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
4208 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
4209 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
4210 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
4211 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
4212 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
4213 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
4214 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
4215 (_mm512_storeu_epi64): Ditto.
4216 (_mm512_cmpge_epi32_mask): Ditto.
4217 (_mm512_cmpge_epu32_mask): Ditto.
4218 (_mm512_cmpge_epi64_mask): Ditto.
4219 (_mm512_cmpge_epu64_mask): Ditto.
4220 (_mm512_cmple_epi32_mask): Ditto.
4221 (_mm512_cmple_epu32_mask): Ditto.
4222 (_mm512_cmple_epi64_mask): Ditto.
4223 (_mm512_cmple_epu64_mask): Ditto.
4224 (_mm512_cmplt_epi32_mask): Ditto.
4225 (_mm512_cmplt_epu32_mask): Ditto.
4226 (_mm512_cmplt_epi64_mask): Ditto.
4227 (_mm512_cmplt_epu64_mask): Ditto.
4228 (_mm512_cmpneq_epi32_mask): Ditto.
4229 (_mm512_cmpneq_epu32_mask): Ditto.
4230 (_mm512_cmpneq_epi64_mask): Ditto.
4231 (_mm512_cmpneq_epu64_mask): Ditto.
4232 (_mm512_expand_pd): Ditto.
4233 (_mm512_expand_ps): Ditto.
4234 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
4235 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
4236 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
4237 * config/i386/i386.c (ix86_builtins): Add
4238 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
4239 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
4240 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
4241 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
4242 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
4243 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
4244 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
4245 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
4246 IX86_BUILTIN_PMOVUSQW512_MEM.
4247 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
4248 __builtin_ia32_pmovsqd512mem_mask,
4249 __builtin_ia32_pmovqd512mem_mask,
4250 __builtin_ia32_pmovusqw512mem_mask,
4251 __builtin_ia32_pmovsqw512mem_mask,
4252 __builtin_ia32_pmovqw512mem_mask,
4253 __builtin_ia32_pmovusdw512mem_mask,
4254 __builtin_ia32_pmovsdw512mem_mask,
4255 __builtin_ia32_pmovdw512mem_mask,
4256 __builtin_ia32_pmovqb512mem_mask,
4257 __builtin_ia32_pmovusqb512mem_mask,
4258 __builtin_ia32_pmovsqb512mem_mask,
4259 __builtin_ia32_pmovusdb512mem_mask,
4260 __builtin_ia32_pmovsdb512mem_mask,
4261 __builtin_ia32_pmovdb512mem_mask.
4262 (bdesc_args): Add __builtin_ia32_expanddf512,
4263 __builtin_ia32_expandsf512.
4264 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
4265 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
4266 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
4267 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
4268 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
4269 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
4270 (avx512f_<code>v8div16qi2_mask_store): This.
4271 (avx512f_expand<mode>): New.
4273 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
4275 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
4277 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4278 (_mm512_prefetch_i32scatter_pd): Ditto.
4279 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4280 (_mm512_prefetch_i64scatter_pd): Ditto.
4281 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4282 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
4283 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4284 (_mm512_prefetch_i32scatter_ps): Ditto.
4285 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4286 (_mm512_prefetch_i64scatter_ps): Ditto.
4287 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4288 * config/i386/i386-builtin-types.def: Define
4289 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
4290 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
4291 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
4292 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
4293 IX86_BUILTIN_SCATTERPFQPD.
4294 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
4295 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
4296 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
4297 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
4298 __builtin_ia32_scatterpfqps.
4299 (ix86_expand_builtin): Expand new built-ins.
4300 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
4301 fix memory access data type.
4302 (*avx512pf_gatherpf<mode>_mask): Ditto.
4303 (*avx512pf_gatherpf<mode>): Ditto.
4304 (avx512pf_scatterpf<mode>): Ditto.
4305 (*avx512pf_scatterpf<mode>_mask): Ditto.
4306 (*avx512pf_scatterpf<mode>): Ditto.
4307 (GATHER_SCATTER_SF_MEM_MODE): New.
4308 (avx512pf_gatherpf<mode>df): Ditto.
4309 (*avx512pf_gatherpf<mode>df_mask): Ditto.
4310 (*avx512pf_scatterpf<mode>df): Ditto.
4312 2014-01-27 Jakub Jelinek <jakub@redhat.com>
4315 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
4316 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
4319 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4321 * common/config/arm/arm-common.c
4322 (arm_rewrite_mcpu): Handle multiple names.
4324 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4326 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
4328 * gimple-builder.h (create_gimple_tmp): Delete.
4330 2014-01-27 Christian Bruel <christian.bruel@st.com>
4332 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
4335 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
4337 * config/pa/pa.md (call): Generate indirect long calls to non-local
4338 functions when outputing 32-bit code.
4339 (call_value): Likewise except for special call to buggy powf function.
4341 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
4342 portable runtime and PIC indirect calls.
4343 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
4344 and PIC call sequences. Use ldo instead of blr to set return register
4345 in PIC call sequence.
4347 2014-01-25 Walter Lee <walt@tilera.com>
4349 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
4350 avoid clobbering a live register.
4352 2014-01-25 Walter Lee <walt@tilera.com>
4354 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
4355 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
4356 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
4357 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
4359 2014-01-25 Walter Lee <walt@tilera.com>
4361 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
4362 arguments on even registers.
4363 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
4365 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
4366 (BIGGEST_ALIGNMENT): Ditto.
4367 (BIGGEST_FIELD_ALIGNMENT): Ditto.
4369 2014-01-25 Walter Lee <walt@tilera.com>
4371 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
4372 insns before bundling.
4373 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
4375 2014-01-25 Walter Lee <walt@tilera.com>
4377 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
4378 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
4379 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
4381 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4383 * config/mips/constraints.md (kl): Delete.
4384 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
4385 define expands, using...
4386 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
4387 instructions for MIPS16.
4388 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
4389 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
4391 2014-01-25 Walter Lee <walt@tilera.com>
4393 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
4397 2014-01-25 Walter Lee <walt@tilera.com>
4399 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
4400 (TARGET_EXPAND_TO_RTL_HOOK): Define.
4402 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
4404 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
4407 2014-01-25 Jakub Jelinek <jakub@redhat.com>
4409 * print-rtl.c (in_call_function_usage): New var.
4410 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
4411 EXPR_LIST mode as mode and not as reg note name.
4414 * cfgloopmanip.c (copy_loop_info): If
4415 loop->warned_aggressive_loop_optimizations, make sure
4416 the flag is set in target loop too.
4418 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
4420 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
4422 * builtins.def: Likewise.
4423 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
4424 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
4425 * ira.c (ira_setup_eliminable_regset): Likewise.
4426 * omp-low.c (gate_expand_omp): Likewise.
4427 (execute_lower_omp): Likewise.
4428 (diagnose_sb_0): Likewise.
4429 (gate_diagnose_omp_blocks): Likewise.
4430 (simd_clone_clauses_extract): Likewise.
4433 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4435 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
4436 correction for little endian...
