1 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
3 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
4 operands[2], not operands[3].
6 2014-02-13 Richard Biener <rguenther@suse.de>
9 * doc/install.texi (ISL): Update recommended version to 0.12.2,
10 mention the possibility of an in-tree build.
11 (CLooG): Update recommended version to 0.18.1, mention the
12 possibility of an in-tree build and clarify that the ISL
13 bundled with CLooG does not work.
15 2014-02-13 Jakub Jelinek <jakub@redhat.com>
18 * expr.c (compress_float_constant): If x is a hard register,
19 extend into a pseudo and then move to x.
21 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
23 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
24 caused by bad second argument to warning_at() with -mhotpatch and
25 nested functions (e.g. with gfortran).
27 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
29 * opts.c (option_name): Remove "enabled by default" rider.
31 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
33 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
35 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
36 Uros Bizjak <ubizjak@gmail.com>
39 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
40 * configure: Regenerated.
42 2014-02-12 Richard Biener <rguenther@suse.de>
44 * vec.c (vec_prefix::calculate_allocation): Move as
45 inline variant to vec.h.
46 (vec_prefix::calculate_allocation_1): New out-of-line version.
47 * vec.h (vec_prefix::calculate_allocation_1): Declare.
48 (vec_prefix::m_has_auto_buf): Rename to ...
49 (vec_prefix::m_using_auto_storage): ... this.
50 (vec_prefix::calculate_allocation): Inline the easy cases
51 and dispatch to calculate_allocation_1 which doesn't need the
53 (va_heap::reserve): Use gcc_checking_assert.
54 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
56 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
58 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
59 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
60 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
62 2014-02-12 Richard Biener <rguenther@suse.de>
64 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
65 when we found a dependence.
67 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
69 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
71 (maybe_fold_stmt): ... into this new function.
72 * omp-low.c (lower_omp): Update comment.
74 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
77 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
80 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
82 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
83 identifiers in comments.
84 (cortexa53_extra_costs): Likewise.
85 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
86 (cortexa7_extra_costs): Likewise.
87 (cortexa12_extra_costs): Likewise.
88 (cortexa15_extra_costs): Likewise.
89 (v7m_extra_costs): Likewise.
91 2014-02-12 Richard Biener <rguenther@suse.de>
94 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
95 of posix_memalign being successful.
96 (lower_stmt): Restrict lowering of posix_memalign to when
97 -ftree-bit-ccp is enabled.
99 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
101 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
103 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
105 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
107 PR rtl-optimization/60116
108 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
109 other_insn once the combination has been validated.
111 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
114 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
116 * ipa-devirt.c: Include demangle.h
117 (odr_violation_reported): New static variable.
118 (add_type_duplicate): Update odr_violations.
119 (maybe_record_node): Add completep parameter; update it.
120 (record_target_from_binfo): Add COMPLETEP parameter;
122 (possible_polymorphic_call_targets_1): Likewise.
123 (struct polymorphic_call_target_d): Add nonconstruction_targets;
124 rename FINAL to COMPLETE.
125 (record_targets_from_bases): Sanity check we found the binfo;
126 fix COMPLETEP updating.
127 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
128 parameter, fix computing of COMPLETEP.
129 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
130 at LTO time do demangling.
131 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
132 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
134 (gimple_get_virt_method_for_binfo): Likewise.
135 * gimple-fold.h (gimple_get_virt_method_for_binfo,
136 gimple_get_virt_method_for_vtable): Update prototypes.
138 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
141 * genautomata.c (add_presence_absence): Fix typo with
142 {final_}presence_list.
144 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
147 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
148 for VSX/Altivec vectors that land in GPR registers.
150 2014-02-11 Richard Henderson <rth@redhat.com>
151 Jakub Jelinek <jakub@redhat.com>
154 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
155 around drhs if type conversion to lacc->type is not useless.
157 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
159 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
161 (cortex-a57.cortex-a53): Likewise.
162 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
164 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
166 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
169 2014-02-11 Renlin Li <Renlin.Li@arm.com>
171 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
172 add_options_for_arm_vfp3.
174 2014-02-11 Jeff Law <law@redhat.com>
177 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
178 object with an undesirable mode.
180 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
183 * config/i386/sol2-9.h: New file.
184 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
185 *-*-solaris2.9*): Use it.
187 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
189 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
190 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
192 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
194 * config/microblaze/microblaze.c: Extend mcpu version format
196 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
198 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
200 2014-02-10 Richard Henderson <rth@redhat.com>
203 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
204 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
205 ms-abi vs -mno-accumulate-outgoing-args.
206 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
207 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
210 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
213 * cfgexpand.c (expand_asm_operands): Attach source location to
214 ASM_INPUT rtx objects.
215 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
217 2014-02-10 Nick Clifton <nickc@redhat.com>
219 * config/mn10300/mn10300.c (popcount): New function.
220 (mn10300_expand_prologue): Include saved registers in stack usage
223 2014-02-10 Jeff Law <law@redhat.com>
226 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
227 when changing the SET_DEST of a prior insn to avoid an input reload.
229 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
231 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
232 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
233 -mcall-openbsd, or -mcall-linux.
234 (CC1_ENDIAN_BIG_SPEC): Remove.
235 (CC1_ENDIAN_LITTLE_SPEC): Remove.
236 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
237 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
238 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
239 and %cc1_endian_default.
240 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
242 2014-02-10 Richard Biener <rguenther@suse.de>
244 PR tree-optimization/60115
245 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
246 MEM_REF handling. Properly verify that the accesses are not
247 out of the objects bound.
249 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
251 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
254 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
256 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
257 proper constants and fix formatting.
258 (possible_polymorphic_call_targets): Fix formatting.
260 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
261 Ilya Tocar <ilya.tocar@intel.com>
263 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
264 (_mm512_loadu_epi32): Renamed into...
265 (_mm512_loadu_si512): This.
266 (_mm512_storeu_epi32): Renamed into...
267 (_mm512_storeu_si512): This.
268 (_mm512_maskz_ceil_ps): Removed.
269 (_mm512_maskz_ceil_pd): Ditto.
270 (_mm512_maskz_floor_ps): Ditto.
271 (_mm512_maskz_floor_pd): Ditto.
272 (_mm512_floor_round_ps): Ditto.
273 (_mm512_floor_round_pd): Ditto.
274 (_mm512_ceil_round_ps): Ditto.
275 (_mm512_ceil_round_pd): Ditto.
276 (_mm512_mask_floor_round_ps): Ditto.
277 (_mm512_mask_floor_round_pd): Ditto.
278 (_mm512_mask_ceil_round_ps): Ditto.
279 (_mm512_mask_ceil_round_pd): Ditto.
280 (_mm512_maskz_floor_round_ps): Ditto.
281 (_mm512_maskz_floor_round_pd): Ditto.
282 (_mm512_maskz_ceil_round_ps): Ditto.
283 (_mm512_maskz_ceil_round_pd): Ditto.
284 (_mm512_expand_pd): Ditto.
285 (_mm512_expand_ps): Ditto.
286 * config/i386/i386.c (ix86_builtins): Remove
287 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
289 * config/i386/predicates.md (const1256_operand): New.
290 (const_1_to_2_operand): Ditto.
291 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
292 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
293 (*avx512pf_gatherpf<mode>sf): Ditto.
294 (avx512pf_gatherpf<mode>df): Ditto.
295 (*avx512pf_gatherpf<mode>df_mask): Ditto.
296 (*avx512pf_gatherpf<mode>df): Ditto.
297 (avx512pf_scatterpf<mode>sf): Ditto.
298 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
299 (*avx512pf_scatterpf<mode>sf): Ditto.
300 (avx512pf_scatterpf<mode>df): Ditto.
301 (*avx512pf_scatterpf<mode>df_mask): Ditto.
302 (*avx512pf_scatterpf<mode>df): Ditto.
303 (avx512f_expand<mode>): Removed.
304 (<shift_insn><mode>3<mask_name>): Change predicate type.
306 2014-02-08 Jakub Jelinek <jakub@redhat.com>
308 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
309 not at the end of datarefs vector use ordered_remove to avoid
310 reordering datarefs vector.
313 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
314 mark local addressable non-static vars as GOVD_PRIVATE
315 instead of GOVD_LOCAL.
316 * omp-low.c (lower_omp_for): Move gimple_bind_vars
317 and BLOCK_VARS of gimple_bind_block to new_stmt rather
321 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
322 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
323 assume_aligned or alloc_align attributes.
324 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
325 arguments. Handle also assume_aligned and alloc_align attributes.
326 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
327 calls to functions with assume_aligned or alloc_align attributes.
328 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
330 2014-02-08 Terry Guo <terry.guo@arm.com>
332 * doc/invoke.texi: Document ARM -march=armv7e-m.
334 2014-02-08 Jakub Jelinek <jakub@redhat.com>
336 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
337 flag on __cilkrts_rethrow builtin.
340 * ipa-cp.c (determine_versionability): Fail at -O0
341 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
342 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
345 2014-02-04 Jakub Jelinek <jakub@redhat.com>
348 * tree-inline.c (copy_forbidden): Fail for
349 __attribute__((optimize (0))) functions.
351 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
353 * varpool.c: Include pointer-set.h.
354 (varpool_remove_unreferenced_decls): Variables in other partitions
355 will not be output; be however careful to not lose information
358 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
360 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
361 lookup in the vtable constructor.
363 2014-02-07 Jeff Law <law@redhat.com>
366 * config/m68k/m68k.md (ashldi_extsi): Turn into a
367 define_insn_and_split.
369 * ipa-inline.c (inline_small_functions): Fix typos.
371 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
373 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
374 (s390_can_use_return_insn): Declare.
375 * config/s390/s390.h (EPILOGUE_USES): Define.
376 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
378 (s390_chunkify_start): Handle return JUMP_LABELs.
379 (s390_early_mach): Emit a main_pool instruction on the entry edge.
380 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
381 (s390_can_use_return_insn): New functions.
382 (s390_fix_long_loop_prediction): Handle conditional returns.
383 (TARGET_SET_UP_BY_PROLOGUE): Define.
384 * config/s390/s390.md (ANY_RETURN): New code iterator.
385 (*creturn, *csimple_return, return, simple_return): New patterns.
387 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
389 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
390 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
391 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
392 REG_CFA_RESTORE list when deciding not to restore a register.
394 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
396 * config/s390/s390.c: Include tree-pass.h and context.h.
397 (s390_early_mach): New function, split out from...
398 (s390_emit_prologue): ...here.
399 (pass_data_s390_early_mach): New pass structure.
400 (pass_s390_early_mach): New class.
401 (s390_option_override): Create and register early_mach pass.
404 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
406 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
407 to match for the exit block.
409 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
411 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
412 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
413 Reject misaligned operands.
415 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
417 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
419 2014-02-07 Richard Biener <rguenther@suse.de>
422 * gimple-low.c (lower_builtin_posix_memalign): New function.
423 (lower_stmt): Call it to lower posix_memalign in a way
424 to make alignment info accessible.
426 2014-02-07 Jakub Jelinek <jakub@redhat.com>
429 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
430 __builtin_setjmp_receiver.
