1 2014-02-14 Jakub Jelinek <jakub@redhat.com>
3 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
5 (get_initial_def_for_induction, vectorizable_induction): Ignore
6 debug stmts when looking for exit_phi.
7 (vectorizable_live_operation): Fix up condition.
9 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
11 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
12 nreverse() because it changes the content of original tree list.
14 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
16 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
17 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
19 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
21 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
24 2014-02-13 Jakub Jelinek <jakub@redhat.com>
27 * dwarf2out.c (gen_subprogram_die): Don't call
28 add_calling_convention_attribute if subr_die is old_die.
30 2014-02-13 Sharad Singhai <singhai@google.com>
32 * doc/optinfo.texi: Fix order of nodes.
34 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
36 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
37 operands[2], not operands[3].
39 2014-02-13 Richard Biener <rguenther@suse.de>
42 * doc/install.texi (ISL): Update recommended version to 0.12.2,
43 mention the possibility of an in-tree build.
44 (CLooG): Update recommended version to 0.18.1, mention the
45 possibility of an in-tree build and clarify that the ISL
46 bundled with CLooG does not work.
48 2014-02-13 Jakub Jelinek <jakub@redhat.com>
51 * expr.c (compress_float_constant): If x is a hard register,
52 extend into a pseudo and then move to x.
54 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
56 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
57 caused by bad second argument to warning_at() with -mhotpatch and
58 nested functions (e.g. with gfortran).
60 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
62 * opts.c (option_name): Remove "enabled by default" rider.
64 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
66 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
68 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
69 Uros Bizjak <ubizjak@gmail.com>
72 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
73 * configure: Regenerated.
75 2014-02-12 Richard Biener <rguenther@suse.de>
77 * vec.c (vec_prefix::calculate_allocation): Move as
78 inline variant to vec.h.
79 (vec_prefix::calculate_allocation_1): New out-of-line version.
80 * vec.h (vec_prefix::calculate_allocation_1): Declare.
81 (vec_prefix::m_has_auto_buf): Rename to ...
82 (vec_prefix::m_using_auto_storage): ... this.
83 (vec_prefix::calculate_allocation): Inline the easy cases
84 and dispatch to calculate_allocation_1 which doesn't need the
86 (va_heap::reserve): Use gcc_checking_assert.
87 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
89 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
91 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
92 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
93 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
95 2014-02-12 Richard Biener <rguenther@suse.de>
97 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
98 when we found a dependence.
100 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
102 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
104 (maybe_fold_stmt): ... into this new function.
105 * omp-low.c (lower_omp): Update comment.
107 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
110 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
113 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
115 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
116 identifiers in comments.
117 (cortexa53_extra_costs): Likewise.
118 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
119 (cortexa7_extra_costs): Likewise.
120 (cortexa12_extra_costs): Likewise.
121 (cortexa15_extra_costs): Likewise.
122 (v7m_extra_costs): Likewise.
124 2014-02-12 Richard Biener <rguenther@suse.de>
127 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
128 of posix_memalign being successful.
129 (lower_stmt): Restrict lowering of posix_memalign to when
130 -ftree-bit-ccp is enabled.
132 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
134 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
136 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
138 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
140 PR rtl-optimization/60116
141 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
142 other_insn once the combination has been validated.
144 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
147 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
149 * ipa-devirt.c: Include demangle.h
150 (odr_violation_reported): New static variable.
151 (add_type_duplicate): Update odr_violations.
152 (maybe_record_node): Add completep parameter; update it.
153 (record_target_from_binfo): Add COMPLETEP parameter;
155 (possible_polymorphic_call_targets_1): Likewise.
156 (struct polymorphic_call_target_d): Add nonconstruction_targets;
157 rename FINAL to COMPLETE.
158 (record_targets_from_bases): Sanity check we found the binfo;
159 fix COMPLETEP updating.
160 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
161 parameter, fix computing of COMPLETEP.
162 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
163 at LTO time do demangling.
164 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
165 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
167 (gimple_get_virt_method_for_binfo): Likewise.
168 * gimple-fold.h (gimple_get_virt_method_for_binfo,
169 gimple_get_virt_method_for_vtable): Update prototypes.
171 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
174 * genautomata.c (add_presence_absence): Fix typo with
175 {final_}presence_list.
177 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
180 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
181 for VSX/Altivec vectors that land in GPR registers.
183 2014-02-11 Richard Henderson <rth@redhat.com>
184 Jakub Jelinek <jakub@redhat.com>
187 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
188 around drhs if type conversion to lacc->type is not useless.
190 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
192 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
194 (cortex-a57.cortex-a53): Likewise.
195 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
197 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
199 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
202 2014-02-11 Renlin Li <Renlin.Li@arm.com>
204 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
205 add_options_for_arm_vfp3.
207 2014-02-11 Jeff Law <law@redhat.com>
210 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
211 object with an undesirable mode.
213 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
216 * config/i386/sol2-9.h: New file.
217 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
218 *-*-solaris2.9*): Use it.
220 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
222 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
223 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
225 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
227 * config/microblaze/microblaze.c: Extend mcpu version format
229 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
231 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
233 2014-02-10 Richard Henderson <rth@redhat.com>
236 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
237 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
238 ms-abi vs -mno-accumulate-outgoing-args.
239 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
240 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
243 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
246 * cfgexpand.c (expand_asm_operands): Attach source location to
247 ASM_INPUT rtx objects.
248 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
250 2014-02-10 Nick Clifton <nickc@redhat.com>
252 * config/mn10300/mn10300.c (popcount): New function.
253 (mn10300_expand_prologue): Include saved registers in stack usage
256 2014-02-10 Jeff Law <law@redhat.com>
259 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
260 when changing the SET_DEST of a prior insn to avoid an input reload.
262 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
264 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
265 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
266 -mcall-openbsd, or -mcall-linux.
267 (CC1_ENDIAN_BIG_SPEC): Remove.
268 (CC1_ENDIAN_LITTLE_SPEC): Remove.
269 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
270 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
271 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
272 and %cc1_endian_default.
273 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
275 2014-02-10 Richard Biener <rguenther@suse.de>
277 PR tree-optimization/60115
278 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
279 MEM_REF handling. Properly verify that the accesses are not
280 out of the objects bound.
282 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
284 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
287 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
289 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
290 proper constants and fix formatting.
291 (possible_polymorphic_call_targets): Fix formatting.
293 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
294 Ilya Tocar <ilya.tocar@intel.com>
296 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
297 (_mm512_loadu_epi32): Renamed into...
298 (_mm512_loadu_si512): This.
299 (_mm512_storeu_epi32): Renamed into...
300 (_mm512_storeu_si512): This.
301 (_mm512_maskz_ceil_ps): Removed.
302 (_mm512_maskz_ceil_pd): Ditto.
303 (_mm512_maskz_floor_ps): Ditto.
304 (_mm512_maskz_floor_pd): Ditto.
305 (_mm512_floor_round_ps): Ditto.
306 (_mm512_floor_round_pd): Ditto.
307 (_mm512_ceil_round_ps): Ditto.
308 (_mm512_ceil_round_pd): Ditto.
309 (_mm512_mask_floor_round_ps): Ditto.
310 (_mm512_mask_floor_round_pd): Ditto.
311 (_mm512_mask_ceil_round_ps): Ditto.
312 (_mm512_mask_ceil_round_pd): Ditto.
313 (_mm512_maskz_floor_round_ps): Ditto.
314 (_mm512_maskz_floor_round_pd): Ditto.
315 (_mm512_maskz_ceil_round_ps): Ditto.
316 (_mm512_maskz_ceil_round_pd): Ditto.
317 (_mm512_expand_pd): Ditto.
318 (_mm512_expand_ps): Ditto.
319 * config/i386/i386.c (ix86_builtins): Remove
320 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
322 * config/i386/predicates.md (const1256_operand): New.
323 (const_1_to_2_operand): Ditto.
324 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
325 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
326 (*avx512pf_gatherpf<mode>sf): Ditto.
327 (avx512pf_gatherpf<mode>df): Ditto.
328 (*avx512pf_gatherpf<mode>df_mask): Ditto.
329 (*avx512pf_gatherpf<mode>df): Ditto.
330 (avx512pf_scatterpf<mode>sf): Ditto.
331 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
332 (*avx512pf_scatterpf<mode>sf): Ditto.
333 (avx512pf_scatterpf<mode>df): Ditto.
334 (*avx512pf_scatterpf<mode>df_mask): Ditto.
335 (*avx512pf_scatterpf<mode>df): Ditto.
336 (avx512f_expand<mode>): Removed.
337 (<shift_insn><mode>3<mask_name>): Change predicate type.
339 2014-02-08 Jakub Jelinek <jakub@redhat.com>
341 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
342 not at the end of datarefs vector use ordered_remove to avoid
343 reordering datarefs vector.
346 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
347 mark local addressable non-static vars as GOVD_PRIVATE
348 instead of GOVD_LOCAL.
349 * omp-low.c (lower_omp_for): Move gimple_bind_vars
350 and BLOCK_VARS of gimple_bind_block to new_stmt rather
354 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
355 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
356 assume_aligned or alloc_align attributes.
357 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
358 arguments. Handle also assume_aligned and alloc_align attributes.
359 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
360 calls to functions with assume_aligned or alloc_align attributes.
361 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
363 2014-02-08 Terry Guo <terry.guo@arm.com>
365 * doc/invoke.texi: Document ARM -march=armv7e-m.
367 2014-02-08 Jakub Jelinek <jakub@redhat.com>
369 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
370 flag on __cilkrts_rethrow builtin.
373 * ipa-cp.c (determine_versionability): Fail at -O0
374 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
375 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
378 2014-02-04 Jakub Jelinek <jakub@redhat.com>
381 * tree-inline.c (copy_forbidden): Fail for
382 __attribute__((optimize (0))) functions.
384 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
386 * varpool.c: Include pointer-set.h.
387 (varpool_remove_unreferenced_decls): Variables in other partitions
388 will not be output; be however careful to not lose information
391 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
393 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
394 lookup in the vtable constructor.
396 2014-02-07 Jeff Law <law@redhat.com>
399 * config/m68k/m68k.md (ashldi_extsi): Turn into a
400 define_insn_and_split.
402 * ipa-inline.c (inline_small_functions): Fix typos.
404 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
406 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
407 (s390_can_use_return_insn): Declare.
408 * config/s390/s390.h (EPILOGUE_USES): Define.
409 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
411 (s390_chunkify_start): Handle return JUMP_LABELs.
412 (s390_early_mach): Emit a main_pool instruction on the entry edge.
413 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
414 (s390_can_use_return_insn): New functions.
415 (s390_fix_long_loop_prediction): Handle conditional returns.
416 (TARGET_SET_UP_BY_PROLOGUE): Define.
417 * config/s390/s390.md (ANY_RETURN): New code iterator.
418 (*creturn, *csimple_return, return, simple_return): New patterns.
420 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
422 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
423 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
424 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
425 REG_CFA_RESTORE list when deciding not to restore a register.
427 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
429 * config/s390/s390.c: Include tree-pass.h and context.h.
430 (s390_early_mach): New function, split out from...
431 (s390_emit_prologue): ...here.
432 (pass_data_s390_early_mach): New pass structure.
433 (pass_s390_early_mach): New class.
