1 2014-02-14 Richard Biener <rguenther@suse.de>
3 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
4 (get_frame_arg): Drop the assert with langhook types_compatible_p.
5 Do not strip INDIRECT_REFs.
7 2014-02-14 Richard Biener <rguenther@suse.de>
10 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
11 DECL_FUNCTION_SPECIFIC_TARGET.
12 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
13 * tree-streamer-out.c (pack_ts_target_option): Remove.
14 (streamer_pack_tree_bitfields): Do not stream
16 (write_ts_function_decl_tree_pointers): Do not stream
17 DECL_FUNCTION_SPECIFIC_TARGET.
18 * tree-streamer-in.c (unpack_ts_target_option): Remove.
19 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
20 (lto_input_ts_function_decl_tree_pointers): Do not stream
21 DECL_FUNCTION_SPECIFIC_TARGET.
23 2014-02-14 Jakub Jelinek <jakub@redhat.com>
25 * tree-vect-loop.c (vect_is_slp_reduction): Don't set
27 (get_initial_def_for_induction, vectorizable_induction): Ignore
28 debug stmts when looking for exit_phi.
29 (vectorizable_live_operation): Fix up condition.
31 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
33 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
34 nreverse() because it changes the content of original tree list.
36 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
38 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
39 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
41 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
43 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
46 2014-02-13 Jakub Jelinek <jakub@redhat.com>
49 * dwarf2out.c (gen_subprogram_die): Don't call
50 add_calling_convention_attribute if subr_die is old_die.
52 2014-02-13 Sharad Singhai <singhai@google.com>
54 * doc/optinfo.texi: Fix order of nodes.
56 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
58 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
59 operands[2], not operands[3].
61 2014-02-13 Richard Biener <rguenther@suse.de>
64 * doc/install.texi (ISL): Update recommended version to 0.12.2,
65 mention the possibility of an in-tree build.
66 (CLooG): Update recommended version to 0.18.1, mention the
67 possibility of an in-tree build and clarify that the ISL
68 bundled with CLooG does not work.
70 2014-02-13 Jakub Jelinek <jakub@redhat.com>
73 * expr.c (compress_float_constant): If x is a hard register,
74 extend into a pseudo and then move to x.
76 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
78 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
79 caused by bad second argument to warning_at() with -mhotpatch and
80 nested functions (e.g. with gfortran).
82 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
84 * opts.c (option_name): Remove "enabled by default" rider.
86 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
88 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
90 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
91 Uros Bizjak <ubizjak@gmail.com>
94 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
95 * configure: Regenerated.
97 2014-02-12 Richard Biener <rguenther@suse.de>
99 * vec.c (vec_prefix::calculate_allocation): Move as
100 inline variant to vec.h.
101 (vec_prefix::calculate_allocation_1): New out-of-line version.
102 * vec.h (vec_prefix::calculate_allocation_1): Declare.
103 (vec_prefix::m_has_auto_buf): Rename to ...
104 (vec_prefix::m_using_auto_storage): ... this.
105 (vec_prefix::calculate_allocation): Inline the easy cases
106 and dispatch to calculate_allocation_1 which doesn't need the
108 (va_heap::reserve): Use gcc_checking_assert.
109 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
110 m_using_auto_storage.
111 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
113 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
114 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
115 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
117 2014-02-12 Richard Biener <rguenther@suse.de>
119 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
120 when we found a dependence.
122 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
124 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
126 (maybe_fold_stmt): ... into this new function.
127 * omp-low.c (lower_omp): Update comment.
129 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
132 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
135 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
137 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
138 identifiers in comments.
139 (cortexa53_extra_costs): Likewise.
140 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
141 (cortexa7_extra_costs): Likewise.
142 (cortexa12_extra_costs): Likewise.
143 (cortexa15_extra_costs): Likewise.
144 (v7m_extra_costs): Likewise.
146 2014-02-12 Richard Biener <rguenther@suse.de>
149 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
150 of posix_memalign being successful.
151 (lower_stmt): Restrict lowering of posix_memalign to when
152 -ftree-bit-ccp is enabled.
154 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
156 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
158 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
160 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
162 PR rtl-optimization/60116
163 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
164 other_insn once the combination has been validated.
166 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
169 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
171 * ipa-devirt.c: Include demangle.h
172 (odr_violation_reported): New static variable.
173 (add_type_duplicate): Update odr_violations.
174 (maybe_record_node): Add completep parameter; update it.
175 (record_target_from_binfo): Add COMPLETEP parameter;
177 (possible_polymorphic_call_targets_1): Likewise.
178 (struct polymorphic_call_target_d): Add nonconstruction_targets;
179 rename FINAL to COMPLETE.
180 (record_targets_from_bases): Sanity check we found the binfo;
181 fix COMPLETEP updating.
182 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
183 parameter, fix computing of COMPLETEP.
184 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
185 at LTO time do demangling.
186 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
187 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
189 (gimple_get_virt_method_for_binfo): Likewise.
190 * gimple-fold.h (gimple_get_virt_method_for_binfo,
191 gimple_get_virt_method_for_vtable): Update prototypes.
193 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
196 * genautomata.c (add_presence_absence): Fix typo with
197 {final_}presence_list.
199 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
202 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
203 for VSX/Altivec vectors that land in GPR registers.
205 2014-02-11 Richard Henderson <rth@redhat.com>
206 Jakub Jelinek <jakub@redhat.com>
209 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
210 around drhs if type conversion to lacc->type is not useless.
212 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
214 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
216 (cortex-a57.cortex-a53): Likewise.
217 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
219 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
221 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
224 2014-02-11 Renlin Li <Renlin.Li@arm.com>
226 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
227 add_options_for_arm_vfp3.
229 2014-02-11 Jeff Law <law@redhat.com>
232 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
233 object with an undesirable mode.
235 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
238 * config/i386/sol2-9.h: New file.
239 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
240 *-*-solaris2.9*): Use it.
242 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
244 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
245 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
247 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
249 * config/microblaze/microblaze.c: Extend mcpu version format
251 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
253 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
255 2014-02-10 Richard Henderson <rth@redhat.com>
258 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
259 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
260 ms-abi vs -mno-accumulate-outgoing-args.
261 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
262 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
265 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
268 * cfgexpand.c (expand_asm_operands): Attach source location to
269 ASM_INPUT rtx objects.
270 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
272 2014-02-10 Nick Clifton <nickc@redhat.com>
274 * config/mn10300/mn10300.c (popcount): New function.
275 (mn10300_expand_prologue): Include saved registers in stack usage
278 2014-02-10 Jeff Law <law@redhat.com>
281 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
282 when changing the SET_DEST of a prior insn to avoid an input reload.
284 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
286 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
287 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
288 -mcall-openbsd, or -mcall-linux.
289 (CC1_ENDIAN_BIG_SPEC): Remove.
290 (CC1_ENDIAN_LITTLE_SPEC): Remove.
291 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
292 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
293 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
294 and %cc1_endian_default.
295 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
297 2014-02-10 Richard Biener <rguenther@suse.de>
299 PR tree-optimization/60115
300 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
301 MEM_REF handling. Properly verify that the accesses are not
302 out of the objects bound.
304 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
306 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
309 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
311 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
312 proper constants and fix formatting.
313 (possible_polymorphic_call_targets): Fix formatting.
315 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
316 Ilya Tocar <ilya.tocar@intel.com>
318 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
319 (_mm512_loadu_epi32): Renamed into...
320 (_mm512_loadu_si512): This.
321 (_mm512_storeu_epi32): Renamed into...
322 (_mm512_storeu_si512): This.
323 (_mm512_maskz_ceil_ps): Removed.
324 (_mm512_maskz_ceil_pd): Ditto.
325 (_mm512_maskz_floor_ps): Ditto.
326 (_mm512_maskz_floor_pd): Ditto.
327 (_mm512_floor_round_ps): Ditto.
328 (_mm512_floor_round_pd): Ditto.
329 (_mm512_ceil_round_ps): Ditto.
330 (_mm512_ceil_round_pd): Ditto.
331 (_mm512_mask_floor_round_ps): Ditto.
332 (_mm512_mask_floor_round_pd): Ditto.
333 (_mm512_mask_ceil_round_ps): Ditto.
334 (_mm512_mask_ceil_round_pd): Ditto.
335 (_mm512_maskz_floor_round_ps): Ditto.
336 (_mm512_maskz_floor_round_pd): Ditto.
337 (_mm512_maskz_ceil_round_ps): Ditto.
338 (_mm512_maskz_ceil_round_pd): Ditto.
339 (_mm512_expand_pd): Ditto.
340 (_mm512_expand_ps): Ditto.
341 * config/i386/i386.c (ix86_builtins): Remove
342 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
344 * config/i386/predicates.md (const1256_operand): New.
345 (const_1_to_2_operand): Ditto.
346 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
347 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
348 (*avx512pf_gatherpf<mode>sf): Ditto.
349 (avx512pf_gatherpf<mode>df): Ditto.
350 (*avx512pf_gatherpf<mode>df_mask): Ditto.
351 (*avx512pf_gatherpf<mode>df): Ditto.
352 (avx512pf_scatterpf<mode>sf): Ditto.
353 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
354 (*avx512pf_scatterpf<mode>sf): Ditto.
355 (avx512pf_scatterpf<mode>df): Ditto.
356 (*avx512pf_scatterpf<mode>df_mask): Ditto.
357 (*avx512pf_scatterpf<mode>df): Ditto.
358 (avx512f_expand<mode>): Removed.
359 (<shift_insn><mode>3<mask_name>): Change predicate type.
361 2014-02-08 Jakub Jelinek <jakub@redhat.com>
363 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
364 not at the end of datarefs vector use ordered_remove to avoid
365 reordering datarefs vector.
368 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
369 mark local addressable non-static vars as GOVD_PRIVATE
370 instead of GOVD_LOCAL.
371 * omp-low.c (lower_omp_for): Move gimple_bind_vars
372 and BLOCK_VARS of gimple_bind_block to new_stmt rather
376 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
377 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
378 assume_aligned or alloc_align attributes.
379 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
380 arguments. Handle also assume_aligned and alloc_align attributes.
381 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
382 calls to functions with assume_aligned or alloc_align attributes.
383 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
385 2014-02-08 Terry Guo <terry.guo@arm.com>
387 * doc/invoke.texi: Document ARM -march=armv7e-m.
389 2014-02-08 Jakub Jelinek <jakub@redhat.com>
391 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
392 flag on __cilkrts_rethrow builtin.