432 2014-02-07 Richard Biener <rguenther@suse.de>
435 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
436 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
437 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
438 Handle BUILT_IN_POSIX_MEMALIGN.
439 (find_func_clobbers): Likewise.
440 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
441 (call_may_clobber_ref_p_1): Likewise.
443 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
446 * ipa-devirt.c (record_target_from_binfo): Remove overactive
449 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
452 * lto-cgraph.c (lto_output_node): Use
453 symtab_get_symbol_partitioning_class.
454 (lto_output_varpool_node): likewise.
455 (symtab_get_symbol_partitioning_class): Move here from
457 * cgraph.h (symbol_partitioning_class): Likewise.
458 (symtab_get_symbol_partitioning_class): Declare.
460 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
462 * ggc.h (ggc_internal_cleared_alloc): New macro.
463 * vec.h (vec_safe_copy): Handle memory stats.
464 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
465 * target-globals.c (save_target_globals): Likewise.
467 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
470 * expr.c (emit_move_resolve_push): Export; be bit more selective
471 on when to clear alias set.
472 * expr.h (emit_move_resolve_push): Declare.
473 * function.h (struct function): Add tail_call_marked.
474 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
475 * config/i386/i386-protos.h (ix86_expand_push): Remove.
476 * config/i386/i386.md (TImode move expander): De not call
478 (FP push expanders): Preserve memory attributes.
479 * config/i386/sse.md (push<mode>1): Remove.
480 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
481 (ix86_expand_push): Remove.
482 * config/i386/mmx.md (push<mode>1): Remove.
484 2014-02-06 Jakub Jelinek <jakub@redhat.com>
486 PR rtl-optimization/60030
487 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
488 lopart with paradoxical subreg before shifting it up by hprec.
490 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
492 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
493 Remove extra newline at end of file.
494 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
495 (arm_issue_rate): Handle cortexa57.
496 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
497 (cortex-a57.cortex-a53): Likewise.
499 2014-02-06 Jakub Jelinek <jakub@redhat.com>
502 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
503 don't record in REG_FRAME_RELATED_EXPR registers not set in that
505 (arm_expand_prologue): Adjust all callers.
506 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
507 info, registers also at the lowest numbered registers side. Use
508 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
512 * var-tracking.c (adjust_mems): Before adding a SET to
513 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
515 2014-02-06 Alan Modra <amodra@gmail.com>
518 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
519 change SDmode to DDmode when lra_in_progress.
521 2014-02-06 Jakub Jelinek <jakub@redhat.com>
524 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
525 free_data_ref on the dr first, and before goto again also set dr
526 to the next dr. For simd_lane_access, free old datarefs[i] before
527 overwriting it. For get_vectype_for_scalar_type failure, don't
528 free_data_ref if simd_lane_access.
530 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
533 * tree.h (opts_for_fn): New inline function.
534 (opt_for_fn): Define.
535 * config/i386/i386.c (ix86_function_regparm): Use
536 opt_for_fn (decl, optimize) instead of optimize.
538 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
540 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
541 for SYMBOL_REF in large memory model.
543 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
545 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
547 (cortex-a57): Likewise.
548 (cortex-a57.cortex-a53): Likewise.
550 2014-02-06 Yury Gribov <y.gribov@samsung.com>
551 Kugan Vivekanandarajah <kuganv@linaro.org>
553 * config/arm/arm.c (arm_vector_alignment_reachable): Check
555 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
557 2014-02-06 Richard Biener <rguenther@suse.de>
559 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
560 set_loop_copy and initialize_original_copy_tables.
562 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
564 * config/aarch64/aarch64-simd.md
565 (aarch64_ashr_simddi): Change QI to SI.
567 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
568 Jakub Jelinek <jakub@redhat.com>
571 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
574 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
576 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
577 CODE_FOR_altivec_vpku[hw]um to
578 CODE_FOR_altivec_vpku[hw]um_direct.
579 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
580 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
581 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
582 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
584 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
586 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
587 generation for -maltivec=be.
588 (altivec_vsumsws): Simplify redundant test.
590 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
592 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
593 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
594 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
595 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
597 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
599 (altivec_vpks<VI_char>ss): Likewise.
600 (altivec_vpks<VI_char>us): Likewise.
601 (altivec_vpku<VI_char>us): Likewise.
602 (altivec_vpku<VI_char>um): Likewise.
603 (altivec_vpku<VI_char>um_direct): New (copy of
604 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
606 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
607 target is little endian and -maltivec=be is not specified.
608 (*altivec_vupkhs<VU_char>_direct): New (copy of
609 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
610 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
611 target is little endian and -maltivec=be is not specified.
612 (*altivec_vupkls<VU_char>_direct): New (copy of
613 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
614 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
615 little endian and -maltivec=be is not specified.
616 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
617 little endian and -maltivec=be is not specified.
619 2014-02-05 Richard Henderson <rth@redhat.com>
622 * combine-stack-adj.c: Revert r206943.
623 * sched-int.h (struct deps_desc): Add last_args_size.
624 * sched-deps.c (init_deps): Initialize it.
625 (sched_analyze_insn): Add OUTPUT dependencies between insns that
626 contain REG_ARGS_SIZE notes.
628 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
630 * lto-cgraph.c (asm_nodes_output): Make global.
631 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
632 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
633 (driver_handle_option): Handle OPT_fwpa.
635 2014-02-05 Jakub Jelinek <jakub@redhat.com>
638 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
639 a comment typo and formatting issue. If odr_hash hasn't been
640 created, return vNULL and set *completep to false.
643 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
644 bb with no successors.
646 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
649 * doc/invoke.texi (-march): Clarify documentation for ARM.
653 2014-02-05 Richard Biener <rguenther@suse.de>
655 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
656 when not vectorizing because of too many alias checks.
657 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
658 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
660 2014-02-05 Nick Clifton <nickc@redhat.com>
662 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
663 accept extended registers in any mode when compiling for the MN10300.
665 2014-02-05 Yury Gribov <y.gribov@samsung.com>
667 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
668 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
669 sanitization attributes.
670 (can_inline_edge_p): Likewise.
671 (sanitize_attrs_match_for_inline_p): New function.
673 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
675 * ipa-prop.c (detect_type_change): Shor circuit testing of
676 type changes on THIS pointer.
678 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
681 * config/pa/pa.c (legitimize_tls_address): Return original address
682 if not passed a SYMBOL_REF rtx.
683 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
685 (pa_emit_move_sequence): Simplify TLS source operands.
686 (pa_legitimate_constant_p): Reject all TLS constants.
687 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
688 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
690 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
692 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
693 groups when we know they are controlled by LTO.
694 * varasm.c (default_binds_local_p_1): If object is in other partition,
695 it will be resolved locally.
697 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
699 * config/host-linux.c (linux_gt_pch_use_address): Don't
700 use SSIZE_MAX because it is not always defined.
702 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
705 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
706 threshold for pseudo splitting.
707 (update_ebb_live_info): Process call argument hard registers and
708 hard registers from insn definition too.
709 (max_small_class_regs_num): New constant.
710 (inherit_in_ebb): Update live hard regs through EBBs. Update
711 reloads_num only for small register classes. Don't split for
714 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
717 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
720 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
722 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
725 2014-02-04 Marek Polacek <polacek@redhat.com>
727 * gdbinit.in (pel): Define.
729 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
731 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
734 2014-02-04 Richard Biener <rguenther@suse.de>
737 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
738 in function context local.
739 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
740 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
741 similar to LTO_imported_decl_ref.
743 2014-02-04 Jakub Jelinek <jakub@redhat.com>
745 PR tree-optimization/60002
746 * cgraphclones.c (build_function_decl_skip_args): Clear
749 PR tree-optimization/60023
750 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
751 false to gsi_replace.
752 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
753 has been in some EH region and vec_stmt could throw, add
754 vec_stmt into the same EH region.
755 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
756 has no lhs, ignore it.
757 * internal-fn.c (expand_MASK_LOAD): Likewise.
760 * tree-inline.c (copy_forbidden): Fail for
761 __attribute__((optimize (0))) functions.
764 * omp-low.c (simd_clone_struct_copy): If from->inbranch
765 is set, copy one less argument.
766 (expand_simd_clones): Don't subtract clone_info->inbranch
767 from simd_clone_struct_alloc argument.
769 PR rtl-optimization/57915
770 * recog.c (simplify_while_replacing): If all unary/binary/relational
771 operation arguments are constant, attempt to simplify those.
774 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
775 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
777 2014-02-04 Richard Biener <rguenther@suse.de>
779 PR tree-optimization/60012
780 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
781 TBAA disambiguation to all DDRs.
783 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
786 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
787 (LINK_SPEC): Use it for -shared, -shared-libgcc.
789 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
792 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
794 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
796 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
797 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
799 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
802 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
803 to figure out targets of polymorphic calls with known decl.
804 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
805 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
806 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
807 (get_polymorphic_call_info): ... here.
808 (get_polymorphic_call_info_from_invariant): New function.
810 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
812 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
813 lookup via vtable pointer; check for type consistency
814 and turn inconsitent facts into UNREACHABLE.
815 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
816 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
817 type inconsistent querries; return UNREACHABLE instead.
819 2014-02-03 Richard Henderson <rth@twiddle.net>
822 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
823 already processed this node.
824 (normalize_one_pred_1): Pass along mark_set.
825 (normalize_one_pred): Create and destroy a pointer_set_t.
826 (normalize_one_pred_chain): Likewise.
828 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
830 PR gcov-profile/58602
831 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
833 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
836 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
837 -fno-devirtualize; try to devirtualize by the knowledge of
838 virtual table pointer given by aggregate propagation.
839 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
840 (ipa_print_node_jump_functions): Dump also offset that
841 is relevant for polymorphic calls.
842 (determine_known_aggregate_parts): Add arg_type parameter; use it
843 instead of determining the type from pointer type.
844 (ipa_compute_jump_functions_for_edge): Update call of
845 determine_known_aggregate_parts.
846 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
847 (gimple_get_virt_method_for_binfo): ... here; simplify using
848 vtable_pointer_value_to_vtable.
849 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
850 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
851 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
852 (vtable_pointer_value_to_vtable): Break out from ...; handle also
854 (vtable_pointer_value_to_binfo): ... here.
855 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
857 2014-02-03 Teresa Johnson <tejohnson@google.com>
859 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
860 redef of outer loop index variable.
862 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
866 * doc/extend.texi (Function Attributes): Typo.
868 2014-02-03 Cong Hou <congh@google.com>
870 PR tree-optimization/60000
871 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
872 if the vectorized statement is a store. A store statement can only
873 appear at the end of pattern statements.
875 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
877 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
878 (ix86_option_override_internal): Default long double to 64-bit for
879 32-bit Bionic and to 128-bit for 64-bit Bionic.
881 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
882 TARGET_LONG_DOUBLE_128 is true.
883 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
885 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
886 (mlong-double-64): Negate -mlong-double-128.
887 (mlong-double-128): New option.
889 * config/i386/i386-c.c (ix86_target_macros): Define
890 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
892 * doc/invoke.texi: Document -mlong-double-128.
894 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
896 PR rtl-optimization/60024
897 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
899 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
901 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
903 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
905 PR rtl-optimization/57662
906 * sel-sched.c (code_motion_path_driver): Do not mark already not
907 existing blocks in the visiting bitmap.
