1 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
3 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
4 proper constants and fix formatting.
5 (possible_polymorphic_call_targets): Fix formatting.
7 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
8 Ilya Tocar <ilya.tocar@intel.com>
10 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
11 (_mm512_loadu_epi32): Renamed into...
12 (_mm512_loadu_si512): This.
13 (_mm512_storeu_epi32): Renamed into...
14 (_mm512_storeu_si512): This.
15 (_mm512_maskz_ceil_ps): Removed.
16 (_mm512_maskz_ceil_pd): Ditto.
17 (_mm512_maskz_floor_ps): Ditto.
18 (_mm512_maskz_floor_pd): Ditto.
19 (_mm512_floor_round_ps): Ditto.
20 (_mm512_floor_round_pd): Ditto.
21 (_mm512_ceil_round_ps): Ditto.
22 (_mm512_ceil_round_pd): Ditto.
23 (_mm512_mask_floor_round_ps): Ditto.
24 (_mm512_mask_floor_round_pd): Ditto.
25 (_mm512_mask_ceil_round_ps): Ditto.
26 (_mm512_mask_ceil_round_pd): Ditto.
27 (_mm512_maskz_floor_round_ps): Ditto.
28 (_mm512_maskz_floor_round_pd): Ditto.
29 (_mm512_maskz_ceil_round_ps): Ditto.
30 (_mm512_maskz_ceil_round_pd): Ditto.
31 (_mm512_expand_pd): Ditto.
32 (_mm512_expand_ps): Ditto.
33 * config/i386/i386.c (ix86_builtins): Remove
34 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
36 * config/i386/predicates.md (const1256_operand): New.
37 (const_1_to_2_operand): Ditto.
38 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
39 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
40 (*avx512pf_gatherpf<mode>sf): Ditto.
41 (avx512pf_gatherpf<mode>df): Ditto.
42 (*avx512pf_gatherpf<mode>df_mask): Ditto.
43 (*avx512pf_gatherpf<mode>df): Ditto.
44 (avx512pf_scatterpf<mode>sf): Ditto.
45 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
46 (*avx512pf_scatterpf<mode>sf): Ditto.
47 (avx512pf_scatterpf<mode>df): Ditto.
48 (*avx512pf_scatterpf<mode>df_mask): Ditto.
49 (*avx512pf_scatterpf<mode>df): Ditto.
50 (avx512f_expand<mode>): Removed.
51 (<shift_insn><mode>3<mask_name>): Change predicate type.
53 2014-02-08 Jakub Jelinek <jakub@redhat.com>
55 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
56 not at the end of datarefs vector use ordered_remove to avoid
57 reordering datarefs vector.
60 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
61 mark local addressable non-static vars as GOVD_PRIVATE
62 instead of GOVD_LOCAL.
63 * omp-low.c (lower_omp_for): Move gimple_bind_vars
64 and BLOCK_VARS of gimple_bind_block to new_stmt rather
68 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
69 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
70 assume_aligned or alloc_align attributes.
71 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
72 arguments. Handle also assume_aligned and alloc_align attributes.
73 (evaluate_stmt): Adjust bit_value_assume_aligned caller.
74 Handle calls to functions with assume_aligned or alloc_align
76 * doc/extend.texi: Document assume_aligned and alloc_align
79 2014-02-08 Terry Guo <terry.guo@arm.com>
81 * doc/invoke.texi: Document ARM -march=armv7e-m.
83 2014-02-08 Jakub Jelinek <jakub@redhat.com>
85 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
86 flag on __cilkrts_rethrow builtin.
89 * ipa-cp.c (determine_versionability): Fail at -O0
90 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
91 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
94 2014-02-04 Jakub Jelinek <jakub@redhat.com>
97 * tree-inline.c (copy_forbidden): Fail for
98 __attribute__((optimize (0))) functions.
100 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
102 * varpool.c: Include pointer-set.h.
103 (varpool_remove_unreferenced_decls): Variables in other partitions
104 will not be output; be however careful to not lose information
107 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
109 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
110 lookup in the vtable constructor.
112 2014-02-07 Jeff Law <law@redhat.com>
115 * config/m68k/m68k.md (ashldi_extsi): Turn into a
116 define_insn_and_split.
118 * ipa-inline.c (inline_small_functions): Fix typos.
120 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
122 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
123 (s390_can_use_return_insn): Declare.
124 * config/s390/s390.h (EPILOGUE_USES): Define.
125 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
127 (s390_chunkify_start): Handle return JUMP_LABELs.
128 (s390_early_mach): Emit a main_pool instruction on the entry edge.
129 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
130 (s390_can_use_return_insn): New functions.
131 (s390_fix_long_loop_prediction): Handle conditional returns.
132 (TARGET_SET_UP_BY_PROLOGUE): Define.
133 * config/s390/s390.md (ANY_RETURN): New code iterator.
134 (*creturn, *csimple_return, return, simple_return): New patterns.
136 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
138 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
139 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
140 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
141 REG_CFA_RESTORE list when deciding not to restore a register.
143 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
145 * config/s390/s390.c: Include tree-pass.h and context.h.
146 (s390_early_mach): New function, split out from...
147 (s390_emit_prologue): ...here.
148 (pass_data_s390_early_mach): New pass structure.
149 (pass_s390_early_mach): New class.
150 (s390_option_override): Create and register early_mach pass.
153 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
155 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
156 to match for the exit block.
158 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
160 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
161 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
162 Reject misaligned operands.
164 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
166 * optabs.c (expand_atomic_compare_and_swap): Allow expander to
169 2014-02-07 Richard Biener <rguenther@suse.de>
172 * gimple-low.c (lower_builtin_posix_memalign): New function.
173 (lower_stmt): Call it to lower posix_memalign in a way
174 to make alignment info accessible.
176 2014-02-07 Jakub Jelinek <jakub@redhat.com>
179 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
180 __builtin_setjmp_receiver.
182 2014-02-07 Richard Biener <rguenther@suse.de>
185 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
186 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
187 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
188 Handle BUILT_IN_POSIX_MEMALIGN.
189 (find_func_clobbers): Likewise.
190 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
191 (call_may_clobber_ref_p_1): Likewise.
193 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
196 * ipa-devirt.c (record_target_from_binfo): Remove overactive sanity check.
198 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
201 * lto-cgraph.c (lto_output_node): Use
202 symtab_get_symbol_partitioning_class.
203 (lto_output_varpool_node): likewise.
204 (symtab_get_symbol_partitioning_class): Move here from
206 * cgraph.h (symbol_partitioning_class): Likewise.
207 (symtab_get_symbol_partitioning_class): Declare.
209 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
211 * ggc.h (ggc_internal_cleared_alloc): New macro.
212 * vec.h (vec_safe_copy): Handle memory stats.
213 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
214 * target-globals.c (save_target_globals): Likewise.
216 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
219 * expr.c (emit_move_resolve_push): Export; be bit more selective
220 on when to clear alias set.
221 * expr.h (emit_move_resolve_push): Declare.
222 * function.h (struct function): Add tail_call_marked.
223 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
224 * config/i386/i386-protos.h (ix86_expand_push): Remove.
225 * config/i386/i386.md (TImode move expander): De not call
227 (FP push expanders): Preserve memory attributes.
228 * config/i386/sse.md (push<mode>1): Remove.
229 * config/i386/i386.c (ix86_expand_vector_move): Handle push
231 (ix86_expand_push): Remove.
232 * config/i386/mmx.md (push<mode>1): Remove.
234 2014-02-06 Jakub Jelinek <jakub@redhat.com>
236 PR rtl-optimization/60030
237 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
238 lopart with paradoxical subreg before shifting it up by hprec.
240 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
242 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
243 Remove extra newline at end of file.
244 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
245 (arm_issue_rate): Handle cortexa57.
246 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
247 (cortex-a57.cortex-a53): Likewise.
249 2014-02-06 Jakub Jelinek <jakub@redhat.com>
252 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
253 don't record in REG_FRAME_RELATED_EXPR registers not set in that
255 (arm_expand_prologue): Adjust all callers.
256 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
257 info, registers also at the lowest numbered registers side. Use
258 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
262 * var-tracking.c (adjust_mems): Before adding a SET
263 to amd->side_effects, adjust it's SET_SRC using
264 simplify_replace_fn_rtx.
266 2014-02-06 Alan Modra <amodra@gmail.com>
269 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
270 change SDmode to DDmode when lra_in_progress.
272 2014-02-06 Jakub Jelinek <jakub@redhat.com>
275 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
276 free_data_ref on the dr first, and before goto again also set dr
277 to the next dr. For simd_lane_access, free old datarefs[i] before
278 overwriting it. For get_vectype_for_scalar_type failure, don't
279 free_data_ref if simd_lane_access.
281 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
284 * tree.h (opts_for_fn): New inline function.
285 (opt_for_fn): Define.
286 * config/i386/i386.c (ix86_function_regparm): Use
287 opt_for_fn (decl, optimize) instead of optimize.
289 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
291 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
292 for SYMBOL_REF in large memory model.
294 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
296 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
298 (cortex-a57): Likewise.
299 (cortex-a57.cortex-a53): Likewise.
301 2014-02-06 Yury Gribov <y.gribov@samsung.com>
302 Kugan Vivekanandarajah <kuganv@linaro.org>
304 * config/arm/arm.c (arm_vector_alignment_reachable): Check
306 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
308 2014-02-06 Richard Biener <rguenther@suse.de>
310 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
311 set_loop_copy and initialize_original_copy_tables.
313 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
315 * config/aarch64/aarch64-simd.md
316 (aarch64_ashr_simddi): Change QI to SI.
318 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
319 Jakub Jelinek <jakub@redhat.com>
322 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
325 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
327 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
328 CODE_FOR_altivec_vpku[hw]um to
329 CODE_FOR_altivec_vpku[hw]um_direct.
330 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
331 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
332 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
333 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
335 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
337 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
338 generation for -maltivec=be.
339 (altivec_vsumsws): Simplify redundant test.
341 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
343 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
344 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
345 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
346 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
348 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
350 (altivec_vpks<VI_char>ss): Likewise.
351 (altivec_vpks<VI_char>us): Likewise.
352 (altivec_vpku<VI_char>us): Likewise.
353 (altivec_vpku<VI_char>um): Likewise.
354 (altivec_vpku<VI_char>um_direct): New (copy of
355 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
357 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
358 target is little endian and -maltivec=be is not specified.
359 (*altivec_vupkhs<VU_char>_direct): New (copy of
360 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal
362 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
363 target is little endian and -maltivec=be is not specified.
364 (*altivec_vupkls<VU_char>_direct): New (copy of
365 altivec_vupkls<VU_char> that always emits vupkls*, for internal
367 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
368 little endian and -maltivec=be is not specified.
369 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
370 little endian and -maltivec=be is not specified.