4437 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
4440 2014-01-24 Jeff Law <law@redhat.com>
4442 PR tree-optimization/59919
4443 * tree-vrp.c (find_assert_locations_1): Do not register asserts
4444 for non-returning calls.
4446 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
4448 * common/config/aarch64/aarch64-common.c
4449 (aarch64_rewrite_mcpu): Handle multiple names.
4450 * config/aarch64/aarch64.h
4451 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
4453 2014-01-24 Dodji Seketeli <dodji@redhat.com>
4455 * input.c (add_file_to_cache_tab): Handle the case where fopen
4458 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
4461 * config/i386/i386.md (pushsf splitter): Get stack adjustment
4462 from push operand if code of push isn't PRE_DEC.
4464 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
4467 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
4468 -mquad-memory-atomic. Update -mquad-memory documentation to say
4469 it is only used for non-atomic loads/stores.
4471 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
4472 -mquad-memory or -mquad-memory-atomic switches.
4474 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
4475 -mquad-memory-atomic to ISA 2.07 support.
4477 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
4478 to separate support of normal quad word memory operations (ldq, stq)
4479 from the atomic quad word memory operations.
4481 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
4482 support to separate non-atomic quad word operations from atomic
4483 quad word operations. Disable non-atomic quad word operations in
4484 little endian mode so that we don't have to swap words after the
4485 load and before the store.
4486 (quad_load_store_p): Add comment about atomic quad word support.
4487 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
4488 options printed with -mdebug=reg.
4490 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
4491 -mquad-memory-atomic as the test for whether we have quad word
4492 atomic instructions.
4493 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
4494 or -mp8-vector are used, allow byte/half-word atomic operations.
4496 * config/rs6000/sync.md (load_lockedti): Insure that the address
4497 is a proper indexed or indirect address for the lqarx instruction.
4498 On little endian systems, swap the hi/lo registers after the lqarx
4500 (load_lockedpti): Use indexed_or_indirect_operand predicate to
4501 insure the address is valid for the lqarx instruction.
4502 (store_conditionalti): Insure that the address is a proper indexed
4503 or indirect address for the stqcrx. instruction. On little endian
4504 systems, swap the hi/lo registers before doing the stqcrx.
4506 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
4507 insure the address is valid for the stqcrx. instruction.
4509 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
4510 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
4511 type of quad memory support is available.
4513 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
4516 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
4517 there is a danger of looping.
4519 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
4521 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
4522 force flag_ira_loop_pressure if set via command line.
4524 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4526 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
4527 (ashr_simd): New builtin handling DI mode.
4528 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
4529 (aarch64_sshr_simddi): New match pattern.
4530 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
4531 (vshrd_n_s64): Likewise.
4532 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
4534 2014-01-23 Nick Clifton <nickc@redhat.com>
4536 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
4537 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
4538 favour of mcu specific scripts.
4539 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
4542 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4543 Alex Velenko <Alex.Velenko@arm.com>
4545 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
4546 (vaddv_s16): Likewise.
4547 (vaddv_s32): Likewise.
4548 (vaddv_u8): Likewise.
4549 (vaddv_u16): Likewise.
4550 (vaddv_u32): Likewise.
4551 (vaddvq_s8): Likewise.
4552 (vaddvq_s16): Likewise.
4553 (vaddvq_s32): Likewise.
4554 (vaddvq_s64): Likewise.
4555 (vaddvq_u8): Likewise.
4556 (vaddvq_u16): Likewise.
4557 (vaddvq_u32): Likewise.
4558 (vaddvq_u64): Likewise.
4559 (vaddv_f32): Likewise.
4560 (vaddvq_f32): Likewise.
4561 (vaddvq_f64): Likewise.
4562 (vmaxv_f32): Likewise.
4563 (vmaxv_s8): Likewise.
4564 (vmaxv_s16): Likewise.
4565 (vmaxv_s32): Likewise.
4566 (vmaxv_u8): Likewise.
4567 (vmaxv_u16): Likewise.
4568 (vmaxv_u32): Likewise.
4569 (vmaxvq_f32): Likewise.
4570 (vmaxvq_f64): Likewise.
4571 (vmaxvq_s8): Likewise.
4572 (vmaxvq_s16): Likewise.
4573 (vmaxvq_s32): Likewise.
4574 (vmaxvq_u8): Likewise.
4575 (vmaxvq_u16): Likewise.
4576 (vmaxvq_u32): Likewise.
4577 (vmaxnmv_f32): Likewise.
4578 (vmaxnmvq_f32): Likewise.
4579 (vmaxnmvq_f64): Likewise.
4580 (vminv_f32): Likewise.
4581 (vminv_s8): Likewise.
4582 (vminv_s16): Likewise.
4583 (vminv_s32): Likewise.
4584 (vminv_u8): Likewise.
4585 (vminv_u16): Likewise.
4586 (vminv_u32): Likewise.
4587 (vminvq_f32): Likewise.
4588 (vminvq_f64): Likewise.
4589 (vminvq_s8): Likewise.
4590 (vminvq_s16): Likewise.
4591 (vminvq_s32): Likewise.
4592 (vminvq_u8): Likewise.
4593 (vminvq_u16): Likewise.
4594 (vminvq_u32): Likewise.
4595 (vminnmv_f32): Likewise.
4596 (vminnmvq_f32): Likewise.
4597 (vminnmvq_f64): Likewise.
4599 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
4601 * config/aarch64/aarch64-simd.md
4602 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
4603 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
4604 (*aarch64_mul3_elt<mode>): Likewise.
4605 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
4606 (*aarch64_mul3_elt_to_64v2df): Likewise.
4607 (*aarch64_mla_elt<mode>): Likewise.
4608 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
4609 (*aarch64_mls_elt<mode>): Likewise.
4610 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
4611 (*aarch64_fma4_elt<mode>): Likewise.
4612 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
4613 (*aarch64_fma4_elt_to_64v2df): Likewise.
4614 (*aarch64_fnma4_elt<mode>): Likewise.
4615 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
4616 (*aarch64_fnma4_elt_to_64v2df): Likewise.
4617 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
4618 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
4619 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4620 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
4621 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
4622 (aarch64_sqdmull_lane<mode>_internal): Likewise.
4623 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
4625 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
4627 * config/aarch64/aarch64-simd.md
4628 (aarch64_be_checked_get_lane<mode>): New define_expand.
4629 * config/aarch64/aarch64-simd-builtins.def
4630 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
4631 New builtin definition.
4632 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
4633 Use new safe be builtin.
4635 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
4637 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
4639 (aarch64_be_st1<mode>): Likewise.
4640 (aarch_ld1<VALL:mode>): Define_expand modified.
4641 (aarch_st1<VALL:mode>): Likewise.
4642 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
4643 (UNSPEC_ST1): Likewise.
4645 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
4647 * config/microblaze/microblaze.md: Add trap insn and attribute
4649 2014-01-23 Dodji Seketeli <dodji@redhat.com>
4651 PR preprocessor/58580
4652 * input.h (location_get_source_line): Take an additional line_size
4654 (void diagnostics_file_cache_fini): Declare new function.