395 * ipa-cp.c (determine_versionability): Fail at -O0
396 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
397 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
400 2014-02-04 Jakub Jelinek <jakub@redhat.com>
403 * tree-inline.c (copy_forbidden): Fail for
404 __attribute__((optimize (0))) functions.
406 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
408 * varpool.c: Include pointer-set.h.
409 (varpool_remove_unreferenced_decls): Variables in other partitions
410 will not be output; be however careful to not lose information
413 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
415 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
416 lookup in the vtable constructor.
418 2014-02-07 Jeff Law <law@redhat.com>
421 * config/m68k/m68k.md (ashldi_extsi): Turn into a
422 define_insn_and_split.
424 * ipa-inline.c (inline_small_functions): Fix typos.
426 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
428 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
429 (s390_can_use_return_insn): Declare.
430 * config/s390/s390.h (EPILOGUE_USES): Define.
431 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
433 (s390_chunkify_start): Handle return JUMP_LABELs.
434 (s390_early_mach): Emit a main_pool instruction on the entry edge.
435 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
436 (s390_can_use_return_insn): New functions.
437 (s390_fix_long_loop_prediction): Handle conditional returns.
438 (TARGET_SET_UP_BY_PROLOGUE): Define.
439 * config/s390/s390.md (ANY_RETURN): New code iterator.
440 (*creturn, *csimple_return, return, simple_return): New patterns.
442 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
444 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
445 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
446 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
447 REG_CFA_RESTORE list when deciding not to restore a register.
449 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
451 * config/s390/s390.c: Include tree-pass.h and context.h.
452 (s390_early_mach): New function, split out from...
453 (s390_emit_prologue): ...here.
454 (pass_data_s390_early_mach): New pass structure.
455 (pass_s390_early_mach): New class.
456 (s390_option_override): Create and register early_mach pass.
459 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
461 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
462 to match for the exit block.
464 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
466 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
467 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
468 Reject misaligned operands.
470 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
472 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
474 2014-02-07 Richard Biener <rguenther@suse.de>
477 * gimple-low.c (lower_builtin_posix_memalign): New function.
478 (lower_stmt): Call it to lower posix_memalign in a way
479 to make alignment info accessible.
481 2014-02-07 Jakub Jelinek <jakub@redhat.com>
484 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
485 __builtin_setjmp_receiver.
487 2014-02-07 Richard Biener <rguenther@suse.de>
490 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
491 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
492 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
493 Handle BUILT_IN_POSIX_MEMALIGN.
494 (find_func_clobbers): Likewise.
495 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
496 (call_may_clobber_ref_p_1): Likewise.
498 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
501 * ipa-devirt.c (record_target_from_binfo): Remove overactive
504 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
507 * lto-cgraph.c (lto_output_node): Use
508 symtab_get_symbol_partitioning_class.
509 (lto_output_varpool_node): likewise.
510 (symtab_get_symbol_partitioning_class): Move here from
512 * cgraph.h (symbol_partitioning_class): Likewise.
513 (symtab_get_symbol_partitioning_class): Declare.
515 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
517 * ggc.h (ggc_internal_cleared_alloc): New macro.
518 * vec.h (vec_safe_copy): Handle memory stats.
519 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
520 * target-globals.c (save_target_globals): Likewise.
522 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
525 * expr.c (emit_move_resolve_push): Export; be bit more selective
526 on when to clear alias set.
527 * expr.h (emit_move_resolve_push): Declare.
528 * function.h (struct function): Add tail_call_marked.
529 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
530 * config/i386/i386-protos.h (ix86_expand_push): Remove.
531 * config/i386/i386.md (TImode move expander): De not call
533 (FP push expanders): Preserve memory attributes.
534 * config/i386/sse.md (push<mode>1): Remove.
535 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
536 (ix86_expand_push): Remove.
537 * config/i386/mmx.md (push<mode>1): Remove.
539 2014-02-06 Jakub Jelinek <jakub@redhat.com>
541 PR rtl-optimization/60030
542 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
543 lopart with paradoxical subreg before shifting it up by hprec.
545 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
547 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
548 Remove extra newline at end of file.
549 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
550 (arm_issue_rate): Handle cortexa57.
551 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
552 (cortex-a57.cortex-a53): Likewise.
554 2014-02-06 Jakub Jelinek <jakub@redhat.com>
557 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
558 don't record in REG_FRAME_RELATED_EXPR registers not set in that
560 (arm_expand_prologue): Adjust all callers.
561 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
562 info, registers also at the lowest numbered registers side. Use
563 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
567 * var-tracking.c (adjust_mems): Before adding a SET to
568 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
570 2014-02-06 Alan Modra <amodra@gmail.com>
573 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
574 change SDmode to DDmode when lra_in_progress.
576 2014-02-06 Jakub Jelinek <jakub@redhat.com>
579 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
580 free_data_ref on the dr first, and before goto again also set dr
581 to the next dr. For simd_lane_access, free old datarefs[i] before
582 overwriting it. For get_vectype_for_scalar_type failure, don't
583 free_data_ref if simd_lane_access.
585 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
588 * tree.h (opts_for_fn): New inline function.
589 (opt_for_fn): Define.
590 * config/i386/i386.c (ix86_function_regparm): Use
591 opt_for_fn (decl, optimize) instead of optimize.
593 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
595 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
596 for SYMBOL_REF in large memory model.
598 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
600 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
602 (cortex-a57): Likewise.
603 (cortex-a57.cortex-a53): Likewise.
605 2014-02-06 Yury Gribov <y.gribov@samsung.com>
606 Kugan Vivekanandarajah <kuganv@linaro.org>
608 * config/arm/arm.c (arm_vector_alignment_reachable): Check
610 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
612 2014-02-06 Richard Biener <rguenther@suse.de>
614 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
615 set_loop_copy and initialize_original_copy_tables.
617 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
619 * config/aarch64/aarch64-simd.md
620 (aarch64_ashr_simddi): Change QI to SI.
622 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
623 Jakub Jelinek <jakub@redhat.com>
626 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
629 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
631 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
632 CODE_FOR_altivec_vpku[hw]um to
633 CODE_FOR_altivec_vpku[hw]um_direct.
634 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
635 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
636 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
637 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
639 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
641 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
642 generation for -maltivec=be.
643 (altivec_vsumsws): Simplify redundant test.
645 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
647 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
648 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
649 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
650 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
652 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
654 (altivec_vpks<VI_char>ss): Likewise.
655 (altivec_vpks<VI_char>us): Likewise.
656 (altivec_vpku<VI_char>us): Likewise.
657 (altivec_vpku<VI_char>um): Likewise.
658 (altivec_vpku<VI_char>um_direct): New (copy of
659 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
661 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
662 target is little endian and -maltivec=be is not specified.
663 (*altivec_vupkhs<VU_char>_direct): New (copy of
664 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
665 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
666 target is little endian and -maltivec=be is not specified.
667 (*altivec_vupkls<VU_char>_direct): New (copy of
668 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
669 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
670 little endian and -maltivec=be is not specified.
671 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
672 little endian and -maltivec=be is not specified.
674 2014-02-05 Richard Henderson <rth@redhat.com>
677 * combine-stack-adj.c: Revert r206943.
678 * sched-int.h (struct deps_desc): Add last_args_size.
679 * sched-deps.c (init_deps): Initialize it.
680 (sched_analyze_insn): Add OUTPUT dependencies between insns that
681 contain REG_ARGS_SIZE notes.
683 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
685 * lto-cgraph.c (asm_nodes_output): Make global.
686 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
687 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
688 (driver_handle_option): Handle OPT_fwpa.
690 2014-02-05 Jakub Jelinek <jakub@redhat.com>
693 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
694 a comment typo and formatting issue. If odr_hash hasn't been
695 created, return vNULL and set *completep to false.
698 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
699 bb with no successors.
701 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
704 * doc/invoke.texi (-march): Clarify documentation for ARM.
708 2014-02-05 Richard Biener <rguenther@suse.de>
710 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
711 when not vectorizing because of too many alias checks.
712 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
713 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
715 2014-02-05 Nick Clifton <nickc@redhat.com>
717 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
718 accept extended registers in any mode when compiling for the MN10300.
720 2014-02-05 Yury Gribov <y.gribov@samsung.com>
722 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
723 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
724 sanitization attributes.
725 (can_inline_edge_p): Likewise.
726 (sanitize_attrs_match_for_inline_p): New function.
728 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
730 * ipa-prop.c (detect_type_change): Shor circuit testing of
731 type changes on THIS pointer.
733 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
736 * config/pa/pa.c (legitimize_tls_address): Return original address
737 if not passed a SYMBOL_REF rtx.
738 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
740 (pa_emit_move_sequence): Simplify TLS source operands.
741 (pa_legitimate_constant_p): Reject all TLS constants.
742 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
743 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
745 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
747 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
748 groups when we know they are controlled by LTO.
749 * varasm.c (default_binds_local_p_1): If object is in other partition,
750 it will be resolved locally.
752 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
754 * config/host-linux.c (linux_gt_pch_use_address): Don't
755 use SSIZE_MAX because it is not always defined.
757 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
760 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
761 threshold for pseudo splitting.
762 (update_ebb_live_info): Process call argument hard registers and
763 hard registers from insn definition too.
764 (max_small_class_regs_num): New constant.
765 (inherit_in_ebb): Update live hard regs through EBBs. Update
766 reloads_num only for small register classes. Don't split for
769 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
772 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
775 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
777 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
780 2014-02-04 Marek Polacek <polacek@redhat.com>
782 * gdbinit.in (pel): Define.
784 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
786 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
789 2014-02-04 Richard Biener <rguenther@suse.de>
792 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
793 in function context local.
794 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
795 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
796 similar to LTO_imported_decl_ref.
798 2014-02-04 Jakub Jelinek <jakub@redhat.com>
800 PR tree-optimization/60002
801 * cgraphclones.c (build_function_decl_skip_args): Clear
804 PR tree-optimization/60023
805 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
806 false to gsi_replace.
807 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
808 has been in some EH region and vec_stmt could throw, add
809 vec_stmt into the same EH region.
810 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
811 has no lhs, ignore it.
812 * internal-fn.c (expand_MASK_LOAD): Likewise.
815 * tree-inline.c (copy_forbidden): Fail for
816 __attribute__((optimize (0))) functions.
819 * omp-low.c (simd_clone_struct_copy): If from->inbranch
820 is set, copy one less argument.
821 (expand_simd_clones): Don't subtract clone_info->inbranch
822 from simd_clone_struct_alloc argument.