372 2014-02-05 Richard Henderson <rth@redhat.com>
375 * combine-stack-adj.c: Revert r206943.
376 * sched-int.h (struct deps_desc): Add last_args_size.
377 * sched-deps.c (init_deps): Initialize it.
378 (sched_analyze_insn): Add OUTPUT dependencies between insns that
379 contain REG_ARGS_SIZE notes.
381 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
383 * lto-cgraph.c (asm_nodes_output): Make global.
384 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
385 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA
387 (driver_handle_option): Handle OPT_fwpa.
389 2014-02-05 Jakub Jelinek <jakub@redhat.com>
392 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
393 a comment typo and formatting issue. If odr_hash hasn't been
394 created, return vNULL and set *completep to false.
397 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
398 bb with no successors.
400 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
403 * doc/invoke.texi (-march): Clarify documentation for ARM.
407 2014-02-05 Richard Biener <rguenther@suse.de>
409 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
410 when not vectorizing because of too many alias checks.
411 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
412 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
414 2014-02-05 Nick Clifton <nickc@redhat.com>
416 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
417 accept extended registers in any mode when compiling for the
420 2014-02-05 Yury Gribov <y.gribov@samsung.com>
422 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
423 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
424 sanitization attributes.
425 (can_inline_edge_p): Likewise.
426 (sanitize_attrs_match_for_inline_p): New function.
428 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
430 * ipa-prop.c (detect_type_change): Shor circuit testing of
431 type changes on THIS pointer.
433 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
436 * config/pa/pa.c (legitimize_tls_address): Return original address
437 if not passed a SYMBOL_REF rtx.
438 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
440 (pa_emit_move_sequence): Simplify TLS source operands.
441 (pa_legitimate_constant_p): Reject all TLS constants.
442 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
443 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
445 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
447 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
448 groups when we know they are controlled by LTO.
449 * varasm.c (default_binds_local_p_1): If object is in other partition,
450 it will be resolved locally.
452 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
454 * gcc/config/host-linux.c (linux_gt_pch_use_address): Don't
455 use SSIZE_MAX because it is not always defined.
457 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
460 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
461 threshold for pseudo splitting.
462 (update_ebb_live_info): Process call argument hard registers and
463 hard registers from insn definition too.
464 (max_small_class_regs_num): New constant.
465 (inherit_in_ebb): Update live hard regs through EBBs. Update
466 reloads_num only for small register classes. Don't split for
469 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
472 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
475 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
477 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default visibility is safe.
479 2014-02-04 Marek Polacek <polacek@redhat.com>
481 * gdbinit.in (pel): Define.
483 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
485 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
488 2014-02-04 Richard Biener <rguenther@suse.de>
491 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
492 in function context local.
493 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
494 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
495 similar to LTO_imported_decl_ref.
497 2014-02-04 Jakub Jelinek <jakub@redhat.com>
499 PR tree-optimization/60002
500 * cgraphclones.c (build_function_decl_skip_args): Clear
503 PR tree-optimization/60023
504 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
505 false to gsi_replace.
506 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
507 has been in some EH region and vec_stmt could throw, add
508 vec_stmt into the same EH region.
509 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
510 has no lhs, ignore it.
511 * internal-fn.c (expand_MASK_LOAD): Likewise.
514 * tree-inline.c (copy_forbidden): Fail for
515 __attribute__((optimize (0))) functions.
518 * omp-low.c (simd_clone_struct_copy): If from->inbranch
519 is set, copy one less argument.
520 (expand_simd_clones): Don't subtract clone_info->inbranch
521 from simd_clone_struct_alloc argument.
523 PR rtl-optimization/57915
524 * recog.c (simplify_while_replacing): If all unary/binary/relational
525 operation arguments are constant, attempt to simplify those.
528 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
529 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
531 2014-02-04 Richard Biener <rguenther@suse.de>
533 PR tree-optimization/60012
534 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
535 TBAA disambiguation to all DDRs.
537 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
540 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
541 (LINK_SPEC): Use it for -shared, -shared-libgcc.
543 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
546 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
548 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
550 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
551 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
553 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
556 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
557 to figure out targets of polymorphic calls with known decl.
558 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
559 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
560 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
561 (get_polymorphic_call_info): ... here.
562 (get_polymorphic_call_info_from_invariant): New function.
564 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
566 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
567 lookup via vtable pointer; check for type consistency
568 and turn inconsitent facts into UNREACHABLE.
569 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
570 * gimple-fold.c (gimple_get_virt_method_for_vtable):
571 Do not ICE on type inconsistent querries; return UNREACHABLE
574 2014-02-03 Richard Henderson <rth@twiddle.net>
577 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
578 already processed this node.
579 (normalize_one_pred_1): Pass along mark_set.
580 (normalize_one_pred): Create and destroy a pointer_set_t.
581 (normalize_one_pred_chain): Likewise.
583 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
585 PR gcov-profile/58602
586 * gcc/gcov-io.c (gcov_open): Open with truncation when mode<0
588 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
591 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on -fno-devirtualize;
592 Try to devirtualize by the knowledge of virtual table pointer given by
593 aggregate propagation.
594 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
595 ipa_print_node_jump_functions): Dump also offset that
596 is relevant for polymorphic calls.
597 (determine_known_aggregate_parts): Add arg_type parameter; use it
598 instead of determining the type from pointer type.
599 (ipa_compute_jump_functions_for_edge): Update call of
600 determine_known_aggregate_parts.
601 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
602 (gimple_get_virt_method_for_binfo): ... here; simplify using
603 vtable_pointer_value_to_vtable.
604 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
605 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
606 to unsigned HOST_WIDE_INT; Use vtable_pointer_value_to_vtable.
607 (vtable_pointer_value_to_vtable): Break out from ...; handle also
609 (vtable_pointer_value_to_binfo): ... here.
610 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
612 2014-02-03 Teresa Johnson <tejohnson@google.com>
614 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
615 redef of outer loop index variable.
617 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
621 * doc/extend.texi (Function Attributes): Typo.
623 2014-02-03 Cong Hou <congh@google.com>
625 PR tree-optimization/60000
626 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
627 if the vectorized statement is a store. A store statement can only
628 appear at the end of pattern statements.
630 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
632 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
633 (ix86_option_override_internal): Default long double to 64-bit for
634 32-bit Bionic and to 128-bit for 64-bit Bionic.
636 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
637 TARGET_LONG_DOUBLE_128 is true.
638 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
640 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
641 (mlong-double-64): Negate -mlong-double-128.
642 (mlong-double-128): New option.
644 * config/i386/i386-c.c (ix86_target_macros): Define
645 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
647 * doc/invoke.texi: Document -mlong-double-128.
649 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
651 PR rtl-optimization/60024
652 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
654 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
656 * doc/invoke.texi: (fprofile-reorder-functions): Fix typo.
658 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
660 PR rtl-optimization/57662
661 * sel-sched.c (code_motion_path_driver): Do not mark already not
662 existing blocks in the visiting bitmap.
664 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
666 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
667 on the insn being emitted.
669 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
670 Will Deacon <will.deacon@arm.com>
672 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
674 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
676 * config/arm/arm-tables.opt: Regenerate.
678 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
680 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
681 for vector types other than V16QImode.
682 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
683 define_expand, and call altivec_expand_vec_perm_le when producing
684 code with little endian element order.
685 (*altivec_vperm_<mode>_internal): New insn having previous
686 behavior of altivec_vperm_<mode>.
687 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
688 altivec_expand_vec_perm_le when producing code with little endian
690 (*altivec_vperm_<mode>_uns_internal): New insn having previous
691 behavior of altivec_vperm_<mode>_uns.
693 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
695 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
696 (altivec_vsumsws): Add handling for -maltivec=be with a little
698 (altivec_vsumsws_direct): New.
699 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
702 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
704 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
705 vtable_pointer_value_to_binfo): New functions.
706 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
707 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
709 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
711 * config/nios2/nios2.md (load_got_register): Initialize GOT
712 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
713 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
715 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
717 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
718 preserverd by passthrough, do not propagate the type.
720 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
722 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
723 (mips_atomic_assign_expand_fenv): New function.
724 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
726 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
728 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
729 (__builtin_mips_set_fcsr): Likewise.
730 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
732 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
733 (mips16_expand_set_fcsr): Likewise.
734 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
735 (mips16_set_fcsr_stub): Likewise.
736 (mips16_get_fcsr_one_only_stub): New class.
737 (mips16_set_fcsr_one_only_stub): Likewise.
738 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
739 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
740 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
741 (hard_float): New availability predicate.
742 (mips_builtins): Add get_fcsr and set_fcsr.
743 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
744 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
745 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
746 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
747 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
750 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
752 * config/mips/mips.c (mips_one_only_stub): New class.
753 (mips_need_mips16_rdhwr_p): Replace with...
754 (mips16_rdhwr_stub): ...this new variable.
755 (mips16_stub_call_address): New function.
756 (mips16_rdhwr_one_only_stub): New class.
757 (mips_expand_thread_pointer): Use mips16_stub_call_address.
758 (mips_output_mips16_rdhwr): Delete.
759 (mips_finish_stub): New function.
760 (mips_code_end): Use it to handle rdhwr stubs.
762 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
765 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
766 when calculating size of integer atomic types.
768 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
770 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
772 2014-02-01 Jakub Jelinek <jakub@redhat.com>
774 PR tree-optimization/60003
775 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
776 * profile.c (branch_prob): Use gimple_call_builtin_p
777 to check for BUILT_IN_SETJMP_RECEIVER.
778 * tree-inline.c (copy_bb): Call notice_special_calls.
780 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
783 * lra-constraints.c (process_alt_operands): Update reload_sum only
786 2014-01-31 Richard Henderson <rth@redhat.com>
789 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
790 until after else_eh is processed.
792 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
794 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
795 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
796 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
797 in smmintrin.h, remove them.
798 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
799 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
800 * config/i386/i386.md (ROUND_SAE): Fix value.
801 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
802 (const48_operand): New.
803 * config/i386/subst.md (round), (round_expand): Use
804 const_4_or_8_to_11_operand.
805 (round_saeonly), (round_saeonly_expand): Use const48_operand.
807 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
809 * config/i386/constraints.md (Yk): Swap meaning with k.
810 * config/i386/i386.md (movhi_internal): Change Yk to k.
811 (movqi_internal): Ditto.
812 (*k<logic><mode>): Ditto.
815 (kandn<mode>): Ditto.
816 (*<code>hi_1): Ditto.
817 (*<code>qi_1): Ditto.
818 (kxnor<mode>): Ditto.
822 (*one_cmplhi2_1): Ditto.
823 (*one_cmplqi2_1): Ditto.
824 * config/i386/sse.md (): Change k to Yk.
825 (avx512f_load<mode>_mask): Ditto.
826 (avx512f_blendm<mode>): Ditto.