4655 * input.c (struct fcache): New type.
4656 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
4657 New static constants.
4658 (diagnostic_file_cache_init, total_lines_num)
4659 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
4660 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
4661 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
4662 (get_next_line, read_next_line, goto_next_line, read_line_num):
4663 New static function definitions.
4664 (diagnostic_file_cache_fini): New function.
4665 (location_get_source_line): Take an additional output line_len
4666 parameter. Re-write using lookup_or_add_file_to_cache_tab and
4668 * diagnostic.c (diagnostic_finish): Call
4669 diagnostic_file_cache_fini.
4670 (adjust_line): Take an additional input parameter for the length
4671 of the line, rather than calculating it with strlen.
4672 (diagnostic_show_locus): Adjust the use of
4673 location_get_source_line and adjust_line with respect to their new
4674 signature. While displaying a line now, do not stop at the first
4675 null byte. Rather, display the zero byte as a space and keep
4676 going until we reach the size of the line.
4677 * Makefile.in: Add vec.o to OBJS-libcommon
4679 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4680 Ilya Tocar <ilya.tocar@intel.com>
4682 * config/i386/avx512fintrin.h (_mm512_kmov): New.
4683 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
4684 (__builtin_ia32_kmov16): Ditto.
4685 * config/i386/i386.md (UNSPEC_KMOV): New.
4688 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
4690 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
4691 (_mm512_storeu_si512): Ditto.
4693 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
4696 * rtl.h (get_referenced_operands): Declare.
4697 * recog.c (get_referenced_operands): New function.
4698 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
4699 operands have been referenced when recording LO_SUM references.
4701 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
4703 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
4705 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4707 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
4708 Enable for generic and recent AMD targets.
4710 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
4712 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
4713 ARG_SIZE note when adjustment was eliminated.
4715 2014-01-22 Jeff Law <law@redhat.com>
4717 PR tree-optimization/59597
4718 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
4719 in file. Accept new argument REGISTERING and use it to modify
4720 dump output appropriately.
4721 (register_jump_thread): Corresponding changes.
4722 (mark_threaded_blocks): Reinstate code to cancel unprofitable
4723 thread paths involving joiner blocks. Add code to dump cancelled
4724 jump threading paths.
4726 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
4728 PR rtl-optimization/59477
4729 * lra-constraints.c (inherit_in_ebb): Process call for living hard
4730 regs. Update reloads_num and potential_reload_hard_regs for all insns.
4732 2014-01-22 Tom Tromey <tromey@redhat.com>
4734 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
4736 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
4738 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4740 PR rtl-optimization/59896
4741 * lra-constraints.c (process_alt_operands): Check unused note for
4742 matched operands of insn with no output reloads.
4744 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
4746 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
4747 (mips_move_from_gpr_cost): Likewise.
4749 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
4751 PR rtl-optimization/59858
4752 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
4753 ira_class_hard_regs_num.
4754 (process_alt_operands): Increase reject for dying matched operand.
4756 2014-01-21 Jakub Jelinek <jakub@redhat.com>
4759 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
4760 smaller than size, perform several stores or loads and stores
4761 at dst + count - size to store or copy all of size bytes, rather
4762 than just last modesize bytes.
4764 2014-01-20 DJ Delorie <dj@redhat.com>
4766 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
4767 that CLOBBERs are REGs before propogating their values.
4769 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
4772 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
4773 (cgraph_inline_failed_type): New function.
4774 * cgraph.h (DEFCIFCODE): Add type.
4775 (cgraph_inline_failed_type_t): New enum.
4776 (cgraph_inline_failed_type): New prototype.
4777 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
4778 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
4779 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
4780 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
4781 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
4782 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
4783 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
4784 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
4785 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
4786 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
4787 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
4788 OPTIMIZATION_MISMATCH.
4789 * tree-inline.c (expand_call_inline): Emit errors during
4790 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
4792 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4795 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
4796 mode attribute in insn output.
4798 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
4800 * output.h (output_constant): Delete.
4801 * varasm.c (output_constant): Make private.
4803 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
4805 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
4807 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4810 * tree.h (fold_builtin_strcat): New prototype.
4811 * builtins.c (fold_builtin_strcat): No longer static. Add len
4812 argument, if non-NULL, don't call c_strlen. Optimize
4813 directly into __builtin_memcpy instead of __builtin_strcpy.
4814 (fold_builtin_2): Adjust fold_builtin_strcat caller.
4815 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
4817 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
4819 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4820 for SImode_address_operand operands, having only a REG argument.
4822 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
4824 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
4825 loader name using mbig-endian.
4826 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
4828 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4830 * doc/invoke.texi (-march): Clarify documentation for AArch64.
4834 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
4836 * config/aarch64/aarch64-protos.h
4837 (aarch64_cannot_change_mode_class_ptr): Declare.
4838 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
4839 aarch64_cannot_change_mode_class_ptr): New.
4840 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
4841 backend hook aarch64_cannot_change_mode_class.
4843 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
4845 * common/config/aarch64/aarch64-common.c
4846 (aarch64_handle_option): Don't handle any option order logic here.
4847 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
4848 selected_cpu, warn on architecture version mismatch.
4849 (aarch64_override_options): Fix parsing order for option strings.
4851 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4852 Iain Sandoe <iain@codesourcery.com>
4855 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
4856 warning. Amend comment to reflect current functionality.
4858 2014-01-20 Richard Biener <rguenther@suse.de>
4861 * builtins.c (fold_builtin_strcat): Remove case better handled
4862 by tree-ssa-strlen.c.
4864 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
4866 * config/aarch64/aarch64.opt
4867 (mcpu, march, mtune): Make case-insensitive.
4869 2014-01-20 Jakub Jelinek <jakub@redhat.com>
4872 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
4873 if operands[1] is a REG or ZERO_EXTEND of a REG.
4875 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
4877 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
4879 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
4881 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
4882 long non-pic millicode calls.
4884 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
4886 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
4888 2014-01-19 Kito Cheng <kito@0xlab.org>
4890 * builtins.c (expand_movstr): Check movstr expand done or fail.
4892 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4893 H.J. Lu <hongjiu.lu@intel.com>
4896 * config/i386/i386.md (*lea<mode>): Zero-extend return register
4897 to DImode for zero-extended addresses.
4899 2014-01-19 Jakub Jelinek <jakub@redhat.com>
4901 PR rtl-optimization/57763
4902 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
4903 on the new indirect jump_insn and increment LABEL_NUSES (label).
4905 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
4909 * config.gcc (x86_archs): New variable.
4910 (x86_64_archs): Likewise.
4911 (x86_cpus): Likewise.
4912 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
4913 --with-arch/--with-cpu= options.
4914 Support --with-arch=/--with-cpu={nehalem,westmere,
4915 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
4917 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4919 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
4920 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
4922 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
4924 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
4926 2014-01-18 Jakub Jelinek <jakub@redhat.com>
4929 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
4930 clear cpp_get_options (parse_in)->warn_unused_macros for
4931 ix86_target_macros_internal with cpp_define.