909 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
911 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
912 on the insn being emitted.
914 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
915 Will Deacon <will.deacon@arm.com>
917 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
919 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
921 * config/arm/arm-tables.opt: Regenerate.
923 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
925 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
926 for vector types other than V16QImode.
927 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
928 define_expand, and call altivec_expand_vec_perm_le when producing
929 code with little endian element order.
930 (*altivec_vperm_<mode>_internal): New insn having previous
931 behavior of altivec_vperm_<mode>.
932 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
933 altivec_expand_vec_perm_le when producing code with little endian
935 (*altivec_vperm_<mode>_uns_internal): New insn having previous
936 behavior of altivec_vperm_<mode>_uns.
938 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
940 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
941 (altivec_vsumsws): Add handling for -maltivec=be with a little
943 (altivec_vsumsws_direct): New.
944 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
947 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
949 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
950 vtable_pointer_value_to_binfo): New functions.
951 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
952 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
954 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
956 * config/nios2/nios2.md (load_got_register): Initialize GOT
957 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
958 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
960 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
962 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
963 preserverd by passthrough, do not propagate the type.
965 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
967 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
968 (mips_atomic_assign_expand_fenv): New function.
969 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
971 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
973 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
974 (__builtin_mips_set_fcsr): Likewise.
975 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
977 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
978 (mips16_expand_set_fcsr): Likewise.
979 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
980 (mips16_set_fcsr_stub): Likewise.
981 (mips16_get_fcsr_one_only_stub): New class.
982 (mips16_set_fcsr_one_only_stub): Likewise.
983 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
984 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
985 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
986 (hard_float): New availability predicate.
987 (mips_builtins): Add get_fcsr and set_fcsr.
988 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
989 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
990 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
991 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
992 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
995 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
997 * config/mips/mips.c (mips_one_only_stub): New class.
998 (mips_need_mips16_rdhwr_p): Replace with...
999 (mips16_rdhwr_stub): ...this new variable.
1000 (mips16_stub_call_address): New function.
1001 (mips16_rdhwr_one_only_stub): New class.
1002 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1003 (mips_output_mips16_rdhwr): Delete.
1004 (mips_finish_stub): New function.
1005 (mips_code_end): Use it to handle rdhwr stubs.
1007 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1010 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1011 when calculating size of integer atomic types.
1013 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1015 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1017 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1019 PR tree-optimization/60003
1020 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1021 * profile.c (branch_prob): Use gimple_call_builtin_p
1022 to check for BUILT_IN_SETJMP_RECEIVER.
1023 * tree-inline.c (copy_bb): Call notice_special_calls.
1025 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1028 * lra-constraints.c (process_alt_operands): Update reload_sum only
1031 2014-01-31 Richard Henderson <rth@redhat.com>
1034 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1035 until after else_eh is processed.
1037 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1039 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1040 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1041 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1042 in smmintrin.h, remove them.
1043 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1044 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1045 * config/i386/i386.md (ROUND_SAE): Fix value.
1046 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1047 (const48_operand): New.
1048 * config/i386/subst.md (round), (round_expand): Use
1049 const_4_or_8_to_11_operand.
1050 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1052 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1054 * config/i386/constraints.md (Yk): Swap meaning with k.
1055 * config/i386/i386.md (movhi_internal): Change Yk to k.
1056 (movqi_internal): Ditto.
1057 (*k<logic><mode>): Ditto.
1060 (kandn<mode>): Ditto.
1061 (*<code>hi_1): Ditto.
1062 (*<code>qi_1): Ditto.
1063 (kxnor<mode>): Ditto.
1064 (kortestzhi): Ditto.
1065 (kortestchi): Ditto.
1067 (*one_cmplhi2_1): Ditto.
1068 (*one_cmplqi2_1): Ditto.
1069 * config/i386/sse.md (): Change k to Yk.
1070 (avx512f_load<mode>_mask): Ditto.
1071 (avx512f_blendm<mode>): Ditto.
1072 (avx512f_store<mode>_mask): Ditto.
1073 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1074 (avx512f_storedqu<mode>_mask): Ditto.
1075 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1077 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1078 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1079 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1080 (avx512f_maskcmp<mode>3): Ditto.
1081 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1082 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1083 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1084 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1085 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1086 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1087 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1088 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1089 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1090 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1091 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1092 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1093 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1094 (vec_extract_lo_<mode>_maskm): Ditto.
1095 (vec_extract_hi_<mode>_maskm): Ditto.
1096 (avx512f_vternlog<mode>_mask): Ditto.
1097 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1098 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1099 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1100 (avx512f_<code>v8div16qi2_mask): Ditto.
1101 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1102 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1103 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1104 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1105 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1106 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1107 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1108 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1109 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1110 (avx512cd_maskb_vec_dupv8di): Ditto.
1111 (avx512cd_maskw_vec_dupv16si): Ditto.
1112 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1113 (avx512f_vpermi2var<mode>3_mask): Ditto.
1114 (avx512f_vpermi2var<mode>3_mask): Ditto.
1115 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1116 (*avx512f_gathersi<mode>): Ditto.
1117 (*avx512f_gathersi<mode>_2): Ditto.
1118 (*avx512f_gatherdi<mode>): Ditto.
1119 (*avx512f_gatherdi<mode>_2): Ditto.
1120 (*avx512f_scattersi<mode>): Ditto.
1121 (*avx512f_scatterdi<mode>): Ditto.
1122 (avx512f_compress<mode>_mask): Ditto.
1123 (avx512f_compressstore<mode>_mask): Ditto.
1124 (avx512f_expand<mode>_mask): Ditto.
1125 * config/i386/subst.md (mask): Change k to Yk.
1126 (mask_scalar_merge): Ditto.
1129 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1131 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1133 2014-01-31 Richard Biener <rguenther@suse.de>
1136 * builtins.c (fold_builtin_memory_op): Make sure to not
1137 use a floating-point mode or a boolean or enumeral type for
1140 2014-01-30 DJ Delorie <dj@redhat.com>
1142 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1143 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1144 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1145 whenever main() has an epilogue.
1147 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1149 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1150 unused variable "field".
1151 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1152 (vsx_mergeh_<mode>): Likewise.
1153 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1154 (altivec_vmrghh): Likewise.
1155 (altivec_vmrghw): Likewise.
1156 (altivec_vmrglb): Likewise.
1157 (altivec_vmrglh): Likewise.
1158 (altivec_vmrglw): Likewise.
1159 (altivec_vspltb): Add missing uses.
1160 (altivec_vsplth): Likewise.
1161 (altivec_vspltw): Likewise.
1162 (altivec_vspltsf): Likewise.
1164 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1167 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1168 frame related instructions.
1170 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1172 PR rtl-optimization/59959
1173 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1174 any reload of register whose subreg is invalid.
1176 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1178 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1179 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1180 Add missing return type - void.
1182 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1184 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1185 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1186 remove element index adjustment for endian (now handled in vsx.md
1188 (altivec_expand_vec_perm_const): Use
1189 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1190 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1191 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1192 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1193 define_expand and a new define_insn *altivec_vspltb_internal;
1194 adjust for -maltivec=be on a little endian target.
1195 (altivec_vspltb_direct): New.
1196 (altivec_vsplth): Divide into a define_expand and a new
1197 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1198 little endian target.
1199 (altivec_vsplth_direct): New.
1200 (altivec_vspltw): Divide into a define_expand and a new
1201 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1202 little endian target.
1203 (altivec_vspltw_direct): New.
1204 (altivec_vspltsf): Divide into a define_expand and a new
1205 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1206 a little endian target.
1208 2014-01-30 Richard Biener <rguenther@suse.de>
1210 PR tree-optimization/59993
1211 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1212 can propagate form the earlier stmt and avoid the transform
1213 when the intermediate result is needed.
1215 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1217 * README.Portability: Fix typo.
1219 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1221 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1222 comparison_operator with ordered_comparison_operator.
1224 2014-01-30 Nick Clifton <nickc@redhat.com>
1226 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1227 Rename to mn10300_store_multiple_regs.
1228 * config/mn10300/mn10300.c: Likewise.
1229 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1230 store_multiple_regs.
1231 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1232 Call mn10300_store_multiple_regs.
1234 2014-01-30 Nick Clifton <nickc@redhat.com>
1235 DJ Delorie <dj@redhat.com>
1237 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1238 %fp 2 to keep registers after it properly word-aligned.
1239 (rl78_alloc_physical_registers_umul): Handle the case where both
1240 input operands are the same.
1242 2014-01-30 Richard Biener <rguenther@suse.de>
1244 PR tree-optimization/59903
1245 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1248 2014-01-30 Jason Merrill <jason@redhat.com>
1251 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1254 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1256 2014-01-30 Richard Biener <rguenther@suse.de>
1258 PR tree-optimization/59951
1259 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1261 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1264 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1265 SFmode to DFmode case.
1267 2014-01-29 DJ Delorie <dj@redhat.com>
1269 * config/msp430/msp430.opt (-minrt): New.
1270 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1272 (ENDFILE_SPEC): Likewise.
1274 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1276 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1277 (estimate_function_body_sizes): Use it.
1279 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1282 * dwarf2out.c (is_cxx_auto): New.
1283 (is_base_type): Use it.
1284 (gen_type_die_with_usage): Likewise.
1286 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1288 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1289 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1290 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1291 -maltivec=be with LE targets.
1292 (vsx_mergeh_<mode>): Likewise.
1293 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1294 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1295 (altivec_vmrghb): Replace with define_expand and new
1296 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1297 (altivec_vmrghb_direct): New define_insn.
1298 (altivec_vmrghh): Replace with define_expand and new
1299 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1300 (altivec_vmrghh_direct): New define_insn.
1301 (altivec_vmrghw): Replace with define_expand and new
1302 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1303 (altivec_vmrghw_direct): New define_insn.
1304 (*altivec_vmrghsf): Adjust for endianness.
1305 (altivec_vmrglb): Replace with define_expand and new
1306 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1307 (altivec_vmrglb_direct): New define_insn.
1308 (altivec_vmrglh): Replace with define_expand and new
1309 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1310 (altivec_vmrglh_direct): New define_insn.
1311 (altivec_vmrglw): Replace with define_expand and new
1312 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1313 (altivec_vmrglw_direct): New define_insn.
1314 (*altivec_vmrglsf): Adjust for endianness.
1315 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1316 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1317 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1318 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1319 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1320 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1321 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1322 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1324 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1326 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1327 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1330 2014-01-29 Richard Biener <rguenther@suse.de>
1332 PR tree-optimization/58742
1333 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1334 associate_pointerplus_align.
1335 (associate_pointerplus_diff): New function.
1336 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1337 and associate_pointerplus_diff.
1339 2014-01-29 Richard Biener <rguenther@suse.de>
1341 * lto-streamer.h (LTO_major_version): Bump to 3.
1342 (LTO_minor_version): Reset to 0.
1344 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1346 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1347 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1348 (arm_file_start): Generate correct asm header for armv7ve.
1349 * config/arm/bpabi.h: Add multilib support for armv7ve.
1350 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1351 and cortex-a15 to armv7ve.
1352 * config/arm/t-aprofile: Add multilib support for armv7ve.