827 (avx512f_store<mode>_mask): Ditto.
828 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
829 (avx512f_storedqu<mode>_mask): Ditto.
830 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
832 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
833 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
834 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
835 (avx512f_maskcmp<mode>3): Ditto.
836 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
837 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
838 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
839 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
840 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
841 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
842 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
843 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
844 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
845 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
846 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
847 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
848 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
849 (vec_extract_lo_<mode>_maskm): Ditto.
850 (vec_extract_hi_<mode>_maskm): Ditto.
851 (avx512f_vternlog<mode>_mask): Ditto.
852 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
853 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
854 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
855 (avx512f_<code>v8div16qi2_mask): Ditto.
856 (avx512f_<code>v8div16qi2_mask_store): Ditto.
857 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
858 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
859 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
860 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
861 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
862 (*avx512pf_gatherpf<mode>df_mask): Ditto.
863 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
864 (*avx512pf_scatterpf<mode>df_mask): Ditto.
865 (avx512cd_maskb_vec_dupv8di): Ditto.
866 (avx512cd_maskw_vec_dupv16si): Ditto.
867 (avx512f_vpermi2var<mode>3_maskz): Ditto.
868 (avx512f_vpermi2var<mode>3_mask): Ditto.
869 (avx512f_vpermi2var<mode>3_mask): Ditto.
870 (avx512f_vpermt2var<mode>3_maskz): Ditto.
871 (*avx512f_gathersi<mode>): Ditto.
872 (*avx512f_gathersi<mode>_2): Ditto.
873 (*avx512f_gatherdi<mode>): Ditto.
874 (*avx512f_gatherdi<mode>_2): Ditto.
875 (*avx512f_scattersi<mode>): Ditto.
876 (*avx512f_scatterdi<mode>): Ditto.
877 (avx512f_compress<mode>_mask): Ditto.
878 (avx512f_compressstore<mode>_mask): Ditto.
879 (avx512f_expand<mode>_mask): Ditto.
880 * config/i386/subst.md (mask): Change k to Yk.
881 (mask_scalar_merge): Ditto.
884 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
886 * doc/extend.texi (Vector Extensions): Document ?: in C++.
888 2014-01-31 Richard Biener <rguenther@suse.de>
891 * builtins.c (fold_builtin_memory_op): Make sure to not
892 use a floating-point mode or a boolean or enumeral type for
895 2014-01-30 DJ Delorie <dj@redhat.com>
897 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
898 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
899 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
900 whenever main() has an epilogue.
902 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
904 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
905 unused variable "field".
906 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
907 (vsx_mergeh_<mode>): Likewise.
908 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
909 (altivec_vmrghh): Likewise.
910 (altivec_vmrghw): Likewise.
911 (altivec_vmrglb): Likewise.
912 (altivec_vmrglh): Likewise.
913 (altivec_vmrglw): Likewise.
914 (altivec_vspltb): Add missing uses.
915 (altivec_vsplth): Likewise.
916 (altivec_vspltw): Likewise.
917 (altivec_vspltsf): Likewise.
919 2014-01-30 Jakub Jelinek <jakub@redhat.com>
922 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
923 frame related instructions.
925 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
927 PR rtl-optimization/59959
928 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
929 any reload of register whose subreg is invalid.
931 2014-01-30 Jakub Jelinek <jakub@redhat.com>
933 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
934 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
935 Add missing return type - void.
937 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
939 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
940 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
941 remove element index adjustment for endian (now handled in vsx.md
943 (altivec_expand_vec_perm_const): Use
944 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
945 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
946 (vsx_xxspltw_<mode>): Adjust element index for little endian.
947 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
948 define_expand and a new define_insn *altivec_vspltb_internal;
949 adjust for -maltivec=be on a little endian target.
950 (altivec_vspltb_direct): New.
951 (altivec_vsplth): Divide into a define_expand and a new
952 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
953 little endian target.
954 (altivec_vsplth_direct): New.
955 (altivec_vspltw): Divide into a define_expand and a new
956 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
957 little endian target.
958 (altivec_vspltw_direct): New.
959 (altivec_vspltsf): Divide into a define_expand and a new
960 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
961 a little endian target.
963 2014-01-30 Richard Biener <rguenther@suse.de>
965 PR tree-optimization/59993
966 * tree-ssa-forwprop.c (associate_pointerplus): Check we
967 can propagate form the earlier stmt and avoid the transform
968 when the intermediate result is needed.
970 2014-01-30 Alangi Derick <alangiderick@gmail.com>
972 * README.Portability: Fix typo.
974 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
976 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
977 comparison_operator with ordered_comparison_operator.
979 2014-01-30 Nick Clifton <nickc@redhat.com>
981 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
982 Rename to mn10300_store_multiple_regs.
983 * config/mn10300/mn10300.c: Likewise.
984 * config/mn10300/mn10300.md (store_movm): Fix typo: call
986 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
987 Call mn10300_store_multiple_regs.
989 2014-01-30 Nick Clifton <nickc@redhat.com>
990 DJ Delorie <dj@redhat.com>
992 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
993 %fp 2 to keep registers after it properly word-aligned.
994 (rl78_alloc_physical_registers_umul): Handle the case where both
995 input operands are the same.
997 2014-01-30 Richard Biener <rguenther@suse.de>
999 PR tree-optimization/59903
1000 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1003 2014-01-30 Jason Merrill <jason@redhat.com>
1006 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1009 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1011 2014-01-30 Richard Biener <rguenther@suse.de>
1013 PR tree-optimization/59951
1014 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1016 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1019 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1020 SFmode to DFmode case.
1022 2014-01-29 DJ Delorie <dj@redhat.com>
1024 * config/msp430/msp430.opt (-minrt): New.
1025 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1027 (ENDFILE_SPEC): Likewise.
1029 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1031 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1032 (estimate_function_body_sizes): Use it.
1034 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1037 * dwarf2out.c (is_cxx_auto): New.
1038 (is_base_type): Use it.
1039 (gen_type_die_with_usage): Likewise.
1041 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1043 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1044 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1045 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1046 -maltivec=be with LE targets.
1047 (vsx_mergeh_<mode>): Likewise.
1048 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1049 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1050 (altivec_vmrghb): Replace with define_expand and new
1051 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1052 (altivec_vmrghb_direct): New define_insn.
1053 (altivec_vmrghh): Replace with define_expand and new
1054 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1055 (altivec_vmrghh_direct): New define_insn.
1056 (altivec_vmrghw): Replace with define_expand and new
1057 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1058 (altivec_vmrghw_direct): New define_insn.
1059 (*altivec_vmrghsf): Adjust for endianness.
1060 (altivec_vmrglb): Replace with define_expand and new
1061 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1062 (altivec_vmrglb_direct): New define_insn.
1063 (altivec_vmrglh): Replace with define_expand and new
1064 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1065 (altivec_vmrglh_direct): New define_insn.
1066 (altivec_vmrglw): Replace with define_expand and new
1067 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1068 (altivec_vmrglw_direct): New define_insn.
1069 (*altivec_vmrglsf): Adjust for endianness.
1070 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1071 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1072 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1073 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1074 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1075 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1076 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1077 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1079 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1081 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1082 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1085 2014-01-29 Richard Biener <rguenther@suse.de>
1087 PR tree-optimization/58742
1088 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1089 associate_pointerplus_align.
1090 (associate_pointerplus_diff): New function.
1091 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1092 and associate_pointerplus_diff.
1094 2014-01-29 Richard Biener <rguenther@suse.de>
1096 * lto-streamer.h (LTO_major_version): Bump to 3.
1097 (LTO_minor_version): Reset to 0.
1099 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1101 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1102 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1103 (arm_file_start): Generate correct asm header for armv7ve.
1104 * config/arm/bpabi.h: Add multilib support for armv7ve.
1105 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1106 and cortex-a15 to armv7ve.
1107 * config/arm/t-aprofile: Add multilib support for armv7ve.
1108 * doc/invoke.texi: Document -march=armv7ve.
1110 2014-01-29 Richard Biener <rguenther@suse.de>
1112 PR tree-optimization/58742
1113 * tree-ssa-forwprop.c (associate_plusminus): Return true
1114 if we changed sth, defer EH cleanup to ...
1115 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1116 (simplify_mult): New function.
1118 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1121 PR tree-optimization/59920
1122 * tree.c (build_common_builtin_nodes): Remove
1123 __builtin_setjmp_dispatcher initialization.
1124 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1125 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1126 instead of gsi_after_labels + manually skipping debug stmts.
1127 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1128 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1129 * tree-inline.c (copy_edges_for_bb): Remove
1130 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1131 argument. Ignore computed_goto_p stmts. Don't call
1132 make_abnormal_goto_edges. If a call might need abnormal edges
1133 for non-local gotos, see if it already has an edge to
1134 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1135 with true argument, don't do anything then, otherwise add
1136 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1137 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1139 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1140 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1141 (lower_stmt): Don't set data->calls_builtin_setjmp.
1142 (lower_builtin_setjmp): Adjust comment.
1143 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1144 * tree-cfg.c (found_computed_goto): Remove.
1145 (factor_computed_gotos): Remove.
1146 (make_goto_expr_edges): Return bool, true for computed gotos.
1147 Don't call make_abnormal_goto_edges.
1148 (build_gimple_cfg): Don't set found_computed_goto, don't call
1149 factor_computed_gotos.
1150 (computed_goto_p): No longer static.
1151 (make_blocks): Don't set found_computed_goto.
1152 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1153 (make_edges): If make_goto_expr_edges returns true, push bb
1154 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1155 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1156 vector. Record mapping between bbs and OpenMP regions if there
1157 are any, adjust make_gimple_omp_edges caller. Call
1158 handle_abnormal_edges.
1159 (make_abnormal_goto_edges): Remove.
1160 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1161 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1162 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1163 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1164 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1165 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1166 filling *region also set *region_idx to (*region)->entry->index.
1169 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1170 For REGs set ORIGINAL_REGNO.
1172 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1174 * doc/md.texi: Mention that a target shouldn't implement
1175 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1178 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1180 PR tree-optimization/59594
1181 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1182 a copy of the datarefs vector rather than the vector itself.
1184 2014-01-28 Jason Merrill <jason@redhat.com>
1187 * dwarf2out.c (auto_die): New static.
1188 (gen_type_die_with_usage): Handle C++1y 'auto'.
1189 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1192 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1195 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1196 (SPEC_X32): Likewise.
1197 (SPEC_64): Likewise.
1198 * config/i386/i386.c (ix86_option_override_internal): Turn off
1199 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1201 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1202 * config/i386/i386.h (TARGET_16BIT): New macro.
1203 (TARGET_16BIT_P): Likewise.
1204 * config/i386/i386.opt: Add m16.
1205 * doc/invoke.texi: Document -m16.