4933 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
4935 * jump.c (delete_related_insns): Keep (use (insn))s.
4936 * reorg.c (redundant_insn): Check for barriers too.
4938 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4940 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
4942 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
4944 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
4945 call to $$dyncall when TARGET_LONG_CALLS is true.
4947 2014-01-17 Jeff Law <law@redhat.com>
4949 * ree.c (combine_set_extension): Temporarily disable test for
4950 changing number of hard registers.
4952 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
4955 * ipa-inline-analysis.c (inline_free_summary):
4956 Do not free summary of aliases.
4958 2014-01-17 Jakub Jelinek <jakub@redhat.com>
4961 * gimplify.c (gimplify_expr): Use create_tmp_var
4962 instead of create_tmp_var_raw. If cond doesn't have
4963 integral type, don't add the IFN_ANNOTATE builtin at all.
4965 2014-01-17 Martin Jambor <mjambor@suse.cz>
4968 * ipa-cp.c (prev_edge_clone): New variable.
4969 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
4970 Also resize prev_edge_clone vector.
4971 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
4972 (ipcp_edge_removal_hook): New function.
4973 (ipcp_driver): Register ipcp_edge_removal_hook.
4975 2014-01-17 Andrew Pinski <apinski@cavium.com>
4976 Steve Ellcey <sellcey@mips.com>
4979 * config/mips/mips.c (mips_print_operand): Check operand mode instead
4982 2014-01-17 Jeff Law <law@redhat.com>
4985 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
4986 so that pass_ccp runs first.
4988 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4990 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
4991 (ix86_adjust_cost): Use !TARGET_XXX.
4992 (do_reorder_for_imul): Likewise.
4993 (swap_top_of_ready_list): Likewise.
4994 (ix86_sched_reorder): Likewise.
4996 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
4998 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4999 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
5000 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
5001 (intel_memset): New. Duplicate slm_memset.
5002 (intel_cost): New. Duplicate slm_cost.
5003 (m_INTEL): New macro.
5004 (processor_target_table): Add "intel".
5005 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
5006 with PROCESSOR_INTEL for "intel".
5007 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
5008 PROCESSOR_SILVERMONT.
5009 (ix86_issue_rate): Likewise.
5010 (ix86_adjust_cost): Likewise.
5011 (ia32_multipass_dfa_lookahead): Likewise.
5012 (swap_top_of_ready_list): Likewise.
5013 (ix86_sched_reorder): Likewise.
5014 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
5015 instead of TARGET_OPT_AGU.
5016 * config/i386/i386.h (TARGET_INTEL): New.
5017 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
5018 (processor_type): Add PROCESSOR_INTEL.
5019 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
5020 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
5022 2014-01-17 Marek Polacek <polacek@redhat.com>
5025 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
5028 2014-01-17 Richard Biener <rguenther@suse.de>
5030 PR tree-optimization/46590
5031 * opts.c (default_options_table): Add entries for
5032 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
5033 all enabled at -O1 but not for -Og.
5034 * common.opt (fbranch-count-reg): Remove Init(1).
5035 (fmove-loop-invariants): Likewise.
5036 (ftree-pta): Likewise.
5038 2014-01-17 Jakub Jelinek <jakub@redhat.com>
5040 * config/i386/i386.c (ix86_data_alignment): For compatibility with
5041 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
5042 decls to at least the GCC 4.8 used alignments.
5045 * tree-nested.c (convert_nonlocal_reference_stmt,
5046 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
5047 of GIMPLE_BIND stmts, adjust associated decls.
5049 2014-01-17 Richard Biener <rguenther@suse.de>
5051 PR tree-optimization/46590
5052 * vec.h (vec<>::bseach): New member function implementing
5053 binary search according to C89 bsearch.
5054 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
5055 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
5056 bitmap pointer again. Make accesses_in_loop a flat array.
5057 (mem_ref_obstack): New global.
5058 (outermost_indep_loop): Adjust for mem_ref->stored changes.
5059 (mark_ref_stored): Likewise.
5060 (ref_indep_loop_p_2): Likewise.
5061 (set_ref_stored_in_loop): New helper function.
5062 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
5063 (memref_free): Adjust.
5064 (record_mem_ref_loc): Simplify.
5065 (gather_mem_refs_stmt): Adjust.
5066 (sort_locs_in_loop_postorder_cmp): New function.
5067 (analyze_memory_references): Sort accesses_in_loop after
5068 loop postorder number.
5069 (find_ref_loc_in_loop_cmp): New function.
5070 (for_all_locs_in_loop): Find relevant cluster of locs in
5071 accesses_in_loop and iterate without recursion.
5072 (execute_sm): Avoid uninit warning.
5073 (struct ref_always_accessed): Simplify.
5074 (ref_always_accessed::operator ()): Likewise.
5075 (ref_always_accessed_p): Likewise.
5076 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
5077 loop postorder numbers here.
5078 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
5081 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
5084 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
5085 on decls for which assemble_alias has been called.
5087 2014-01-17 Nick Clifton <nickc@redhat.com>
5089 * config/msp430/msp430.opt: (mcpu): New option.
5090 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
5091 (msp430_option_override): Parse target_cpu. If the MCU name
5092 matches a generic string, clear target_mcu.
5093 (msp430_attr): Allow numeric interrupt values up to 63.
5094 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
5095 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
5097 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
5099 * config/msp430/msp430.md (popm): Use %J rather than %I.
5100 (addsi3): Use msp430_nonimmediate_operand for operand 2.
5101 (addhi_cy_i): Use immediate_operand for operand 2.
5102 * doc/invoke.texi: Document -mcpu option.
5104 2014-01-17 Richard Biener <rguenther@suse.de>
5106 PR rtl-optimization/38518
5107 * df.h (df_analyze_loop): Declare.
5108 * df-core.c: Include cfgloop.h.
5109 (df_analyze_1): Split out main part of df_analyze.
5110 (df_analyze): Adjust.
5111 (loop_inverted_post_order_compute): New function.
5112 (loop_post_order_compute): Likewise.
5113 (df_analyze_loop): New function avoiding whole-function
5115 * loop-invariant.c (find_defs): Use df_analyze_loop.
5116 (find_invariants): Adjust.
5117 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
5119 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
5121 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
5122 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
5124 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
5126 * ipa-ref.c (ipa_remove_stmt_references): Fix references
5127 traversal when removing references.
5129 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
5132 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
5134 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
5137 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
5138 pushing a reload for an autoinc when we had previously reloaded an
5139 inner part of the address.
5141 2014-01-16 Jakub Jelinek <jakub@redhat.com>
5143 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
5145 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
5146 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
5147 when not giving up or versioning for alias only because of
5149 (vect_analyze_data_ref_dependences): Set to true.
5150 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
5152 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
5153 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
5157 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
5160 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
5161 operand 0 predicate for gathers, use a new pseudo as subtarget.