1353 * doc/invoke.texi: Document -march=armv7ve.
1355 2014-01-29 Richard Biener <rguenther@suse.de>
1357 PR tree-optimization/58742
1358 * tree-ssa-forwprop.c (associate_plusminus): Return true
1359 if we changed sth, defer EH cleanup to ...
1360 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1361 (simplify_mult): New function.
1363 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1366 PR tree-optimization/59920
1367 * tree.c (build_common_builtin_nodes): Remove
1368 __builtin_setjmp_dispatcher initialization.
1369 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1370 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1371 instead of gsi_after_labels + manually skipping debug stmts.
1372 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1373 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1374 * tree-inline.c (copy_edges_for_bb): Remove
1375 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1376 argument. Ignore computed_goto_p stmts. Don't call
1377 make_abnormal_goto_edges. If a call might need abnormal edges
1378 for non-local gotos, see if it already has an edge to
1379 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1380 with true argument, don't do anything then, otherwise add
1381 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1382 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1384 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1385 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1386 (lower_stmt): Don't set data->calls_builtin_setjmp.
1387 (lower_builtin_setjmp): Adjust comment.
1388 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1389 * tree-cfg.c (found_computed_goto): Remove.
1390 (factor_computed_gotos): Remove.
1391 (make_goto_expr_edges): Return bool, true for computed gotos.
1392 Don't call make_abnormal_goto_edges.
1393 (build_gimple_cfg): Don't set found_computed_goto, don't call
1394 factor_computed_gotos.
1395 (computed_goto_p): No longer static.
1396 (make_blocks): Don't set found_computed_goto.
1397 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1398 (make_edges): If make_goto_expr_edges returns true, push bb
1399 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1400 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1401 vector. Record mapping between bbs and OpenMP regions if there
1402 are any, adjust make_gimple_omp_edges caller. Call
1403 handle_abnormal_edges.
1404 (make_abnormal_goto_edges): Remove.
1405 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1406 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1407 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1408 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1409 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1410 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1411 filling *region also set *region_idx to (*region)->entry->index.
1414 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1415 For REGs set ORIGINAL_REGNO.
1417 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1419 * doc/md.texi: Mention that a target shouldn't implement
1420 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1423 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1425 PR tree-optimization/59594
1426 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1427 a copy of the datarefs vector rather than the vector itself.
1429 2014-01-28 Jason Merrill <jason@redhat.com>
1432 * dwarf2out.c (auto_die): New static.
1433 (gen_type_die_with_usage): Handle C++1y 'auto'.
1434 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1437 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1440 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1441 (SPEC_X32): Likewise.
1442 (SPEC_64): Likewise.
1443 * config/i386/i386.c (ix86_option_override_internal): Turn off
1444 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1446 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1447 * config/i386/i386.h (TARGET_16BIT): New macro.
1448 (TARGET_16BIT_P): Likewise.
1449 * config/i386/i386.opt: Add m16.
1450 * doc/invoke.texi: Document -m16.
1452 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1454 PR preprocessor/59935
1455 * input.c (location_get_source_line): Bail out on when line number
1456 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1458 2014-01-28 Richard Biener <rguenther@suse.de>
1460 PR tree-optimization/58742
1461 * tree-ssa-forwprop.c (associate_plusminus): Handle
1462 pointer subtraction of the form (T)(P + A) - (T)P.
1464 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1466 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1469 2014-01-28 Richard Biener <rguenther@suse.de>
1472 2014-01-28 Richard Biener <rguenther@suse.de>
1474 PR rtl-optimization/45364
1475 PR rtl-optimization/59890
1476 * var-tracking.c (local_get_addr_clear_given_value): Handle
1477 already cleared slot.
1478 (val_reset): Handle not allocated local_get_addr_cache.
1479 (vt_find_locations): Use post-order on the inverted CFG.
1481 2014-01-28 Richard Biener <rguenther@suse.de>
1483 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1485 2014-01-28 Richard Biener <rguenther@suse.de>
1487 PR rtl-optimization/45364
1488 PR rtl-optimization/59890
1489 * var-tracking.c (local_get_addr_clear_given_value): Handle
1490 already cleared slot.
1491 (val_reset): Handle not allocated local_get_addr_cache.
1492 (vt_find_locations): Use post-order on the inverted CFG.
1494 2014-01-28 Alan Modra <amodra@gmail.com>
1496 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1497 * configure.ac <recursive call for build != host>: Define
1498 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1499 and LD_FOR_BUILD too.
1500 * configure: Regenerate.
1502 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1504 * config/i386/i386.c (get_builtin_code_for_version): Separate
1505 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1506 Broadwell from Haswell.
1508 2014-01-27 Steve Ellcey <sellcey@mips.com>
1510 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1511 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1512 * config/mips/mips.c (mips_option_override): Change setting
1514 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1515 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1516 Change from Mask to Var.
1518 2014-01-27 Jeff Law <law@redhat.com>
1520 * ipa-inline.c (inline_small_functions): Fix typo.
1522 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1524 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1525 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1526 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1527 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1528 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1529 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1530 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1531 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1532 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1533 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1534 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1535 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1536 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1537 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1538 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1539 (_mm512_storeu_epi64): Ditto.
1540 (_mm512_cmpge_epi32_mask): Ditto.
1541 (_mm512_cmpge_epu32_mask): Ditto.
1542 (_mm512_cmpge_epi64_mask): Ditto.
1543 (_mm512_cmpge_epu64_mask): Ditto.
1544 (_mm512_cmple_epi32_mask): Ditto.
1545 (_mm512_cmple_epu32_mask): Ditto.
1546 (_mm512_cmple_epi64_mask): Ditto.
1547 (_mm512_cmple_epu64_mask): Ditto.
1548 (_mm512_cmplt_epi32_mask): Ditto.
1549 (_mm512_cmplt_epu32_mask): Ditto.
1550 (_mm512_cmplt_epi64_mask): Ditto.
1551 (_mm512_cmplt_epu64_mask): Ditto.
1552 (_mm512_cmpneq_epi32_mask): Ditto.
1553 (_mm512_cmpneq_epu32_mask): Ditto.
1554 (_mm512_cmpneq_epi64_mask): Ditto.
1555 (_mm512_cmpneq_epu64_mask): Ditto.
1556 (_mm512_expand_pd): Ditto.
1557 (_mm512_expand_ps): Ditto.
1558 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1559 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1560 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1561 * config/i386/i386.c (ix86_builtins): Add
1562 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1563 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1564 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1565 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1566 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1567 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1568 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1569 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1570 IX86_BUILTIN_PMOVUSQW512_MEM.
1571 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1572 __builtin_ia32_pmovsqd512mem_mask,
1573 __builtin_ia32_pmovqd512mem_mask,
1574 __builtin_ia32_pmovusqw512mem_mask,
1575 __builtin_ia32_pmovsqw512mem_mask,
1576 __builtin_ia32_pmovqw512mem_mask,
1577 __builtin_ia32_pmovusdw512mem_mask,
1578 __builtin_ia32_pmovsdw512mem_mask,
1579 __builtin_ia32_pmovdw512mem_mask,
1580 __builtin_ia32_pmovqb512mem_mask,
1581 __builtin_ia32_pmovusqb512mem_mask,
1582 __builtin_ia32_pmovsqb512mem_mask,
1583 __builtin_ia32_pmovusdb512mem_mask,
1584 __builtin_ia32_pmovsdb512mem_mask,
1585 __builtin_ia32_pmovdb512mem_mask.
1586 (bdesc_args): Add __builtin_ia32_expanddf512,
1587 __builtin_ia32_expandsf512.
1588 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1589 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1590 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1591 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1592 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1593 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1594 (avx512f_<code>v8div16qi2_mask_store): This.
1595 (avx512f_expand<mode>): New.
1597 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1599 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1601 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1602 (_mm512_prefetch_i32scatter_pd): Ditto.
1603 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1604 (_mm512_prefetch_i64scatter_pd): Ditto.
1605 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1606 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1607 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1608 (_mm512_prefetch_i32scatter_ps): Ditto.
1609 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1610 (_mm512_prefetch_i64scatter_ps): Ditto.
1611 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1612 * config/i386/i386-builtin-types.def: Define
1613 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1614 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1615 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1616 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1617 IX86_BUILTIN_SCATTERPFQPD.
1618 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1619 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1620 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1621 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1622 __builtin_ia32_scatterpfqps.
1623 (ix86_expand_builtin): Expand new built-ins.
1624 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1625 fix memory access data type.
1626 (*avx512pf_gatherpf<mode>_mask): Ditto.
1627 (*avx512pf_gatherpf<mode>): Ditto.
1628 (avx512pf_scatterpf<mode>): Ditto.
1629 (*avx512pf_scatterpf<mode>_mask): Ditto.
1630 (*avx512pf_scatterpf<mode>): Ditto.
1631 (GATHER_SCATTER_SF_MEM_MODE): New.
1632 (avx512pf_gatherpf<mode>df): Ditto.
1633 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1634 (*avx512pf_scatterpf<mode>df): Ditto.
1636 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1639 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1640 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1643 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1645 * common/config/arm/arm-common.c
1646 (arm_rewrite_mcpu): Handle multiple names.
1648 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1650 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1652 * gimple-builder.h (create_gimple_tmp): Delete.
1654 2014-01-27 Christian Bruel <christian.bruel@st.com>
1656 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1659 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1661 * config/pa/pa.md (call): Generate indirect long calls to non-local
1662 functions when outputing 32-bit code.
1663 (call_value): Likewise except for special call to buggy powf function.
1665 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1666 portable runtime and PIC indirect calls.
1667 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1668 and PIC call sequences. Use ldo instead of blr to set return register
1669 in PIC call sequence.
1671 2014-01-25 Walter Lee <walt@tilera.com>
1673 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1674 avoid clobbering a live register.
1676 2014-01-25 Walter Lee <walt@tilera.com>
1678 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1679 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1680 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1681 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1683 2014-01-25 Walter Lee <walt@tilera.com>
1685 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1686 arguments on even registers.
1687 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1689 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1690 (BIGGEST_ALIGNMENT): Ditto.
1691 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1693 2014-01-25 Walter Lee <walt@tilera.com>
1695 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1696 insns before bundling.
1697 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1699 2014-01-25 Walter Lee <walt@tilera.com>
1701 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1702 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1703 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1705 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1707 * config/mips/constraints.md (kl): Delete.
1708 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1709 define expands, using...
1710 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1711 instructions for MIPS16.
1712 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1713 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1715 2014-01-25 Walter Lee <walt@tilera.com>
1717 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1721 2014-01-25 Walter Lee <walt@tilera.com>
1723 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1724 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1726 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1728 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1731 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1733 * print-rtl.c (in_call_function_usage): New var.
1734 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1735 EXPR_LIST mode as mode and not as reg note name.
1738 * cfgloopmanip.c (copy_loop_info): If
1739 loop->warned_aggressive_loop_optimizations, make sure
1740 the flag is set in target loop too.
1742 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1744 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1746 * builtins.def: Likewise.
1747 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1748 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1749 * ira.c (ira_setup_eliminable_regset): Likewise.
1750 * omp-low.c (gate_expand_omp): Likewise.
1751 (execute_lower_omp): Likewise.