824 PR rtl-optimization/57915
825 * recog.c (simplify_while_replacing): If all unary/binary/relational
826 operation arguments are constant, attempt to simplify those.
829 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
830 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
832 2014-02-04 Richard Biener <rguenther@suse.de>
834 PR tree-optimization/60012
835 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
836 TBAA disambiguation to all DDRs.
838 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
841 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
842 (LINK_SPEC): Use it for -shared, -shared-libgcc.
844 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
847 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
849 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
851 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
852 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
854 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
857 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
858 to figure out targets of polymorphic calls with known decl.
859 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
860 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
861 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
862 (get_polymorphic_call_info): ... here.
863 (get_polymorphic_call_info_from_invariant): New function.
865 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
867 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
868 lookup via vtable pointer; check for type consistency
869 and turn inconsitent facts into UNREACHABLE.
870 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
871 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
872 type inconsistent querries; return UNREACHABLE instead.
874 2014-02-03 Richard Henderson <rth@twiddle.net>
877 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
878 already processed this node.
879 (normalize_one_pred_1): Pass along mark_set.
880 (normalize_one_pred): Create and destroy a pointer_set_t.
881 (normalize_one_pred_chain): Likewise.
883 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
885 PR gcov-profile/58602
886 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
888 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
891 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
892 -fno-devirtualize; try to devirtualize by the knowledge of
893 virtual table pointer given by aggregate propagation.
894 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
895 (ipa_print_node_jump_functions): Dump also offset that
896 is relevant for polymorphic calls.
897 (determine_known_aggregate_parts): Add arg_type parameter; use it
898 instead of determining the type from pointer type.
899 (ipa_compute_jump_functions_for_edge): Update call of
900 determine_known_aggregate_parts.
901 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
902 (gimple_get_virt_method_for_binfo): ... here; simplify using
903 vtable_pointer_value_to_vtable.
904 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
905 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
906 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
907 (vtable_pointer_value_to_vtable): Break out from ...; handle also
909 (vtable_pointer_value_to_binfo): ... here.
910 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
912 2014-02-03 Teresa Johnson <tejohnson@google.com>
914 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
915 redef of outer loop index variable.
917 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
921 * doc/extend.texi (Function Attributes): Typo.
923 2014-02-03 Cong Hou <congh@google.com>
925 PR tree-optimization/60000
926 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
927 if the vectorized statement is a store. A store statement can only
928 appear at the end of pattern statements.
930 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
932 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
933 (ix86_option_override_internal): Default long double to 64-bit for
934 32-bit Bionic and to 128-bit for 64-bit Bionic.
936 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
937 TARGET_LONG_DOUBLE_128 is true.
938 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
940 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
941 (mlong-double-64): Negate -mlong-double-128.
942 (mlong-double-128): New option.
944 * config/i386/i386-c.c (ix86_target_macros): Define
945 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
947 * doc/invoke.texi: Document -mlong-double-128.
949 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
951 PR rtl-optimization/60024
952 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
954 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
956 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
958 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
960 PR rtl-optimization/57662
961 * sel-sched.c (code_motion_path_driver): Do not mark already not
962 existing blocks in the visiting bitmap.
964 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
966 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
967 on the insn being emitted.
969 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
970 Will Deacon <will.deacon@arm.com>
972 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
974 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
976 * config/arm/arm-tables.opt: Regenerate.
978 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
980 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
981 for vector types other than V16QImode.
982 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
983 define_expand, and call altivec_expand_vec_perm_le when producing
984 code with little endian element order.
985 (*altivec_vperm_<mode>_internal): New insn having previous
986 behavior of altivec_vperm_<mode>.
987 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
988 altivec_expand_vec_perm_le when producing code with little endian
990 (*altivec_vperm_<mode>_uns_internal): New insn having previous
991 behavior of altivec_vperm_<mode>_uns.
993 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
995 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
996 (altivec_vsumsws): Add handling for -maltivec=be with a little
998 (altivec_vsumsws_direct): New.
999 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1000 gen_altivec_vsumsws.
1002 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1004 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1005 vtable_pointer_value_to_binfo): New functions.
1006 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1007 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1009 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1011 * config/nios2/nios2.md (load_got_register): Initialize GOT
1012 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1013 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1015 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1017 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1018 preserverd by passthrough, do not propagate the type.
1020 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1022 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1023 (mips_atomic_assign_expand_fenv): New function.
1024 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1026 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1028 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1029 (__builtin_mips_set_fcsr): Likewise.
1030 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1031 MIPS_USI_FTYPE_VOID.
1032 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1033 (mips16_expand_set_fcsr): Likewise.
1034 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1035 (mips16_set_fcsr_stub): Likewise.
1036 (mips16_get_fcsr_one_only_stub): New class.
1037 (mips16_set_fcsr_one_only_stub): Likewise.
1038 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1039 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1040 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1041 (hard_float): New availability predicate.
1042 (mips_builtins): Add get_fcsr and set_fcsr.
1043 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1044 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1045 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1046 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1047 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1050 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1052 * config/mips/mips.c (mips_one_only_stub): New class.
1053 (mips_need_mips16_rdhwr_p): Replace with...
1054 (mips16_rdhwr_stub): ...this new variable.
1055 (mips16_stub_call_address): New function.
1056 (mips16_rdhwr_one_only_stub): New class.
1057 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1058 (mips_output_mips16_rdhwr): Delete.
1059 (mips_finish_stub): New function.
1060 (mips_code_end): Use it to handle rdhwr stubs.
1062 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1065 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1066 when calculating size of integer atomic types.
1068 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1070 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1072 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1074 PR tree-optimization/60003
1075 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1076 * profile.c (branch_prob): Use gimple_call_builtin_p
1077 to check for BUILT_IN_SETJMP_RECEIVER.
1078 * tree-inline.c (copy_bb): Call notice_special_calls.
1080 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1083 * lra-constraints.c (process_alt_operands): Update reload_sum only
1086 2014-01-31 Richard Henderson <rth@redhat.com>
1089 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1090 until after else_eh is processed.
1092 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1094 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1095 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1096 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1097 in smmintrin.h, remove them.
1098 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1099 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1100 * config/i386/i386.md (ROUND_SAE): Fix value.
1101 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1102 (const48_operand): New.
1103 * config/i386/subst.md (round), (round_expand): Use
1104 const_4_or_8_to_11_operand.
1105 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1107 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1109 * config/i386/constraints.md (Yk): Swap meaning with k.
1110 * config/i386/i386.md (movhi_internal): Change Yk to k.
1111 (movqi_internal): Ditto.
1112 (*k<logic><mode>): Ditto.
1115 (kandn<mode>): Ditto.
1116 (*<code>hi_1): Ditto.
1117 (*<code>qi_1): Ditto.
1118 (kxnor<mode>): Ditto.
1119 (kortestzhi): Ditto.
1120 (kortestchi): Ditto.
1122 (*one_cmplhi2_1): Ditto.
1123 (*one_cmplqi2_1): Ditto.
1124 * config/i386/sse.md (): Change k to Yk.
1125 (avx512f_load<mode>_mask): Ditto.
1126 (avx512f_blendm<mode>): Ditto.
1127 (avx512f_store<mode>_mask): Ditto.
1128 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1129 (avx512f_storedqu<mode>_mask): Ditto.
1130 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1132 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1133 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1134 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1135 (avx512f_maskcmp<mode>3): Ditto.
1136 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1137 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1138 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1139 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1140 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1141 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1142 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1143 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1144 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1145 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1146 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1147 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1148 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1149 (vec_extract_lo_<mode>_maskm): Ditto.
1150 (vec_extract_hi_<mode>_maskm): Ditto.
1151 (avx512f_vternlog<mode>_mask): Ditto.
1152 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1153 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1154 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1155 (avx512f_<code>v8div16qi2_mask): Ditto.
1156 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1157 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1158 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1159 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1160 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1161 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1162 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1163 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1164 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1165 (avx512cd_maskb_vec_dupv8di): Ditto.
1166 (avx512cd_maskw_vec_dupv16si): Ditto.
1167 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1168 (avx512f_vpermi2var<mode>3_mask): Ditto.
1169 (avx512f_vpermi2var<mode>3_mask): Ditto.
1170 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1171 (*avx512f_gathersi<mode>): Ditto.
1172 (*avx512f_gathersi<mode>_2): Ditto.
1173 (*avx512f_gatherdi<mode>): Ditto.
1174 (*avx512f_gatherdi<mode>_2): Ditto.
1175 (*avx512f_scattersi<mode>): Ditto.
1176 (*avx512f_scatterdi<mode>): Ditto.
1177 (avx512f_compress<mode>_mask): Ditto.
1178 (avx512f_compressstore<mode>_mask): Ditto.
1179 (avx512f_expand<mode>_mask): Ditto.
1180 * config/i386/subst.md (mask): Change k to Yk.
1181 (mask_scalar_merge): Ditto.
1184 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1186 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1188 2014-01-31 Richard Biener <rguenther@suse.de>
1191 * builtins.c (fold_builtin_memory_op): Make sure to not
1192 use a floating-point mode or a boolean or enumeral type for
1195 2014-01-30 DJ Delorie <dj@redhat.com>
1197 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1198 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1199 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1200 whenever main() has an epilogue.
1202 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1204 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1205 unused variable "field".
1206 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1207 (vsx_mergeh_<mode>): Likewise.
1208 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1209 (altivec_vmrghh): Likewise.
1210 (altivec_vmrghw): Likewise.
1211 (altivec_vmrglb): Likewise.
1212 (altivec_vmrglh): Likewise.
1213 (altivec_vmrglw): Likewise.
1214 (altivec_vspltb): Add missing uses.
1215 (altivec_vsplth): Likewise.
1216 (altivec_vspltw): Likewise.
1217 (altivec_vspltsf): Likewise.
1219 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1222 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1223 frame related instructions.
1225 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1227 PR rtl-optimization/59959
1228 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1229 any reload of register whose subreg is invalid.
1231 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1233 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1234 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1235 Add missing return type - void.
1237 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1239 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1240 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1241 remove element index adjustment for endian (now handled in vsx.md
1243 (altivec_expand_vec_perm_const): Use
1244 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1245 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1246 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1247 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1248 define_expand and a new define_insn *altivec_vspltb_internal;
1249 adjust for -maltivec=be on a little endian target.
1250 (altivec_vspltb_direct): New.
1251 (altivec_vsplth): Divide into a define_expand and a new
1252 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1253 little endian target.