5163 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5166 * lra-constraints.c (process_alt_operands): Add printing debug info.
5167 Check absence of input/output reloads for matched operands too.
5169 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
5171 PR rtl-optimization/59835
5172 * ira.c (ira_init_register_move_cost): Increase cost for
5175 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
5177 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
5179 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
5182 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
5183 non-register objects. Use gen_(high/low)part more consistently.
5186 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
5189 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
5190 endian support, remove tests for WORDS_BIG_ENDIAN.
5191 (p8_mfvsrd_3_<mode>): Likewise.
5192 (reload_gpr_from_vsx<mode>): Likewise.
5193 (reload_gpr_from_vsxsf): Likewise.
5194 (p8_mfvsrd_4_disf): Likewise.
5196 2014-01-16 Richard Biener <rguenther@suse.de>
5198 PR rtl-optimization/46590
5199 * lcm.c (compute_antinout_edge): Use postorder iteration.
5200 (compute_laterin): Use inverted postorder iteration.
5202 2014-01-16 Nick Clifton <nickc@redhat.com>
5205 * varasm.c (output_constant): Return the number of bytes actually
5207 (output_constructor_array_range): Update the field size with the
5208 number of bytes emitted by output_constant.
5209 (output_constructor_regular_field): Likewise. Also do not
5210 complain if the total number of bytes emitted is now greater
5211 than the expected fieldpos.
5212 * output.h (output_constant): Update prototype and descriptive comment.
5214 2014-01-16 Marek Polacek <polacek@redhat.com>
5217 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
5218 it is error_mark_node.
5220 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
5222 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
5223 VALID_AVX256_REG_OR_OI_MODE.
5225 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
5227 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
5228 current procedure should be profiled.
5230 2014-01-15 Andrew Pinski <apinski@cavium.com>
5232 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
5233 of moving from/to the STACK_REG register class.
5235 2014-01-15 Richard Henderson <rth@redhat.com>
5238 * reginfo.c (global_regs_decl): Globalize.
5239 * rtl.h (global_regs_decl): Declare.
5240 * ira.c (do_reload): Diagnose frame_pointer_needed and it
5241 reserved via global_regs.
5243 2014-01-15 Teresa Johnson <tejohnson@google.com>
5245 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
5247 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
5249 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
5250 and vmulosh rather than call gen_vec_widen_smult_*.
5251 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
5252 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
5253 (vec_widen_smult_even_v16qi): Likewise.
5254 (vec_widen_umult_even_v8hi): Likewise.
5255 (vec_widen_smult_even_v8hi): Likewise.
5256 (vec_widen_umult_odd_v16qi): Likewise.
5257 (vec_widen_smult_odd_v16qi): Likewise.
5258 (vec_widen_umult_odd_v8hi): Likewise.
5259 (vec_widen_smult_odd_v8hi): Likewise.
5260 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
5261 vmuloub rather than call gen_vec_widen_umult_*.
5262 (vec_widen_umult_lo_v16qi): Likewise.
5263 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
5264 vmulosb rather than call gen_vec_widen_smult_*.
5265 (vec_widen_smult_lo_v16qi): Likewise.
5266 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
5267 rather than call gen_vec_widen_umult_*.
5268 (vec_widen_umult_lo_v8hi): Likewise.
5269 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
5270 rather than call gen_vec_widen_smult_*.
5271 (vec_widen_smult_lo_v8hi): Likewise.
5273 2014-01-15 Jeff Law <law@redhat.com>
5275 PR tree-optimization/59747
5276 * ree.c (find_and_remove_re): Properly handle case where a second
5277 eliminated extension requires widening a copy created for elimination
5278 of a prior extension.
5279 (combine_set_extension): Ensure that the number of hard regs needed
5280 for a destination register does not change when we widen it.
5282 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
5284 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
5285 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
5286 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
5287 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
5288 (avr-*-rtems*): Likewise.
5289 (bfin*-rtems*): Likewise.
5290 (moxie-*-rtems*): Likewise.
5291 (h8300-*-rtems*): Likewise.
5292 (i[34567]86-*-rtems*): Likewise.
5293 (lm32-*-rtems*): Likewise.
5294 (m32r-*-rtems*): Likewise.
5295 (m68k-*-rtems*): Likewise.
5296 (microblaze*-*-rtems*): Likewise.
5297 (mips*-*-rtems*): Likewise.
5298 (powerpc-*-rtems*): Likewise.
5299 (sh-*-rtems*): Likewise.
5300 (sparc-*-rtems*): Likewise.
5301 (sparc64-*-rtems*): Likewise.
5302 (v850-*-rtems*): Likewise.
5303 (m32c-*-rtems*): Likewise.
5305 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
5307 PR rtl-optimization/59511
5308 * ira.c (ira_init_register_move_cost): Use memory costs for some
5309 cases of register move cost calculations.
5310 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
5311 instead of BB frequency.
5312 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
5313 * lra-assigns.c (find_hard_regno_for): Ditto.
5315 2014-01-15 Richard Biener <rguenther@suse.de>
5317 PR tree-optimization/59822
5318 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
5319 (vectorizable_load): Use it to hoist defs of uses of invariant
5320 loads out of the loop.
5322 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
5323 Kugan Vivekanandarajah <kuganv@linaro.org>
5326 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
5329 2014-01-15 Richard Biener <rguenther@suse.de>
5331 PR rtl-optimization/59802
5332 * lcm.c (compute_available): Use inverted postorder to seed
5333 the initial worklist.
5335 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5338 * config/s390/s390.c (s390_preferred_reload_class): Don't return
5339 ADDR_REGS for invalid symrefs in non-PIC code.
5341 2014-01-15 Jakub Jelinek <jakub@redhat.com>
5344 * builtins.c (determine_block_size): Initialize *probable_max_size
5345 even if len_rtx is CONST_INT.
5347 2014-01-14 Andrew Pinski <apinski@cavium.com>
5349 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
5350 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
5351 (cortexa53_tunings): Likewise.
5352 (aarch64_sched_issue_rate): New function.
5353 (TARGET_SCHED_ISSUE_RATE): Define.
5355 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5357 * ira-costs.c (find_costs_and_classes): Add missed
5358 ira_init_register_move_cost_if_necessary.
5360 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
5363 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
5365 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
5368 * config/i386/i386.c (type_natural_mode): Add a bool parameter
5369 to indicate if type is used for function return value. Warn ABI
5370 change if the vector mode isn't available for function return value.
5371 (ix86_function_arg_advance): Pass false to type_natural_mode.
5372 (ix86_function_arg): Likewise.
5373 (ix86_gimplify_va_arg): Likewise.
5374 (function_arg_32): Don't warn ABI change.
5375 (ix86_function_value): Pass true to type_natural_mode.
5376 (ix86_return_in_memory): Likewise.
5377 (ix86_struct_value_rtx): Removed.
5378 (TARGET_STRUCT_VALUE_RTX): Likewise.
5380 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
5382 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
5383 converting a conditional jump into a conditional return.