1752 (diagnose_sb_0): Likewise.
1753 (gate_diagnose_omp_blocks): Likewise.
1754 (simd_clone_clauses_extract): Likewise.
1757 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1759 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1760 correction for little endian...
1761 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1764 2014-01-24 Jeff Law <law@redhat.com>
1766 PR tree-optimization/59919
1767 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1768 for non-returning calls.
1770 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1772 * common/config/aarch64/aarch64-common.c
1773 (aarch64_rewrite_mcpu): Handle multiple names.
1774 * config/aarch64/aarch64.h
1775 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1777 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1779 * input.c (add_file_to_cache_tab): Handle the case where fopen
1782 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1785 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1786 from push operand if code of push isn't PRE_DEC.
1788 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1791 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1792 -mquad-memory-atomic. Update -mquad-memory documentation to say
1793 it is only used for non-atomic loads/stores.
1795 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1796 -mquad-memory or -mquad-memory-atomic switches.
1798 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1799 -mquad-memory-atomic to ISA 2.07 support.
1801 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1802 to separate support of normal quad word memory operations (ldq, stq)
1803 from the atomic quad word memory operations.
1805 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1806 support to separate non-atomic quad word operations from atomic
1807 quad word operations. Disable non-atomic quad word operations in
1808 little endian mode so that we don't have to swap words after the
1809 load and before the store.
1810 (quad_load_store_p): Add comment about atomic quad word support.
1811 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1812 options printed with -mdebug=reg.
1814 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1815 -mquad-memory-atomic as the test for whether we have quad word
1816 atomic instructions.
1817 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1818 or -mp8-vector are used, allow byte/half-word atomic operations.
1820 * config/rs6000/sync.md (load_lockedti): Insure that the address
1821 is a proper indexed or indirect address for the lqarx instruction.
1822 On little endian systems, swap the hi/lo registers after the lqarx
1824 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1825 insure the address is valid for the lqarx instruction.
1826 (store_conditionalti): Insure that the address is a proper indexed
1827 or indirect address for the stqcrx. instruction. On little endian
1828 systems, swap the hi/lo registers before doing the stqcrx.
1830 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1831 insure the address is valid for the stqcrx. instruction.
1833 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1834 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1835 type of quad memory support is available.
1837 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1840 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1841 there is a danger of looping.
1843 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1846 force flag_ira_loop_pressure if set via command line.
1848 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1850 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1851 (ashr_simd): New builtin handling DI mode.
1852 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1853 (aarch64_sshr_simddi): New match pattern.
1854 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1855 (vshrd_n_s64): Likewise.
1856 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1858 2014-01-23 Nick Clifton <nickc@redhat.com>
1860 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1861 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1862 favour of mcu specific scripts.
1863 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1866 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1867 Alex Velenko <Alex.Velenko@arm.com>
1869 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1870 (vaddv_s16): Likewise.
1871 (vaddv_s32): Likewise.
1872 (vaddv_u8): Likewise.
1873 (vaddv_u16): Likewise.
1874 (vaddv_u32): Likewise.
1875 (vaddvq_s8): Likewise.
1876 (vaddvq_s16): Likewise.
1877 (vaddvq_s32): Likewise.
1878 (vaddvq_s64): Likewise.
1879 (vaddvq_u8): Likewise.
1880 (vaddvq_u16): Likewise.
1881 (vaddvq_u32): Likewise.
1882 (vaddvq_u64): Likewise.
1883 (vaddv_f32): Likewise.
1884 (vaddvq_f32): Likewise.
1885 (vaddvq_f64): Likewise.
1886 (vmaxv_f32): Likewise.
1887 (vmaxv_s8): Likewise.
1888 (vmaxv_s16): Likewise.
1889 (vmaxv_s32): Likewise.
1890 (vmaxv_u8): Likewise.
1891 (vmaxv_u16): Likewise.
1892 (vmaxv_u32): Likewise.
1893 (vmaxvq_f32): Likewise.
1894 (vmaxvq_f64): Likewise.
1895 (vmaxvq_s8): Likewise.
1896 (vmaxvq_s16): Likewise.
1897 (vmaxvq_s32): Likewise.
1898 (vmaxvq_u8): Likewise.
1899 (vmaxvq_u16): Likewise.
1900 (vmaxvq_u32): Likewise.
1901 (vmaxnmv_f32): Likewise.
1902 (vmaxnmvq_f32): Likewise.
1903 (vmaxnmvq_f64): Likewise.
1904 (vminv_f32): Likewise.
1905 (vminv_s8): Likewise.
1906 (vminv_s16): Likewise.
1907 (vminv_s32): Likewise.
1908 (vminv_u8): Likewise.
1909 (vminv_u16): Likewise.
1910 (vminv_u32): Likewise.
1911 (vminvq_f32): Likewise.
1912 (vminvq_f64): Likewise.
1913 (vminvq_s8): Likewise.
1914 (vminvq_s16): Likewise.
1915 (vminvq_s32): Likewise.
1916 (vminvq_u8): Likewise.
1917 (vminvq_u16): Likewise.
1918 (vminvq_u32): Likewise.
1919 (vminnmv_f32): Likewise.
1920 (vminnmvq_f32): Likewise.
1921 (vminnmvq_f64): Likewise.
1923 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1925 * config/aarch64/aarch64-simd.md
1926 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1927 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1928 (*aarch64_mul3_elt<mode>): Likewise.
1929 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1930 (*aarch64_mul3_elt_to_64v2df): Likewise.
1931 (*aarch64_mla_elt<mode>): Likewise.
1932 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1933 (*aarch64_mls_elt<mode>): Likewise.
1934 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1935 (*aarch64_fma4_elt<mode>): Likewise.
1936 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1937 (*aarch64_fma4_elt_to_64v2df): Likewise.
1938 (*aarch64_fnma4_elt<mode>): Likewise.
1939 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1940 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1941 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1942 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1943 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1944 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1945 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1946 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1947 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1949 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1951 * config/aarch64/aarch64-simd.md
1952 (aarch64_be_checked_get_lane<mode>): New define_expand.
1953 * config/aarch64/aarch64-simd-builtins.def
1954 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1955 New builtin definition.
1956 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1957 Use new safe be builtin.
1959 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1961 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1963 (aarch64_be_st1<mode>): Likewise.
1964 (aarch_ld1<VALL:mode>): Define_expand modified.
1965 (aarch_st1<VALL:mode>): Likewise.
1966 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
1967 (UNSPEC_ST1): Likewise.
1969 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
1971 * config/microblaze/microblaze.md: Add trap insn and attribute
1973 2014-01-23 Dodji Seketeli <dodji@redhat.com>
1975 PR preprocessor/58580
1976 * input.h (location_get_source_line): Take an additional line_size
1978 (void diagnostics_file_cache_fini): Declare new function.
1979 * input.c (struct fcache): New type.
1980 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
1981 New static constants.
1982 (diagnostic_file_cache_init, total_lines_num)
1983 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
1984 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
1985 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
1986 (get_next_line, read_next_line, goto_next_line, read_line_num):
1987 New static function definitions.
1988 (diagnostic_file_cache_fini): New function.
1989 (location_get_source_line): Take an additional output line_len
1990 parameter. Re-write using lookup_or_add_file_to_cache_tab and
1992 * diagnostic.c (diagnostic_finish): Call
1993 diagnostic_file_cache_fini.
1994 (adjust_line): Take an additional input parameter for the length
1995 of the line, rather than calculating it with strlen.
1996 (diagnostic_show_locus): Adjust the use of
1997 location_get_source_line and adjust_line with respect to their new
1998 signature. While displaying a line now, do not stop at the first
1999 null byte. Rather, display the zero byte as a space and keep
2000 going until we reach the size of the line.
2001 * Makefile.in: Add vec.o to OBJS-libcommon
2003 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2004 Ilya Tocar <ilya.tocar@intel.com>
2006 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2007 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2008 (__builtin_ia32_kmov16): Ditto.
2009 * config/i386/i386.md (UNSPEC_KMOV): New.
2012 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2014 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2015 (_mm512_storeu_si512): Ditto.
2017 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2020 * rtl.h (get_referenced_operands): Declare.
2021 * recog.c (get_referenced_operands): New function.
2022 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2023 operands have been referenced when recording LO_SUM references.
2025 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2027 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2029 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2031 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2032 Enable for generic and recent AMD targets.
2034 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2036 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2037 ARG_SIZE note when adjustment was eliminated.
2039 2014-01-22 Jeff Law <law@redhat.com>
2041 PR tree-optimization/59597
2042 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2043 in file. Accept new argument REGISTERING and use it to modify
2044 dump output appropriately.
2045 (register_jump_thread): Corresponding changes.
2046 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2047 thread paths involving joiner blocks. Add code to dump cancelled
2048 jump threading paths.
2050 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2052 PR rtl-optimization/59477
2053 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2054 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2056 2014-01-22 Tom Tromey <tromey@redhat.com>
2058 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2060 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2062 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2064 PR rtl-optimization/59896
2065 * lra-constraints.c (process_alt_operands): Check unused note for
2066 matched operands of insn with no output reloads.
2068 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2070 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2071 (mips_move_from_gpr_cost): Likewise.
2073 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2075 PR rtl-optimization/59858
2076 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2077 ira_class_hard_regs_num.
2078 (process_alt_operands): Increase reject for dying matched operand.
2080 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2083 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2084 smaller than size, perform several stores or loads and stores
2085 at dst + count - size to store or copy all of size bytes, rather
2086 than just last modesize bytes.
2088 2014-01-20 DJ Delorie <dj@redhat.com>
2090 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2091 that CLOBBERs are REGs before propogating their values.
2093 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2096 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2097 (cgraph_inline_failed_type): New function.
2098 * cgraph.h (DEFCIFCODE): Add type.
2099 (cgraph_inline_failed_type_t): New enum.
2100 (cgraph_inline_failed_type): New prototype.
2101 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2102 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2103 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2104 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2105 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2106 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2107 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2108 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2109 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2110 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2111 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2112 OPTIMIZATION_MISMATCH.
2113 * tree-inline.c (expand_call_inline): Emit errors during
2114 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2116 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2119 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2120 mode attribute in insn output.
2122 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2124 * output.h (output_constant): Delete.
2125 * varasm.c (output_constant): Make private.
2127 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2129 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2131 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2134 * tree.h (fold_builtin_strcat): New prototype.
2135 * builtins.c (fold_builtin_strcat): No longer static. Add len
2136 argument, if non-NULL, don't call c_strlen. Optimize
2137 directly into __builtin_memcpy instead of __builtin_strcpy.
2138 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2139 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2141 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2143 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2144 for SImode_address_operand operands, having only a REG argument.
2146 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2148 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2149 loader name using mbig-endian.
2150 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2152 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2154 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2158 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2160 * config/aarch64/aarch64-protos.h
2161 (aarch64_cannot_change_mode_class_ptr): Declare.
2162 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2163 aarch64_cannot_change_mode_class_ptr): New.
2164 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2165 backend hook aarch64_cannot_change_mode_class.
2167 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2169 * common/config/aarch64/aarch64-common.c
2170 (aarch64_handle_option): Don't handle any option order logic here.
2171 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2172 selected_cpu, warn on architecture version mismatch.