1254 (altivec_vsplth_direct): New.
1255 (altivec_vspltw): Divide into a define_expand and a new
1256 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1257 little endian target.
1258 (altivec_vspltw_direct): New.
1259 (altivec_vspltsf): Divide into a define_expand and a new
1260 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1261 a little endian target.
1263 2014-01-30 Richard Biener <rguenther@suse.de>
1265 PR tree-optimization/59993
1266 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1267 can propagate form the earlier stmt and avoid the transform
1268 when the intermediate result is needed.
1270 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1272 * README.Portability: Fix typo.
1274 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1276 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1277 comparison_operator with ordered_comparison_operator.
1279 2014-01-30 Nick Clifton <nickc@redhat.com>
1281 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1282 Rename to mn10300_store_multiple_regs.
1283 * config/mn10300/mn10300.c: Likewise.
1284 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1285 store_multiple_regs.
1286 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1287 Call mn10300_store_multiple_regs.
1289 2014-01-30 Nick Clifton <nickc@redhat.com>
1290 DJ Delorie <dj@redhat.com>
1292 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1293 %fp 2 to keep registers after it properly word-aligned.
1294 (rl78_alloc_physical_registers_umul): Handle the case where both
1295 input operands are the same.
1297 2014-01-30 Richard Biener <rguenther@suse.de>
1299 PR tree-optimization/59903
1300 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1303 2014-01-30 Jason Merrill <jason@redhat.com>
1306 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1309 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1311 2014-01-30 Richard Biener <rguenther@suse.de>
1313 PR tree-optimization/59951
1314 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1316 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1319 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1320 SFmode to DFmode case.
1322 2014-01-29 DJ Delorie <dj@redhat.com>
1324 * config/msp430/msp430.opt (-minrt): New.
1325 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1327 (ENDFILE_SPEC): Likewise.
1329 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1331 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1332 (estimate_function_body_sizes): Use it.
1334 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1337 * dwarf2out.c (is_cxx_auto): New.
1338 (is_base_type): Use it.
1339 (gen_type_die_with_usage): Likewise.
1341 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1343 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1344 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1345 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1346 -maltivec=be with LE targets.
1347 (vsx_mergeh_<mode>): Likewise.
1348 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1349 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1350 (altivec_vmrghb): Replace with define_expand and new
1351 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1352 (altivec_vmrghb_direct): New define_insn.
1353 (altivec_vmrghh): Replace with define_expand and new
1354 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1355 (altivec_vmrghh_direct): New define_insn.
1356 (altivec_vmrghw): Replace with define_expand and new
1357 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1358 (altivec_vmrghw_direct): New define_insn.
1359 (*altivec_vmrghsf): Adjust for endianness.
1360 (altivec_vmrglb): Replace with define_expand and new
1361 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1362 (altivec_vmrglb_direct): New define_insn.
1363 (altivec_vmrglh): Replace with define_expand and new
1364 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1365 (altivec_vmrglh_direct): New define_insn.
1366 (altivec_vmrglw): Replace with define_expand and new
1367 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1368 (altivec_vmrglw_direct): New define_insn.
1369 (*altivec_vmrglsf): Adjust for endianness.
1370 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1371 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1372 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1373 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1374 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1375 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1376 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1377 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1379 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1381 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1382 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1385 2014-01-29 Richard Biener <rguenther@suse.de>
1387 PR tree-optimization/58742
1388 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1389 associate_pointerplus_align.
1390 (associate_pointerplus_diff): New function.
1391 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1392 and associate_pointerplus_diff.
1394 2014-01-29 Richard Biener <rguenther@suse.de>
1396 * lto-streamer.h (LTO_major_version): Bump to 3.
1397 (LTO_minor_version): Reset to 0.
1399 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1401 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1402 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1403 (arm_file_start): Generate correct asm header for armv7ve.
1404 * config/arm/bpabi.h: Add multilib support for armv7ve.
1405 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1406 and cortex-a15 to armv7ve.
1407 * config/arm/t-aprofile: Add multilib support for armv7ve.
1408 * doc/invoke.texi: Document -march=armv7ve.
1410 2014-01-29 Richard Biener <rguenther@suse.de>
1412 PR tree-optimization/58742
1413 * tree-ssa-forwprop.c (associate_plusminus): Return true
1414 if we changed sth, defer EH cleanup to ...
1415 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1416 (simplify_mult): New function.
1418 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1421 PR tree-optimization/59920
1422 * tree.c (build_common_builtin_nodes): Remove
1423 __builtin_setjmp_dispatcher initialization.
1424 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1425 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1426 instead of gsi_after_labels + manually skipping debug stmts.
1427 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1428 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1429 * tree-inline.c (copy_edges_for_bb): Remove
1430 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1431 argument. Ignore computed_goto_p stmts. Don't call
1432 make_abnormal_goto_edges. If a call might need abnormal edges
1433 for non-local gotos, see if it already has an edge to
1434 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1435 with true argument, don't do anything then, otherwise add
1436 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1437 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1439 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1440 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1441 (lower_stmt): Don't set data->calls_builtin_setjmp.
1442 (lower_builtin_setjmp): Adjust comment.
1443 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1444 * tree-cfg.c (found_computed_goto): Remove.
1445 (factor_computed_gotos): Remove.
1446 (make_goto_expr_edges): Return bool, true for computed gotos.
1447 Don't call make_abnormal_goto_edges.
1448 (build_gimple_cfg): Don't set found_computed_goto, don't call
1449 factor_computed_gotos.
1450 (computed_goto_p): No longer static.
1451 (make_blocks): Don't set found_computed_goto.
1452 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1453 (make_edges): If make_goto_expr_edges returns true, push bb
1454 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1455 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1456 vector. Record mapping between bbs and OpenMP regions if there
1457 are any, adjust make_gimple_omp_edges caller. Call
1458 handle_abnormal_edges.
1459 (make_abnormal_goto_edges): Remove.
1460 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1461 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1462 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1463 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1464 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1465 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1466 filling *region also set *region_idx to (*region)->entry->index.
1469 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1470 For REGs set ORIGINAL_REGNO.
1472 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1474 * doc/md.texi: Mention that a target shouldn't implement
1475 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1478 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1480 PR tree-optimization/59594
1481 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1482 a copy of the datarefs vector rather than the vector itself.
1484 2014-01-28 Jason Merrill <jason@redhat.com>
1487 * dwarf2out.c (auto_die): New static.
1488 (gen_type_die_with_usage): Handle C++1y 'auto'.
1489 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1492 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1495 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1496 (SPEC_X32): Likewise.
1497 (SPEC_64): Likewise.
1498 * config/i386/i386.c (ix86_option_override_internal): Turn off
1499 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1501 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1502 * config/i386/i386.h (TARGET_16BIT): New macro.
1503 (TARGET_16BIT_P): Likewise.
1504 * config/i386/i386.opt: Add m16.
1505 * doc/invoke.texi: Document -m16.
1507 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1509 PR preprocessor/59935
1510 * input.c (location_get_source_line): Bail out on when line number
1511 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1513 2014-01-28 Richard Biener <rguenther@suse.de>
1515 PR tree-optimization/58742
1516 * tree-ssa-forwprop.c (associate_plusminus): Handle
1517 pointer subtraction of the form (T)(P + A) - (T)P.
1519 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1521 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1524 2014-01-28 Richard Biener <rguenther@suse.de>
1527 2014-01-28 Richard Biener <rguenther@suse.de>
1529 PR rtl-optimization/45364
1530 PR rtl-optimization/59890
1531 * var-tracking.c (local_get_addr_clear_given_value): Handle
1532 already cleared slot.
1533 (val_reset): Handle not allocated local_get_addr_cache.
1534 (vt_find_locations): Use post-order on the inverted CFG.
1536 2014-01-28 Richard Biener <rguenther@suse.de>
1538 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1540 2014-01-28 Richard Biener <rguenther@suse.de>
1542 PR rtl-optimization/45364
1543 PR rtl-optimization/59890
1544 * var-tracking.c (local_get_addr_clear_given_value): Handle
1545 already cleared slot.
1546 (val_reset): Handle not allocated local_get_addr_cache.
1547 (vt_find_locations): Use post-order on the inverted CFG.
1549 2014-01-28 Alan Modra <amodra@gmail.com>
1551 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1552 * configure.ac <recursive call for build != host>: Define
1553 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1554 and LD_FOR_BUILD too.
1555 * configure: Regenerate.
1557 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1559 * config/i386/i386.c (get_builtin_code_for_version): Separate
1560 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1561 Broadwell from Haswell.
1563 2014-01-27 Steve Ellcey <sellcey@mips.com>
1565 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1566 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1567 * config/mips/mips.c (mips_option_override): Change setting
1569 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1570 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1571 Change from Mask to Var.
1573 2014-01-27 Jeff Law <law@redhat.com>
1575 * ipa-inline.c (inline_small_functions): Fix typo.
1577 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1579 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1580 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1581 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1582 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1583 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1584 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1585 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1586 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1587 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1588 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1589 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1590 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1591 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1592 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1593 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1594 (_mm512_storeu_epi64): Ditto.
1595 (_mm512_cmpge_epi32_mask): Ditto.
1596 (_mm512_cmpge_epu32_mask): Ditto.
1597 (_mm512_cmpge_epi64_mask): Ditto.
1598 (_mm512_cmpge_epu64_mask): Ditto.
1599 (_mm512_cmple_epi32_mask): Ditto.
1600 (_mm512_cmple_epu32_mask): Ditto.
1601 (_mm512_cmple_epi64_mask): Ditto.
1602 (_mm512_cmple_epu64_mask): Ditto.
1603 (_mm512_cmplt_epi32_mask): Ditto.
1604 (_mm512_cmplt_epu32_mask): Ditto.
1605 (_mm512_cmplt_epi64_mask): Ditto.
1606 (_mm512_cmplt_epu64_mask): Ditto.
1607 (_mm512_cmpneq_epi32_mask): Ditto.
1608 (_mm512_cmpneq_epu32_mask): Ditto.
1609 (_mm512_cmpneq_epi64_mask): Ditto.
1610 (_mm512_cmpneq_epu64_mask): Ditto.
1611 (_mm512_expand_pd): Ditto.
1612 (_mm512_expand_ps): Ditto.
1613 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1614 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1615 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1616 * config/i386/i386.c (ix86_builtins): Add
1617 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1618 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1619 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1620 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1621 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1622 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1623 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1624 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1625 IX86_BUILTIN_PMOVUSQW512_MEM.