5385 2014-01-14 Richard Biener <rguenther@suse.de>
5387 PR tree-optimization/58921
5388 PR tree-optimization/59006
5389 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
5390 hoisting invariant stmts.
5391 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
5392 invariant loads on the preheader edge if possible.
5394 2014-01-14 Joey Ye <joey.ye@arm.com>
5396 * doc/plugin.texi (Building GCC plugins): Update to C++.
5398 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
5400 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
5401 (_mm_rcp28_round_ss): Ditto.
5402 (_mm_rsqrt28_round_sd): Ditto.
5403 (_mm_rsqrt28_round_ss): Ditto.
5404 (_mm_rcp28_sd): Ditto.
5405 (_mm_rcp28_ss): Ditto.
5406 (_mm_rsqrt28_sd): Ditto.
5407 (_mm_rsqrt28_ss): Ditto.
5408 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
5409 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
5410 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
5411 (IX86_BUILTIN_RCP28SD): Ditto.
5412 (IX86_BUILTIN_RCP28SS): Ditto.
5413 (IX86_BUILTIN_RSQRT28SD): Ditto.
5414 (IX86_BUILTIN_RSQRT28SS): Ditto.
5415 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
5416 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
5417 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
5418 (ix86_expand_special_args_builtin): Expand new FTYPE.
5419 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
5420 (srcp14<mode>): Make insn unary.
5421 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
5422 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
5423 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
5424 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
5425 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
5426 Fix rounding: make it SAE only.
5427 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
5429 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
5431 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
5432 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
5433 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
5434 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
5435 (round_saeonly_mask_scalar_operand4): Ditto.
5436 (round_saeonly_mask_scalar_op3): Ditto.
5437 (round_saeonly_mask_scalar_op4): Ditto.
5439 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5441 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5442 Implement -maltivec=be for vec_insert and vec_extract.
5444 2014-01-10 DJ Delorie <dj@redhat.com>
5446 * config/msp430/msp430.md (call_internal): Don't allow memory
5447 references with SP as the base register.
5448 (call_value_internal): Likewise.
5449 * config/msp430/constraints.md (Yc): New. For memory references
5450 that don't use SP as a base register.
5452 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
5453 "an integer without a # prefix"
5454 * config/msp430/msp430.md (epilogue_helper): Use it.
5456 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5459 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
5460 AVX512F gather builtins.
5461 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
5462 on gather decls with INTEGER_TYPE masktype.
5463 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
5464 directly into the builtin rather than hoisting it before loop.
5466 PR tree-optimization/59387
5467 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
5468 (scev_const_prop): If folded_casts and type has undefined overflow,
5469 use force_gimple_operand instead of force_gimple_operand_gsi and
5470 for each added stmt if it is assign with
5471 arith_code_with_undefined_signed_overflow, call
5472 rewrite_to_defined_overflow.
5473 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
5474 gimple-fold.h instead.
5475 (arith_code_with_undefined_signed_overflow,
5476 rewrite_to_defined_overflow): Moved to ...
5477 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
5478 rewrite_to_defined_overflow): ... here. No longer static.
5479 Include gimplify-me.h.
5480 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
5481 rewrite_to_defined_overflow): New prototypes.
5483 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5485 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
5487 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
5489 * builtins.c (get_object_alignment_2): Minor tweak.
5490 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
5492 2014-01-13 Christian Bruel <christian.bruel@st.com>
5494 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
5495 optimized non constant lengths.
5497 2014-01-13 Jakub Jelinek <jakub@redhat.com>
5500 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
5501 load as __atomic_load_N if possible.
5503 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5505 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
5507 (rs6000_expand_builtin): Adjust call.
5509 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
5512 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
5513 * config/rs6000/rs6000.c: Include target-globals.h.
5514 (rs6000_set_current_function): Instead of doing target_reinit
5515 unconditionally, use save_target_globals_default_opts and
5516 restore_target_globals.
5518 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
5520 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
5521 (rs6000_expand_builtin): Handle mffs and mtfsf.
5522 (rs6000_init_builtins): Define mffs and mtfsf.
5523 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
5524 (rs6000_mffs): New pattern.
5525 (rs6000_mtfsf): New pattern.
5527 2014-01-11 Bin Cheng <bin.cheng@arm.com>
5529 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
5530 Start narrowing with START. Apply candidate-use pair
5531 and check overall cost in narrowing.
5532 (iv_ca_prune): Pass new argument.
5534 2014-01-10 Jeff Law <law@redhat.com>
5537 * ree.c (combine_reaching_defs): Ensure the defining statement
5538 occurs before the extension when optimizing extensions with
5539 different source and destination hard registers.
5541 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5544 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
5545 vtables into the type inheritance graph.
5547 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5549 PR rtl-optimization/59754
5550 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
5551 modes in the REGNO != REGNO case.
5553 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5555 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
5557 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5559 PR tree-optimization/59745
5560 * tree-predcom.c (tree_predictive_commoning_loop): Call
5561 free_affine_expand_cache if giving up because components is NULL.
5563 * target-globals.c (save_target_globals): Allocate < 4KB structs using
5564 GC in payload of target_globals struct instead of allocating them on
5565 the heap and the larger structs separately using GC.
5566 * target-globals.h (struct target_globals): Make regs, hard_regs,
5567 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
5568 of GTY((skip)) and change type to void *.
5569 (reset_target_globals): Cast loads from those fields to corresponding
5572 2014-01-10 Steve Ellcey <sellcey@mips.com>
5575 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
5576 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
5577 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
5579 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
5582 * aarch64-modes.def (CC_Zmode): New flags mode.
5583 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
5584 represents an equality.
5585 (aarch64_get_condition_code): Handle CC_Zmode.
5586 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
5588 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
5590 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
5591 extraction in good case.
5593 2014-01-10 Richard Biener <rguenther@suse.de>
5595 PR tree-optimization/59374
5596 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
5597 checking after SLP discovery. Mark stmts not participating
5598 in any SLP instance properly.
5600 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5602 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
5603 when handling a SET rtx.
5605 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5607 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
5608 (cortex-a57): Likewise.
5609 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
5611 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5613 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
5614 non-iwmmxt builtins.
5616 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
5620 * ipa-devirt.c record_target_from_binfo): Take as argument
5621 stack of binfos and lookup matching one for virtual inheritance.
5622 (possible_polymorphic_call_targets_1): Update.
5624 2014-01-10 Huacai Chen <chenhc@lemote.com>
5626 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
5627 kernel strings for Loongson-2E/2F/3A.
5629 2014-01-10 Jakub Jelinek <jakub@redhat.com>
5632 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
5633 is_gimple_call before calling gimple_call_internal_p.
5635 2014-01-09 Steve Ellcey <sellcey@mips.com>
5637 * Makefile.in (TREE_FLOW_H): Remove.
5638 (TREE_SSA_H): Add file names from tree-flow.h.
5639 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
5640 * tree.h: Remove tree-flow.h reference.
5641 * hash-table.h: Remove tree-flow.h reference.
5642 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
5643 reference with tree-ssa-loop.h.