434 (s390_option_override): Create and register early_mach pass.
437 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
439 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
440 to match for the exit block.
442 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
444 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
445 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
446 Reject misaligned operands.
448 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
450 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
452 2014-02-07 Richard Biener <rguenther@suse.de>
455 * gimple-low.c (lower_builtin_posix_memalign): New function.
456 (lower_stmt): Call it to lower posix_memalign in a way
457 to make alignment info accessible.
459 2014-02-07 Jakub Jelinek <jakub@redhat.com>
462 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
463 __builtin_setjmp_receiver.
465 2014-02-07 Richard Biener <rguenther@suse.de>
468 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
469 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
470 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
471 Handle BUILT_IN_POSIX_MEMALIGN.
472 (find_func_clobbers): Likewise.
473 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
474 (call_may_clobber_ref_p_1): Likewise.
476 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
479 * ipa-devirt.c (record_target_from_binfo): Remove overactive
482 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
485 * lto-cgraph.c (lto_output_node): Use
486 symtab_get_symbol_partitioning_class.
487 (lto_output_varpool_node): likewise.
488 (symtab_get_symbol_partitioning_class): Move here from
490 * cgraph.h (symbol_partitioning_class): Likewise.
491 (symtab_get_symbol_partitioning_class): Declare.
493 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
495 * ggc.h (ggc_internal_cleared_alloc): New macro.
496 * vec.h (vec_safe_copy): Handle memory stats.
497 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
498 * target-globals.c (save_target_globals): Likewise.
500 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
503 * expr.c (emit_move_resolve_push): Export; be bit more selective
504 on when to clear alias set.
505 * expr.h (emit_move_resolve_push): Declare.
506 * function.h (struct function): Add tail_call_marked.
507 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
508 * config/i386/i386-protos.h (ix86_expand_push): Remove.
509 * config/i386/i386.md (TImode move expander): De not call
511 (FP push expanders): Preserve memory attributes.
512 * config/i386/sse.md (push<mode>1): Remove.
513 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
514 (ix86_expand_push): Remove.
515 * config/i386/mmx.md (push<mode>1): Remove.
517 2014-02-06 Jakub Jelinek <jakub@redhat.com>
519 PR rtl-optimization/60030
520 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
521 lopart with paradoxical subreg before shifting it up by hprec.
523 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
525 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
526 Remove extra newline at end of file.
527 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
528 (arm_issue_rate): Handle cortexa57.
529 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
530 (cortex-a57.cortex-a53): Likewise.
532 2014-02-06 Jakub Jelinek <jakub@redhat.com>
535 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
536 don't record in REG_FRAME_RELATED_EXPR registers not set in that
538 (arm_expand_prologue): Adjust all callers.
539 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
540 info, registers also at the lowest numbered registers side. Use
541 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
545 * var-tracking.c (adjust_mems): Before adding a SET to
546 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
548 2014-02-06 Alan Modra <amodra@gmail.com>
551 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
552 change SDmode to DDmode when lra_in_progress.
554 2014-02-06 Jakub Jelinek <jakub@redhat.com>
557 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
558 free_data_ref on the dr first, and before goto again also set dr
559 to the next dr. For simd_lane_access, free old datarefs[i] before
560 overwriting it. For get_vectype_for_scalar_type failure, don't
561 free_data_ref if simd_lane_access.
563 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
566 * tree.h (opts_for_fn): New inline function.
567 (opt_for_fn): Define.
568 * config/i386/i386.c (ix86_function_regparm): Use
569 opt_for_fn (decl, optimize) instead of optimize.
571 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
573 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
574 for SYMBOL_REF in large memory model.
576 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
578 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
580 (cortex-a57): Likewise.
581 (cortex-a57.cortex-a53): Likewise.
583 2014-02-06 Yury Gribov <y.gribov@samsung.com>
584 Kugan Vivekanandarajah <kuganv@linaro.org>
586 * config/arm/arm.c (arm_vector_alignment_reachable): Check
588 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
590 2014-02-06 Richard Biener <rguenther@suse.de>
592 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
593 set_loop_copy and initialize_original_copy_tables.
595 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
597 * config/aarch64/aarch64-simd.md
598 (aarch64_ashr_simddi): Change QI to SI.
600 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
601 Jakub Jelinek <jakub@redhat.com>
604 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
607 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
609 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
610 CODE_FOR_altivec_vpku[hw]um to
611 CODE_FOR_altivec_vpku[hw]um_direct.
612 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
613 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
614 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
615 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
617 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
619 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
620 generation for -maltivec=be.
621 (altivec_vsumsws): Simplify redundant test.
623 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
625 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
626 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
627 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
628 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
630 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
632 (altivec_vpks<VI_char>ss): Likewise.
633 (altivec_vpks<VI_char>us): Likewise.
634 (altivec_vpku<VI_char>us): Likewise.
635 (altivec_vpku<VI_char>um): Likewise.
636 (altivec_vpku<VI_char>um_direct): New (copy of
637 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
639 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
640 target is little endian and -maltivec=be is not specified.
641 (*altivec_vupkhs<VU_char>_direct): New (copy of
642 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
643 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
644 target is little endian and -maltivec=be is not specified.
645 (*altivec_vupkls<VU_char>_direct): New (copy of
646 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
647 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
648 little endian and -maltivec=be is not specified.
649 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
650 little endian and -maltivec=be is not specified.
652 2014-02-05 Richard Henderson <rth@redhat.com>
655 * combine-stack-adj.c: Revert r206943.
656 * sched-int.h (struct deps_desc): Add last_args_size.
657 * sched-deps.c (init_deps): Initialize it.
658 (sched_analyze_insn): Add OUTPUT dependencies between insns that
659 contain REG_ARGS_SIZE notes.
661 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
663 * lto-cgraph.c (asm_nodes_output): Make global.
664 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
665 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
666 (driver_handle_option): Handle OPT_fwpa.
668 2014-02-05 Jakub Jelinek <jakub@redhat.com>
671 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
672 a comment typo and formatting issue. If odr_hash hasn't been
673 created, return vNULL and set *completep to false.
676 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
677 bb with no successors.
679 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
682 * doc/invoke.texi (-march): Clarify documentation for ARM.
686 2014-02-05 Richard Biener <rguenther@suse.de>
688 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
689 when not vectorizing because of too many alias checks.
690 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
691 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
693 2014-02-05 Nick Clifton <nickc@redhat.com>
695 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
696 accept extended registers in any mode when compiling for the MN10300.
698 2014-02-05 Yury Gribov <y.gribov@samsung.com>
700 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
701 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
702 sanitization attributes.
703 (can_inline_edge_p): Likewise.
704 (sanitize_attrs_match_for_inline_p): New function.
706 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
708 * ipa-prop.c (detect_type_change): Shor circuit testing of
709 type changes on THIS pointer.
711 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
714 * config/pa/pa.c (legitimize_tls_address): Return original address
715 if not passed a SYMBOL_REF rtx.
716 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
718 (pa_emit_move_sequence): Simplify TLS source operands.
719 (pa_legitimate_constant_p): Reject all TLS constants.
720 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
721 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
723 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
725 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
726 groups when we know they are controlled by LTO.
727 * varasm.c (default_binds_local_p_1): If object is in other partition,
728 it will be resolved locally.
730 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
732 * config/host-linux.c (linux_gt_pch_use_address): Don't
733 use SSIZE_MAX because it is not always defined.
735 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
738 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
739 threshold for pseudo splitting.
740 (update_ebb_live_info): Process call argument hard registers and
741 hard registers from insn definition too.
742 (max_small_class_regs_num): New constant.
743 (inherit_in_ebb): Update live hard regs through EBBs. Update
744 reloads_num only for small register classes. Don't split for
747 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
750 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
753 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
755 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
758 2014-02-04 Marek Polacek <polacek@redhat.com>
760 * gdbinit.in (pel): Define.
762 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
764 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
767 2014-02-04 Richard Biener <rguenther@suse.de>
770 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
771 in function context local.
772 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
773 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
774 similar to LTO_imported_decl_ref.
776 2014-02-04 Jakub Jelinek <jakub@redhat.com>
778 PR tree-optimization/60002
779 * cgraphclones.c (build_function_decl_skip_args): Clear
782 PR tree-optimization/60023
783 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
784 false to gsi_replace.
785 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
786 has been in some EH region and vec_stmt could throw, add
787 vec_stmt into the same EH region.
788 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
789 has no lhs, ignore it.
790 * internal-fn.c (expand_MASK_LOAD): Likewise.
793 * tree-inline.c (copy_forbidden): Fail for
794 __attribute__((optimize (0))) functions.
797 * omp-low.c (simd_clone_struct_copy): If from->inbranch
798 is set, copy one less argument.
799 (expand_simd_clones): Don't subtract clone_info->inbranch
800 from simd_clone_struct_alloc argument.
802 PR rtl-optimization/57915
803 * recog.c (simplify_while_replacing): If all unary/binary/relational
804 operation arguments are constant, attempt to simplify those.
807 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
808 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
810 2014-02-04 Richard Biener <rguenther@suse.de>
812 PR tree-optimization/60012
813 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
814 TBAA disambiguation to all DDRs.
816 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
819 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
820 (LINK_SPEC): Use it for -shared, -shared-libgcc.
822 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
825 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
827 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
829 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
830 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
832 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
835 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
836 to figure out targets of polymorphic calls with known decl.
837 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
838 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
839 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
840 (get_polymorphic_call_info): ... here.
841 (get_polymorphic_call_info_from_invariant): New function.
843 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
845 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
846 lookup via vtable pointer; check for type consistency
847 and turn inconsitent facts into UNREACHABLE.
848 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
849 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
850 type inconsistent querries; return UNREACHABLE instead.
852 2014-02-03 Richard Henderson <rth@twiddle.net>
855 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
856 already processed this node.
857 (normalize_one_pred_1): Pass along mark_set.
858 (normalize_one_pred): Create and destroy a pointer_set_t.
859 (normalize_one_pred_chain): Likewise.
861 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
863 PR gcov-profile/58602
864 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
866 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
869 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
870 -fno-devirtualize; try to devirtualize by the knowledge of
871 virtual table pointer given by aggregate propagation.
872 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
873 (ipa_print_node_jump_functions): Dump also offset that
874 is relevant for polymorphic calls.
875 (determine_known_aggregate_parts): Add arg_type parameter; use it
876 instead of determining the type from pointer type.
877 (ipa_compute_jump_functions_for_edge): Update call of
878 determine_known_aggregate_parts.
879 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
880 (gimple_get_virt_method_for_binfo): ... here; simplify using
881 vtable_pointer_value_to_vtable.
882 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
883 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
884 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
885 (vtable_pointer_value_to_vtable): Break out from ...; handle also
887 (vtable_pointer_value_to_binfo): ... here.
888 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
890 2014-02-03 Teresa Johnson <tejohnson@google.com>
892 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
893 redef of outer loop index variable.
895 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
899 * doc/extend.texi (Function Attributes): Typo.
901 2014-02-03 Cong Hou <congh@google.com>
903 PR tree-optimization/60000
904 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
905 if the vectorized statement is a store. A store statement can only
906 appear at the end of pattern statements.
908 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
910 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
911 (ix86_option_override_internal): Default long double to 64-bit for
912 32-bit Bionic and to 128-bit for 64-bit Bionic.