1626 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1627 __builtin_ia32_pmovsqd512mem_mask,
1628 __builtin_ia32_pmovqd512mem_mask,
1629 __builtin_ia32_pmovusqw512mem_mask,
1630 __builtin_ia32_pmovsqw512mem_mask,
1631 __builtin_ia32_pmovqw512mem_mask,
1632 __builtin_ia32_pmovusdw512mem_mask,
1633 __builtin_ia32_pmovsdw512mem_mask,
1634 __builtin_ia32_pmovdw512mem_mask,
1635 __builtin_ia32_pmovqb512mem_mask,
1636 __builtin_ia32_pmovusqb512mem_mask,
1637 __builtin_ia32_pmovsqb512mem_mask,
1638 __builtin_ia32_pmovusdb512mem_mask,
1639 __builtin_ia32_pmovsdb512mem_mask,
1640 __builtin_ia32_pmovdb512mem_mask.
1641 (bdesc_args): Add __builtin_ia32_expanddf512,
1642 __builtin_ia32_expandsf512.
1643 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1644 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1645 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1646 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1647 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1648 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1649 (avx512f_<code>v8div16qi2_mask_store): This.
1650 (avx512f_expand<mode>): New.
1652 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1654 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1656 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1657 (_mm512_prefetch_i32scatter_pd): Ditto.
1658 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1659 (_mm512_prefetch_i64scatter_pd): Ditto.
1660 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1661 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1662 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1663 (_mm512_prefetch_i32scatter_ps): Ditto.
1664 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1665 (_mm512_prefetch_i64scatter_ps): Ditto.
1666 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1667 * config/i386/i386-builtin-types.def: Define
1668 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1669 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1670 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1671 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1672 IX86_BUILTIN_SCATTERPFQPD.
1673 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1674 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1675 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1676 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1677 __builtin_ia32_scatterpfqps.
1678 (ix86_expand_builtin): Expand new built-ins.
1679 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1680 fix memory access data type.
1681 (*avx512pf_gatherpf<mode>_mask): Ditto.
1682 (*avx512pf_gatherpf<mode>): Ditto.
1683 (avx512pf_scatterpf<mode>): Ditto.
1684 (*avx512pf_scatterpf<mode>_mask): Ditto.
1685 (*avx512pf_scatterpf<mode>): Ditto.
1686 (GATHER_SCATTER_SF_MEM_MODE): New.
1687 (avx512pf_gatherpf<mode>df): Ditto.
1688 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1689 (*avx512pf_scatterpf<mode>df): Ditto.
1691 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1694 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1695 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1698 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1700 * common/config/arm/arm-common.c
1701 (arm_rewrite_mcpu): Handle multiple names.
1703 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1705 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1707 * gimple-builder.h (create_gimple_tmp): Delete.
1709 2014-01-27 Christian Bruel <christian.bruel@st.com>
1711 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1714 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1716 * config/pa/pa.md (call): Generate indirect long calls to non-local
1717 functions when outputing 32-bit code.
1718 (call_value): Likewise except for special call to buggy powf function.
1720 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1721 portable runtime and PIC indirect calls.
1722 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1723 and PIC call sequences. Use ldo instead of blr to set return register
1724 in PIC call sequence.
1726 2014-01-25 Walter Lee <walt@tilera.com>
1728 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1729 avoid clobbering a live register.
1731 2014-01-25 Walter Lee <walt@tilera.com>
1733 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1734 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1735 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1736 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1738 2014-01-25 Walter Lee <walt@tilera.com>
1740 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1741 arguments on even registers.
1742 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1744 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1745 (BIGGEST_ALIGNMENT): Ditto.
1746 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1748 2014-01-25 Walter Lee <walt@tilera.com>
1750 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1751 insns before bundling.
1752 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1754 2014-01-25 Walter Lee <walt@tilera.com>
1756 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1757 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1758 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1760 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1762 * config/mips/constraints.md (kl): Delete.
1763 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1764 define expands, using...
1765 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1766 instructions for MIPS16.
1767 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1768 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1770 2014-01-25 Walter Lee <walt@tilera.com>
1772 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1776 2014-01-25 Walter Lee <walt@tilera.com>
1778 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1779 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1781 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1783 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1786 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1788 * print-rtl.c (in_call_function_usage): New var.
1789 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1790 EXPR_LIST mode as mode and not as reg note name.
1793 * cfgloopmanip.c (copy_loop_info): If
1794 loop->warned_aggressive_loop_optimizations, make sure
1795 the flag is set in target loop too.
1797 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1799 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1801 * builtins.def: Likewise.
1802 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1803 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1804 * ira.c (ira_setup_eliminable_regset): Likewise.
1805 * omp-low.c (gate_expand_omp): Likewise.
1806 (execute_lower_omp): Likewise.
1807 (diagnose_sb_0): Likewise.
1808 (gate_diagnose_omp_blocks): Likewise.
1809 (simd_clone_clauses_extract): Likewise.
1812 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1814 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1815 correction for little endian...
1816 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1819 2014-01-24 Jeff Law <law@redhat.com>
1821 PR tree-optimization/59919
1822 * tree-vrp.c (find_assert_locations_1): Do not register asserts
1823 for non-returning calls.
1825 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
1827 * common/config/aarch64/aarch64-common.c
1828 (aarch64_rewrite_mcpu): Handle multiple names.
1829 * config/aarch64/aarch64.h
1830 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1832 2014-01-24 Dodji Seketeli <dodji@redhat.com>
1834 * input.c (add_file_to_cache_tab): Handle the case where fopen
1837 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
1840 * config/i386/i386.md (pushsf splitter): Get stack adjustment
1841 from push operand if code of push isn't PRE_DEC.
1843 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
1846 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
1847 -mquad-memory-atomic. Update -mquad-memory documentation to say
1848 it is only used for non-atomic loads/stores.
1850 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
1851 -mquad-memory or -mquad-memory-atomic switches.
1853 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
1854 -mquad-memory-atomic to ISA 2.07 support.
1856 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
1857 to separate support of normal quad word memory operations (ldq, stq)
1858 from the atomic quad word memory operations.
1860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1861 support to separate non-atomic quad word operations from atomic
1862 quad word operations. Disable non-atomic quad word operations in
1863 little endian mode so that we don't have to swap words after the
1864 load and before the store.
1865 (quad_load_store_p): Add comment about atomic quad word support.
1866 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
1867 options printed with -mdebug=reg.
1869 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
1870 -mquad-memory-atomic as the test for whether we have quad word
1871 atomic instructions.
1872 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
1873 or -mp8-vector are used, allow byte/half-word atomic operations.
1875 * config/rs6000/sync.md (load_lockedti): Insure that the address
1876 is a proper indexed or indirect address for the lqarx instruction.
1877 On little endian systems, swap the hi/lo registers after the lqarx
1879 (load_lockedpti): Use indexed_or_indirect_operand predicate to
1880 insure the address is valid for the lqarx instruction.
1881 (store_conditionalti): Insure that the address is a proper indexed
1882 or indirect address for the stqcrx. instruction. On little endian
1883 systems, swap the hi/lo registers before doing the stqcrx.
1885 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
1886 insure the address is valid for the stqcrx. instruction.
1888 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
1889 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
1890 type of quad memory support is available.
1892 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
1895 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
1896 there is a danger of looping.
1898 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
1900 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1901 force flag_ira_loop_pressure if set via command line.
1903 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
1905 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
1906 (ashr_simd): New builtin handling DI mode.
1907 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
1908 (aarch64_sshr_simddi): New match pattern.
1909 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
1910 (vshrd_n_s64): Likewise.
1911 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
1913 2014-01-23 Nick Clifton <nickc@redhat.com>
1915 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
1916 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
1917 favour of mcu specific scripts.
1918 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
1921 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1922 Alex Velenko <Alex.Velenko@arm.com>
1924 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
1925 (vaddv_s16): Likewise.
1926 (vaddv_s32): Likewise.
1927 (vaddv_u8): Likewise.
1928 (vaddv_u16): Likewise.
1929 (vaddv_u32): Likewise.
1930 (vaddvq_s8): Likewise.
1931 (vaddvq_s16): Likewise.
1932 (vaddvq_s32): Likewise.
1933 (vaddvq_s64): Likewise.
1934 (vaddvq_u8): Likewise.
1935 (vaddvq_u16): Likewise.
1936 (vaddvq_u32): Likewise.
1937 (vaddvq_u64): Likewise.
1938 (vaddv_f32): Likewise.
1939 (vaddvq_f32): Likewise.
1940 (vaddvq_f64): Likewise.
1941 (vmaxv_f32): Likewise.
1942 (vmaxv_s8): Likewise.
1943 (vmaxv_s16): Likewise.
1944 (vmaxv_s32): Likewise.
1945 (vmaxv_u8): Likewise.
1946 (vmaxv_u16): Likewise.
1947 (vmaxv_u32): Likewise.
1948 (vmaxvq_f32): Likewise.
1949 (vmaxvq_f64): Likewise.
1950 (vmaxvq_s8): Likewise.
1951 (vmaxvq_s16): Likewise.
1952 (vmaxvq_s32): Likewise.
1953 (vmaxvq_u8): Likewise.
1954 (vmaxvq_u16): Likewise.
1955 (vmaxvq_u32): Likewise.
1956 (vmaxnmv_f32): Likewise.
1957 (vmaxnmvq_f32): Likewise.
1958 (vmaxnmvq_f64): Likewise.
1959 (vminv_f32): Likewise.
1960 (vminv_s8): Likewise.
1961 (vminv_s16): Likewise.
1962 (vminv_s32): Likewise.
1963 (vminv_u8): Likewise.
1964 (vminv_u16): Likewise.
1965 (vminv_u32): Likewise.
1966 (vminvq_f32): Likewise.
1967 (vminvq_f64): Likewise.
1968 (vminvq_s8): Likewise.
1969 (vminvq_s16): Likewise.
1970 (vminvq_s32): Likewise.
1971 (vminvq_u8): Likewise.
1972 (vminvq_u16): Likewise.
1973 (vminvq_u32): Likewise.
1974 (vminnmv_f32): Likewise.
1975 (vminnmvq_f32): Likewise.
1976 (vminnmvq_f64): Likewise.
1978 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
1980 * config/aarch64/aarch64-simd.md
1981 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
1982 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
1983 (*aarch64_mul3_elt<mode>): Likewise.
1984 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
1985 (*aarch64_mul3_elt_to_64v2df): Likewise.