5645 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5647 * doc/invoke.texi: Add -maltivec={be,le} options, and document
5648 default element-order behavior for -maltivec.
5649 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
5650 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
5651 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
5652 when targeting big endian, at least for now.
5653 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
5655 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5658 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
5659 var satisfies use_register_for_decl, just take into account type
5660 alignment, rather than decl alignment.
5662 PR tree-optimization/59622
5663 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
5664 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
5665 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
5666 Don't devirtualize for inplace at all. For targets.length () == 1,
5667 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
5669 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5671 * config/i386/i386.md (cpu): Remove the unused btver1.
5673 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
5675 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
5677 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5680 * tree-core.h (struct target_globals): New forward declaration.
5681 (struct tree_target_option): Add globals field.
5682 * tree.h (TREE_TARGET_GLOBALS): Define.
5683 (prepare_target_option_nodes_for_pch): New prototype.
5684 * target-globals.h (struct target_globals): Define even if
5686 * tree.c (prepare_target_option_node_for_pch,
5687 prepare_target_option_nodes_for_pch): New functions.
5688 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
5689 * config/i386/i386.c: Include target-globals.h.
5690 (ix86_set_current_function): Instead of doing target_reinit
5691 unconditionally, use save_target_globals_default_opts and
5692 restore_target_globals.
5694 2014-01-09 Richard Biener <rguenther@suse.de>
5696 PR tree-optimization/59715
5697 * tree-cfg.h (split_critical_edges): Declare.
5698 * tree-cfg.c (split_critical_edges): Export.
5699 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
5701 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
5703 * cfgexpand.c (expand_stack_vars): Optionally disable
5704 asan stack protection.
5705 (expand_used_vars): Likewise.
5706 (partition_stack_vars): Likewise.
5707 * asan.c (asan_emit_stack_protection): Optionally disable
5708 after return stack usage.
5709 (instrument_derefs): Optionally disable memory access instrumentation.
5710 (instrument_builtin_call): Likewise.
5711 (instrument_strlen_call): Likewise.
5712 (asan_protect_global): Optionally disable global variables protection.
5713 * doc/invoke.texi: Added doc for new options.
5714 * params.def: Added new options.
5715 * params.h: Likewise.
5717 2014-01-09 Jakub Jelinek <jakub@redhat.com>
5719 PR rtl-optimization/59724
5720 * ifcvt.c (cond_exec_process_if_block): Don't call
5721 flow_find_head_matching_sequence with 0 longest_match.
5722 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
5723 non-active insns if !stop_after.
5724 (try_head_merge_bb): Revert 2014-01-07 changes.
5726 2014-01-08 Jeff Law <law@redhat.com>
5728 * ree.c (get_sub_rtx): New function, extracted from...
5729 (merge_def_and_ext): Here.
5730 (combine_reaching_defs): Use get_sub_rtx.
5732 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
5734 * cgraph.h (varpool_variable_node): Do not choke on null node.
5736 2014-01-08 Catherine Moore <clm@codesourcery.com>
5738 * config/mips/mips.md (simple_return): Attempt to use JRC
5740 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
5742 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5744 PR rtl-optimization/59137
5745 * reorg.c (steal_delay_list_from_target): Call update_block for
5747 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
5749 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5751 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5752 two duplicate entries.
5754 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
5757 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
5759 * config/mips/mips.c (mips_truncated_op_cost): New function.
5760 (mips_rtx_costs): Adjust test for BADDU.
5761 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
5763 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
5765 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
5766 (*baddu_si): ...this new pattern.
5768 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5771 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
5773 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
5776 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
5778 (expand_expr, expand_normal): Adjust.
5779 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
5780 inner_reference_p. Use inner_reference_p to expand inner references.
5781 (store_expr): Adjust.
5782 * cfgexpand.c (expand_call_stmt): Adjust.
5784 2014-01-08 Rong Xu <xur@google.com>
5786 * gcov-io.c (gcov_var): Move from gcov-io.h.
5787 (gcov_position): Ditto.
5788 (gcov_is_error): Ditto.
5789 (gcov_rewrite): Ditto.
5790 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
5791 only part to libgcc/libgcov.h.
5793 2014-01-08 Marek Polacek <polacek@redhat.com>
5796 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
5798 2014-01-08 Marek Polacek <polacek@redhat.com>
5801 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
5803 2014-01-08 Jakub Jelinek <jakub@redhat.com>
5805 PR rtl-optimization/59649
5806 * stor-layout.c (get_mode_bounds): For BImode return
5807 0 and STORE_FLAG_VALUE.
5809 2014-01-08 Richard Biener <rguenther@suse.de>
5812 * gimple.h (is_gimple_builtin_call): Remove.
5813 (gimple_builtin_call_types_compatible_p): New.
5814 (gimple_call_builtin_p): New overload.
5815 * gimple.c (is_gimple_builtin_call): Remove.
5816 (validate_call): Rename to ...
5817 (gimple_builtin_call_types_compatible_p): ... this and export. Also
5819 (validate_type): New static function.
5820 (gimple_call_builtin_p): New overload and adjust.
5821 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
5822 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
5823 (gimple_fold_stmt_to_constant_1): Likewise.
5824 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
5826 2014-01-08 Richard Biener <rguenther@suse.de>
5829 * gimplify.c (gimplify_expr): Gimplify register-register type
5830 VIEW_CONVERT_EXPRs to separate stmts.
5832 2014-01-07 Jeff Law <law@redhat.com>
5835 * ree.c (combine_set_extension): Handle case where source
5836 and destination registers in an extension insn are different.
5837 (combine_reaching_defs): Allow source and destination registers
5838 in extension to be different under limited circumstances.
5839 (add_removable_extension): Remove restriction that the
5840 source and destination registers in the extension are the same.
5841 (find_and_remove_re): Emit a copy from the extension's
5842 destination to its source after the defining insn if
5843 the source and destination registers are different.
5846 * ifcvt.c (merge_if_block): If we are merging a block with more than
5847 one successor with a block with no successors, remove any BARRIER
5848 after the second block.
5850 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
5852 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
5854 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
5857 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
5858 for 14-bit register offsets when INT14_OK_STRICT is false.
5860 2014-01-07 Roland Stigge <stigge@antcom.de>
5861 Michael Meissner <meissner@linux.vnet.ibm.com>
5864 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
5865 Only check TFmode for SPE constants. Don't check TImode or TDmode.
5867 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
5869 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
5872 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
5874 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
5875 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
5876 rtx is const0_rtx or not.
5878 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
5881 * target-globals.c (save_target_globals): Remove this_fn_optab
5883 * toplev.c: Include optabs.h.
5884 (target_reinit): Temporarily restore the global options if another
5885 set of options are in force.
5887 2014-01-07 Jakub Jelinek <jakub@redhat.com>
5889 PR rtl-optimization/58668
5890 * cfgcleanup.c (flow_find_cross_jump): Don't count
5891 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
5892 to determine what is counted.
5893 (flow_find_head_matching_sequence): Use active_insn_p to determine
5895 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
5897 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
5898 determine what is counted.