914 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
915 TARGET_LONG_DOUBLE_128 is true.
916 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
918 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
919 (mlong-double-64): Negate -mlong-double-128.
920 (mlong-double-128): New option.
922 * config/i386/i386-c.c (ix86_target_macros): Define
923 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
925 * doc/invoke.texi: Document -mlong-double-128.
927 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
929 PR rtl-optimization/60024
930 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
932 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
934 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
936 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
938 PR rtl-optimization/57662
939 * sel-sched.c (code_motion_path_driver): Do not mark already not
940 existing blocks in the visiting bitmap.
942 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
944 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
945 on the insn being emitted.
947 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
948 Will Deacon <will.deacon@arm.com>
950 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
952 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
954 * config/arm/arm-tables.opt: Regenerate.
956 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
958 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
959 for vector types other than V16QImode.
960 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
961 define_expand, and call altivec_expand_vec_perm_le when producing
962 code with little endian element order.
963 (*altivec_vperm_<mode>_internal): New insn having previous
964 behavior of altivec_vperm_<mode>.
965 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
966 altivec_expand_vec_perm_le when producing code with little endian
968 (*altivec_vperm_<mode>_uns_internal): New insn having previous
969 behavior of altivec_vperm_<mode>_uns.
971 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
973 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
974 (altivec_vsumsws): Add handling for -maltivec=be with a little
976 (altivec_vsumsws_direct): New.
977 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
980 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
982 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
983 vtable_pointer_value_to_binfo): New functions.
984 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
985 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
987 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
989 * config/nios2/nios2.md (load_got_register): Initialize GOT
990 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
991 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
993 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
995 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
996 preserverd by passthrough, do not propagate the type.
998 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1000 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1001 (mips_atomic_assign_expand_fenv): New function.
1002 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1004 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1006 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1007 (__builtin_mips_set_fcsr): Likewise.
1008 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1009 MIPS_USI_FTYPE_VOID.
1010 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1011 (mips16_expand_set_fcsr): Likewise.
1012 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1013 (mips16_set_fcsr_stub): Likewise.
1014 (mips16_get_fcsr_one_only_stub): New class.
1015 (mips16_set_fcsr_one_only_stub): Likewise.
1016 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1017 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1018 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1019 (hard_float): New availability predicate.
1020 (mips_builtins): Add get_fcsr and set_fcsr.
1021 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1022 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1023 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1024 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1025 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1028 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1030 * config/mips/mips.c (mips_one_only_stub): New class.
1031 (mips_need_mips16_rdhwr_p): Replace with...
1032 (mips16_rdhwr_stub): ...this new variable.
1033 (mips16_stub_call_address): New function.
1034 (mips16_rdhwr_one_only_stub): New class.
1035 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1036 (mips_output_mips16_rdhwr): Delete.
1037 (mips_finish_stub): New function.
1038 (mips_code_end): Use it to handle rdhwr stubs.
1040 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1043 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1044 when calculating size of integer atomic types.
1046 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1048 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1050 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1052 PR tree-optimization/60003
1053 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1054 * profile.c (branch_prob): Use gimple_call_builtin_p
1055 to check for BUILT_IN_SETJMP_RECEIVER.
1056 * tree-inline.c (copy_bb): Call notice_special_calls.
1058 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1061 * lra-constraints.c (process_alt_operands): Update reload_sum only
1064 2014-01-31 Richard Henderson <rth@redhat.com>
1067 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1068 until after else_eh is processed.
1070 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1072 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1073 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1074 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1075 in smmintrin.h, remove them.
1076 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1077 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1078 * config/i386/i386.md (ROUND_SAE): Fix value.
1079 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1080 (const48_operand): New.
1081 * config/i386/subst.md (round), (round_expand): Use
1082 const_4_or_8_to_11_operand.
1083 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1085 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1087 * config/i386/constraints.md (Yk): Swap meaning with k.
1088 * config/i386/i386.md (movhi_internal): Change Yk to k.
1089 (movqi_internal): Ditto.
1090 (*k<logic><mode>): Ditto.
1093 (kandn<mode>): Ditto.
1094 (*<code>hi_1): Ditto.
1095 (*<code>qi_1): Ditto.
1096 (kxnor<mode>): Ditto.
1097 (kortestzhi): Ditto.
1098 (kortestchi): Ditto.
1100 (*one_cmplhi2_1): Ditto.
1101 (*one_cmplqi2_1): Ditto.
1102 * config/i386/sse.md (): Change k to Yk.
1103 (avx512f_load<mode>_mask): Ditto.
1104 (avx512f_blendm<mode>): Ditto.
1105 (avx512f_store<mode>_mask): Ditto.
1106 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1107 (avx512f_storedqu<mode>_mask): Ditto.
1108 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1110 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1111 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1112 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1113 (avx512f_maskcmp<mode>3): Ditto.
1114 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1115 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1116 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1117 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1118 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1119 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1120 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1121 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1122 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1123 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1124 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1125 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1126 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1127 (vec_extract_lo_<mode>_maskm): Ditto.
1128 (vec_extract_hi_<mode>_maskm): Ditto.
1129 (avx512f_vternlog<mode>_mask): Ditto.
1130 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1131 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1132 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1133 (avx512f_<code>v8div16qi2_mask): Ditto.
1134 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1135 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1136 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1137 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1138 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1139 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1140 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1141 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1142 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1143 (avx512cd_maskb_vec_dupv8di): Ditto.
1144 (avx512cd_maskw_vec_dupv16si): Ditto.
1145 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1146 (avx512f_vpermi2var<mode>3_mask): Ditto.
1147 (avx512f_vpermi2var<mode>3_mask): Ditto.
1148 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1149 (*avx512f_gathersi<mode>): Ditto.
1150 (*avx512f_gathersi<mode>_2): Ditto.
1151 (*avx512f_gatherdi<mode>): Ditto.
1152 (*avx512f_gatherdi<mode>_2): Ditto.
1153 (*avx512f_scattersi<mode>): Ditto.
1154 (*avx512f_scatterdi<mode>): Ditto.
1155 (avx512f_compress<mode>_mask): Ditto.
1156 (avx512f_compressstore<mode>_mask): Ditto.
1157 (avx512f_expand<mode>_mask): Ditto.
1158 * config/i386/subst.md (mask): Change k to Yk.
1159 (mask_scalar_merge): Ditto.
1162 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1164 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1166 2014-01-31 Richard Biener <rguenther@suse.de>
1169 * builtins.c (fold_builtin_memory_op): Make sure to not
1170 use a floating-point mode or a boolean or enumeral type for
1173 2014-01-30 DJ Delorie <dj@redhat.com>
1175 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1176 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1177 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1178 whenever main() has an epilogue.
1180 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1182 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1183 unused variable "field".
1184 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1185 (vsx_mergeh_<mode>): Likewise.
1186 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1187 (altivec_vmrghh): Likewise.
1188 (altivec_vmrghw): Likewise.
1189 (altivec_vmrglb): Likewise.
1190 (altivec_vmrglh): Likewise.
1191 (altivec_vmrglw): Likewise.
1192 (altivec_vspltb): Add missing uses.
1193 (altivec_vsplth): Likewise.
1194 (altivec_vspltw): Likewise.
1195 (altivec_vspltsf): Likewise.
1197 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1200 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1201 frame related instructions.
1203 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1205 PR rtl-optimization/59959
1206 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1207 any reload of register whose subreg is invalid.
1209 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1211 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1212 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1213 Add missing return type - void.
1215 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1217 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1218 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1219 remove element index adjustment for endian (now handled in vsx.md
1221 (altivec_expand_vec_perm_const): Use
1222 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1223 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1224 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1225 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1226 define_expand and a new define_insn *altivec_vspltb_internal;
1227 adjust for -maltivec=be on a little endian target.
1228 (altivec_vspltb_direct): New.
1229 (altivec_vsplth): Divide into a define_expand and a new
1230 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1231 little endian target.
1232 (altivec_vsplth_direct): New.
1233 (altivec_vspltw): Divide into a define_expand and a new
1234 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1235 little endian target.
1236 (altivec_vspltw_direct): New.
1237 (altivec_vspltsf): Divide into a define_expand and a new
1238 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1239 a little endian target.
1241 2014-01-30 Richard Biener <rguenther@suse.de>
1243 PR tree-optimization/59993
1244 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1245 can propagate form the earlier stmt and avoid the transform
1246 when the intermediate result is needed.
1248 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1250 * README.Portability: Fix typo.
1252 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1254 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1255 comparison_operator with ordered_comparison_operator.
1257 2014-01-30 Nick Clifton <nickc@redhat.com>
1259 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1260 Rename to mn10300_store_multiple_regs.
1261 * config/mn10300/mn10300.c: Likewise.
1262 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1263 store_multiple_regs.
1264 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1265 Call mn10300_store_multiple_regs.
1267 2014-01-30 Nick Clifton <nickc@redhat.com>
1268 DJ Delorie <dj@redhat.com>
1270 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1271 %fp 2 to keep registers after it properly word-aligned.
1272 (rl78_alloc_physical_registers_umul): Handle the case where both
1273 input operands are the same.
1275 2014-01-30 Richard Biener <rguenther@suse.de>
1277 PR tree-optimization/59903
1278 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1281 2014-01-30 Jason Merrill <jason@redhat.com>
1284 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1287 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1289 2014-01-30 Richard Biener <rguenther@suse.de>
1291 PR tree-optimization/59951
1292 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1294 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1297 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1298 SFmode to DFmode case.
1300 2014-01-29 DJ Delorie <dj@redhat.com>
1302 * config/msp430/msp430.opt (-minrt): New.
1303 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1305 (ENDFILE_SPEC): Likewise.
1307 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1309 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1310 (estimate_function_body_sizes): Use it.
1312 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1315 * dwarf2out.c (is_cxx_auto): New.
1316 (is_base_type): Use it.
1317 (gen_type_die_with_usage): Likewise.
1319 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1321 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1322 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1323 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1324 -maltivec=be with LE targets.
1325 (vsx_mergeh_<mode>): Likewise.
1326 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1327 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1328 (altivec_vmrghb): Replace with define_expand and new
1329 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1330 (altivec_vmrghb_direct): New define_insn.
1331 (altivec_vmrghh): Replace with define_expand and new
1332 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1333 (altivec_vmrghh_direct): New define_insn.
1334 (altivec_vmrghw): Replace with define_expand and new
1335 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1336 (altivec_vmrghw_direct): New define_insn.
1337 (*altivec_vmrghsf): Adjust for endianness.
1338 (altivec_vmrglb): Replace with define_expand and new
1339 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1340 (altivec_vmrglb_direct): New define_insn.
1341 (altivec_vmrglh): Replace with define_expand and new
1342 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1343 (altivec_vmrglh_direct): New define_insn.
1344 (altivec_vmrglw): Replace with define_expand and new
1345 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1346 (altivec_vmrglw_direct): New define_insn.
1347 (*altivec_vmrglsf): Adjust for endianness.
1348 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1349 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1350 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1351 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1352 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1353 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1354 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1355 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1357 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1359 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1360 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1363 2014-01-29 Richard Biener <rguenther@suse.de>
1365 PR tree-optimization/58742
1366 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1367 associate_pointerplus_align.
1368 (associate_pointerplus_diff): New function.