1207 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1209 PR preprocessor/59935
1210 * input.c (location_get_source_line): Bail out on when line number
1211 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1213 2014-01-28 Richard Biener <rguenther@suse.de>
1215 PR tree-optimization/58742
1216 * tree-ssa-forwprop.c (associate_plusminus): Handle
1217 pointer subtraction of the form (T)(P + A) - (T)P.
1219 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1221 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1224 2014-01-28 Richard Biener <rguenther@suse.de>
1227 2014-01-28 Richard Biener <rguenther@suse.de>
1229 PR rtl-optimization/45364
1230 PR rtl-optimization/59890
1231 * var-tracking.c (local_get_addr_clear_given_value): Handle
1232 already cleared slot.
1233 (val_reset): Handle not allocated local_get_addr_cache.
1234 (vt_find_locations): Use post-order on the inverted CFG.
1236 2014-01-28 Richard Biener <rguenther@suse.de>
1238 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1240 2014-01-28 Richard Biener <rguenther@suse.de>
1242 PR rtl-optimization/45364
1243 PR rtl-optimization/59890
1244 * var-tracking.c (local_get_addr_clear_given_value): Handle
1245 already cleared slot.
1246 (val_reset): Handle not allocated local_get_addr_cache.
1247 (vt_find_locations): Use post-order on the inverted CFG.
1249 2014-01-28 Alan Modra <amodra@gmail.com>
1251 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1252 * configure.ac <recursive call for build != host>: Define
1253 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1254 and LD_FOR_BUILD too.
1255 * configure: Regenerate.
1257 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1259 * config/i386/i386.c (get_builtin_code_for_version): Separate
1260 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1261 Broadwell from Haswell.
1263 2014-01-27 Steve Ellcey <sellcey@mips.com>
1265 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1266 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1267 * config/mips/mips.c (mips_option_override): Change setting
1269 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1270 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1271 Change from Mask to Var.
1273 2014-01-27 Jeff Law <law@redhat.com>
1275 * ipa-inline.c (inline_small_functions): Fix typo.
1277 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1279 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1280 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1281 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1282 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1283 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1284 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1285 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1286 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1287 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1288 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1289 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1290 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1291 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1292 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1293 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1294 (_mm512_storeu_epi64): Ditto.
1295 (_mm512_cmpge_epi32_mask): Ditto.
1296 (_mm512_cmpge_epu32_mask): Ditto.
1297 (_mm512_cmpge_epi64_mask): Ditto.
1298 (_mm512_cmpge_epu64_mask): Ditto.
1299 (_mm512_cmple_epi32_mask): Ditto.
1300 (_mm512_cmple_epu32_mask): Ditto.
1301 (_mm512_cmple_epi64_mask): Ditto.
1302 (_mm512_cmple_epu64_mask): Ditto.
1303 (_mm512_cmplt_epi32_mask): Ditto.
1304 (_mm512_cmplt_epu32_mask): Ditto.
1305 (_mm512_cmplt_epi64_mask): Ditto.
1306 (_mm512_cmplt_epu64_mask): Ditto.
1307 (_mm512_cmpneq_epi32_mask): Ditto.
1308 (_mm512_cmpneq_epu32_mask): Ditto.
1309 (_mm512_cmpneq_epi64_mask): Ditto.
1310 (_mm512_cmpneq_epu64_mask): Ditto.
1311 (_mm512_expand_pd): Ditto.
1312 (_mm512_expand_ps): Ditto.
1313 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1314 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1315 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1316 * config/i386/i386.c (ix86_builtins): Add
1317 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1318 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1319 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1320 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1321 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1322 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1323 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1324 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1325 IX86_BUILTIN_PMOVUSQW512_MEM.
1326 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1327 __builtin_ia32_pmovsqd512mem_mask,
1328 __builtin_ia32_pmovqd512mem_mask,
1329 __builtin_ia32_pmovusqw512mem_mask,
1330 __builtin_ia32_pmovsqw512mem_mask,
1331 __builtin_ia32_pmovqw512mem_mask,
1332 __builtin_ia32_pmovusdw512mem_mask,
1333 __builtin_ia32_pmovsdw512mem_mask,
1334 __builtin_ia32_pmovdw512mem_mask,
1335 __builtin_ia32_pmovqb512mem_mask,
1336 __builtin_ia32_pmovusqb512mem_mask,
1337 __builtin_ia32_pmovsqb512mem_mask,
1338 __builtin_ia32_pmovusdb512mem_mask,
1339 __builtin_ia32_pmovsdb512mem_mask,
1340 __builtin_ia32_pmovdb512mem_mask.
1341 (bdesc_args): Add __builtin_ia32_expanddf512,
1342 __builtin_ia32_expandsf512.
1343 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1344 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1345 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1346 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1347 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1348 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1349 (avx512f_<code>v8div16qi2_mask_store): This.
1350 (avx512f_expand<mode>): New.
1352 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1354 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1356 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1357 (_mm512_prefetch_i32scatter_pd): Ditto.
1358 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1359 (_mm512_prefetch_i64scatter_pd): Ditto.
1360 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1361 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1362 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1363 (_mm512_prefetch_i32scatter_ps): Ditto.
1364 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1365 (_mm512_prefetch_i64scatter_ps): Ditto.
1366 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1367 * config/i386/i386-builtin-types.def: Define
1368 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1369 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1370 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1371 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1372 IX86_BUILTIN_SCATTERPFQPD.
1373 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1374 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1375 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1376 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1377 __builtin_ia32_scatterpfqps.
1378 (ix86_expand_builtin): Expand new built-ins.
1379 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1380 fix memory access data type.
1381 (*avx512pf_gatherpf<mode>_mask): Ditto.
1382 (*avx512pf_gatherpf<mode>): Ditto.
1383 (avx512pf_scatterpf<mode>): Ditto.
1384 (*avx512pf_scatterpf<mode>_mask): Ditto.
1385 (*avx512pf_scatterpf<mode>): Ditto.
1386 (GATHER_SCATTER_SF_MEM_MODE): New.
1387 (avx512pf_gatherpf<mode>df): Ditto.
1388 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1389 (*avx512pf_scatterpf<mode>df): Ditto.
1391 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1394 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1395 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1398 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1400 * common/config/arm/arm-common.c
1401 (arm_rewrite_mcpu): Handle multiple names.
1403 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1405 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1407 * gimple-builder.h (create_gimple_tmp): Delete.
1409 2014-01-27 Christian Bruel <christian.bruel@st.com>
1411 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1414 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1416 * config/pa/pa.md (call): Generate indirect long calls to non-local
1417 functions when outputing 32-bit code.
1418 (call_value): Likewise except for special call to buggy powf function.
1420 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1421 portable runtime and PIC indirect calls.
1422 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1423 and PIC call sequences. Use ldo instead of blr to set return register
1424 in PIC call sequence.
1426 2014-01-25 Walter Lee <walt@tilera.com>
1428 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1429 avoid clobbering a live register.
1431 2014-01-25 Walter Lee <walt@tilera.com>
1433 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1434 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1435 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1436 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1438 2014-01-25 Walter Lee <walt@tilera.com>
1440 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1441 arguments on even registers.
1442 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1444 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1445 (BIGGEST_ALIGNMENT): Ditto.
1446 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1448 2014-01-25 Walter Lee <walt@tilera.com>
1450 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1451 insns before bundling.
1452 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1454 2014-01-25 Walter Lee <walt@tilera.com>
1456 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1457 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1458 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1460 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1462 * config/mips/constraints.md (kl): Delete.
1463 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1464 define expands, using...
1465 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1466 instructions for MIPS16.
1467 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1468 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1470 2014-01-25 Walter Lee <walt@tilera.com>
1472 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1476 2014-01-25 Walter Lee <walt@tilera.com>
1478 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1479 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1481 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1483 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1486 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1488 * print-rtl.c (in_call_function_usage): New var.
1489 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1490 EXPR_LIST mode as mode and not as reg note name.
1493 * cfgloopmanip.c (copy_loop_info): If
1494 loop->warned_aggressive_loop_optimizations, make sure
1495 the flag is set in target loop too.
1497 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1499 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1501 * builtins.def: Likewise.
1502 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1503 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1504 * ira.c (ira_setup_eliminable_regset): Likewise.
1505 * omp-low.c (gate_expand_omp): Likewise.
1506 (execute_lower_omp): Likewise.
1507 (diagnose_sb_0): Likewise.
1508 (gate_diagnose_omp_blocks): Likewise.
1509 (simd_clone_clauses_extract): Likewise.
1512 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1514 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1515 correction for little endian...
1516 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1519 2014-01-24 Jeff Law <law@redhat.com>
1521 PR tree-optimization/59919
1522 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1523 for non-returning calls.
1525 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1527 * common/config/aarch64/aarch64-common.c
1528 (aarch64_rewrite_mcpu): Handle multiple names.
1529 * config/aarch64/aarch64.h
1530 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1532 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1534 * input.c (add_file_to_cache_tab): Handle the case where fopen
1537 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1540 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1541 from push operand if code of push isn't PRE_DEC.
1543 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1546 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1547 -mquad-memory-atomic. Update -mquad-memory documentation to say
1548 it is only used for non-atomic loads/stores.
1550 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1551 -mquad-memory or -mquad-memory-atomic switches.
1553 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1554 -mquad-memory-atomic to ISA 2.07 support.
1556 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1557 to separate support of normal quad word memory operations (ldq, stq)
1558 from the atomic quad word memory operations.
1560 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1561 support to separate non-atomic quad word operations from atomic
1562 quad word operations. Disable non-atomic quad word operations in
1563 little endian mode so that we don't have to swap words after the
1564 load and before the store.
1565 (quad_load_store_p): Add comment about atomic quad word support.
1566 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1567 options printed with -mdebug=reg.
1569 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1570 -mquad-memory-atomic as the test for whether we have quad word
1571 atomic instructions.
1572 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1573 or -mp8-vector are used, allow byte/half-word atomic operations.
1575 * config/rs6000/sync.md (load_lockedti): Insure that the address
1576 is a proper indexed or indirect address for the lqarx instruction.
1577 On little endian systems, swap the hi/lo registers after the lqarx
1579 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1580 insure the address is valid for the lqarx instruction.
1581 (store_conditionalti): Insure that the address is a proper indexed
1582 or indirect address for the stqcrx. instruction. On little endian
1583 systems, swap the hi/lo registers before doing the stqcrx.
1585 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1586 insure the address is valid for the stqcrx. instruction.
1588 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1589 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1590 type of quad memory support is available.
1592 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1595 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1596 there is a danger of looping.
1598 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1600 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1601 force flag_ira_loop_pressure if set via command line.
1603 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1605 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1606 (ashr_simd): New builtin handling DI mode.
1607 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1608 (aarch64_sshr_simddi): New match pattern.
1609 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1610 (vshrd_n_s64): Likewise.