5900 PR tree-optimization/59643
5901 * tree-predcom.c (split_data_refs_to_components): If one dr is
5902 read and one write, determine_offset fails and the write isn't
5903 in the bad component, just put the read into the bad component.
5905 2014-01-07 Mike Stump <mikestump@comcast.net>
5906 Jakub Jelinek <jakub@redhat.com>
5909 * tree-core.h (struct tree_optimization_option): Change optabs
5910 type from unsigned char * to void *.
5911 * optabs.c (init_tree_optimization_optabs): Adjust
5912 TREE_OPTIMIZATION_OPTABS initialization.
5914 2014-01-06 Jakub Jelinek <jakub@redhat.com>
5917 * config/i386/i386.h (struct machine_function): Add
5918 no_drap_save_restore field.
5919 * config/i386/i386.c (ix86_save_reg): Use
5920 !cfun->machine->no_drap_save_restore instead of
5921 crtl->stack_realign_needed.
5922 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
5923 this function clears frame_pointer_needed. Set
5924 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
5925 and DRAP reg is needed.
5927 2014-01-06 Marek Polacek <polacek@redhat.com>
5930 * doc/implement-c.texi: Mention that other integer types are
5931 permitted as bit-field types in strictly conforming mode.
5933 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
5935 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
5938 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
5940 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
5942 2014-01-06 Martin Jambor <mjambor@suse.cz>
5945 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
5947 * ipa-prop.c (ipa_print_node_params): Fix indentation.
5949 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
5953 * var-tracking.c (add_stores): Preserve the value of the source even if
5954 we don't record the store.
5956 2014-01-06 Terry Guo <terry.guo@arm.com>
5958 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
5960 2014-01-05 Iain Sandoe <iain@codesourcery.com>
5963 * config/darwin.c (darwin_function_section): Adjust return values to
5964 correspond to optimisation changes made in r206070.
5966 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
5968 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
5969 from prefetch_block tune setting.
5970 (nocona_cost): Correct size of prefetch block to 64.
5972 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
5974 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
5975 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
5976 used to save the static chain register in the computation of the offset
5977 from which the FP registers need to be restored.
5979 2014-01-04 Jakub Jelinek <jakub@redhat.com>
5981 PR tree-optimization/59519
5982 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
5983 ICE if get_current_def (current_new_name) is already non-NULL, as long
5984 as it is a phi result of some other phi in *new_exit_bb that has
5987 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
5988 or vmovdqu* for misaligned_operand.
5989 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
5990 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
5991 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
5992 aligned_mem for AVX512F masked aligned load and store builtins and for
5995 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
5997 PR tree-optimization/59651
5998 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
5999 Address range for negative step should be added by TYPE_SIZE_UNIT.
6001 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
6003 * config/m68k/m68k.c (handle_move_double): Handle pushes with
6004 overlapping registers also for registers other than the stack pointer.
6006 2014-01-03 Marek Polacek <polacek@redhat.com>
6009 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
6012 2014-01-03 Jakub Jelinek <jakub@redhat.com>
6015 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
6018 * config/i386/i386.md (MODE_SIZE): New mode attribute.
6019 (push splitter): Use <P:MODE_SIZE> instead of
6020 GET_MODE_SIZE (<P:MODE>mode).
6021 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
6022 (mov -1, reg peephole2): Likewise.
6023 * config/i386/sse.md (*mov<mode>_internal,
6024 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
6025 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
6026 *<code><mode>3, *andnot<mode>3<mask_name>,
6027 <mask_codefor><code><mode>3<mask_name>): Likewise.
6028 * config/i386/subst.md (mask_mode512bit_condition,
6029 sd_mask_mode512bit_condition): Likewise.
6031 2014-01-02 Xinliang David Li <davidxl@google.com>
6033 PR tree-optimization/59303
6034 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
6035 (dump_predicates): Better output format.
6036 (pred_equal_p): New function.
6037 (is_neq_relop_p): Ditto.
6038 (is_neq_zero_form_p): Ditto.
6039 (pred_expr_equal_p): Ditto.
6040 (pred_neg_p): Ditto.
6041 (simplify_pred): Ditto.
6042 (simplify_preds_2): Ditto.
6043 (simplify_preds_3): Ditto.
6044 (simplify_preds_4): Ditto.
6045 (simplify_preds): Ditto.
6047 (push_to_worklist): Ditto.
6048 (get_pred_info_from_cmp): Ditto.
6049 (is_degenerated_phi): Ditto.
6050 (normalize_one_pred_1): Ditto.
6051 (normalize_one_pred): Ditto.
6052 (normalize_one_pred_chain): Ditto.
6053 (normalize_preds): Ditto.
6054 (normalize_cond_1): Remove function.
6055 (normalize_cond): Ditto.
6056 (is_gcond_subset_of): Ditto.
6057 (is_subset_of_any): Ditto.
6058 (is_or_set_subset_of): Ditto.
6059 (is_and_set_subset_of): Ditto.
6060 (is_norm_cond_subset_of): Ditto.
6061 (pred_chain_length_cmp): Ditto.
6062 (convert_control_dep_chain_into_preds): Type change.
6063 (find_predicates): Ditto.
6064 (find_def_preds): Ditto.
6065 (destroy_predicates_vecs): Ditto.
6066 (find_matching_predicates_in_rest_chains): Ditto.
6067 (use_pred_not_overlap_with_undef_path_pred): Ditto.
6068 (is_pred_expr_subset): Ditto.
6069 (is_pred_chain_subset_of): Ditto.
6070 (is_included_in): Ditto.
6071 (is_superset_of): Ditto.
6073 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6075 Update copyright years.
6077 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
6079 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
6080 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
6081 config/arc/arc.md, config/arc/arc.opt,
6082 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
6083 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
6084 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
6085 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
6086 config/linux-protos.h, config/linux.c, config/winnt-c.c,
6087 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
6088 vtable-verify.c, vtable-verify.h: Use the standard form for the
6091 2014-01-02 Tobias Burnus <burnus@net-b.de>
6093 * gcc.c (process_command): Update copyright notice dates.
6094 * gcov-dump.c: Ditto.
6096 * doc/cpp.texi: Bump @copying's copyright year.
6097 * doc/cppinternals.texi: Ditto.
6098 * doc/gcc.texi: Ditto.
6099 * doc/gccint.texi: Ditto.
6100 * doc/gcov.texi: Ditto.
6101 * doc/install.texi: Ditto.
6102 * doc/invoke.texi: Ditto.
6104 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
6106 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
6108 2014-01-01 Jakub Jelinek <jakub@redhat.com>
6110 * config/i386/sse.md (*mov<mode>_internal): Guard
6111 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
6113 PR rtl-optimization/59647
6114 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
6115 new_rtx into UNSIGNED_FLOAT rtxes.
6117 Copyright (C) 2014 Free Software Foundation, Inc.
6119 Copying and distribution of this file, with or without modification,
6120 are permitted in any medium without royalty provided the copyright
6121 notice and this notice are preserved.