1369 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1370 and associate_pointerplus_diff.
1372 2014-01-29 Richard Biener <rguenther@suse.de>
1374 * lto-streamer.h (LTO_major_version): Bump to 3.
1375 (LTO_minor_version): Reset to 0.
1377 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1379 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1380 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1381 (arm_file_start): Generate correct asm header for armv7ve.
1382 * config/arm/bpabi.h: Add multilib support for armv7ve.
1383 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1384 and cortex-a15 to armv7ve.
1385 * config/arm/t-aprofile: Add multilib support for armv7ve.
1386 * doc/invoke.texi: Document -march=armv7ve.
1388 2014-01-29 Richard Biener <rguenther@suse.de>
1390 PR tree-optimization/58742
1391 * tree-ssa-forwprop.c (associate_plusminus): Return true
1392 if we changed sth, defer EH cleanup to ...
1393 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1394 (simplify_mult): New function.
1396 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1399 PR tree-optimization/59920
1400 * tree.c (build_common_builtin_nodes): Remove
1401 __builtin_setjmp_dispatcher initialization.
1402 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1403 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1404 instead of gsi_after_labels + manually skipping debug stmts.
1405 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1406 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1407 * tree-inline.c (copy_edges_for_bb): Remove
1408 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1409 argument. Ignore computed_goto_p stmts. Don't call
1410 make_abnormal_goto_edges. If a call might need abnormal edges
1411 for non-local gotos, see if it already has an edge to
1412 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1413 with true argument, don't do anything then, otherwise add
1414 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1415 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1417 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1418 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1419 (lower_stmt): Don't set data->calls_builtin_setjmp.
1420 (lower_builtin_setjmp): Adjust comment.
1421 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1422 * tree-cfg.c (found_computed_goto): Remove.
1423 (factor_computed_gotos): Remove.
1424 (make_goto_expr_edges): Return bool, true for computed gotos.
1425 Don't call make_abnormal_goto_edges.
1426 (build_gimple_cfg): Don't set found_computed_goto, don't call
1427 factor_computed_gotos.
1428 (computed_goto_p): No longer static.
1429 (make_blocks): Don't set found_computed_goto.
1430 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1431 (make_edges): If make_goto_expr_edges returns true, push bb
1432 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1433 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1434 vector. Record mapping between bbs and OpenMP regions if there
1435 are any, adjust make_gimple_omp_edges caller. Call
1436 handle_abnormal_edges.
1437 (make_abnormal_goto_edges): Remove.
1438 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1439 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1440 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1441 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1442 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1443 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1444 filling *region also set *region_idx to (*region)->entry->index.
1447 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1448 For REGs set ORIGINAL_REGNO.
1450 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1452 * doc/md.texi: Mention that a target shouldn't implement
1453 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1456 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1458 PR tree-optimization/59594
1459 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1460 a copy of the datarefs vector rather than the vector itself.
1462 2014-01-28 Jason Merrill <jason@redhat.com>
1465 * dwarf2out.c (auto_die): New static.
1466 (gen_type_die_with_usage): Handle C++1y 'auto'.
1467 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1470 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1473 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1474 (SPEC_X32): Likewise.
1475 (SPEC_64): Likewise.
1476 * config/i386/i386.c (ix86_option_override_internal): Turn off
1477 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1479 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1480 * config/i386/i386.h (TARGET_16BIT): New macro.
1481 (TARGET_16BIT_P): Likewise.
1482 * config/i386/i386.opt: Add m16.
1483 * doc/invoke.texi: Document -m16.
1485 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1487 PR preprocessor/59935
1488 * input.c (location_get_source_line): Bail out on when line number
1489 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1491 2014-01-28 Richard Biener <rguenther@suse.de>
1493 PR tree-optimization/58742
1494 * tree-ssa-forwprop.c (associate_plusminus): Handle
1495 pointer subtraction of the form (T)(P + A) - (T)P.
1497 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1499 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1502 2014-01-28 Richard Biener <rguenther@suse.de>
1505 2014-01-28 Richard Biener <rguenther@suse.de>
1507 PR rtl-optimization/45364
1508 PR rtl-optimization/59890
1509 * var-tracking.c (local_get_addr_clear_given_value): Handle
1510 already cleared slot.
1511 (val_reset): Handle not allocated local_get_addr_cache.
1512 (vt_find_locations): Use post-order on the inverted CFG.
1514 2014-01-28 Richard Biener <rguenther@suse.de>
1516 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1518 2014-01-28 Richard Biener <rguenther@suse.de>
1520 PR rtl-optimization/45364
1521 PR rtl-optimization/59890
1522 * var-tracking.c (local_get_addr_clear_given_value): Handle
1523 already cleared slot.
1524 (val_reset): Handle not allocated local_get_addr_cache.
1525 (vt_find_locations): Use post-order on the inverted CFG.
1527 2014-01-28 Alan Modra <amodra@gmail.com>
1529 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1530 * configure.ac <recursive call for build != host>: Define
1531 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1532 and LD_FOR_BUILD too.
1533 * configure: Regenerate.
1535 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1537 * config/i386/i386.c (get_builtin_code_for_version): Separate
1538 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1539 Broadwell from Haswell.
1541 2014-01-27 Steve Ellcey <sellcey@mips.com>
1543 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1544 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1545 * config/mips/mips.c (mips_option_override): Change setting
1547 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1548 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1549 Change from Mask to Var.
1551 2014-01-27 Jeff Law <law@redhat.com>
1553 * ipa-inline.c (inline_small_functions): Fix typo.
1555 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1557 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1558 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1559 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1560 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1561 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1562 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1563 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1564 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1565 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1566 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1567 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1568 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1569 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1570 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1571 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1572 (_mm512_storeu_epi64): Ditto.
1573 (_mm512_cmpge_epi32_mask): Ditto.
1574 (_mm512_cmpge_epu32_mask): Ditto.
1575 (_mm512_cmpge_epi64_mask): Ditto.
1576 (_mm512_cmpge_epu64_mask): Ditto.
1577 (_mm512_cmple_epi32_mask): Ditto.
1578 (_mm512_cmple_epu32_mask): Ditto.
1579 (_mm512_cmple_epi64_mask): Ditto.
1580 (_mm512_cmple_epu64_mask): Ditto.
1581 (_mm512_cmplt_epi32_mask): Ditto.
1582 (_mm512_cmplt_epu32_mask): Ditto.
1583 (_mm512_cmplt_epi64_mask): Ditto.
1584 (_mm512_cmplt_epu64_mask): Ditto.
1585 (_mm512_cmpneq_epi32_mask): Ditto.
1586 (_mm512_cmpneq_epu32_mask): Ditto.
1587 (_mm512_cmpneq_epi64_mask): Ditto.
1588 (_mm512_cmpneq_epu64_mask): Ditto.
1589 (_mm512_expand_pd): Ditto.
1590 (_mm512_expand_ps): Ditto.
1591 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1592 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1593 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1594 * config/i386/i386.c (ix86_builtins): Add
1595 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1596 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1597 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1598 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1599 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1600 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1601 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1602 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1603 IX86_BUILTIN_PMOVUSQW512_MEM.
1604 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1605 __builtin_ia32_pmovsqd512mem_mask,
1606 __builtin_ia32_pmovqd512mem_mask,
1607 __builtin_ia32_pmovusqw512mem_mask,
1608 __builtin_ia32_pmovsqw512mem_mask,
1609 __builtin_ia32_pmovqw512mem_mask,
1610 __builtin_ia32_pmovusdw512mem_mask,
1611 __builtin_ia32_pmovsdw512mem_mask,
1612 __builtin_ia32_pmovdw512mem_mask,
1613 __builtin_ia32_pmovqb512mem_mask,
1614 __builtin_ia32_pmovusqb512mem_mask,
1615 __builtin_ia32_pmovsqb512mem_mask,
1616 __builtin_ia32_pmovusdb512mem_mask,
1617 __builtin_ia32_pmovsdb512mem_mask,
1618 __builtin_ia32_pmovdb512mem_mask.
1619 (bdesc_args): Add __builtin_ia32_expanddf512,
1620 __builtin_ia32_expandsf512.
1621 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1622 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1623 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1624 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1625 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1626 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1627 (avx512f_<code>v8div16qi2_mask_store): This.
1628 (avx512f_expand<mode>): New.
1630 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1632 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1634 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1635 (_mm512_prefetch_i32scatter_pd): Ditto.
1636 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1637 (_mm512_prefetch_i64scatter_pd): Ditto.
1638 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1639 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1640 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1641 (_mm512_prefetch_i32scatter_ps): Ditto.
1642 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1643 (_mm512_prefetch_i64scatter_ps): Ditto.
1644 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1645 * config/i386/i386-builtin-types.def: Define
1646 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1647 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1648 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1649 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1650 IX86_BUILTIN_SCATTERPFQPD.
1651 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1652 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1653 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1654 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1655 __builtin_ia32_scatterpfqps.
1656 (ix86_expand_builtin): Expand new built-ins.
1657 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1658 fix memory access data type.
1659 (*avx512pf_gatherpf<mode>_mask): Ditto.
1660 (*avx512pf_gatherpf<mode>): Ditto.
1661 (avx512pf_scatterpf<mode>): Ditto.
1662 (*avx512pf_scatterpf<mode>_mask): Ditto.
1663 (*avx512pf_scatterpf<mode>): Ditto.
1664 (GATHER_SCATTER_SF_MEM_MODE): New.
1665 (avx512pf_gatherpf<mode>df): Ditto.
1666 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1667 (*avx512pf_scatterpf<mode>df): Ditto.
1669 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1672 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1673 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1676 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1678 * common/config/arm/arm-common.c
1679 (arm_rewrite_mcpu): Handle multiple names.
1681 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1683 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1685 * gimple-builder.h (create_gimple_tmp): Delete.
1687 2014-01-27 Christian Bruel <christian.bruel@st.com>
1689 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1692 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1694 * config/pa/pa.md (call): Generate indirect long calls to non-local
1695 functions when outputing 32-bit code.
1696 (call_value): Likewise except for special call to buggy powf function.
1698 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1699 portable runtime and PIC indirect calls.
1700 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1701 and PIC call sequences. Use ldo instead of blr to set return register
1702 in PIC call sequence.
1704 2014-01-25 Walter Lee <walt@tilera.com>
1706 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1707 avoid clobbering a live register.
1709 2014-01-25 Walter Lee <walt@tilera.com>
1711 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1712 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1713 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1714 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1716 2014-01-25 Walter Lee <walt@tilera.com>
1718 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1719 arguments on even registers.
1720 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1722 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1723 (BIGGEST_ALIGNMENT): Ditto.
1724 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1726 2014-01-25 Walter Lee <walt@tilera.com>
1728 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1729 insns before bundling.
1730 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1732 2014-01-25 Walter Lee <walt@tilera.com>
1734 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1735 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1736 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1738 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1740 * config/mips/constraints.md (kl): Delete.
1741 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1742 define expands, using...
1743 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1744 instructions for MIPS16.
1745 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1746 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1748 2014-01-25 Walter Lee <walt@tilera.com>
1750 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1754 2014-01-25 Walter Lee <walt@tilera.com>
1756 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1757 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1759 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1761 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1764 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1766 * print-rtl.c (in_call_function_usage): New var.