1986 (*aarch64_mla_elt<mode>): Likewise.
1987 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1988 (*aarch64_mls_elt<mode>): Likewise.
1989 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1990 (*aarch64_fma4_elt<mode>): Likewise.
1991 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1992 (*aarch64_fma4_elt_to_64v2df): Likewise.
1993 (*aarch64_fnma4_elt<mode>): Likewise.
1994 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1995 (*aarch64_fnma4_elt_to_64v2df): Likewise.
1996 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1997 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1998 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1999 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2000 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2001 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2002 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2004 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2006 * config/aarch64/aarch64-simd.md
2007 (aarch64_be_checked_get_lane<mode>): New define_expand.
2008 * config/aarch64/aarch64-simd-builtins.def
2009 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2010 New builtin definition.
2011 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2012 Use new safe be builtin.
2014 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2016 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2018 (aarch64_be_st1<mode>): Likewise.
2019 (aarch_ld1<VALL:mode>): Define_expand modified.
2020 (aarch_st1<VALL:mode>): Likewise.
2021 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2022 (UNSPEC_ST1): Likewise.
2024 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2026 * config/microblaze/microblaze.md: Add trap insn and attribute
2028 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2030 PR preprocessor/58580
2031 * input.h (location_get_source_line): Take an additional line_size
2033 (void diagnostics_file_cache_fini): Declare new function.
2034 * input.c (struct fcache): New type.
2035 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2036 New static constants.
2037 (diagnostic_file_cache_init, total_lines_num)
2038 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2039 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2040 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2041 (get_next_line, read_next_line, goto_next_line, read_line_num):
2042 New static function definitions.
2043 (diagnostic_file_cache_fini): New function.
2044 (location_get_source_line): Take an additional output line_len
2045 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2047 * diagnostic.c (diagnostic_finish): Call
2048 diagnostic_file_cache_fini.
2049 (adjust_line): Take an additional input parameter for the length
2050 of the line, rather than calculating it with strlen.
2051 (diagnostic_show_locus): Adjust the use of
2052 location_get_source_line and adjust_line with respect to their new
2053 signature. While displaying a line now, do not stop at the first
2054 null byte. Rather, display the zero byte as a space and keep
2055 going until we reach the size of the line.
2056 * Makefile.in: Add vec.o to OBJS-libcommon
2058 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2059 Ilya Tocar <ilya.tocar@intel.com>
2061 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2062 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2063 (__builtin_ia32_kmov16): Ditto.
2064 * config/i386/i386.md (UNSPEC_KMOV): New.
2067 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2069 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2070 (_mm512_storeu_si512): Ditto.
2072 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2075 * rtl.h (get_referenced_operands): Declare.
2076 * recog.c (get_referenced_operands): New function.
2077 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2078 operands have been referenced when recording LO_SUM references.
2080 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2082 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2084 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2086 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2087 Enable for generic and recent AMD targets.
2089 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2091 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2092 ARG_SIZE note when adjustment was eliminated.
2094 2014-01-22 Jeff Law <law@redhat.com>
2096 PR tree-optimization/59597
2097 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2098 in file. Accept new argument REGISTERING and use it to modify
2099 dump output appropriately.
2100 (register_jump_thread): Corresponding changes.
2101 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2102 thread paths involving joiner blocks. Add code to dump cancelled
2103 jump threading paths.
2105 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2107 PR rtl-optimization/59477
2108 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2109 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2111 2014-01-22 Tom Tromey <tromey@redhat.com>
2113 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2115 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2117 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2119 PR rtl-optimization/59896
2120 * lra-constraints.c (process_alt_operands): Check unused note for
2121 matched operands of insn with no output reloads.
2123 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2125 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2126 (mips_move_from_gpr_cost): Likewise.
2128 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2130 PR rtl-optimization/59858
2131 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2132 ira_class_hard_regs_num.
2133 (process_alt_operands): Increase reject for dying matched operand.
2135 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2138 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2139 smaller than size, perform several stores or loads and stores
2140 at dst + count - size to store or copy all of size bytes, rather
2141 than just last modesize bytes.
2143 2014-01-20 DJ Delorie <dj@redhat.com>
2145 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2146 that CLOBBERs are REGs before propogating their values.
2148 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2151 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2152 (cgraph_inline_failed_type): New function.
2153 * cgraph.h (DEFCIFCODE): Add type.
2154 (cgraph_inline_failed_type_t): New enum.
2155 (cgraph_inline_failed_type): New prototype.
2156 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2157 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2158 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2159 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2160 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2161 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2162 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2163 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2164 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2165 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2166 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2167 OPTIMIZATION_MISMATCH.
2168 * tree-inline.c (expand_call_inline): Emit errors during
2169 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2171 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2174 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2175 mode attribute in insn output.
2177 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2179 * output.h (output_constant): Delete.
2180 * varasm.c (output_constant): Make private.
2182 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2184 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2186 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2189 * tree.h (fold_builtin_strcat): New prototype.
2190 * builtins.c (fold_builtin_strcat): No longer static. Add len
2191 argument, if non-NULL, don't call c_strlen. Optimize
2192 directly into __builtin_memcpy instead of __builtin_strcpy.
2193 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2194 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2196 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2198 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2199 for SImode_address_operand operands, having only a REG argument.
2201 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2203 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2204 loader name using mbig-endian.
2205 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2207 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2209 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2213 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2215 * config/aarch64/aarch64-protos.h
2216 (aarch64_cannot_change_mode_class_ptr): Declare.
2217 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2218 aarch64_cannot_change_mode_class_ptr): New.
2219 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2220 backend hook aarch64_cannot_change_mode_class.
2222 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2224 * common/config/aarch64/aarch64-common.c
2225 (aarch64_handle_option): Don't handle any option order logic here.
2226 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2227 selected_cpu, warn on architecture version mismatch.
2228 (aarch64_override_options): Fix parsing order for option strings.
2230 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2231 Iain Sandoe <iain@codesourcery.com>
2234 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2235 warning. Amend comment to reflect current functionality.
2237 2014-01-20 Richard Biener <rguenther@suse.de>
2240 * builtins.c (fold_builtin_strcat): Remove case better handled
2241 by tree-ssa-strlen.c.
2243 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2245 * config/aarch64/aarch64.opt
2246 (mcpu, march, mtune): Make case-insensitive.
2248 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2251 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2252 if operands[1] is a REG or ZERO_EXTEND of a REG.
2254 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2256 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2258 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2260 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2261 long non-pic millicode calls.
2263 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2265 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2267 2014-01-19 Kito Cheng <kito@0xlab.org>
2269 * builtins.c (expand_movstr): Check movstr expand done or fail.
2271 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2272 H.J. Lu <hongjiu.lu@intel.com>
2275 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2276 to DImode for zero-extended addresses.
2278 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2280 PR rtl-optimization/57763
2281 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2282 on the new indirect jump_insn and increment LABEL_NUSES (label).
2284 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2288 * config.gcc (x86_archs): New variable.
2289 (x86_64_archs): Likewise.
2290 (x86_cpus): Likewise.
2291 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2292 --with-arch/--with-cpu= options.
2293 Support --with-arch=/--with-cpu={nehalem,westmere,
2294 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2296 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2298 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2299 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2301 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2303 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2305 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2308 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2309 clear cpp_get_options (parse_in)->warn_unused_macros for
2310 ix86_target_macros_internal with cpp_define.
2312 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2314 * jump.c (delete_related_insns): Keep (use (insn))s.
2315 * reorg.c (redundant_insn): Check for barriers too.
2317 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2319 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2321 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2323 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2324 call to $$dyncall when TARGET_LONG_CALLS is true.
2326 2014-01-17 Jeff Law <law@redhat.com>
2328 * ree.c (combine_set_extension): Temporarily disable test for
2329 changing number of hard registers.
2331 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2334 * ipa-inline-analysis.c (inline_free_summary):
2335 Do not free summary of aliases.
2337 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2340 * gimplify.c (gimplify_expr): Use create_tmp_var
2341 instead of create_tmp_var_raw. If cond doesn't have
2342 integral type, don't add the IFN_ANNOTATE builtin at all.
2344 2014-01-17 Martin Jambor <mjambor@suse.cz>
2347 * ipa-cp.c (prev_edge_clone): New variable.
2348 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2349 Also resize prev_edge_clone vector.
2350 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2351 (ipcp_edge_removal_hook): New function.
2352 (ipcp_driver): Register ipcp_edge_removal_hook.
2354 2014-01-17 Andrew Pinski <apinski@cavium.com>
2355 Steve Ellcey <sellcey@mips.com>
2358 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2361 2014-01-17 Jeff Law <law@redhat.com>
2364 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2365 so that pass_ccp runs first.
2367 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2369 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2370 (ix86_adjust_cost): Use !TARGET_XXX.
2371 (do_reorder_for_imul): Likewise.
2372 (swap_top_of_ready_list): Likewise.
2373 (ix86_sched_reorder): Likewise.
2375 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2377 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2378 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2379 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2380 (intel_memset): New. Duplicate slm_memset.
2381 (intel_cost): New. Duplicate slm_cost.
2382 (m_INTEL): New macro.
2383 (processor_target_table): Add "intel".
2384 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2385 with PROCESSOR_INTEL for "intel".
2386 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2387 PROCESSOR_SILVERMONT.
2388 (ix86_issue_rate): Likewise.
2389 (ix86_adjust_cost): Likewise.
2390 (ia32_multipass_dfa_lookahead): Likewise.
2391 (swap_top_of_ready_list): Likewise.
2392 (ix86_sched_reorder): Likewise.
2393 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2394 instead of TARGET_OPT_AGU.
2395 * config/i386/i386.h (TARGET_INTEL): New.
2396 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2397 (processor_type): Add PROCESSOR_INTEL.
2398 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2399 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2401 2014-01-17 Marek Polacek <polacek@redhat.com>
2404 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2407 2014-01-17 Richard Biener <rguenther@suse.de>
2409 PR tree-optimization/46590
2410 * opts.c (default_options_table): Add entries for
2411 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2412 all enabled at -O1 but not for -Og.
2413 * common.opt (fbranch-count-reg): Remove Init(1).
2414 (fmove-loop-invariants): Likewise.
2415 (ftree-pta): Likewise.
2417 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2419 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2420 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2421 decls to at least the GCC 4.8 used alignments.
2424 * tree-nested.c (convert_nonlocal_reference_stmt,
2425 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2426 of GIMPLE_BIND stmts, adjust associated decls.