2173 (aarch64_override_options): Fix parsing order for option strings.
2175 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2176 Iain Sandoe <iain@codesourcery.com>
2179 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2180 warning. Amend comment to reflect current functionality.
2182 2014-01-20 Richard Biener <rguenther@suse.de>
2185 * builtins.c (fold_builtin_strcat): Remove case better handled
2186 by tree-ssa-strlen.c.
2188 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2190 * config/aarch64/aarch64.opt
2191 (mcpu, march, mtune): Make case-insensitive.
2193 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2196 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2197 if operands[1] is a REG or ZERO_EXTEND of a REG.
2199 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2201 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2203 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2205 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2206 long non-pic millicode calls.
2208 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2210 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2212 2014-01-19 Kito Cheng <kito@0xlab.org>
2214 * builtins.c (expand_movstr): Check movstr expand done or fail.
2216 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2217 H.J. Lu <hongjiu.lu@intel.com>
2220 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2221 to DImode for zero-extended addresses.
2223 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2225 PR rtl-optimization/57763
2226 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2227 on the new indirect jump_insn and increment LABEL_NUSES (label).
2229 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2233 * config.gcc (x86_archs): New variable.
2234 (x86_64_archs): Likewise.
2235 (x86_cpus): Likewise.
2236 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2237 --with-arch/--with-cpu= options.
2238 Support --with-arch=/--with-cpu={nehalem,westmere,
2239 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2241 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2243 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2244 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2246 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2248 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2250 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2253 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2254 clear cpp_get_options (parse_in)->warn_unused_macros for
2255 ix86_target_macros_internal with cpp_define.
2257 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2259 * jump.c (delete_related_insns): Keep (use (insn))s.
2260 * reorg.c (redundant_insn): Check for barriers too.
2262 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2264 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2266 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2268 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2269 call to $$dyncall when TARGET_LONG_CALLS is true.
2271 2014-01-17 Jeff Law <law@redhat.com>
2273 * ree.c (combine_set_extension): Temporarily disable test for
2274 changing number of hard registers.
2276 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2279 * ipa-inline-analysis.c (inline_free_summary):
2280 Do not free summary of aliases.
2282 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2285 * gimplify.c (gimplify_expr): Use create_tmp_var
2286 instead of create_tmp_var_raw. If cond doesn't have
2287 integral type, don't add the IFN_ANNOTATE builtin at all.
2289 2014-01-17 Martin Jambor <mjambor@suse.cz>
2292 * ipa-cp.c (prev_edge_clone): New variable.
2293 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2294 Also resize prev_edge_clone vector.
2295 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2296 (ipcp_edge_removal_hook): New function.
2297 (ipcp_driver): Register ipcp_edge_removal_hook.
2299 2014-01-17 Andrew Pinski <apinski@cavium.com>
2300 Steve Ellcey <sellcey@mips.com>
2303 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2306 2014-01-17 Jeff Law <law@redhat.com>
2309 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2310 so that pass_ccp runs first.
2312 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2314 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2315 (ix86_adjust_cost): Use !TARGET_XXX.
2316 (do_reorder_for_imul): Likewise.
2317 (swap_top_of_ready_list): Likewise.
2318 (ix86_sched_reorder): Likewise.
2320 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2322 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2323 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2324 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2325 (intel_memset): New. Duplicate slm_memset.
2326 (intel_cost): New. Duplicate slm_cost.
2327 (m_INTEL): New macro.
2328 (processor_target_table): Add "intel".
2329 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2330 with PROCESSOR_INTEL for "intel".
2331 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2332 PROCESSOR_SILVERMONT.
2333 (ix86_issue_rate): Likewise.
2334 (ix86_adjust_cost): Likewise.
2335 (ia32_multipass_dfa_lookahead): Likewise.
2336 (swap_top_of_ready_list): Likewise.
2337 (ix86_sched_reorder): Likewise.
2338 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2339 instead of TARGET_OPT_AGU.
2340 * config/i386/i386.h (TARGET_INTEL): New.
2341 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2342 (processor_type): Add PROCESSOR_INTEL.
2343 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2344 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2346 2014-01-17 Marek Polacek <polacek@redhat.com>
2349 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2352 2014-01-17 Richard Biener <rguenther@suse.de>
2354 PR tree-optimization/46590
2355 * opts.c (default_options_table): Add entries for
2356 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2357 all enabled at -O1 but not for -Og.
2358 * common.opt (fbranch-count-reg): Remove Init(1).
2359 (fmove-loop-invariants): Likewise.
2360 (ftree-pta): Likewise.
2362 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2364 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2365 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2366 decls to at least the GCC 4.8 used alignments.
2369 * tree-nested.c (convert_nonlocal_reference_stmt,
2370 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2371 of GIMPLE_BIND stmts, adjust associated decls.
2373 2014-01-17 Richard Biener <rguenther@suse.de>
2375 PR tree-optimization/46590
2376 * vec.h (vec<>::bseach): New member function implementing
2377 binary search according to C89 bsearch.
2378 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2379 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2380 bitmap pointer again. Make accesses_in_loop a flat array.
2381 (mem_ref_obstack): New global.
2382 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2383 (mark_ref_stored): Likewise.
2384 (ref_indep_loop_p_2): Likewise.
2385 (set_ref_stored_in_loop): New helper function.
2386 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2387 (memref_free): Adjust.
2388 (record_mem_ref_loc): Simplify.
2389 (gather_mem_refs_stmt): Adjust.
2390 (sort_locs_in_loop_postorder_cmp): New function.
2391 (analyze_memory_references): Sort accesses_in_loop after
2392 loop postorder number.
2393 (find_ref_loc_in_loop_cmp): New function.
2394 (for_all_locs_in_loop): Find relevant cluster of locs in
2395 accesses_in_loop and iterate without recursion.
2396 (execute_sm): Avoid uninit warning.
2397 (struct ref_always_accessed): Simplify.
2398 (ref_always_accessed::operator ()): Likewise.
2399 (ref_always_accessed_p): Likewise.
2400 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2401 loop postorder numbers here.
2402 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2405 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2408 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2409 on decls for which assemble_alias has been called.
2411 2014-01-17 Nick Clifton <nickc@redhat.com>
2413 * config/msp430/msp430.opt: (mcpu): New option.
2414 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2415 (msp430_option_override): Parse target_cpu. If the MCU name
2416 matches a generic string, clear target_mcu.
2417 (msp430_attr): Allow numeric interrupt values up to 63.
2418 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2419 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2421 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2423 * config/msp430/msp430.md (popm): Use %J rather than %I.
2424 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2425 (addhi_cy_i): Use immediate_operand for operand 2.
2426 * doc/invoke.texi: Document -mcpu option.
2428 2014-01-17 Richard Biener <rguenther@suse.de>
2430 PR rtl-optimization/38518
2431 * df.h (df_analyze_loop): Declare.
2432 * df-core.c: Include cfgloop.h.
2433 (df_analyze_1): Split out main part of df_analyze.
2434 (df_analyze): Adjust.
2435 (loop_inverted_post_order_compute): New function.
2436 (loop_post_order_compute): Likewise.
2437 (df_analyze_loop): New function avoiding whole-function
2439 * loop-invariant.c (find_defs): Use df_analyze_loop.
2440 (find_invariants): Adjust.
2441 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2443 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2445 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2446 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2448 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2450 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2451 traversal when removing references.
2453 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2456 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2458 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2461 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2462 pushing a reload for an autoinc when we had previously reloaded an
2463 inner part of the address.
2465 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2467 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2469 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2470 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2471 when not giving up or versioning for alias only because of
2473 (vect_analyze_data_ref_dependences): Set to true.
2474 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2476 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2477 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2481 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2484 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2485 operand 0 predicate for gathers, use a new pseudo as subtarget.
2487 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2490 * lra-constraints.c (process_alt_operands): Add printing debug info.
2491 Check absence of input/output reloads for matched operands too.
2493 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2495 PR rtl-optimization/59835
2496 * ira.c (ira_init_register_move_cost): Increase cost for
2499 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2501 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2503 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2506 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2507 non-register objects. Use gen_(high/low)part more consistently.
2510 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2513 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2514 endian support, remove tests for WORDS_BIG_ENDIAN.
2515 (p8_mfvsrd_3_<mode>): Likewise.
2516 (reload_gpr_from_vsx<mode>): Likewise.
2517 (reload_gpr_from_vsxsf): Likewise.
2518 (p8_mfvsrd_4_disf): Likewise.
2520 2014-01-16 Richard Biener <rguenther@suse.de>
2522 PR rtl-optimization/46590
2523 * lcm.c (compute_antinout_edge): Use postorder iteration.
2524 (compute_laterin): Use inverted postorder iteration.
2526 2014-01-16 Nick Clifton <nickc@redhat.com>
2529 * varasm.c (output_constant): Return the number of bytes actually
2531 (output_constructor_array_range): Update the field size with the
2532 number of bytes emitted by output_constant.
2533 (output_constructor_regular_field): Likewise. Also do not
2534 complain if the total number of bytes emitted is now greater
2535 than the expected fieldpos.
2536 * output.h (output_constant): Update prototype and descriptive comment.
2538 2014-01-16 Marek Polacek <polacek@redhat.com>
2541 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2542 it is error_mark_node.
2544 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2546 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2547 VALID_AVX256_REG_OR_OI_MODE.
2549 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2551 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2552 current procedure should be profiled.
2554 2014-01-15 Andrew Pinski <apinski@cavium.com>
2556 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2557 of moving from/to the STACK_REG register class.
2559 2014-01-15 Richard Henderson <rth@redhat.com>
2562 * reginfo.c (global_regs_decl): Globalize.
2563 * rtl.h (global_regs_decl): Declare.
2564 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2565 reserved via global_regs.
2567 2014-01-15 Teresa Johnson <tejohnson@google.com>
2569 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2571 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2573 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2574 and vmulosh rather than call gen_vec_widen_smult_*.
2575 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2576 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2577 (vec_widen_smult_even_v16qi): Likewise.
2578 (vec_widen_umult_even_v8hi): Likewise.
2579 (vec_widen_smult_even_v8hi): Likewise.
2580 (vec_widen_umult_odd_v16qi): Likewise.
2581 (vec_widen_smult_odd_v16qi): Likewise.
2582 (vec_widen_umult_odd_v8hi): Likewise.
2583 (vec_widen_smult_odd_v8hi): Likewise.
2584 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2585 vmuloub rather than call gen_vec_widen_umult_*.
2586 (vec_widen_umult_lo_v16qi): Likewise.
2587 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2588 vmulosb rather than call gen_vec_widen_smult_*.
2589 (vec_widen_smult_lo_v16qi): Likewise.
2590 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2591 rather than call gen_vec_widen_umult_*.
2592 (vec_widen_umult_lo_v8hi): Likewise.
2593 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2594 rather than call gen_vec_widen_smult_*.
2595 (vec_widen_smult_lo_v8hi): Likewise.
2597 2014-01-15 Jeff Law <law@redhat.com>
2599 PR tree-optimization/59747
2600 * ree.c (find_and_remove_re): Properly handle case where a second
2601 eliminated extension requires widening a copy created for elimination
2602 of a prior extension.