1767 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1768 EXPR_LIST mode as mode and not as reg note name.
1771 * cfgloopmanip.c (copy_loop_info): If
1772 loop->warned_aggressive_loop_optimizations, make sure
1773 the flag is set in target loop too.
1775 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1777 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1779 * builtins.def: Likewise.
1780 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1781 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1782 * ira.c (ira_setup_eliminable_regset): Likewise.
1783 * omp-low.c (gate_expand_omp): Likewise.
1784 (execute_lower_omp): Likewise.
1785 (diagnose_sb_0): Likewise.
1786 (gate_diagnose_omp_blocks): Likewise.
1787 (simd_clone_clauses_extract): Likewise.
1790 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1792 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1793 correction for little endian...
1794 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1797 2014-01-24 Jeff Law <law@redhat.com>
1799 PR tree-optimization/59919
1800 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1801 for non-returning calls.
1803 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1805 * common/config/aarch64/aarch64-common.c
1806 (aarch64_rewrite_mcpu): Handle multiple names.
1807 * config/aarch64/aarch64.h
1808 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1810 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1812 * input.c (add_file_to_cache_tab): Handle the case where fopen
1815 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1818 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1819 from push operand if code of push isn't PRE_DEC.
1821 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1824 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1825 -mquad-memory-atomic. Update -mquad-memory documentation to say
1826 it is only used for non-atomic loads/stores.
1828 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1829 -mquad-memory or -mquad-memory-atomic switches.
1831 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1832 -mquad-memory-atomic to ISA 2.07 support.
1834 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1835 to separate support of normal quad word memory operations (ldq, stq)
1836 from the atomic quad word memory operations.
1838 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1839 support to separate non-atomic quad word operations from atomic
1840 quad word operations. Disable non-atomic quad word operations in
1841 little endian mode so that we don't have to swap words after the
1842 load and before the store.
1843 (quad_load_store_p): Add comment about atomic quad word support.
1844 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1845 options printed with -mdebug=reg.
1847 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1848 -mquad-memory-atomic as the test for whether we have quad word
1849 atomic instructions.
1850 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1851 or -mp8-vector are used, allow byte/half-word atomic operations.
1853 * config/rs6000/sync.md (load_lockedti): Insure that the address
1854 is a proper indexed or indirect address for the lqarx instruction.
1855 On little endian systems, swap the hi/lo registers after the lqarx
1857 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1858 insure the address is valid for the lqarx instruction.
1859 (store_conditionalti): Insure that the address is a proper indexed
1860 or indirect address for the stqcrx. instruction. On little endian
1861 systems, swap the hi/lo registers before doing the stqcrx.
1863 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1864 insure the address is valid for the stqcrx. instruction.
1866 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1867 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1868 type of quad memory support is available.
1870 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1873 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1874 there is a danger of looping.
1876 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1879 force flag_ira_loop_pressure if set via command line.
1881 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1883 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1884 (ashr_simd): New builtin handling DI mode.
1885 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1886 (aarch64_sshr_simddi): New match pattern.
1887 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1888 (vshrd_n_s64): Likewise.
1889 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1891 2014-01-23 Nick Clifton <nickc@redhat.com>
1893 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1894 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1895 favour of mcu specific scripts.
1896 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1899 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1900 Alex Velenko <Alex.Velenko@arm.com>
1902 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1903 (vaddv_s16): Likewise.
1904 (vaddv_s32): Likewise.
1905 (vaddv_u8): Likewise.
1906 (vaddv_u16): Likewise.
1907 (vaddv_u32): Likewise.
1908 (vaddvq_s8): Likewise.
1909 (vaddvq_s16): Likewise.
1910 (vaddvq_s32): Likewise.
1911 (vaddvq_s64): Likewise.
1912 (vaddvq_u8): Likewise.
1913 (vaddvq_u16): Likewise.
1914 (vaddvq_u32): Likewise.
1915 (vaddvq_u64): Likewise.
1916 (vaddv_f32): Likewise.
1917 (vaddvq_f32): Likewise.
1918 (vaddvq_f64): Likewise.
1919 (vmaxv_f32): Likewise.
1920 (vmaxv_s8): Likewise.
1921 (vmaxv_s16): Likewise.
1922 (vmaxv_s32): Likewise.
1923 (vmaxv_u8): Likewise.
1924 (vmaxv_u16): Likewise.
1925 (vmaxv_u32): Likewise.
1926 (vmaxvq_f32): Likewise.
1927 (vmaxvq_f64): Likewise.
1928 (vmaxvq_s8): Likewise.
1929 (vmaxvq_s16): Likewise.
1930 (vmaxvq_s32): Likewise.
1931 (vmaxvq_u8): Likewise.
1932 (vmaxvq_u16): Likewise.
1933 (vmaxvq_u32): Likewise.
1934 (vmaxnmv_f32): Likewise.
1935 (vmaxnmvq_f32): Likewise.
1936 (vmaxnmvq_f64): Likewise.
1937 (vminv_f32): Likewise.
1938 (vminv_s8): Likewise.
1939 (vminv_s16): Likewise.
1940 (vminv_s32): Likewise.
1941 (vminv_u8): Likewise.
1942 (vminv_u16): Likewise.
1943 (vminv_u32): Likewise.
1944 (vminvq_f32): Likewise.
1945 (vminvq_f64): Likewise.
1946 (vminvq_s8): Likewise.
1947 (vminvq_s16): Likewise.
1948 (vminvq_s32): Likewise.
1949 (vminvq_u8): Likewise.
1950 (vminvq_u16): Likewise.
1951 (vminvq_u32): Likewise.
1952 (vminnmv_f32): Likewise.
1953 (vminnmvq_f32): Likewise.
1954 (vminnmvq_f64): Likewise.
1956 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1958 * config/aarch64/aarch64-simd.md
1959 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1960 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1961 (*aarch64_mul3_elt<mode>): Likewise.
1962 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1963 (*aarch64_mul3_elt_to_64v2df): Likewise.
1964 (*aarch64_mla_elt<mode>): Likewise.
1965 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1966 (*aarch64_mls_elt<mode>): Likewise.
1967 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1968 (*aarch64_fma4_elt<mode>): Likewise.
1969 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1970 (*aarch64_fma4_elt_to_64v2df): Likewise.
1971 (*aarch64_fnma4_elt<mode>): Likewise.
1972 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1973 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1974 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1975 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1976 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1977 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1978 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1979 (aarch64_sqdmull_lane<mode>_internal): Likewise.
1980 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1982 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
1984 * config/aarch64/aarch64-simd.md
1985 (aarch64_be_checked_get_lane<mode>): New define_expand.
1986 * config/aarch64/aarch64-simd-builtins.def
1987 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
1988 New builtin definition.
1989 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
1990 Use new safe be builtin.
1992 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1994 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
1996 (aarch64_be_st1<mode>): Likewise.
1997 (aarch_ld1<VALL:mode>): Define_expand modified.
1998 (aarch_st1<VALL:mode>): Likewise.
1999 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2000 (UNSPEC_ST1): Likewise.
2002 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2004 * config/microblaze/microblaze.md: Add trap insn and attribute
2006 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2008 PR preprocessor/58580
2009 * input.h (location_get_source_line): Take an additional line_size
2011 (void diagnostics_file_cache_fini): Declare new function.
2012 * input.c (struct fcache): New type.
2013 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2014 New static constants.
2015 (diagnostic_file_cache_init, total_lines_num)
2016 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2017 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2018 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2019 (get_next_line, read_next_line, goto_next_line, read_line_num):
2020 New static function definitions.
2021 (diagnostic_file_cache_fini): New function.
2022 (location_get_source_line): Take an additional output line_len
2023 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2025 * diagnostic.c (diagnostic_finish): Call
2026 diagnostic_file_cache_fini.
2027 (adjust_line): Take an additional input parameter for the length
2028 of the line, rather than calculating it with strlen.
2029 (diagnostic_show_locus): Adjust the use of
2030 location_get_source_line and adjust_line with respect to their new
2031 signature. While displaying a line now, do not stop at the first
2032 null byte. Rather, display the zero byte as a space and keep
2033 going until we reach the size of the line.
2034 * Makefile.in: Add vec.o to OBJS-libcommon
2036 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2037 Ilya Tocar <ilya.tocar@intel.com>
2039 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2040 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2041 (__builtin_ia32_kmov16): Ditto.
2042 * config/i386/i386.md (UNSPEC_KMOV): New.
2045 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2047 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2048 (_mm512_storeu_si512): Ditto.
2050 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2053 * rtl.h (get_referenced_operands): Declare.
2054 * recog.c (get_referenced_operands): New function.
2055 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2056 operands have been referenced when recording LO_SUM references.
2058 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2060 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2062 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2064 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2065 Enable for generic and recent AMD targets.
2067 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2069 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2070 ARG_SIZE note when adjustment was eliminated.
2072 2014-01-22 Jeff Law <law@redhat.com>
2074 PR tree-optimization/59597
2075 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2076 in file. Accept new argument REGISTERING and use it to modify
2077 dump output appropriately.
2078 (register_jump_thread): Corresponding changes.
2079 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2080 thread paths involving joiner blocks. Add code to dump cancelled
2081 jump threading paths.
2083 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2085 PR rtl-optimization/59477
2086 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2087 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2089 2014-01-22 Tom Tromey <tromey@redhat.com>
2091 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2093 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2095 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2097 PR rtl-optimization/59896
2098 * lra-constraints.c (process_alt_operands): Check unused note for
2099 matched operands of insn with no output reloads.
2101 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2103 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2104 (mips_move_from_gpr_cost): Likewise.
2106 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2108 PR rtl-optimization/59858
2109 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2110 ira_class_hard_regs_num.
2111 (process_alt_operands): Increase reject for dying matched operand.
2113 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2116 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2117 smaller than size, perform several stores or loads and stores
2118 at dst + count - size to store or copy all of size bytes, rather
2119 than just last modesize bytes.
2121 2014-01-20 DJ Delorie <dj@redhat.com>
2123 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2124 that CLOBBERs are REGs before propogating their values.
2126 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2129 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2130 (cgraph_inline_failed_type): New function.
2131 * cgraph.h (DEFCIFCODE): Add type.
2132 (cgraph_inline_failed_type_t): New enum.
2133 (cgraph_inline_failed_type): New prototype.
2134 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2135 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2136 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2137 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2138 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2139 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2140 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2141 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2142 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2143 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2144 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2145 OPTIMIZATION_MISMATCH.
2146 * tree-inline.c (expand_call_inline): Emit errors during
2147 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2149 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2152 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2153 mode attribute in insn output.
2155 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2157 * output.h (output_constant): Delete.
2158 * varasm.c (output_constant): Make private.
2160 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2162 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2164 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2167 * tree.h (fold_builtin_strcat): New prototype.
2168 * builtins.c (fold_builtin_strcat): No longer static. Add len
2169 argument, if non-NULL, don't call c_strlen. Optimize
2170 directly into __builtin_memcpy instead of __builtin_strcpy.
2171 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2172 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2174 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2176 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2177 for SImode_address_operand operands, having only a REG argument.
2179 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2181 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2182 loader name using mbig-endian.
2183 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2185 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2187 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2191 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2193 * config/aarch64/aarch64-protos.h
2194 (aarch64_cannot_change_mode_class_ptr): Declare.