1611 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1613 2014-01-23 Nick Clifton <nickc@redhat.com>
1615 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1616 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1617 favour of mcu specific scripts.
1618 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1621 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1622 Alex Velenko <Alex.Velenko@arm.com>
1624 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1625 (vaddv_s16): Likewise.
1626 (vaddv_s32): Likewise.
1627 (vaddv_u8): Likewise.
1628 (vaddv_u16): Likewise.
1629 (vaddv_u32): Likewise.
1630 (vaddvq_s8): Likewise.
1631 (vaddvq_s16): Likewise.
1632 (vaddvq_s32): Likewise.
1633 (vaddvq_s64): Likewise.
1634 (vaddvq_u8): Likewise.
1635 (vaddvq_u16): Likewise.
1636 (vaddvq_u32): Likewise.
1637 (vaddvq_u64): Likewise.
1638 (vaddv_f32): Likewise.
1639 (vaddvq_f32): Likewise.
1640 (vaddvq_f64): Likewise.
1641 (vmaxv_f32): Likewise.
1642 (vmaxv_s8): Likewise.
1643 (vmaxv_s16): Likewise.
1644 (vmaxv_s32): Likewise.
1645 (vmaxv_u8): Likewise.
1646 (vmaxv_u16): Likewise.
1647 (vmaxv_u32): Likewise.
1648 (vmaxvq_f32): Likewise.
1649 (vmaxvq_f64): Likewise.
1650 (vmaxvq_s8): Likewise.
1651 (vmaxvq_s16): Likewise.
1652 (vmaxvq_s32): Likewise.
1653 (vmaxvq_u8): Likewise.
1654 (vmaxvq_u16): Likewise.
1655 (vmaxvq_u32): Likewise.
1656 (vmaxnmv_f32): Likewise.
1657 (vmaxnmvq_f32): Likewise.
1658 (vmaxnmvq_f64): Likewise.
1659 (vminv_f32): Likewise.
1660 (vminv_s8): Likewise.
1661 (vminv_s16): Likewise.
1662 (vminv_s32): Likewise.
1663 (vminv_u8): Likewise.
1664 (vminv_u16): Likewise.
1665 (vminv_u32): Likewise.
1666 (vminvq_f32): Likewise.
1667 (vminvq_f64): Likewise.
1668 (vminvq_s8): Likewise.
1669 (vminvq_s16): Likewise.
1670 (vminvq_s32): Likewise.
1671 (vminvq_u8): Likewise.
1672 (vminvq_u16): Likewise.
1673 (vminvq_u32): Likewise.
1674 (vminnmv_f32): Likewise.
1675 (vminnmvq_f32): Likewise.
1676 (vminnmvq_f64): Likewise.
1678 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1680 * config/aarch64/aarch64-simd.md
1681 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1682 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1683 (*aarch64_mul3_elt<mode>): Likewise.
1684 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1685 (*aarch64_mul3_elt_to_64v2df): Likewise.
1686 (*aarch64_mla_elt<mode>): Likewise.
1687 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1688 (*aarch64_mls_elt<mode>): Likewise.
1689 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1690 (*aarch64_fma4_elt<mode>): Likewise.
1691 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1692 (*aarch64_fma4_elt_to_64v2df): Likewise.
1693 (*aarch64_fnma4_elt<mode>): Likewise.
1694 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1695 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1696 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1697 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1698 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1699 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1700 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1701 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1702 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1704 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1706 * config/aarch64/aarch64-simd.md
1707 (aarch64_be_checked_get_lane<mode>): New define_expand.
1708 * config/aarch64/aarch64-simd-builtins.def
1709 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1710 New builtin definition.
1711 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1712 Use new safe be builtin.
1714 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1716 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1718 (aarch64_be_st1<mode>): Likewise.
1719 (aarch_ld1<VALL:mode>): Define_expand modified.
1720 (aarch_st1<VALL:mode>): Likewise.
1721 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1722 (UNSPEC_ST1): Likewise.
1724 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1726 * config/microblaze/microblaze.md: Add trap insn and attribute
1728 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1730 PR preprocessor/58580
1731 * input.h (location_get_source_line): Take an additional line_size
1733 (void diagnostics_file_cache_fini): Declare new function.
1734 * input.c (struct fcache): New type.
1735 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1736 New static constants.
1737 (diagnostic_file_cache_init, total_lines_num)
1738 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1739 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1740 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1741 (get_next_line, read_next_line, goto_next_line, read_line_num):
1742 New static function definitions.
1743 (diagnostic_file_cache_fini): New function.
1744 (location_get_source_line): Take an additional output line_len
1745 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1747 * diagnostic.c (diagnostic_finish): Call
1748 diagnostic_file_cache_fini.
1749 (adjust_line): Take an additional input parameter for the length
1750 of the line, rather than calculating it with strlen.
1751 (diagnostic_show_locus): Adjust the use of
1752 location_get_source_line and adjust_line with respect to their new
1753 signature. While displaying a line now, do not stop at the first
1754 null byte. Rather, display the zero byte as a space and keep
1755 going until we reach the size of the line.
1756 * Makefile.in: Add vec.o to OBJS-libcommon
1758 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1759 Ilya Tocar <ilya.tocar@intel.com>
1761 * config/i386/avx512fintrin.h (_mm512_kmov): New.
1762 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
1763 (__builtin_ia32_kmov16): Ditto.
1764 * config/i386/i386.md (UNSPEC_KMOV): New.
1767 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
1769 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
1770 (_mm512_storeu_si512): Ditto.
1772 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
1775 * rtl.h (get_referenced_operands): Declare.
1776 * recog.c (get_referenced_operands): New function.
1777 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
1778 operands have been referenced when recording LO_SUM references.
1780 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
1782 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
1784 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1786 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
1787 Enable for generic and recent AMD targets.
1789 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
1791 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
1792 ARG_SIZE note when adjustment was eliminated.
1794 2014-01-22 Jeff Law <law@redhat.com>
1796 PR tree-optimization/59597
1797 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
1798 in file. Accept new argument REGISTERING and use it to modify
1799 dump output appropriately.
1800 (register_jump_thread): Corresponding changes.
1801 (mark_threaded_blocks): Reinstate code to cancel unprofitable
1802 thread paths involving joiner blocks. Add code to dump cancelled
1803 jump threading paths.
1805 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
1807 PR rtl-optimization/59477
1808 * lra-constraints.c (inherit_in_ebb): Process call for living hard
1809 regs. Update reloads_num and potential_reload_hard_regs for all insns.
1811 2014-01-22 Tom Tromey <tromey@redhat.com>
1813 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
1815 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
1817 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1819 PR rtl-optimization/59896
1820 * lra-constraints.c (process_alt_operands): Check unused note for
1821 matched operands of insn with no output reloads.
1823 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
1825 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
1826 (mips_move_from_gpr_cost): Likewise.
1828 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
1830 PR rtl-optimization/59858
1831 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
1832 ira_class_hard_regs_num.
1833 (process_alt_operands): Increase reject for dying matched operand.
1835 2014-01-21 Jakub Jelinek <jakub@redhat.com>
1838 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
1839 smaller than size, perform several stores or loads and stores
1840 at dst + count - size to store or copy all of size bytes, rather
1841 than just last modesize bytes.
1843 2014-01-20 DJ Delorie <dj@redhat.com>
1845 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
1846 that CLOBBERs are REGs before propogating their values.
1848 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
1851 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
1852 (cgraph_inline_failed_type): New function.
1853 * cgraph.h (DEFCIFCODE): Add type.
1854 (cgraph_inline_failed_type_t): New enum.
1855 (cgraph_inline_failed_type): New prototype.
1856 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
1857 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
1858 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
1859 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
1860 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
1861 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
1862 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
1863 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
1864 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
1865 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
1866 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
1867 OPTIMIZATION_MISMATCH.
1868 * tree-inline.c (expand_call_inline): Emit errors during
1869 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
1871 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1874 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
1875 mode attribute in insn output.
1877 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
1879 * output.h (output_constant): Delete.
1880 * varasm.c (output_constant): Make private.
1882 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
1884 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
1886 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1889 * tree.h (fold_builtin_strcat): New prototype.
1890 * builtins.c (fold_builtin_strcat): No longer static. Add len
1891 argument, if non-NULL, don't call c_strlen. Optimize
1892 directly into __builtin_memcpy instead of __builtin_strcpy.
1893 (fold_builtin_2): Adjust fold_builtin_strcat caller.
1894 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
1896 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
1898 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1899 for SImode_address_operand operands, having only a REG argument.
1901 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
1903 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
1904 loader name using mbig-endian.
1905 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
1907 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1909 * doc/invoke.texi (-march): Clarify documentation for AArch64.
1913 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
1915 * config/aarch64/aarch64-protos.h
1916 (aarch64_cannot_change_mode_class_ptr): Declare.
1917 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
1918 aarch64_cannot_change_mode_class_ptr): New.
1919 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
1920 backend hook aarch64_cannot_change_mode_class.
1922 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
1924 * common/config/aarch64/aarch64-common.c
1925 (aarch64_handle_option): Don't handle any option order logic here.
1926 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
1927 selected_cpu, warn on architecture version mismatch.
1928 (aarch64_override_options): Fix parsing order for option strings.
1930 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1931 Iain Sandoe <iain@codesourcery.com>
1934 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
1935 warning. Amend comment to reflect current functionality.
1937 2014-01-20 Richard Biener <rguenther@suse.de>
1940 * builtins.c (fold_builtin_strcat): Remove case better handled
1941 by tree-ssa-strlen.c.
1943 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
1945 * config/aarch64/aarch64.opt
1946 (mcpu, march, mtune): Make case-insensitive.
1948 2014-01-20 Jakub Jelinek <jakub@redhat.com>
1951 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
1952 if operands[1] is a REG or ZERO_EXTEND of a REG.
1954 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
1956 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
1958 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
1960 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
1961 long non-pic millicode calls.
1963 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
1965 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
1967 2014-01-19 Kito Cheng <kito@0xlab.org>
1969 * builtins.c (expand_movstr): Check movstr expand done or fail.
1971 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1972 H.J. Lu <hongjiu.lu@intel.com>
1975 * config/i386/i386.md (*lea<mode>): Zero-extend return register
1976 to DImode for zero-extended addresses.
1978 2014-01-19 Jakub Jelinek <jakub@redhat.com>
1980 PR rtl-optimization/57763
1981 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
1982 on the new indirect jump_insn and increment LABEL_NUSES (label).
1984 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
1988 * config.gcc (x86_archs): New variable.
1989 (x86_64_archs): Likewise.
1990 (x86_cpus): Likewise.
1991 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
1992 --with-arch/--with-cpu= options.
1993 Support --with-arch=/--with-cpu={nehalem,westmere,
1994 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
1996 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
1998 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
1999 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2001 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2003 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2005 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2008 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2009 clear cpp_get_options (parse_in)->warn_unused_macros for
2010 ix86_target_macros_internal with cpp_define.