2603 (combine_set_extension): Ensure that the number of hard regs needed
2604 for a destination register does not change when we widen it.
2606 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2608 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2609 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2610 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2611 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2612 (avr-*-rtems*): Likewise.
2613 (bfin*-rtems*): Likewise.
2614 (moxie-*-rtems*): Likewise.
2615 (h8300-*-rtems*): Likewise.
2616 (i[34567]86-*-rtems*): Likewise.
2617 (lm32-*-rtems*): Likewise.
2618 (m32r-*-rtems*): Likewise.
2619 (m68k-*-rtems*): Likewise.
2620 (microblaze*-*-rtems*): Likewise.
2621 (mips*-*-rtems*): Likewise.
2622 (powerpc-*-rtems*): Likewise.
2623 (sh-*-rtems*): Likewise.
2624 (sparc-*-rtems*): Likewise.
2625 (sparc64-*-rtems*): Likewise.
2626 (v850-*-rtems*): Likewise.
2627 (m32c-*-rtems*): Likewise.
2629 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2631 PR rtl-optimization/59511
2632 * ira.c (ira_init_register_move_cost): Use memory costs for some
2633 cases of register move cost calculations.
2634 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2635 instead of BB frequency.
2636 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2637 * lra-assigns.c (find_hard_regno_for): Ditto.
2639 2014-01-15 Richard Biener <rguenther@suse.de>
2641 PR tree-optimization/59822
2642 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2643 (vectorizable_load): Use it to hoist defs of uses of invariant
2644 loads out of the loop.
2646 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2647 Kugan Vivekanandarajah <kuganv@linaro.org>
2650 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2653 2014-01-15 Richard Biener <rguenther@suse.de>
2655 PR rtl-optimization/59802
2656 * lcm.c (compute_available): Use inverted postorder to seed
2657 the initial worklist.
2659 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2662 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2663 ADDR_REGS for invalid symrefs in non-PIC code.
2665 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2668 * builtins.c (determine_block_size): Initialize *probable_max_size
2669 even if len_rtx is CONST_INT.
2671 2014-01-14 Andrew Pinski <apinski@cavium.com>
2673 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2674 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2675 (cortexa53_tunings): Likewise.
2676 (aarch64_sched_issue_rate): New function.
2677 (TARGET_SCHED_ISSUE_RATE): Define.
2679 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2681 * ira-costs.c (find_costs_and_classes): Add missed
2682 ira_init_register_move_cost_if_necessary.
2684 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2687 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2689 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2692 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2693 to indicate if type is used for function return value. Warn ABI
2694 change if the vector mode isn't available for function return value.
2695 (ix86_function_arg_advance): Pass false to type_natural_mode.
2696 (ix86_function_arg): Likewise.
2697 (ix86_gimplify_va_arg): Likewise.
2698 (function_arg_32): Don't warn ABI change.
2699 (ix86_function_value): Pass true to type_natural_mode.
2700 (ix86_return_in_memory): Likewise.
2701 (ix86_struct_value_rtx): Removed.
2702 (TARGET_STRUCT_VALUE_RTX): Likewise.
2704 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2706 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2707 converting a conditional jump into a conditional return.
2709 2014-01-14 Richard Biener <rguenther@suse.de>
2711 PR tree-optimization/58921
2712 PR tree-optimization/59006
2713 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2714 hoisting invariant stmts.
2715 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2716 invariant loads on the preheader edge if possible.
2718 2014-01-14 Joey Ye <joey.ye@arm.com>
2720 * doc/plugin.texi (Building GCC plugins): Update to C++.
2722 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2724 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2725 (_mm_rcp28_round_ss): Ditto.
2726 (_mm_rsqrt28_round_sd): Ditto.
2727 (_mm_rsqrt28_round_ss): Ditto.
2728 (_mm_rcp28_sd): Ditto.
2729 (_mm_rcp28_ss): Ditto.
2730 (_mm_rsqrt28_sd): Ditto.
2731 (_mm_rsqrt28_ss): Ditto.
2732 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2733 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2734 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2735 (IX86_BUILTIN_RCP28SD): Ditto.
2736 (IX86_BUILTIN_RCP28SS): Ditto.
2737 (IX86_BUILTIN_RSQRT28SD): Ditto.
2738 (IX86_BUILTIN_RSQRT28SS): Ditto.
2739 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2740 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2741 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2742 (ix86_expand_special_args_builtin): Expand new FTYPE.
2743 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2744 (srcp14<mode>): Make insn unary.
2745 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2746 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2747 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2748 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2749 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2750 Fix rounding: make it SAE only.
2751 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2753 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2755 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2756 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2757 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2758 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2759 (round_saeonly_mask_scalar_operand4): Ditto.
2760 (round_saeonly_mask_scalar_op3): Ditto.
2761 (round_saeonly_mask_scalar_op4): Ditto.
2763 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2765 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2766 Implement -maltivec=be for vec_insert and vec_extract.
2768 2014-01-10 DJ Delorie <dj@redhat.com>
2770 * config/msp430/msp430.md (call_internal): Don't allow memory
2771 references with SP as the base register.
2772 (call_value_internal): Likewise.
2773 * config/msp430/constraints.md (Yc): New. For memory references
2774 that don't use SP as a base register.
2776 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2777 "an integer without a # prefix"
2778 * config/msp430/msp430.md (epilogue_helper): Use it.
2780 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2783 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2784 AVX512F gather builtins.
2785 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2786 on gather decls with INTEGER_TYPE masktype.
2787 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2788 directly into the builtin rather than hoisting it before loop.
2790 PR tree-optimization/59387
2791 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2792 (scev_const_prop): If folded_casts and type has undefined overflow,
2793 use force_gimple_operand instead of force_gimple_operand_gsi and
2794 for each added stmt if it is assign with
2795 arith_code_with_undefined_signed_overflow, call
2796 rewrite_to_defined_overflow.
2797 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2798 gimple-fold.h instead.
2799 (arith_code_with_undefined_signed_overflow,
2800 rewrite_to_defined_overflow): Moved to ...
2801 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2802 rewrite_to_defined_overflow): ... here. No longer static.
2803 Include gimplify-me.h.
2804 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2805 rewrite_to_defined_overflow): New prototypes.
2807 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2809 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2811 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2813 * builtins.c (get_object_alignment_2): Minor tweak.
2814 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2816 2014-01-13 Christian Bruel <christian.bruel@st.com>
2818 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2819 optimized non constant lengths.
2821 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2824 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2825 load as __atomic_load_N if possible.
2827 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2829 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2831 (rs6000_expand_builtin): Adjust call.
2833 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2836 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2837 * config/rs6000/rs6000.c: Include target-globals.h.
2838 (rs6000_set_current_function): Instead of doing target_reinit
2839 unconditionally, use save_target_globals_default_opts and
2840 restore_target_globals.
2842 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2844 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2845 (rs6000_expand_builtin): Handle mffs and mtfsf.
2846 (rs6000_init_builtins): Define mffs and mtfsf.
2847 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2848 (rs6000_mffs): New pattern.
2849 (rs6000_mtfsf): New pattern.
2851 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2853 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2854 Start narrowing with START. Apply candidate-use pair
2855 and check overall cost in narrowing.
2856 (iv_ca_prune): Pass new argument.
2858 2014-01-10 Jeff Law <law@redhat.com>
2861 * ree.c (combine_reaching_defs): Ensure the defining statement
2862 occurs before the extension when optimizing extensions with
2863 different source and destination hard registers.
2865 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2868 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2869 vtables into the type inheritance graph.
2871 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2873 PR rtl-optimization/59754
2874 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2875 modes in the REGNO != REGNO case.
2877 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2879 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2881 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2883 PR tree-optimization/59745
2884 * tree-predcom.c (tree_predictive_commoning_loop): Call
2885 free_affine_expand_cache if giving up because components is NULL.
2887 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2888 GC in payload of target_globals struct instead of allocating them on
2889 the heap and the larger structs separately using GC.
2890 * target-globals.h (struct target_globals): Make regs, hard_regs,
2891 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2892 of GTY((skip)) and change type to void *.
2893 (reset_target_globals): Cast loads from those fields to corresponding
2896 2014-01-10 Steve Ellcey <sellcey@mips.com>
2899 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2900 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2901 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2903 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2906 * aarch64-modes.def (CC_Zmode): New flags mode.
2907 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2908 represents an equality.
2909 (aarch64_get_condition_code): Handle CC_Zmode.
2910 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2912 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2914 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2915 extraction in good case.
2917 2014-01-10 Richard Biener <rguenther@suse.de>
2919 PR tree-optimization/59374
2920 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2921 checking after SLP discovery. Mark stmts not participating
2922 in any SLP instance properly.
2924 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2926 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2927 when handling a SET rtx.
2929 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2931 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2932 (cortex-a57): Likewise.
2933 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2935 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2937 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2938 non-iwmmxt builtins.
2940 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2944 * ipa-devirt.c record_target_from_binfo): Take as argument
2945 stack of binfos and lookup matching one for virtual inheritance.
2946 (possible_polymorphic_call_targets_1): Update.
2948 2014-01-10 Huacai Chen <chenhc@lemote.com>
2950 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2951 kernel strings for Loongson-2E/2F/3A.
2953 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2956 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2957 is_gimple_call before calling gimple_call_internal_p.
2959 2014-01-09 Steve Ellcey <sellcey@mips.com>
2961 * Makefile.in (TREE_FLOW_H): Remove.
2962 (TREE_SSA_H): Add file names from tree-flow.h.
2963 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2964 * tree.h: Remove tree-flow.h reference.
2965 * hash-table.h: Remove tree-flow.h reference.
2966 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
2967 reference with tree-ssa-loop.h.
2969 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2971 * doc/invoke.texi: Add -maltivec={be,le} options, and document
2972 default element-order behavior for -maltivec.
2973 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
2974 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
2975 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
2976 when targeting big endian, at least for now.
2977 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
2979 2014-01-09 Jakub Jelinek <jakub@redhat.com>
2982 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
2983 var satisfies use_register_for_decl, just take into account type
2984 alignment, rather than decl alignment.
2986 PR tree-optimization/59622
2987 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
2988 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
2989 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
2990 Don't devirtualize for inplace at all. For targets.length () == 1,
2991 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
2993 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2995 * config/i386/i386.md (cpu): Remove the unused btver1.
2997 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
2999 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3001 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3004 * tree-core.h (struct target_globals): New forward declaration.
3005 (struct tree_target_option): Add globals field.
3006 * tree.h (TREE_TARGET_GLOBALS): Define.
3007 (prepare_target_option_nodes_for_pch): New prototype.
3008 * target-globals.h (struct target_globals): Define even if
3010 * tree.c (prepare_target_option_node_for_pch,
3011 prepare_target_option_nodes_for_pch): New functions.
3012 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3013 * config/i386/i386.c: Include target-globals.h.
3014 (ix86_set_current_function): Instead of doing target_reinit
3015 unconditionally, use save_target_globals_default_opts and
3016 restore_target_globals.
3018 2014-01-09 Richard Biener <rguenther@suse.de>
3020 PR tree-optimization/59715
3021 * tree-cfg.h (split_critical_edges): Declare.
3022 * tree-cfg.c (split_critical_edges): Export.