2195 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2196 aarch64_cannot_change_mode_class_ptr): New.
2197 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2198 backend hook aarch64_cannot_change_mode_class.
2200 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2202 * common/config/aarch64/aarch64-common.c
2203 (aarch64_handle_option): Don't handle any option order logic here.
2204 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2205 selected_cpu, warn on architecture version mismatch.
2206 (aarch64_override_options): Fix parsing order for option strings.
2208 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2209 Iain Sandoe <iain@codesourcery.com>
2212 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2213 warning. Amend comment to reflect current functionality.
2215 2014-01-20 Richard Biener <rguenther@suse.de>
2218 * builtins.c (fold_builtin_strcat): Remove case better handled
2219 by tree-ssa-strlen.c.
2221 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2223 * config/aarch64/aarch64.opt
2224 (mcpu, march, mtune): Make case-insensitive.
2226 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2229 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2230 if operands[1] is a REG or ZERO_EXTEND of a REG.
2232 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2234 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2236 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2238 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2239 long non-pic millicode calls.
2241 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2243 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2245 2014-01-19 Kito Cheng <kito@0xlab.org>
2247 * builtins.c (expand_movstr): Check movstr expand done or fail.
2249 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2250 H.J. Lu <hongjiu.lu@intel.com>
2253 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2254 to DImode for zero-extended addresses.
2256 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2258 PR rtl-optimization/57763
2259 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2260 on the new indirect jump_insn and increment LABEL_NUSES (label).
2262 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2266 * config.gcc (x86_archs): New variable.
2267 (x86_64_archs): Likewise.
2268 (x86_cpus): Likewise.
2269 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2270 --with-arch/--with-cpu= options.
2271 Support --with-arch=/--with-cpu={nehalem,westmere,
2272 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2274 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2276 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2277 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2279 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2281 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2283 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2286 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2287 clear cpp_get_options (parse_in)->warn_unused_macros for
2288 ix86_target_macros_internal with cpp_define.
2290 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2292 * jump.c (delete_related_insns): Keep (use (insn))s.
2293 * reorg.c (redundant_insn): Check for barriers too.
2295 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2297 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2299 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2301 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2302 call to $$dyncall when TARGET_LONG_CALLS is true.
2304 2014-01-17 Jeff Law <law@redhat.com>
2306 * ree.c (combine_set_extension): Temporarily disable test for
2307 changing number of hard registers.
2309 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2312 * ipa-inline-analysis.c (inline_free_summary):
2313 Do not free summary of aliases.
2315 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2318 * gimplify.c (gimplify_expr): Use create_tmp_var
2319 instead of create_tmp_var_raw. If cond doesn't have
2320 integral type, don't add the IFN_ANNOTATE builtin at all.
2322 2014-01-17 Martin Jambor <mjambor@suse.cz>
2325 * ipa-cp.c (prev_edge_clone): New variable.
2326 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2327 Also resize prev_edge_clone vector.
2328 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2329 (ipcp_edge_removal_hook): New function.
2330 (ipcp_driver): Register ipcp_edge_removal_hook.
2332 2014-01-17 Andrew Pinski <apinski@cavium.com>
2333 Steve Ellcey <sellcey@mips.com>
2336 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2339 2014-01-17 Jeff Law <law@redhat.com>
2342 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2343 so that pass_ccp runs first.
2345 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2347 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2348 (ix86_adjust_cost): Use !TARGET_XXX.
2349 (do_reorder_for_imul): Likewise.
2350 (swap_top_of_ready_list): Likewise.
2351 (ix86_sched_reorder): Likewise.
2353 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2355 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2356 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2357 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2358 (intel_memset): New. Duplicate slm_memset.
2359 (intel_cost): New. Duplicate slm_cost.
2360 (m_INTEL): New macro.
2361 (processor_target_table): Add "intel".
2362 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2363 with PROCESSOR_INTEL for "intel".
2364 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2365 PROCESSOR_SILVERMONT.
2366 (ix86_issue_rate): Likewise.
2367 (ix86_adjust_cost): Likewise.
2368 (ia32_multipass_dfa_lookahead): Likewise.
2369 (swap_top_of_ready_list): Likewise.
2370 (ix86_sched_reorder): Likewise.
2371 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2372 instead of TARGET_OPT_AGU.
2373 * config/i386/i386.h (TARGET_INTEL): New.
2374 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2375 (processor_type): Add PROCESSOR_INTEL.
2376 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2377 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2379 2014-01-17 Marek Polacek <polacek@redhat.com>
2382 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2385 2014-01-17 Richard Biener <rguenther@suse.de>
2387 PR tree-optimization/46590
2388 * opts.c (default_options_table): Add entries for
2389 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2390 all enabled at -O1 but not for -Og.
2391 * common.opt (fbranch-count-reg): Remove Init(1).
2392 (fmove-loop-invariants): Likewise.
2393 (ftree-pta): Likewise.
2395 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2397 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2398 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2399 decls to at least the GCC 4.8 used alignments.
2402 * tree-nested.c (convert_nonlocal_reference_stmt,
2403 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2404 of GIMPLE_BIND stmts, adjust associated decls.
2406 2014-01-17 Richard Biener <rguenther@suse.de>
2408 PR tree-optimization/46590
2409 * vec.h (vec<>::bseach): New member function implementing
2410 binary search according to C89 bsearch.
2411 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2412 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2413 bitmap pointer again. Make accesses_in_loop a flat array.
2414 (mem_ref_obstack): New global.
2415 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2416 (mark_ref_stored): Likewise.
2417 (ref_indep_loop_p_2): Likewise.
2418 (set_ref_stored_in_loop): New helper function.
2419 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2420 (memref_free): Adjust.
2421 (record_mem_ref_loc): Simplify.
2422 (gather_mem_refs_stmt): Adjust.
2423 (sort_locs_in_loop_postorder_cmp): New function.
2424 (analyze_memory_references): Sort accesses_in_loop after
2425 loop postorder number.
2426 (find_ref_loc_in_loop_cmp): New function.
2427 (for_all_locs_in_loop): Find relevant cluster of locs in
2428 accesses_in_loop and iterate without recursion.
2429 (execute_sm): Avoid uninit warning.
2430 (struct ref_always_accessed): Simplify.
2431 (ref_always_accessed::operator ()): Likewise.
2432 (ref_always_accessed_p): Likewise.
2433 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2434 loop postorder numbers here.
2435 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2438 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2441 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2442 on decls for which assemble_alias has been called.
2444 2014-01-17 Nick Clifton <nickc@redhat.com>
2446 * config/msp430/msp430.opt: (mcpu): New option.
2447 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2448 (msp430_option_override): Parse target_cpu. If the MCU name
2449 matches a generic string, clear target_mcu.
2450 (msp430_attr): Allow numeric interrupt values up to 63.
2451 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2452 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2454 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2456 * config/msp430/msp430.md (popm): Use %J rather than %I.
2457 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2458 (addhi_cy_i): Use immediate_operand for operand 2.
2459 * doc/invoke.texi: Document -mcpu option.
2461 2014-01-17 Richard Biener <rguenther@suse.de>
2463 PR rtl-optimization/38518
2464 * df.h (df_analyze_loop): Declare.
2465 * df-core.c: Include cfgloop.h.
2466 (df_analyze_1): Split out main part of df_analyze.
2467 (df_analyze): Adjust.
2468 (loop_inverted_post_order_compute): New function.
2469 (loop_post_order_compute): Likewise.
2470 (df_analyze_loop): New function avoiding whole-function
2472 * loop-invariant.c (find_defs): Use df_analyze_loop.
2473 (find_invariants): Adjust.
2474 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2476 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2478 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2479 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2481 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2483 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2484 traversal when removing references.
2486 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2489 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2491 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2494 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2495 pushing a reload for an autoinc when we had previously reloaded an
2496 inner part of the address.
2498 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2500 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2502 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2503 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2504 when not giving up or versioning for alias only because of
2506 (vect_analyze_data_ref_dependences): Set to true.
2507 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2509 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2510 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2514 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2517 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2518 operand 0 predicate for gathers, use a new pseudo as subtarget.
2520 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2523 * lra-constraints.c (process_alt_operands): Add printing debug info.
2524 Check absence of input/output reloads for matched operands too.
2526 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2528 PR rtl-optimization/59835
2529 * ira.c (ira_init_register_move_cost): Increase cost for
2532 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2534 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2536 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2539 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2540 non-register objects. Use gen_(high/low)part more consistently.
2543 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2546 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2547 endian support, remove tests for WORDS_BIG_ENDIAN.
2548 (p8_mfvsrd_3_<mode>): Likewise.
2549 (reload_gpr_from_vsx<mode>): Likewise.
2550 (reload_gpr_from_vsxsf): Likewise.
2551 (p8_mfvsrd_4_disf): Likewise.
2553 2014-01-16 Richard Biener <rguenther@suse.de>
2555 PR rtl-optimization/46590
2556 * lcm.c (compute_antinout_edge): Use postorder iteration.
2557 (compute_laterin): Use inverted postorder iteration.
2559 2014-01-16 Nick Clifton <nickc@redhat.com>
2562 * varasm.c (output_constant): Return the number of bytes actually
2564 (output_constructor_array_range): Update the field size with the
2565 number of bytes emitted by output_constant.
2566 (output_constructor_regular_field): Likewise. Also do not
2567 complain if the total number of bytes emitted is now greater
2568 than the expected fieldpos.
2569 * output.h (output_constant): Update prototype and descriptive comment.
2571 2014-01-16 Marek Polacek <polacek@redhat.com>
2574 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2575 it is error_mark_node.
2577 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2579 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2580 VALID_AVX256_REG_OR_OI_MODE.
2582 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2584 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2585 current procedure should be profiled.
2587 2014-01-15 Andrew Pinski <apinski@cavium.com>
2589 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2590 of moving from/to the STACK_REG register class.
2592 2014-01-15 Richard Henderson <rth@redhat.com>
2595 * reginfo.c (global_regs_decl): Globalize.
2596 * rtl.h (global_regs_decl): Declare.
2597 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2598 reserved via global_regs.
2600 2014-01-15 Teresa Johnson <tejohnson@google.com>
2602 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2604 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2606 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2607 and vmulosh rather than call gen_vec_widen_smult_*.
2608 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2609 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2610 (vec_widen_smult_even_v16qi): Likewise.
2611 (vec_widen_umult_even_v8hi): Likewise.
2612 (vec_widen_smult_even_v8hi): Likewise.
2613 (vec_widen_umult_odd_v16qi): Likewise.
2614 (vec_widen_smult_odd_v16qi): Likewise.
2615 (vec_widen_umult_odd_v8hi): Likewise.
2616 (vec_widen_smult_odd_v8hi): Likewise.
2617 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2618 vmuloub rather than call gen_vec_widen_umult_*.
2619 (vec_widen_umult_lo_v16qi): Likewise.
2620 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2621 vmulosb rather than call gen_vec_widen_smult_*.
2622 (vec_widen_smult_lo_v16qi): Likewise.
2623 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2624 rather than call gen_vec_widen_umult_*.
2625 (vec_widen_umult_lo_v8hi): Likewise.
2626 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2627 rather than call gen_vec_widen_smult_*.
2628 (vec_widen_smult_lo_v8hi): Likewise.