2012 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2014 * jump.c (delete_related_insns): Keep (use (insn))s.
2015 * reorg.c (redundant_insn): Check for barriers too.
2017 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2019 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2021 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2023 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2024 call to $$dyncall when TARGET_LONG_CALLS is true.
2026 2014-01-17 Jeff Law <law@redhat.com>
2028 * ree.c (combine_set_extension): Temporarily disable test for
2029 changing number of hard registers.
2031 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2034 * ipa-inline-analysis.c (inline_free_summary):
2035 Do not free summary of aliases.
2037 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2040 * gimplify.c (gimplify_expr): Use create_tmp_var
2041 instead of create_tmp_var_raw. If cond doesn't have
2042 integral type, don't add the IFN_ANNOTATE builtin at all.
2044 2014-01-17 Martin Jambor <mjambor@suse.cz>
2047 * ipa-cp.c (prev_edge_clone): New variable.
2048 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2049 Also resize prev_edge_clone vector.
2050 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2051 (ipcp_edge_removal_hook): New function.
2052 (ipcp_driver): Register ipcp_edge_removal_hook.
2054 2014-01-17 Andrew Pinski <apinski@cavium.com>
2055 Steve Ellcey <sellcey@mips.com>
2058 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2061 2014-01-17 Jeff Law <law@redhat.com>
2064 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2065 so that pass_ccp runs first.
2067 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2069 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2070 (ix86_adjust_cost): Use !TARGET_XXX.
2071 (do_reorder_for_imul): Likewise.
2072 (swap_top_of_ready_list): Likewise.
2073 (ix86_sched_reorder): Likewise.
2075 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2077 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2078 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2079 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2080 (intel_memset): New. Duplicate slm_memset.
2081 (intel_cost): New. Duplicate slm_cost.
2082 (m_INTEL): New macro.
2083 (processor_target_table): Add "intel".
2084 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2085 with PROCESSOR_INTEL for "intel".
2086 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2087 PROCESSOR_SILVERMONT.
2088 (ix86_issue_rate): Likewise.
2089 (ix86_adjust_cost): Likewise.
2090 (ia32_multipass_dfa_lookahead): Likewise.
2091 (swap_top_of_ready_list): Likewise.
2092 (ix86_sched_reorder): Likewise.
2093 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2094 instead of TARGET_OPT_AGU.
2095 * config/i386/i386.h (TARGET_INTEL): New.
2096 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2097 (processor_type): Add PROCESSOR_INTEL.
2098 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2099 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2101 2014-01-17 Marek Polacek <polacek@redhat.com>
2104 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2107 2014-01-17 Richard Biener <rguenther@suse.de>
2109 PR tree-optimization/46590
2110 * opts.c (default_options_table): Add entries for
2111 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2112 all enabled at -O1 but not for -Og.
2113 * common.opt (fbranch-count-reg): Remove Init(1).
2114 (fmove-loop-invariants): Likewise.
2115 (ftree-pta): Likewise.
2117 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2119 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2120 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2121 decls to at least the GCC 4.8 used alignments.
2124 * tree-nested.c (convert_nonlocal_reference_stmt,
2125 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2126 of GIMPLE_BIND stmts, adjust associated decls.
2128 2014-01-17 Richard Biener <rguenther@suse.de>
2130 PR tree-optimization/46590
2131 * vec.h (vec<>::bseach): New member function implementing
2132 binary search according to C89 bsearch.
2133 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2134 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2135 bitmap pointer again. Make accesses_in_loop a flat array.
2136 (mem_ref_obstack): New global.
2137 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2138 (mark_ref_stored): Likewise.
2139 (ref_indep_loop_p_2): Likewise.
2140 (set_ref_stored_in_loop): New helper function.
2141 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2142 (memref_free): Adjust.
2143 (record_mem_ref_loc): Simplify.
2144 (gather_mem_refs_stmt): Adjust.
2145 (sort_locs_in_loop_postorder_cmp): New function.
2146 (analyze_memory_references): Sort accesses_in_loop after
2147 loop postorder number.
2148 (find_ref_loc_in_loop_cmp): New function.
2149 (for_all_locs_in_loop): Find relevant cluster of locs in
2150 accesses_in_loop and iterate without recursion.
2151 (execute_sm): Avoid uninit warning.
2152 (struct ref_always_accessed): Simplify.
2153 (ref_always_accessed::operator ()): Likewise.
2154 (ref_always_accessed_p): Likewise.
2155 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2156 loop postorder numbers here.
2157 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2160 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2163 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2164 on decls for which assemble_alias has been called.
2166 2014-01-17 Nick Clifton <nickc@redhat.com>
2168 * config/msp430/msp430.opt: (mcpu): New option.
2169 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2170 (msp430_option_override): Parse target_cpu. If the MCU name
2171 matches a generic string, clear target_mcu.
2172 (msp430_attr): Allow numeric interrupt values up to 63.
2173 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2174 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2176 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2178 * config/msp430/msp430.md (popm): Use %J rather than %I.
2179 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2180 (addhi_cy_i): Use immediate_operand for operand 2.
2181 * doc/invoke.texi: Document -mcpu option.
2183 2014-01-17 Richard Biener <rguenther@suse.de>
2185 PR rtl-optimization/38518
2186 * df.h (df_analyze_loop): Declare.
2187 * df-core.c: Include cfgloop.h.
2188 (df_analyze_1): Split out main part of df_analyze.
2189 (df_analyze): Adjust.
2190 (loop_inverted_post_order_compute): New function.
2191 (loop_post_order_compute): Likewise.
2192 (df_analyze_loop): New function avoiding whole-function
2194 * loop-invariant.c (find_defs): Use df_analyze_loop.
2195 (find_invariants): Adjust.
2196 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2198 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2200 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2201 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2203 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2205 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2206 traversal when removing references.
2208 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2211 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2213 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2216 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2217 pushing a reload for an autoinc when we had previously reloaded an
2218 inner part of the address.
2220 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2222 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2224 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2225 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2226 when not giving up or versioning for alias only because of
2228 (vect_analyze_data_ref_dependences): Set to true.
2229 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2231 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2232 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2236 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2239 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2240 operand 0 predicate for gathers, use a new pseudo as subtarget.
2242 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2245 * lra-constraints.c (process_alt_operands): Add printing debug info.
2246 Check absence of input/output reloads for matched operands too.
2248 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2250 PR rtl-optimization/59835
2251 * ira.c (ira_init_register_move_cost): Increase cost for
2254 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2256 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2258 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2261 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2262 non-register objects. Use gen_(high/low)part more consistently.
2265 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2268 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2269 endian support, remove tests for WORDS_BIG_ENDIAN.
2270 (p8_mfvsrd_3_<mode>): Likewise.
2271 (reload_gpr_from_vsx<mode>): Likewise.
2272 (reload_gpr_from_vsxsf): Likewise.
2273 (p8_mfvsrd_4_disf): Likewise.
2275 2014-01-16 Richard Biener <rguenther@suse.de>
2277 PR rtl-optimization/46590
2278 * lcm.c (compute_antinout_edge): Use postorder iteration.
2279 (compute_laterin): Use inverted postorder iteration.
2281 2014-01-16 Nick Clifton <nickc@redhat.com>
2284 * varasm.c (output_constant): Return the number of bytes actually
2286 (output_constructor_array_range): Update the field size with the
2287 number of bytes emitted by output_constant.
2288 (output_constructor_regular_field): Likewise. Also do not
2289 complain if the total number of bytes emitted is now greater
2290 than the expected fieldpos.
2291 * output.h (output_constant): Update prototype and descriptive comment.
2293 2014-01-16 Marek Polacek <polacek@redhat.com>
2296 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2297 it is error_mark_node.
2299 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2301 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2302 VALID_AVX256_REG_OR_OI_MODE.
2304 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2306 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2307 current procedure should be profiled.
2309 2014-01-15 Andrew Pinski <apinski@cavium.com>
2311 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2312 of moving from/to the STACK_REG register class.
2314 2014-01-15 Richard Henderson <rth@redhat.com>
2317 * reginfo.c (global_regs_decl): Globalize.
2318 * rtl.h (global_regs_decl): Declare.
2319 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2320 reserved via global_regs.
2322 2014-01-15 Teresa Johnson <tejohnson@google.com>
2324 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2326 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2328 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2329 and vmulosh rather than call gen_vec_widen_smult_*.
2330 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2331 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2332 (vec_widen_smult_even_v16qi): Likewise.
2333 (vec_widen_umult_even_v8hi): Likewise.
2334 (vec_widen_smult_even_v8hi): Likewise.
2335 (vec_widen_umult_odd_v16qi): Likewise.
2336 (vec_widen_smult_odd_v16qi): Likewise.
2337 (vec_widen_umult_odd_v8hi): Likewise.
2338 (vec_widen_smult_odd_v8hi): Likewise.
2339 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2340 vmuloub rather than call gen_vec_widen_umult_*.
2341 (vec_widen_umult_lo_v16qi): Likewise.
2342 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2343 vmulosb rather than call gen_vec_widen_smult_*.
2344 (vec_widen_smult_lo_v16qi): Likewise.
2345 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2346 rather than call gen_vec_widen_umult_*.
2347 (vec_widen_umult_lo_v8hi): Likewise.
2348 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2349 rather than call gen_vec_widen_smult_*.
2350 (vec_widen_smult_lo_v8hi): Likewise.
2352 2014-01-15 Jeff Law <law@redhat.com>
2354 PR tree-optimization/59747
2355 * ree.c (find_and_remove_re): Properly handle case where a second
2356 eliminated extension requires widening a copy created for elimination
2357 of a prior extension.
2358 (combine_set_extension): Ensure that the number of hard regs needed
2359 for a destination register does not change when we widen it.
2361 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2363 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2364 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2365 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2366 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2367 (avr-*-rtems*): Likewise.
2368 (bfin*-rtems*): Likewise.
2369 (moxie-*-rtems*): Likewise.
2370 (h8300-*-rtems*): Likewise.
2371 (i[34567]86-*-rtems*): Likewise.
2372 (lm32-*-rtems*): Likewise.
2373 (m32r-*-rtems*): Likewise.
2374 (m68k-*-rtems*): Likewise.
2375 (microblaze*-*-rtems*): Likewise.
2376 (mips*-*-rtems*): Likewise.
2377 (powerpc-*-rtems*): Likewise.
2378 (sh-*-rtems*): Likewise.
2379 (sparc-*-rtems*): Likewise.
2380 (sparc64-*-rtems*): Likewise.
2381 (v850-*-rtems*): Likewise.
2382 (m32c-*-rtems*): Likewise.
2384 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2386 PR rtl-optimization/59511
2387 * ira.c (ira_init_register_move_cost): Use memory costs for some
2388 cases of register move cost calculations.