3023 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3025 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3027 * cfgexpand.c (expand_stack_vars): Optionally disable
3028 asan stack protection.
3029 (expand_used_vars): Likewise.
3030 (partition_stack_vars): Likewise.
3031 * asan.c (asan_emit_stack_protection): Optionally disable
3032 after return stack usage.
3033 (instrument_derefs): Optionally disable memory access instrumentation.
3034 (instrument_builtin_call): Likewise.
3035 (instrument_strlen_call): Likewise.
3036 (asan_protect_global): Optionally disable global variables protection.
3037 * doc/invoke.texi: Added doc for new options.
3038 * params.def: Added new options.
3039 * params.h: Likewise.
3041 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3043 PR rtl-optimization/59724
3044 * ifcvt.c (cond_exec_process_if_block): Don't call
3045 flow_find_head_matching_sequence with 0 longest_match.
3046 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3047 non-active insns if !stop_after.
3048 (try_head_merge_bb): Revert 2014-01-07 changes.
3050 2014-01-08 Jeff Law <law@redhat.com>
3052 * ree.c (get_sub_rtx): New function, extracted from...
3053 (merge_def_and_ext): Here.
3054 (combine_reaching_defs): Use get_sub_rtx.
3056 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3058 * cgraph.h (varpool_variable_node): Do not choke on null node.
3060 2014-01-08 Catherine Moore <clm@codesourcery.com>
3062 * config/mips/mips.md (simple_return): Attempt to use JRC
3064 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3066 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3068 PR rtl-optimization/59137
3069 * reorg.c (steal_delay_list_from_target): Call update_block for
3071 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3073 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3075 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3076 two duplicate entries.
3078 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3081 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3083 * config/mips/mips.c (mips_truncated_op_cost): New function.
3084 (mips_rtx_costs): Adjust test for BADDU.
3085 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3087 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3089 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3090 (*baddu_si): ...this new pattern.
3092 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3095 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3097 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3100 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3102 (expand_expr, expand_normal): Adjust.
3103 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3104 inner_reference_p. Use inner_reference_p to expand inner references.
3105 (store_expr): Adjust.
3106 * cfgexpand.c (expand_call_stmt): Adjust.
3108 2014-01-08 Rong Xu <xur@google.com>
3110 * gcov-io.c (gcov_var): Move from gcov-io.h.
3111 (gcov_position): Ditto.
3112 (gcov_is_error): Ditto.
3113 (gcov_rewrite): Ditto.
3114 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3115 only part to libgcc/libgcov.h.
3117 2014-01-08 Marek Polacek <polacek@redhat.com>
3120 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3122 2014-01-08 Marek Polacek <polacek@redhat.com>
3125 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3127 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3129 PR rtl-optimization/59649
3130 * stor-layout.c (get_mode_bounds): For BImode return
3131 0 and STORE_FLAG_VALUE.
3133 2014-01-08 Richard Biener <rguenther@suse.de>
3136 * gimple.h (is_gimple_builtin_call): Remove.
3137 (gimple_builtin_call_types_compatible_p): New.
3138 (gimple_call_builtin_p): New overload.
3139 * gimple.c (is_gimple_builtin_call): Remove.
3140 (validate_call): Rename to ...
3141 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3143 (validate_type): New static function.
3144 (gimple_call_builtin_p): New overload and adjust.
3145 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3146 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3147 (gimple_fold_stmt_to_constant_1): Likewise.
3148 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3150 2014-01-08 Richard Biener <rguenther@suse.de>
3153 * gimplify.c (gimplify_expr): Gimplify register-register type
3154 VIEW_CONVERT_EXPRs to separate stmts.
3156 2014-01-07 Jeff Law <law@redhat.com>
3159 * ree.c (combine_set_extension): Handle case where source
3160 and destination registers in an extension insn are different.
3161 (combine_reaching_defs): Allow source and destination registers
3162 in extension to be different under limited circumstances.
3163 (add_removable_extension): Remove restriction that the
3164 source and destination registers in the extension are the same.
3165 (find_and_remove_re): Emit a copy from the extension's
3166 destination to its source after the defining insn if
3167 the source and destination registers are different.
3170 * ifcvt.c (merge_if_block): If we are merging a block with more than
3171 one successor with a block with no successors, remove any BARRIER
3172 after the second block.
3174 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3176 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3178 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3181 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3182 for 14-bit register offsets when INT14_OK_STRICT is false.
3184 2014-01-07 Roland Stigge <stigge@antcom.de>
3185 Michael Meissner <meissner@linux.vnet.ibm.com>
3188 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3189 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3191 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3193 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3196 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3198 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3199 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3200 rtx is const0_rtx or not.
3202 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3205 * target-globals.c (save_target_globals): Remove this_fn_optab
3207 * toplev.c: Include optabs.h.
3208 (target_reinit): Temporarily restore the global options if another
3209 set of options are in force.
3211 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3213 PR rtl-optimization/58668
3214 * cfgcleanup.c (flow_find_cross_jump): Don't count
3215 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3216 to determine what is counted.
3217 (flow_find_head_matching_sequence): Use active_insn_p to determine
3219 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3221 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3222 determine what is counted.
3224 PR tree-optimization/59643
3225 * tree-predcom.c (split_data_refs_to_components): If one dr is
3226 read and one write, determine_offset fails and the write isn't
3227 in the bad component, just put the read into the bad component.
3229 2014-01-07 Mike Stump <mikestump@comcast.net>
3230 Jakub Jelinek <jakub@redhat.com>
3233 * tree-core.h (struct tree_optimization_option): Change optabs
3234 type from unsigned char * to void *.
3235 * optabs.c (init_tree_optimization_optabs): Adjust
3236 TREE_OPTIMIZATION_OPTABS initialization.
3238 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3241 * config/i386/i386.h (struct machine_function): Add
3242 no_drap_save_restore field.
3243 * config/i386/i386.c (ix86_save_reg): Use
3244 !cfun->machine->no_drap_save_restore instead of
3245 crtl->stack_realign_needed.
3246 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3247 this function clears frame_pointer_needed. Set
3248 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3249 and DRAP reg is needed.
3251 2014-01-06 Marek Polacek <polacek@redhat.com>
3254 * doc/implement-c.texi: Mention that other integer types are
3255 permitted as bit-field types in strictly conforming mode.
3257 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3259 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3262 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3264 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3266 2014-01-06 Martin Jambor <mjambor@suse.cz>
3269 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3271 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3273 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3277 * var-tracking.c (add_stores): Preserve the value of the source even if
3278 we don't record the store.
3280 2014-01-06 Terry Guo <terry.guo@arm.com>
3282 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3284 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3287 * config/darwin.c (darwin_function_section): Adjust return values to
3288 correspond to optimisation changes made in r206070.
3290 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3292 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3293 from prefetch_block tune setting.
3294 (nocona_cost): Correct size of prefetch block to 64.
3296 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3298 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3299 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3300 used to save the static chain register in the computation of the offset
3301 from which the FP registers need to be restored.
3303 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3305 PR tree-optimization/59519
3306 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3307 ICE if get_current_def (current_new_name) is already non-NULL, as long
3308 as it is a phi result of some other phi in *new_exit_bb that has
3311 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3312 or vmovdqu* for misaligned_operand.
3313 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3314 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3315 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3316 aligned_mem for AVX512F masked aligned load and store builtins and for
3319 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3321 PR tree-optimization/59651
3322 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3323 Address range for negative step should be added by TYPE_SIZE_UNIT.
3325 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3327 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3328 overlapping registers also for registers other than the stack pointer.
3330 2014-01-03 Marek Polacek <polacek@redhat.com>
3333 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3336 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3339 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3342 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3343 (push splitter): Use <P:MODE_SIZE> instead of
3344 GET_MODE_SIZE (<P:MODE>mode).
3345 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3346 (mov -1, reg peephole2): Likewise.
3347 * config/i386/sse.md (*mov<mode>_internal,
3348 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3349 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3350 *<code><mode>3, *andnot<mode>3<mask_name>,
3351 <mask_codefor><code><mode>3<mask_name>): Likewise.
3352 * config/i386/subst.md (mask_mode512bit_condition,
3353 sd_mask_mode512bit_condition): Likewise.
3355 2014-01-02 Xinliang David Li <davidxl@google.com>
3357 PR tree-optimization/59303
3358 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3359 (dump_predicates): Better output format.
3360 (pred_equal_p): New function.
3361 (is_neq_relop_p): Ditto.
3362 (is_neq_zero_form_p): Ditto.
3363 (pred_expr_equal_p): Ditto.
3364 (pred_neg_p): Ditto.
3365 (simplify_pred): Ditto.
3366 (simplify_preds_2): Ditto.
3367 (simplify_preds_3): Ditto.
3368 (simplify_preds_4): Ditto.
3369 (simplify_preds): Ditto.
3371 (push_to_worklist): Ditto.
3372 (get_pred_info_from_cmp): Ditto.
3373 (is_degenerated_phi): Ditto.
3374 (normalize_one_pred_1): Ditto.
3375 (normalize_one_pred): Ditto.
3376 (normalize_one_pred_chain): Ditto.
3377 (normalize_preds): Ditto.
3378 (normalize_cond_1): Remove function.
3379 (normalize_cond): Ditto.
3380 (is_gcond_subset_of): Ditto.
3381 (is_subset_of_any): Ditto.
3382 (is_or_set_subset_of): Ditto.
3383 (is_and_set_subset_of): Ditto.
3384 (is_norm_cond_subset_of): Ditto.
3385 (pred_chain_length_cmp): Ditto.
3386 (convert_control_dep_chain_into_preds): Type change.
3387 (find_predicates): Ditto.
3388 (find_def_preds): Ditto.
3389 (destroy_predicates_vecs): Ditto.
3390 (find_matching_predicates_in_rest_chains): Ditto.
3391 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3392 (is_pred_expr_subset): Ditto.
3393 (is_pred_chain_subset_of): Ditto.
3394 (is_included_in): Ditto.
3395 (is_superset_of): Ditto.
3397 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3399 Update copyright years.
3401 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3403 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3404 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3405 config/arc/arc.md, config/arc/arc.opt,
3406 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3407 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3408 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3409 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3410 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3411 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3412 vtable-verify.c, vtable-verify.h: Use the standard form for the
3415 2014-01-02 Tobias Burnus <burnus@net-b.de>
3417 * gcc.c (process_command): Update copyright notice dates.
3418 * gcov-dump.c: Ditto.
3420 * doc/cpp.texi: Bump @copying's copyright year.
3421 * doc/cppinternals.texi: Ditto.
3422 * doc/gcc.texi: Ditto.
3423 * doc/gccint.texi: Ditto.
3424 * doc/gcov.texi: Ditto.
3425 * doc/install.texi: Ditto.
3426 * doc/invoke.texi: Ditto.
3428 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3430 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3432 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3434 * config/i386/sse.md (*mov<mode>_internal): Guard
3435 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3437 PR rtl-optimization/59647
3438 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3439 new_rtx into UNSIGNED_FLOAT rtxes.
3441 Copyright (C) 2014 Free Software Foundation, Inc.
3443 Copying and distribution of this file, with or without modification,
3444 are permitted in any medium without royalty provided the copyright
3445 notice and this notice are preserved.