2630 2014-01-15 Jeff Law <law@redhat.com>
2632 PR tree-optimization/59747
2633 * ree.c (find_and_remove_re): Properly handle case where a second
2634 eliminated extension requires widening a copy created for elimination
2635 of a prior extension.
2636 (combine_set_extension): Ensure that the number of hard regs needed
2637 for a destination register does not change when we widen it.
2639 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2641 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2642 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2643 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2644 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2645 (avr-*-rtems*): Likewise.
2646 (bfin*-rtems*): Likewise.
2647 (moxie-*-rtems*): Likewise.
2648 (h8300-*-rtems*): Likewise.
2649 (i[34567]86-*-rtems*): Likewise.
2650 (lm32-*-rtems*): Likewise.
2651 (m32r-*-rtems*): Likewise.
2652 (m68k-*-rtems*): Likewise.
2653 (microblaze*-*-rtems*): Likewise.
2654 (mips*-*-rtems*): Likewise.
2655 (powerpc-*-rtems*): Likewise.
2656 (sh-*-rtems*): Likewise.
2657 (sparc-*-rtems*): Likewise.
2658 (sparc64-*-rtems*): Likewise.
2659 (v850-*-rtems*): Likewise.
2660 (m32c-*-rtems*): Likewise.
2662 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2664 PR rtl-optimization/59511
2665 * ira.c (ira_init_register_move_cost): Use memory costs for some
2666 cases of register move cost calculations.
2667 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2668 instead of BB frequency.
2669 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2670 * lra-assigns.c (find_hard_regno_for): Ditto.
2672 2014-01-15 Richard Biener <rguenther@suse.de>
2674 PR tree-optimization/59822
2675 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2676 (vectorizable_load): Use it to hoist defs of uses of invariant
2677 loads out of the loop.
2679 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2680 Kugan Vivekanandarajah <kuganv@linaro.org>
2683 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2686 2014-01-15 Richard Biener <rguenther@suse.de>
2688 PR rtl-optimization/59802
2689 * lcm.c (compute_available): Use inverted postorder to seed
2690 the initial worklist.
2692 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2695 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2696 ADDR_REGS for invalid symrefs in non-PIC code.
2698 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2701 * builtins.c (determine_block_size): Initialize *probable_max_size
2702 even if len_rtx is CONST_INT.
2704 2014-01-14 Andrew Pinski <apinski@cavium.com>
2706 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2707 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2708 (cortexa53_tunings): Likewise.
2709 (aarch64_sched_issue_rate): New function.
2710 (TARGET_SCHED_ISSUE_RATE): Define.
2712 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2714 * ira-costs.c (find_costs_and_classes): Add missed
2715 ira_init_register_move_cost_if_necessary.
2717 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2720 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2722 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2725 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2726 to indicate if type is used for function return value. Warn ABI
2727 change if the vector mode isn't available for function return value.
2728 (ix86_function_arg_advance): Pass false to type_natural_mode.
2729 (ix86_function_arg): Likewise.
2730 (ix86_gimplify_va_arg): Likewise.
2731 (function_arg_32): Don't warn ABI change.
2732 (ix86_function_value): Pass true to type_natural_mode.
2733 (ix86_return_in_memory): Likewise.
2734 (ix86_struct_value_rtx): Removed.
2735 (TARGET_STRUCT_VALUE_RTX): Likewise.
2737 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2739 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2740 converting a conditional jump into a conditional return.
2742 2014-01-14 Richard Biener <rguenther@suse.de>
2744 PR tree-optimization/58921
2745 PR tree-optimization/59006
2746 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2747 hoisting invariant stmts.
2748 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2749 invariant loads on the preheader edge if possible.
2751 2014-01-14 Joey Ye <joey.ye@arm.com>
2753 * doc/plugin.texi (Building GCC plugins): Update to C++.
2755 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2757 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2758 (_mm_rcp28_round_ss): Ditto.
2759 (_mm_rsqrt28_round_sd): Ditto.
2760 (_mm_rsqrt28_round_ss): Ditto.
2761 (_mm_rcp28_sd): Ditto.
2762 (_mm_rcp28_ss): Ditto.
2763 (_mm_rsqrt28_sd): Ditto.
2764 (_mm_rsqrt28_ss): Ditto.
2765 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2766 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2767 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2768 (IX86_BUILTIN_RCP28SD): Ditto.
2769 (IX86_BUILTIN_RCP28SS): Ditto.
2770 (IX86_BUILTIN_RSQRT28SD): Ditto.
2771 (IX86_BUILTIN_RSQRT28SS): Ditto.
2772 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2773 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2774 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2775 (ix86_expand_special_args_builtin): Expand new FTYPE.
2776 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2777 (srcp14<mode>): Make insn unary.
2778 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2779 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2780 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2781 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2782 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2783 Fix rounding: make it SAE only.
2784 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2786 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2788 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2789 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2790 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2791 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2792 (round_saeonly_mask_scalar_operand4): Ditto.
2793 (round_saeonly_mask_scalar_op3): Ditto.
2794 (round_saeonly_mask_scalar_op4): Ditto.
2796 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2798 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2799 Implement -maltivec=be for vec_insert and vec_extract.
2801 2014-01-10 DJ Delorie <dj@redhat.com>
2803 * config/msp430/msp430.md (call_internal): Don't allow memory
2804 references with SP as the base register.
2805 (call_value_internal): Likewise.
2806 * config/msp430/constraints.md (Yc): New. For memory references
2807 that don't use SP as a base register.
2809 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2810 "an integer without a # prefix"
2811 * config/msp430/msp430.md (epilogue_helper): Use it.
2813 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2816 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2817 AVX512F gather builtins.
2818 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2819 on gather decls with INTEGER_TYPE masktype.
2820 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2821 directly into the builtin rather than hoisting it before loop.
2823 PR tree-optimization/59387
2824 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2825 (scev_const_prop): If folded_casts and type has undefined overflow,
2826 use force_gimple_operand instead of force_gimple_operand_gsi and
2827 for each added stmt if it is assign with
2828 arith_code_with_undefined_signed_overflow, call
2829 rewrite_to_defined_overflow.
2830 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2831 gimple-fold.h instead.
2832 (arith_code_with_undefined_signed_overflow,
2833 rewrite_to_defined_overflow): Moved to ...
2834 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2835 rewrite_to_defined_overflow): ... here. No longer static.
2836 Include gimplify-me.h.
2837 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2838 rewrite_to_defined_overflow): New prototypes.
2840 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2842 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2844 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2846 * builtins.c (get_object_alignment_2): Minor tweak.
2847 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2849 2014-01-13 Christian Bruel <christian.bruel@st.com>
2851 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2852 optimized non constant lengths.
2854 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2857 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2858 load as __atomic_load_N if possible.
2860 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2862 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2864 (rs6000_expand_builtin): Adjust call.
2866 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2869 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2870 * config/rs6000/rs6000.c: Include target-globals.h.
2871 (rs6000_set_current_function): Instead of doing target_reinit
2872 unconditionally, use save_target_globals_default_opts and
2873 restore_target_globals.
2875 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2877 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2878 (rs6000_expand_builtin): Handle mffs and mtfsf.
2879 (rs6000_init_builtins): Define mffs and mtfsf.
2880 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2881 (rs6000_mffs): New pattern.
2882 (rs6000_mtfsf): New pattern.
2884 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2886 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2887 Start narrowing with START. Apply candidate-use pair
2888 and check overall cost in narrowing.
2889 (iv_ca_prune): Pass new argument.
2891 2014-01-10 Jeff Law <law@redhat.com>
2894 * ree.c (combine_reaching_defs): Ensure the defining statement
2895 occurs before the extension when optimizing extensions with
2896 different source and destination hard registers.
2898 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2901 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2902 vtables into the type inheritance graph.
2904 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2906 PR rtl-optimization/59754
2907 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2908 modes in the REGNO != REGNO case.
2910 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2912 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2914 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2916 PR tree-optimization/59745
2917 * tree-predcom.c (tree_predictive_commoning_loop): Call
2918 free_affine_expand_cache if giving up because components is NULL.
2920 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2921 GC in payload of target_globals struct instead of allocating them on
2922 the heap and the larger structs separately using GC.
2923 * target-globals.h (struct target_globals): Make regs, hard_regs,
2924 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2925 of GTY((skip)) and change type to void *.
2926 (reset_target_globals): Cast loads from those fields to corresponding
2929 2014-01-10 Steve Ellcey <sellcey@mips.com>
2932 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2933 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2934 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2936 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2939 * aarch64-modes.def (CC_Zmode): New flags mode.
2940 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2941 represents an equality.
2942 (aarch64_get_condition_code): Handle CC_Zmode.
2943 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2945 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2947 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2948 extraction in good case.
2950 2014-01-10 Richard Biener <rguenther@suse.de>
2952 PR tree-optimization/59374
2953 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2954 checking after SLP discovery. Mark stmts not participating
2955 in any SLP instance properly.
2957 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2959 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2960 when handling a SET rtx.
2962 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2964 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2965 (cortex-a57): Likewise.
2966 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2968 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2970 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2971 non-iwmmxt builtins.
2973 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2977 * ipa-devirt.c record_target_from_binfo): Take as argument
2978 stack of binfos and lookup matching one for virtual inheritance.
2979 (possible_polymorphic_call_targets_1): Update.
2981 2014-01-10 Huacai Chen <chenhc@lemote.com>
2983 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
2984 kernel strings for Loongson-2E/2F/3A.
2986 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2989 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
2990 is_gimple_call before calling gimple_call_internal_p.
2992 2014-01-09 Steve Ellcey <sellcey@mips.com>
2994 * Makefile.in (TREE_FLOW_H): Remove.
2995 (TREE_SSA_H): Add file names from tree-flow.h.
2996 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
2997 * tree.h: Remove tree-flow.h reference.
2998 * hash-table.h: Remove tree-flow.h reference.
2999 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3000 reference with tree-ssa-loop.h.
3002 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3004 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3005 default element-order behavior for -maltivec.
3006 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3007 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3008 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3009 when targeting big endian, at least for now.
3010 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3012 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3015 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3016 var satisfies use_register_for_decl, just take into account type
3017 alignment, rather than decl alignment.
3019 PR tree-optimization/59622
3020 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3021 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3022 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3023 Don't devirtualize for inplace at all. For targets.length () == 1,
3024 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3026 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3028 * config/i386/i386.md (cpu): Remove the unused btver1.
3030 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3032 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3034 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3037 * tree-core.h (struct target_globals): New forward declaration.
3038 (struct tree_target_option): Add globals field.
3039 * tree.h (TREE_TARGET_GLOBALS): Define.
3040 (prepare_target_option_nodes_for_pch): New prototype.
3041 * target-globals.h (struct target_globals): Define even if
3043 * tree.c (prepare_target_option_node_for_pch,
3044 prepare_target_option_nodes_for_pch): New functions.
3045 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3046 * config/i386/i386.c: Include target-globals.h.
3047 (ix86_set_current_function): Instead of doing target_reinit
3048 unconditionally, use save_target_globals_default_opts and
3049 restore_target_globals.
3051 2014-01-09 Richard Biener <rguenther@suse.de>
3053 PR tree-optimization/59715
3054 * tree-cfg.h (split_critical_edges): Declare.