2389 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2390 instead of BB frequency.
2391 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2392 * lra-assigns.c (find_hard_regno_for): Ditto.
2394 2014-01-15 Richard Biener <rguenther@suse.de>
2396 PR tree-optimization/59822
2397 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2398 (vectorizable_load): Use it to hoist defs of uses of invariant
2399 loads out of the loop.
2401 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2402 Kugan Vivekanandarajah <kuganv@linaro.org>
2405 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2408 2014-01-15 Richard Biener <rguenther@suse.de>
2410 PR rtl-optimization/59802
2411 * lcm.c (compute_available): Use inverted postorder to seed
2412 the initial worklist.
2414 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2417 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2418 ADDR_REGS for invalid symrefs in non-PIC code.
2420 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2423 * builtins.c (determine_block_size): Initialize *probable_max_size
2424 even if len_rtx is CONST_INT.
2426 2014-01-14 Andrew Pinski <apinski@cavium.com>
2428 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2429 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2430 (cortexa53_tunings): Likewise.
2431 (aarch64_sched_issue_rate): New function.
2432 (TARGET_SCHED_ISSUE_RATE): Define.
2434 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2436 * ira-costs.c (find_costs_and_classes): Add missed
2437 ira_init_register_move_cost_if_necessary.
2439 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2442 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2444 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2447 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2448 to indicate if type is used for function return value. Warn ABI
2449 change if the vector mode isn't available for function return value.
2450 (ix86_function_arg_advance): Pass false to type_natural_mode.
2451 (ix86_function_arg): Likewise.
2452 (ix86_gimplify_va_arg): Likewise.
2453 (function_arg_32): Don't warn ABI change.
2454 (ix86_function_value): Pass true to type_natural_mode.
2455 (ix86_return_in_memory): Likewise.
2456 (ix86_struct_value_rtx): Removed.
2457 (TARGET_STRUCT_VALUE_RTX): Likewise.
2459 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2461 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2462 converting a conditional jump into a conditional return.
2464 2014-01-14 Richard Biener <rguenther@suse.de>
2466 PR tree-optimization/58921
2467 PR tree-optimization/59006
2468 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2469 hoisting invariant stmts.
2470 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2471 invariant loads on the preheader edge if possible.
2473 2014-01-14 Joey Ye <joey.ye@arm.com>
2475 * doc/plugin.texi (Building GCC plugins): Update to C++.
2477 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2479 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2480 (_mm_rcp28_round_ss): Ditto.
2481 (_mm_rsqrt28_round_sd): Ditto.
2482 (_mm_rsqrt28_round_ss): Ditto.
2483 (_mm_rcp28_sd): Ditto.
2484 (_mm_rcp28_ss): Ditto.
2485 (_mm_rsqrt28_sd): Ditto.
2486 (_mm_rsqrt28_ss): Ditto.
2487 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2488 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2489 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2490 (IX86_BUILTIN_RCP28SD): Ditto.
2491 (IX86_BUILTIN_RCP28SS): Ditto.
2492 (IX86_BUILTIN_RSQRT28SD): Ditto.
2493 (IX86_BUILTIN_RSQRT28SS): Ditto.
2494 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2495 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2496 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2497 (ix86_expand_special_args_builtin): Expand new FTYPE.
2498 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2499 (srcp14<mode>): Make insn unary.
2500 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2501 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2502 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2503 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2504 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2505 Fix rounding: make it SAE only.
2506 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2508 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2510 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2511 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2512 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2513 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2514 (round_saeonly_mask_scalar_operand4): Ditto.
2515 (round_saeonly_mask_scalar_op3): Ditto.
2516 (round_saeonly_mask_scalar_op4): Ditto.
2518 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2520 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2521 Implement -maltivec=be for vec_insert and vec_extract.
2523 2014-01-10 DJ Delorie <dj@redhat.com>
2525 * config/msp430/msp430.md (call_internal): Don't allow memory
2526 references with SP as the base register.
2527 (call_value_internal): Likewise.
2528 * config/msp430/constraints.md (Yc): New. For memory references
2529 that don't use SP as a base register.
2531 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2532 "an integer without a # prefix"
2533 * config/msp430/msp430.md (epilogue_helper): Use it.
2535 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2538 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2539 AVX512F gather builtins.
2540 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2541 on gather decls with INTEGER_TYPE masktype.
2542 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2543 directly into the builtin rather than hoisting it before loop.
2545 PR tree-optimization/59387
2546 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2547 (scev_const_prop): If folded_casts and type has undefined overflow,
2548 use force_gimple_operand instead of force_gimple_operand_gsi and
2549 for each added stmt if it is assign with
2550 arith_code_with_undefined_signed_overflow, call
2551 rewrite_to_defined_overflow.
2552 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2553 gimple-fold.h instead.
2554 (arith_code_with_undefined_signed_overflow,
2555 rewrite_to_defined_overflow): Moved to ...
2556 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2557 rewrite_to_defined_overflow): ... here. No longer static.
2558 Include gimplify-me.h.
2559 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2560 rewrite_to_defined_overflow): New prototypes.
2562 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2564 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2566 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2568 * builtins.c (get_object_alignment_2): Minor tweak.
2569 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2571 2014-01-13 Christian Bruel <christian.bruel@st.com>
2573 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2574 optimized non constant lengths.
2576 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2579 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2580 load as __atomic_load_N if possible.
2582 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2584 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2586 (rs6000_expand_builtin): Adjust call.
2588 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2591 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2592 * config/rs6000/rs6000.c: Include target-globals.h.
2593 (rs6000_set_current_function): Instead of doing target_reinit
2594 unconditionally, use save_target_globals_default_opts and
2595 restore_target_globals.
2597 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2599 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2600 (rs6000_expand_builtin): Handle mffs and mtfsf.
2601 (rs6000_init_builtins): Define mffs and mtfsf.
2602 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2603 (rs6000_mffs): New pattern.
2604 (rs6000_mtfsf): New pattern.
2606 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2608 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2609 Start narrowing with START. Apply candidate-use pair
2610 and check overall cost in narrowing.
2611 (iv_ca_prune): Pass new argument.
2613 2014-01-10 Jeff Law <law@redhat.com>
2616 * ree.c (combine_reaching_defs): Ensure the defining statement
2617 occurs before the extension when optimizing extensions with
2618 different source and destination hard registers.
2620 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2623 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2624 vtables into the type inheritance graph.
2626 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2628 PR rtl-optimization/59754
2629 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2630 modes in the REGNO != REGNO case.
2632 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2634 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2636 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2638 PR tree-optimization/59745
2639 * tree-predcom.c (tree_predictive_commoning_loop): Call
2640 free_affine_expand_cache if giving up because components is NULL.
2642 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2643 GC in payload of target_globals struct instead of allocating them on
2644 the heap and the larger structs separately using GC.
2645 * target-globals.h (struct target_globals): Make regs, hard_regs,
2646 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2647 of GTY((skip)) and change type to void *.
2648 (reset_target_globals): Cast loads from those fields to corresponding
2651 2014-01-10 Steve Ellcey <sellcey@mips.com>
2654 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2655 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2656 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2658 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2661 * aarch64-modes.def (CC_Zmode): New flags mode.
2662 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2663 represents an equality.
2664 (aarch64_get_condition_code): Handle CC_Zmode.
2665 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2667 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2669 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2670 extraction in good case.
2672 2014-01-10 Richard Biener <rguenther@suse.de>
2674 PR tree-optimization/59374
2675 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2676 checking after SLP discovery. Mark stmts not participating
2677 in any SLP instance properly.
2679 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2681 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2682 when handling a SET rtx.
2684 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2686 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2687 (cortex-a57): Likewise.
2688 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2690 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2692 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2693 non-iwmmxt builtins.
2695 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2699 * ipa-devirt.c record_target_from_binfo): Take as argument
2700 stack of binfos and lookup matching one for virtual inheritance.
2701 (possible_polymorphic_call_targets_1): Update.
2703 2014-01-10 Huacai Chen <chenhc@lemote.com>
2705 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2706 kernel strings for Loongson-2E/2F/3A.
2708 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2711 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2712 is_gimple_call before calling gimple_call_internal_p.
2714 2014-01-09 Steve Ellcey <sellcey@mips.com>
2716 * Makefile.in (TREE_FLOW_H): Remove.
2717 (TREE_SSA_H): Add file names from tree-flow.h.
2718 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2719 * tree.h: Remove tree-flow.h reference.
2720 * hash-table.h: Remove tree-flow.h reference.
2721 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2722 reference with tree-ssa-loop.h.
2724 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2726 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2727 default element-order behavior for -maltivec.
2728 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2730 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2731 when targeting big endian, at least for now.
2732 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2734 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2737 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2738 var satisfies use_register_for_decl, just take into account type
2739 alignment, rather than decl alignment.
2741 PR tree-optimization/59622
2742 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2743 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2744 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2745 Don't devirtualize for inplace at all. For targets.length () == 1,
2746 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2748 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2750 * config/i386/i386.md (cpu): Remove the unused btver1.
2752 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2754 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
2756 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2759 * tree-core.h (struct target_globals): New forward declaration.
2760 (struct tree_target_option): Add globals field.
2761 * tree.h (TREE_TARGET_GLOBALS): Define.
2762 (prepare_target_option_nodes_for_pch): New prototype.
2763 * target-globals.h (struct target_globals): Define even if
2765 * tree.c (prepare_target_option_node_for_pch,
2766 prepare_target_option_nodes_for_pch): New functions.
2767 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
2768 * config/i386/i386.c: Include target-globals.h.
2769 (ix86_set_current_function): Instead of doing target_reinit
2770 unconditionally, use save_target_globals_default_opts and
2771 restore_target_globals.
2773 2014-01-09 Richard Biener <rguenther@suse.de>
2775 PR tree-optimization/59715
2776 * tree-cfg.h (split_critical_edges): Declare.
2777 * tree-cfg.c (split_critical_edges): Export.
2778 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
2780 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
2782 * cfgexpand.c (expand_stack_vars): Optionally disable
2783 asan stack protection.
2784 (expand_used_vars): Likewise.
2785 (partition_stack_vars): Likewise.
2786 * asan.c (asan_emit_stack_protection): Optionally disable
2787 after return stack usage.
2788 (instrument_derefs): Optionally disable memory access instrumentation.
2789 (instrument_builtin_call): Likewise.
2790 (instrument_strlen_call): Likewise.
2791 (asan_protect_global): Optionally disable global variables protection.
2792 * doc/invoke.texi: Added doc for new options.
2793 * params.def: Added new options.
2794 * params.h: Likewise.
2796 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2798 PR rtl-optimization/59724
2799 * ifcvt.c (cond_exec_process_if_block): Don't call
2800 flow_find_head_matching_sequence with 0 longest_match.