2428 2014-01-17 Richard Biener <rguenther@suse.de>
2430 PR tree-optimization/46590
2431 * vec.h (vec<>::bseach): New member function implementing
2432 binary search according to C89 bsearch.
2433 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2434 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2435 bitmap pointer again. Make accesses_in_loop a flat array.
2436 (mem_ref_obstack): New global.
2437 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2438 (mark_ref_stored): Likewise.
2439 (ref_indep_loop_p_2): Likewise.
2440 (set_ref_stored_in_loop): New helper function.
2441 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2442 (memref_free): Adjust.
2443 (record_mem_ref_loc): Simplify.
2444 (gather_mem_refs_stmt): Adjust.
2445 (sort_locs_in_loop_postorder_cmp): New function.
2446 (analyze_memory_references): Sort accesses_in_loop after
2447 loop postorder number.
2448 (find_ref_loc_in_loop_cmp): New function.
2449 (for_all_locs_in_loop): Find relevant cluster of locs in
2450 accesses_in_loop and iterate without recursion.
2451 (execute_sm): Avoid uninit warning.
2452 (struct ref_always_accessed): Simplify.
2453 (ref_always_accessed::operator ()): Likewise.
2454 (ref_always_accessed_p): Likewise.
2455 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2456 loop postorder numbers here.
2457 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2460 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2463 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2464 on decls for which assemble_alias has been called.
2466 2014-01-17 Nick Clifton <nickc@redhat.com>
2468 * config/msp430/msp430.opt: (mcpu): New option.
2469 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2470 (msp430_option_override): Parse target_cpu. If the MCU name
2471 matches a generic string, clear target_mcu.
2472 (msp430_attr): Allow numeric interrupt values up to 63.
2473 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2474 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2476 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2478 * config/msp430/msp430.md (popm): Use %J rather than %I.
2479 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2480 (addhi_cy_i): Use immediate_operand for operand 2.
2481 * doc/invoke.texi: Document -mcpu option.
2483 2014-01-17 Richard Biener <rguenther@suse.de>
2485 PR rtl-optimization/38518
2486 * df.h (df_analyze_loop): Declare.
2487 * df-core.c: Include cfgloop.h.
2488 (df_analyze_1): Split out main part of df_analyze.
2489 (df_analyze): Adjust.
2490 (loop_inverted_post_order_compute): New function.
2491 (loop_post_order_compute): Likewise.
2492 (df_analyze_loop): New function avoiding whole-function
2494 * loop-invariant.c (find_defs): Use df_analyze_loop.
2495 (find_invariants): Adjust.
2496 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2498 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2500 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2501 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2503 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2505 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2506 traversal when removing references.
2508 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2511 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2513 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2516 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2517 pushing a reload for an autoinc when we had previously reloaded an
2518 inner part of the address.
2520 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2522 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2524 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2525 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2526 when not giving up or versioning for alias only because of
2528 (vect_analyze_data_ref_dependences): Set to true.
2529 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2531 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2532 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2536 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2539 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2540 operand 0 predicate for gathers, use a new pseudo as subtarget.
2542 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2545 * lra-constraints.c (process_alt_operands): Add printing debug info.
2546 Check absence of input/output reloads for matched operands too.
2548 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2550 PR rtl-optimization/59835
2551 * ira.c (ira_init_register_move_cost): Increase cost for
2554 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2556 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2558 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2561 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2562 non-register objects. Use gen_(high/low)part more consistently.
2565 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2568 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2569 endian support, remove tests for WORDS_BIG_ENDIAN.
2570 (p8_mfvsrd_3_<mode>): Likewise.
2571 (reload_gpr_from_vsx<mode>): Likewise.
2572 (reload_gpr_from_vsxsf): Likewise.
2573 (p8_mfvsrd_4_disf): Likewise.
2575 2014-01-16 Richard Biener <rguenther@suse.de>
2577 PR rtl-optimization/46590
2578 * lcm.c (compute_antinout_edge): Use postorder iteration.
2579 (compute_laterin): Use inverted postorder iteration.
2581 2014-01-16 Nick Clifton <nickc@redhat.com>
2584 * varasm.c (output_constant): Return the number of bytes actually
2586 (output_constructor_array_range): Update the field size with the
2587 number of bytes emitted by output_constant.
2588 (output_constructor_regular_field): Likewise. Also do not
2589 complain if the total number of bytes emitted is now greater
2590 than the expected fieldpos.
2591 * output.h (output_constant): Update prototype and descriptive comment.
2593 2014-01-16 Marek Polacek <polacek@redhat.com>
2596 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2597 it is error_mark_node.
2599 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2601 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2602 VALID_AVX256_REG_OR_OI_MODE.
2604 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2606 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2607 current procedure should be profiled.
2609 2014-01-15 Andrew Pinski <apinski@cavium.com>
2611 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2612 of moving from/to the STACK_REG register class.
2614 2014-01-15 Richard Henderson <rth@redhat.com>
2617 * reginfo.c (global_regs_decl): Globalize.
2618 * rtl.h (global_regs_decl): Declare.
2619 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2620 reserved via global_regs.
2622 2014-01-15 Teresa Johnson <tejohnson@google.com>
2624 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2626 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2628 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2629 and vmulosh rather than call gen_vec_widen_smult_*.
2630 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2631 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2632 (vec_widen_smult_even_v16qi): Likewise.
2633 (vec_widen_umult_even_v8hi): Likewise.
2634 (vec_widen_smult_even_v8hi): Likewise.
2635 (vec_widen_umult_odd_v16qi): Likewise.
2636 (vec_widen_smult_odd_v16qi): Likewise.
2637 (vec_widen_umult_odd_v8hi): Likewise.
2638 (vec_widen_smult_odd_v8hi): Likewise.
2639 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2640 vmuloub rather than call gen_vec_widen_umult_*.
2641 (vec_widen_umult_lo_v16qi): Likewise.
2642 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2643 vmulosb rather than call gen_vec_widen_smult_*.
2644 (vec_widen_smult_lo_v16qi): Likewise.
2645 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2646 rather than call gen_vec_widen_umult_*.
2647 (vec_widen_umult_lo_v8hi): Likewise.
2648 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2649 rather than call gen_vec_widen_smult_*.
2650 (vec_widen_smult_lo_v8hi): Likewise.
2652 2014-01-15 Jeff Law <law@redhat.com>
2654 PR tree-optimization/59747
2655 * ree.c (find_and_remove_re): Properly handle case where a second
2656 eliminated extension requires widening a copy created for elimination
2657 of a prior extension.
2658 (combine_set_extension): Ensure that the number of hard regs needed
2659 for a destination register does not change when we widen it.
2661 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2663 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2664 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2665 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2666 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2667 (avr-*-rtems*): Likewise.
2668 (bfin*-rtems*): Likewise.
2669 (moxie-*-rtems*): Likewise.
2670 (h8300-*-rtems*): Likewise.
2671 (i[34567]86-*-rtems*): Likewise.
2672 (lm32-*-rtems*): Likewise.
2673 (m32r-*-rtems*): Likewise.
2674 (m68k-*-rtems*): Likewise.
2675 (microblaze*-*-rtems*): Likewise.
2676 (mips*-*-rtems*): Likewise.
2677 (powerpc-*-rtems*): Likewise.
2678 (sh-*-rtems*): Likewise.
2679 (sparc-*-rtems*): Likewise.
2680 (sparc64-*-rtems*): Likewise.
2681 (v850-*-rtems*): Likewise.
2682 (m32c-*-rtems*): Likewise.
2684 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2686 PR rtl-optimization/59511
2687 * ira.c (ira_init_register_move_cost): Use memory costs for some
2688 cases of register move cost calculations.
2689 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2690 instead of BB frequency.
2691 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2692 * lra-assigns.c (find_hard_regno_for): Ditto.
2694 2014-01-15 Richard Biener <rguenther@suse.de>
2696 PR tree-optimization/59822
2697 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2698 (vectorizable_load): Use it to hoist defs of uses of invariant
2699 loads out of the loop.
2701 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2702 Kugan Vivekanandarajah <kuganv@linaro.org>
2705 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2708 2014-01-15 Richard Biener <rguenther@suse.de>
2710 PR rtl-optimization/59802
2711 * lcm.c (compute_available): Use inverted postorder to seed
2712 the initial worklist.
2714 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2717 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2718 ADDR_REGS for invalid symrefs in non-PIC code.
2720 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2723 * builtins.c (determine_block_size): Initialize *probable_max_size
2724 even if len_rtx is CONST_INT.
2726 2014-01-14 Andrew Pinski <apinski@cavium.com>
2728 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2729 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2730 (cortexa53_tunings): Likewise.
2731 (aarch64_sched_issue_rate): New function.
2732 (TARGET_SCHED_ISSUE_RATE): Define.
2734 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2736 * ira-costs.c (find_costs_and_classes): Add missed
2737 ira_init_register_move_cost_if_necessary.
2739 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2742 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2744 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2747 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2748 to indicate if type is used for function return value. Warn ABI
2749 change if the vector mode isn't available for function return value.
2750 (ix86_function_arg_advance): Pass false to type_natural_mode.
2751 (ix86_function_arg): Likewise.
2752 (ix86_gimplify_va_arg): Likewise.
2753 (function_arg_32): Don't warn ABI change.
2754 (ix86_function_value): Pass true to type_natural_mode.
2755 (ix86_return_in_memory): Likewise.
2756 (ix86_struct_value_rtx): Removed.
2757 (TARGET_STRUCT_VALUE_RTX): Likewise.
2759 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2761 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2762 converting a conditional jump into a conditional return.
2764 2014-01-14 Richard Biener <rguenther@suse.de>
2766 PR tree-optimization/58921
2767 PR tree-optimization/59006
2768 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2769 hoisting invariant stmts.
2770 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2771 invariant loads on the preheader edge if possible.
2773 2014-01-14 Joey Ye <joey.ye@arm.com>
2775 * doc/plugin.texi (Building GCC plugins): Update to C++.
2777 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2779 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2780 (_mm_rcp28_round_ss): Ditto.
2781 (_mm_rsqrt28_round_sd): Ditto.
2782 (_mm_rsqrt28_round_ss): Ditto.
2783 (_mm_rcp28_sd): Ditto.
2784 (_mm_rcp28_ss): Ditto.
2785 (_mm_rsqrt28_sd): Ditto.
2786 (_mm_rsqrt28_ss): Ditto.
2787 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2788 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2789 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2790 (IX86_BUILTIN_RCP28SD): Ditto.
2791 (IX86_BUILTIN_RCP28SS): Ditto.