3055 * tree-cfg.c (split_critical_edges): Export.
3056 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3058 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3060 * cfgexpand.c (expand_stack_vars): Optionally disable
3061 asan stack protection.
3062 (expand_used_vars): Likewise.
3063 (partition_stack_vars): Likewise.
3064 * asan.c (asan_emit_stack_protection): Optionally disable
3065 after return stack usage.
3066 (instrument_derefs): Optionally disable memory access instrumentation.
3067 (instrument_builtin_call): Likewise.
3068 (instrument_strlen_call): Likewise.
3069 (asan_protect_global): Optionally disable global variables protection.
3070 * doc/invoke.texi: Added doc for new options.
3071 * params.def: Added new options.
3072 * params.h: Likewise.
3074 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3076 PR rtl-optimization/59724
3077 * ifcvt.c (cond_exec_process_if_block): Don't call
3078 flow_find_head_matching_sequence with 0 longest_match.
3079 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3080 non-active insns if !stop_after.
3081 (try_head_merge_bb): Revert 2014-01-07 changes.
3083 2014-01-08 Jeff Law <law@redhat.com>
3085 * ree.c (get_sub_rtx): New function, extracted from...
3086 (merge_def_and_ext): Here.
3087 (combine_reaching_defs): Use get_sub_rtx.
3089 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3091 * cgraph.h (varpool_variable_node): Do not choke on null node.
3093 2014-01-08 Catherine Moore <clm@codesourcery.com>
3095 * config/mips/mips.md (simple_return): Attempt to use JRC
3097 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3099 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3101 PR rtl-optimization/59137
3102 * reorg.c (steal_delay_list_from_target): Call update_block for
3104 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3106 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3108 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3109 two duplicate entries.
3111 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3114 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3116 * config/mips/mips.c (mips_truncated_op_cost): New function.
3117 (mips_rtx_costs): Adjust test for BADDU.
3118 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3120 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3122 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3123 (*baddu_si): ...this new pattern.
3125 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3128 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3130 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3133 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3135 (expand_expr, expand_normal): Adjust.
3136 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3137 inner_reference_p. Use inner_reference_p to expand inner references.
3138 (store_expr): Adjust.
3139 * cfgexpand.c (expand_call_stmt): Adjust.
3141 2014-01-08 Rong Xu <xur@google.com>
3143 * gcov-io.c (gcov_var): Move from gcov-io.h.
3144 (gcov_position): Ditto.
3145 (gcov_is_error): Ditto.
3146 (gcov_rewrite): Ditto.
3147 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3148 only part to libgcc/libgcov.h.
3150 2014-01-08 Marek Polacek <polacek@redhat.com>
3153 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3155 2014-01-08 Marek Polacek <polacek@redhat.com>
3158 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3160 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3162 PR rtl-optimization/59649
3163 * stor-layout.c (get_mode_bounds): For BImode return
3164 0 and STORE_FLAG_VALUE.
3166 2014-01-08 Richard Biener <rguenther@suse.de>
3169 * gimple.h (is_gimple_builtin_call): Remove.
3170 (gimple_builtin_call_types_compatible_p): New.
3171 (gimple_call_builtin_p): New overload.
3172 * gimple.c (is_gimple_builtin_call): Remove.
3173 (validate_call): Rename to ...
3174 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3176 (validate_type): New static function.
3177 (gimple_call_builtin_p): New overload and adjust.
3178 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3179 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3180 (gimple_fold_stmt_to_constant_1): Likewise.
3181 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3183 2014-01-08 Richard Biener <rguenther@suse.de>
3186 * gimplify.c (gimplify_expr): Gimplify register-register type
3187 VIEW_CONVERT_EXPRs to separate stmts.
3189 2014-01-07 Jeff Law <law@redhat.com>
3192 * ree.c (combine_set_extension): Handle case where source
3193 and destination registers in an extension insn are different.
3194 (combine_reaching_defs): Allow source and destination registers
3195 in extension to be different under limited circumstances.
3196 (add_removable_extension): Remove restriction that the
3197 source and destination registers in the extension are the same.
3198 (find_and_remove_re): Emit a copy from the extension's
3199 destination to its source after the defining insn if
3200 the source and destination registers are different.
3203 * ifcvt.c (merge_if_block): If we are merging a block with more than
3204 one successor with a block with no successors, remove any BARRIER
3205 after the second block.
3207 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3209 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3211 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3214 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3215 for 14-bit register offsets when INT14_OK_STRICT is false.
3217 2014-01-07 Roland Stigge <stigge@antcom.de>
3218 Michael Meissner <meissner@linux.vnet.ibm.com>
3221 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3222 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3224 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3226 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3229 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3231 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3232 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3233 rtx is const0_rtx or not.
3235 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3238 * target-globals.c (save_target_globals): Remove this_fn_optab
3240 * toplev.c: Include optabs.h.
3241 (target_reinit): Temporarily restore the global options if another
3242 set of options are in force.
3244 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3246 PR rtl-optimization/58668
3247 * cfgcleanup.c (flow_find_cross_jump): Don't count
3248 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3249 to determine what is counted.
3250 (flow_find_head_matching_sequence): Use active_insn_p to determine
3252 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3254 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3255 determine what is counted.
3257 PR tree-optimization/59643
3258 * tree-predcom.c (split_data_refs_to_components): If one dr is
3259 read and one write, determine_offset fails and the write isn't
3260 in the bad component, just put the read into the bad component.
3262 2014-01-07 Mike Stump <mikestump@comcast.net>
3263 Jakub Jelinek <jakub@redhat.com>
3266 * tree-core.h (struct tree_optimization_option): Change optabs
3267 type from unsigned char * to void *.
3268 * optabs.c (init_tree_optimization_optabs): Adjust
3269 TREE_OPTIMIZATION_OPTABS initialization.
3271 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3274 * config/i386/i386.h (struct machine_function): Add
3275 no_drap_save_restore field.
3276 * config/i386/i386.c (ix86_save_reg): Use
3277 !cfun->machine->no_drap_save_restore instead of
3278 crtl->stack_realign_needed.
3279 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3280 this function clears frame_pointer_needed. Set
3281 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3282 and DRAP reg is needed.
3284 2014-01-06 Marek Polacek <polacek@redhat.com>
3287 * doc/implement-c.texi: Mention that other integer types are
3288 permitted as bit-field types in strictly conforming mode.
3290 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3292 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3295 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3297 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3299 2014-01-06 Martin Jambor <mjambor@suse.cz>
3302 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3304 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3306 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3310 * var-tracking.c (add_stores): Preserve the value of the source even if
3311 we don't record the store.
3313 2014-01-06 Terry Guo <terry.guo@arm.com>
3315 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3317 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3320 * config/darwin.c (darwin_function_section): Adjust return values to
3321 correspond to optimisation changes made in r206070.
3323 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3325 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3326 from prefetch_block tune setting.
3327 (nocona_cost): Correct size of prefetch block to 64.
3329 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3331 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3332 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3333 used to save the static chain register in the computation of the offset
3334 from which the FP registers need to be restored.
3336 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3338 PR tree-optimization/59519
3339 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3340 ICE if get_current_def (current_new_name) is already non-NULL, as long
3341 as it is a phi result of some other phi in *new_exit_bb that has
3344 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3345 or vmovdqu* for misaligned_operand.
3346 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3347 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3348 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3349 aligned_mem for AVX512F masked aligned load and store builtins and for
3352 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3354 PR tree-optimization/59651
3355 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3356 Address range for negative step should be added by TYPE_SIZE_UNIT.
3358 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3360 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3361 overlapping registers also for registers other than the stack pointer.
3363 2014-01-03 Marek Polacek <polacek@redhat.com>
3366 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3369 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3372 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3375 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3376 (push splitter): Use <P:MODE_SIZE> instead of
3377 GET_MODE_SIZE (<P:MODE>mode).
3378 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3379 (mov -1, reg peephole2): Likewise.
3380 * config/i386/sse.md (*mov<mode>_internal,
3381 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3382 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3383 *<code><mode>3, *andnot<mode>3<mask_name>,
3384 <mask_codefor><code><mode>3<mask_name>): Likewise.
3385 * config/i386/subst.md (mask_mode512bit_condition,
3386 sd_mask_mode512bit_condition): Likewise.
3388 2014-01-02 Xinliang David Li <davidxl@google.com>
3390 PR tree-optimization/59303
3391 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3392 (dump_predicates): Better output format.
3393 (pred_equal_p): New function.
3394 (is_neq_relop_p): Ditto.
3395 (is_neq_zero_form_p): Ditto.
3396 (pred_expr_equal_p): Ditto.
3397 (pred_neg_p): Ditto.
3398 (simplify_pred): Ditto.
3399 (simplify_preds_2): Ditto.
3400 (simplify_preds_3): Ditto.
3401 (simplify_preds_4): Ditto.
3402 (simplify_preds): Ditto.
3404 (push_to_worklist): Ditto.
3405 (get_pred_info_from_cmp): Ditto.
3406 (is_degenerated_phi): Ditto.
3407 (normalize_one_pred_1): Ditto.
3408 (normalize_one_pred): Ditto.
3409 (normalize_one_pred_chain): Ditto.
3410 (normalize_preds): Ditto.
3411 (normalize_cond_1): Remove function.
3412 (normalize_cond): Ditto.
3413 (is_gcond_subset_of): Ditto.
3414 (is_subset_of_any): Ditto.
3415 (is_or_set_subset_of): Ditto.
3416 (is_and_set_subset_of): Ditto.
3417 (is_norm_cond_subset_of): Ditto.
3418 (pred_chain_length_cmp): Ditto.
3419 (convert_control_dep_chain_into_preds): Type change.
3420 (find_predicates): Ditto.
3421 (find_def_preds): Ditto.
3422 (destroy_predicates_vecs): Ditto.
3423 (find_matching_predicates_in_rest_chains): Ditto.
3424 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3425 (is_pred_expr_subset): Ditto.
3426 (is_pred_chain_subset_of): Ditto.
3427 (is_included_in): Ditto.
3428 (is_superset_of): Ditto.
3430 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3432 Update copyright years.
3434 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3436 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3437 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3438 config/arc/arc.md, config/arc/arc.opt,
3439 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3440 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3441 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3442 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3443 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3444 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3445 vtable-verify.c, vtable-verify.h: Use the standard form for the
3448 2014-01-02 Tobias Burnus <burnus@net-b.de>
3450 * gcc.c (process_command): Update copyright notice dates.
3451 * gcov-dump.c: Ditto.
3453 * doc/cpp.texi: Bump @copying's copyright year.
3454 * doc/cppinternals.texi: Ditto.
3455 * doc/gcc.texi: Ditto.
3456 * doc/gccint.texi: Ditto.
3457 * doc/gcov.texi: Ditto.
3458 * doc/install.texi: Ditto.
3459 * doc/invoke.texi: Ditto.
3461 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3463 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3465 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3467 * config/i386/sse.md (*mov<mode>_internal): Guard
3468 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3470 PR rtl-optimization/59647
3471 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3472 new_rtx into UNSIGNED_FLOAT rtxes.
3474 Copyright (C) 2014 Free Software Foundation, Inc.
3476 Copying and distribution of this file, with or without modification,
3477 are permitted in any medium without royalty provided the copyright
3478 notice and this notice are preserved.