2801 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
2802 non-active insns if !stop_after.
2803 (try_head_merge_bb): Revert 2014-01-07 changes.
2805 2014-01-08 Jeff Law <law@redhat.com>
2807 * ree.c (get_sub_rtx): New function, extracted from...
2808 (merge_def_and_ext): Here.
2809 (combine_reaching_defs): Use get_sub_rtx.
2811 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
2813 * cgraph.h (varpool_variable_node): Do not choke on null node.
2815 2014-01-08 Catherine Moore <clm@codesourcery.com>
2817 * config/mips/mips.md (simple_return): Attempt to use JRC
2819 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
2821 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2823 PR rtl-optimization/59137
2824 * reorg.c (steal_delay_list_from_target): Call update_block for
2826 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
2828 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2830 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
2831 two duplicate entries.
2833 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
2836 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
2838 * config/mips/mips.c (mips_truncated_op_cost): New function.
2839 (mips_rtx_costs): Adjust test for BADDU.
2840 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
2842 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
2844 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
2845 (*baddu_si): ...this new pattern.
2847 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2850 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
2852 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
2855 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
2857 (expand_expr, expand_normal): Adjust.
2858 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
2859 inner_reference_p. Use inner_reference_p to expand inner references.
2860 (store_expr): Adjust.
2861 * cfgexpand.c (expand_call_stmt): Adjust.
2863 2014-01-08 Rong Xu <xur@google.com>
2865 * gcov-io.c (gcov_var): Move from gcov-io.h.
2866 (gcov_position): Ditto.
2867 (gcov_is_error): Ditto.
2868 (gcov_rewrite): Ditto.
2869 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
2870 only part to libgcc/libgcov.h.
2872 2014-01-08 Marek Polacek <polacek@redhat.com>
2875 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
2877 2014-01-08 Marek Polacek <polacek@redhat.com>
2880 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
2882 2014-01-08 Jakub Jelinek <jakub@redhat.com>
2884 PR rtl-optimization/59649
2885 * stor-layout.c (get_mode_bounds): For BImode return
2886 0 and STORE_FLAG_VALUE.
2888 2014-01-08 Richard Biener <rguenther@suse.de>
2891 * gimple.h (is_gimple_builtin_call): Remove.
2892 (gimple_builtin_call_types_compatible_p): New.
2893 (gimple_call_builtin_p): New overload.
2894 * gimple.c (is_gimple_builtin_call): Remove.
2895 (validate_call): Rename to ...
2896 (gimple_builtin_call_types_compatible_p): ... this and export. Also
2898 (validate_type): New static function.
2899 (gimple_call_builtin_p): New overload and adjust.
2900 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
2901 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
2902 (gimple_fold_stmt_to_constant_1): Likewise.
2903 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
2905 2014-01-08 Richard Biener <rguenther@suse.de>
2908 * gimplify.c (gimplify_expr): Gimplify register-register type
2909 VIEW_CONVERT_EXPRs to separate stmts.
2911 2014-01-07 Jeff Law <law@redhat.com>
2914 * ree.c (combine_set_extension): Handle case where source
2915 and destination registers in an extension insn are different.
2916 (combine_reaching_defs): Allow source and destination registers
2917 in extension to be different under limited circumstances.
2918 (add_removable_extension): Remove restriction that the
2919 source and destination registers in the extension are the same.
2920 (find_and_remove_re): Emit a copy from the extension's
2921 destination to its source after the defining insn if
2922 the source and destination registers are different.
2925 * ifcvt.c (merge_if_block): If we are merging a block with more than
2926 one successor with a block with no successors, remove any BARRIER
2927 after the second block.
2929 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
2931 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
2933 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
2936 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
2937 for 14-bit register offsets when INT14_OK_STRICT is false.
2939 2014-01-07 Roland Stigge <stigge@antcom.de>
2940 Michael Meissner <meissner@linux.vnet.ibm.com>
2943 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
2944 Only check TFmode for SPE constants. Don't check TImode or TDmode.
2946 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
2948 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
2951 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
2953 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
2954 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
2955 rtx is const0_rtx or not.
2957 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
2960 * target-globals.c (save_target_globals): Remove this_fn_optab
2962 * toplev.c: Include optabs.h.
2963 (target_reinit): Temporarily restore the global options if another
2964 set of options are in force.
2966 2014-01-07 Jakub Jelinek <jakub@redhat.com>
2968 PR rtl-optimization/58668
2969 * cfgcleanup.c (flow_find_cross_jump): Don't count
2970 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
2971 to determine what is counted.
2972 (flow_find_head_matching_sequence): Use active_insn_p to determine
2974 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
2976 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
2977 determine what is counted.
2979 PR tree-optimization/59643
2980 * tree-predcom.c (split_data_refs_to_components): If one dr is
2981 read and one write, determine_offset fails and the write isn't
2982 in the bad component, just put the read into the bad component.
2984 2014-01-07 Mike Stump <mikestump@comcast.net>
2985 Jakub Jelinek <jakub@redhat.com>
2988 * tree-core.h (struct tree_optimization_option): Change optabs
2989 type from unsigned char * to void *.
2990 * optabs.c (init_tree_optimization_optabs): Adjust
2991 TREE_OPTIMIZATION_OPTABS initialization.
2993 2014-01-06 Jakub Jelinek <jakub@redhat.com>
2996 * config/i386/i386.h (struct machine_function): Add
2997 no_drap_save_restore field.
2998 * config/i386/i386.c (ix86_save_reg): Use
2999 !cfun->machine->no_drap_save_restore instead of
3000 crtl->stack_realign_needed.
3001 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3002 this function clears frame_pointer_needed. Set
3003 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3004 and DRAP reg is needed.
3006 2014-01-06 Marek Polacek <polacek@redhat.com>
3009 * doc/implement-c.texi: Mention that other integer types are
3010 permitted as bit-field types in strictly conforming mode.
3012 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3014 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3017 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3019 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3021 2014-01-06 Martin Jambor <mjambor@suse.cz>
3024 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3026 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3028 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3032 * var-tracking.c (add_stores): Preserve the value of the source even if
3033 we don't record the store.
3035 2014-01-06 Terry Guo <terry.guo@arm.com>
3037 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3039 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3042 * config/darwin.c (darwin_function_section): Adjust return values to
3043 correspond to optimisation changes made in r206070.
3045 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3047 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3048 from prefetch_block tune setting.
3049 (nocona_cost): Correct size of prefetch block to 64.
3051 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3053 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3054 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3055 used to save the static chain register in the computation of the offset
3056 from which the FP registers need to be restored.
3058 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3060 PR tree-optimization/59519
3061 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3062 ICE if get_current_def (current_new_name) is already non-NULL, as long
3063 as it is a phi result of some other phi in *new_exit_bb that has
3066 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3067 or vmovdqu* for misaligned_operand.
3068 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3069 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3070 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3071 aligned_mem for AVX512F masked aligned load and store builtins and for
3074 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3076 PR tree-optimization/59651
3077 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3078 Address range for negative step should be added by TYPE_SIZE_UNIT.
3080 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3082 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3083 overlapping registers also for registers other than the stack pointer.
3085 2014-01-03 Marek Polacek <polacek@redhat.com>
3088 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3091 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3094 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3097 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3098 (push splitter): Use <P:MODE_SIZE> instead of
3099 GET_MODE_SIZE (<P:MODE>mode).
3100 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3101 (mov -1, reg peephole2): Likewise.
3102 * config/i386/sse.md (*mov<mode>_internal,
3103 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3104 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3105 *<code><mode>3, *andnot<mode>3<mask_name>,
3106 <mask_codefor><code><mode>3<mask_name>): Likewise.
3107 * config/i386/subst.md (mask_mode512bit_condition,
3108 sd_mask_mode512bit_condition): Likewise.
3110 2014-01-02 Xinliang David Li <davidxl@google.com>
3112 PR tree-optimization/59303
3113 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3114 (dump_predicates): Better output format.
3115 (pred_equal_p): New function.
3116 (is_neq_relop_p): Ditto.
3117 (is_neq_zero_form_p): Ditto.
3118 (pred_expr_equal_p): Ditto.
3119 (pred_neg_p): Ditto.
3120 (simplify_pred): Ditto.
3121 (simplify_preds_2): Ditto.
3122 (simplify_preds_3): Ditto.
3123 (simplify_preds_4): Ditto.
3124 (simplify_preds): Ditto.
3126 (push_to_worklist): Ditto.
3127 (get_pred_info_from_cmp): Ditto.
3128 (is_degenerated_phi): Ditto.
3129 (normalize_one_pred_1): Ditto.
3130 (normalize_one_pred): Ditto.
3131 (normalize_one_pred_chain): Ditto.
3132 (normalize_preds): Ditto.
3133 (normalize_cond_1): Remove function.
3134 (normalize_cond): Ditto.
3135 (is_gcond_subset_of): Ditto.
3136 (is_subset_of_any): Ditto.
3137 (is_or_set_subset_of): Ditto.
3138 (is_and_set_subset_of): Ditto.
3139 (is_norm_cond_subset_of): Ditto.
3140 (pred_chain_length_cmp): Ditto.
3141 (convert_control_dep_chain_into_preds): Type change.
3142 (find_predicates): Ditto.
3143 (find_def_preds): Ditto.
3144 (destroy_predicates_vecs): Ditto.
3145 (find_matching_predicates_in_rest_chains): Ditto.
3146 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3147 (is_pred_expr_subset): Ditto.
3148 (is_pred_chain_subset_of): Ditto.
3149 (is_included_in): Ditto.
3150 (is_superset_of): Ditto.
3152 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3154 Update copyright years.
3156 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3158 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3159 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3160 config/arc/arc.md, config/arc/arc.opt,
3161 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3162 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3163 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3164 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3165 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3166 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3167 vtable-verify.c, vtable-verify.h: Use the standard form for the
3170 2014-01-02 Tobias Burnus <burnus@net-b.de>
3172 * gcc.c (process_command): Update copyright notice dates.
3173 * gcov-dump.c: Ditto.
3175 * doc/cpp.texi: Bump @copying's copyright year.
3176 * doc/cppinternals.texi: Ditto.
3177 * doc/gcc.texi: Ditto.
3178 * doc/gccint.texi: Ditto.
3179 * doc/gcov.texi: Ditto.
3180 * doc/install.texi: Ditto.
3181 * doc/invoke.texi: Ditto.
3183 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3185 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3187 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3189 * config/i386/sse.md (*mov<mode>_internal): Guard
3190 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3192 PR rtl-optimization/59647
3193 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3194 new_rtx into UNSIGNED_FLOAT rtxes.
3196 Copyright (C) 2014 Free Software Foundation, Inc.
3198 Copying and distribution of this file, with or without modification,
3199 are permitted in any medium without royalty provided the copyright
3200 notice and this notice are preserved.