2792 (IX86_BUILTIN_RSQRT28SD): Ditto.
2793 (IX86_BUILTIN_RSQRT28SS): Ditto.
2794 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2795 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2796 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2797 (ix86_expand_special_args_builtin): Expand new FTYPE.
2798 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2799 (srcp14<mode>): Make insn unary.
2800 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2801 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2802 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2803 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2804 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2805 Fix rounding: make it SAE only.
2806 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2808 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2810 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2811 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2812 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2813 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2814 (round_saeonly_mask_scalar_operand4): Ditto.
2815 (round_saeonly_mask_scalar_op3): Ditto.
2816 (round_saeonly_mask_scalar_op4): Ditto.
2818 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2820 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2821 Implement -maltivec=be for vec_insert and vec_extract.
2823 2014-01-10 DJ Delorie <dj@redhat.com>
2825 * config/msp430/msp430.md (call_internal): Don't allow memory
2826 references with SP as the base register.
2827 (call_value_internal): Likewise.
2828 * config/msp430/constraints.md (Yc): New. For memory references
2829 that don't use SP as a base register.
2831 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
2832 "an integer without a # prefix"
2833 * config/msp430/msp430.md (epilogue_helper): Use it.
2835 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2838 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
2839 AVX512F gather builtins.
2840 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
2841 on gather decls with INTEGER_TYPE masktype.
2842 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
2843 directly into the builtin rather than hoisting it before loop.
2845 PR tree-optimization/59387
2846 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
2847 (scev_const_prop): If folded_casts and type has undefined overflow,
2848 use force_gimple_operand instead of force_gimple_operand_gsi and
2849 for each added stmt if it is assign with
2850 arith_code_with_undefined_signed_overflow, call
2851 rewrite_to_defined_overflow.
2852 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
2853 gimple-fold.h instead.
2854 (arith_code_with_undefined_signed_overflow,
2855 rewrite_to_defined_overflow): Moved to ...
2856 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
2857 rewrite_to_defined_overflow): ... here. No longer static.
2858 Include gimplify-me.h.
2859 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
2860 rewrite_to_defined_overflow): New prototypes.
2862 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2864 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
2866 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
2868 * builtins.c (get_object_alignment_2): Minor tweak.
2869 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
2871 2014-01-13 Christian Bruel <christian.bruel@st.com>
2873 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
2874 optimized non constant lengths.
2876 2014-01-13 Jakub Jelinek <jakub@redhat.com>
2879 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
2880 load as __atomic_load_N if possible.
2882 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2884 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
2886 (rs6000_expand_builtin): Adjust call.
2888 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
2891 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
2892 * config/rs6000/rs6000.c: Include target-globals.h.
2893 (rs6000_set_current_function): Instead of doing target_reinit
2894 unconditionally, use save_target_globals_default_opts and
2895 restore_target_globals.
2897 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
2899 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
2900 (rs6000_expand_builtin): Handle mffs and mtfsf.
2901 (rs6000_init_builtins): Define mffs and mtfsf.
2902 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
2903 (rs6000_mffs): New pattern.
2904 (rs6000_mtfsf): New pattern.
2906 2014-01-11 Bin Cheng <bin.cheng@arm.com>
2908 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
2909 Start narrowing with START. Apply candidate-use pair
2910 and check overall cost in narrowing.
2911 (iv_ca_prune): Pass new argument.
2913 2014-01-10 Jeff Law <law@redhat.com>
2916 * ree.c (combine_reaching_defs): Ensure the defining statement
2917 occurs before the extension when optimizing extensions with
2918 different source and destination hard registers.
2920 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2923 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
2924 vtables into the type inheritance graph.
2926 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2928 PR rtl-optimization/59754
2929 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
2930 modes in the REGNO != REGNO case.
2932 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2934 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
2936 2014-01-10 Jakub Jelinek <jakub@redhat.com>
2938 PR tree-optimization/59745
2939 * tree-predcom.c (tree_predictive_commoning_loop): Call
2940 free_affine_expand_cache if giving up because components is NULL.
2942 * target-globals.c (save_target_globals): Allocate < 4KB structs using
2943 GC in payload of target_globals struct instead of allocating them on
2944 the heap and the larger structs separately using GC.
2945 * target-globals.h (struct target_globals): Make regs, hard_regs,
2946 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
2947 of GTY((skip)) and change type to void *.
2948 (reset_target_globals): Cast loads from those fields to corresponding
2951 2014-01-10 Steve Ellcey <sellcey@mips.com>
2954 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
2955 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
2956 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
2958 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
2961 * aarch64-modes.def (CC_Zmode): New flags mode.
2962 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
2963 represents an equality.
2964 (aarch64_get_condition_code): Handle CC_Zmode.
2965 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
2967 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2969 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
2970 extraction in good case.
2972 2014-01-10 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/59374
2975 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
2976 checking after SLP discovery. Mark stmts not participating
2977 in any SLP instance properly.
2979 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2981 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
2982 when handling a SET rtx.
2984 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2986 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
2987 (cortex-a57): Likewise.
2988 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
2990 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2992 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
2993 non-iwmmxt builtins.
2995 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
2999 * ipa-devirt.c record_target_from_binfo): Take as argument
3000 stack of binfos and lookup matching one for virtual inheritance.
3001 (possible_polymorphic_call_targets_1): Update.
3003 2014-01-10 Huacai Chen <chenhc@lemote.com>
3005 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3006 kernel strings for Loongson-2E/2F/3A.
3008 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3011 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3012 is_gimple_call before calling gimple_call_internal_p.
3014 2014-01-09 Steve Ellcey <sellcey@mips.com>
3016 * Makefile.in (TREE_FLOW_H): Remove.
3017 (TREE_SSA_H): Add file names from tree-flow.h.
3018 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3019 * tree.h: Remove tree-flow.h reference.
3020 * hash-table.h: Remove tree-flow.h reference.
3021 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3022 reference with tree-ssa-loop.h.
3024 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3026 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3027 default element-order behavior for -maltivec.
3028 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3030 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3031 when targeting big endian, at least for now.
3032 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3034 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3037 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3038 var satisfies use_register_for_decl, just take into account type
3039 alignment, rather than decl alignment.
3041 PR tree-optimization/59622
3042 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3043 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3044 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3045 Don't devirtualize for inplace at all. For targets.length () == 1,
3046 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3048 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3050 * config/i386/i386.md (cpu): Remove the unused btver1.
3052 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3054 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3056 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3059 * tree-core.h (struct target_globals): New forward declaration.
3060 (struct tree_target_option): Add globals field.
3061 * tree.h (TREE_TARGET_GLOBALS): Define.
3062 (prepare_target_option_nodes_for_pch): New prototype.
3063 * target-globals.h (struct target_globals): Define even if
3065 * tree.c (prepare_target_option_node_for_pch,
3066 prepare_target_option_nodes_for_pch): New functions.
3067 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3068 * config/i386/i386.c: Include target-globals.h.
3069 (ix86_set_current_function): Instead of doing target_reinit
3070 unconditionally, use save_target_globals_default_opts and
3071 restore_target_globals.
3073 2014-01-09 Richard Biener <rguenther@suse.de>
3075 PR tree-optimization/59715
3076 * tree-cfg.h (split_critical_edges): Declare.
3077 * tree-cfg.c (split_critical_edges): Export.
3078 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3080 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3082 * cfgexpand.c (expand_stack_vars): Optionally disable
3083 asan stack protection.
3084 (expand_used_vars): Likewise.
3085 (partition_stack_vars): Likewise.
3086 * asan.c (asan_emit_stack_protection): Optionally disable
3087 after return stack usage.
3088 (instrument_derefs): Optionally disable memory access instrumentation.
3089 (instrument_builtin_call): Likewise.
3090 (instrument_strlen_call): Likewise.
3091 (asan_protect_global): Optionally disable global variables protection.
3092 * doc/invoke.texi: Added doc for new options.
3093 * params.def: Added new options.
3094 * params.h: Likewise.
3096 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3098 PR rtl-optimization/59724
3099 * ifcvt.c (cond_exec_process_if_block): Don't call
3100 flow_find_head_matching_sequence with 0 longest_match.
3101 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3102 non-active insns if !stop_after.
3103 (try_head_merge_bb): Revert 2014-01-07 changes.
3105 2014-01-08 Jeff Law <law@redhat.com>
3107 * ree.c (get_sub_rtx): New function, extracted from...
3108 (merge_def_and_ext): Here.
3109 (combine_reaching_defs): Use get_sub_rtx.
3111 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3113 * cgraph.h (varpool_variable_node): Do not choke on null node.
3115 2014-01-08 Catherine Moore <clm@codesourcery.com>
3117 * config/mips/mips.md (simple_return): Attempt to use JRC
3119 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3121 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3123 PR rtl-optimization/59137
3124 * reorg.c (steal_delay_list_from_target): Call update_block for
3126 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3128 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3130 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3131 two duplicate entries.
3133 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3136 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3138 * config/mips/mips.c (mips_truncated_op_cost): New function.
3139 (mips_rtx_costs): Adjust test for BADDU.
3140 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3142 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3144 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3145 (*baddu_si): ...this new pattern.
3147 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3150 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3152 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3155 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3157 (expand_expr, expand_normal): Adjust.
3158 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3159 inner_reference_p. Use inner_reference_p to expand inner references.
3160 (store_expr): Adjust.
3161 * cfgexpand.c (expand_call_stmt): Adjust.
3163 2014-01-08 Rong Xu <xur@google.com>
3165 * gcov-io.c (gcov_var): Move from gcov-io.h.
3166 (gcov_position): Ditto.
3167 (gcov_is_error): Ditto.
3168 (gcov_rewrite): Ditto.
3169 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3170 only part to libgcc/libgcov.h.
3172 2014-01-08 Marek Polacek <polacek@redhat.com>
3175 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3177 2014-01-08 Marek Polacek <polacek@redhat.com>
3180 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3182 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3184 PR rtl-optimization/59649
3185 * stor-layout.c (get_mode_bounds): For BImode return
3186 0 and STORE_FLAG_VALUE.
3188 2014-01-08 Richard Biener <rguenther@suse.de>
3191 * gimple.h (is_gimple_builtin_call): Remove.
3192 (gimple_builtin_call_types_compatible_p): New.
3193 (gimple_call_builtin_p): New overload.
3194 * gimple.c (is_gimple_builtin_call): Remove.
3195 (validate_call): Rename to ...
3196 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3198 (validate_type): New static function.
3199 (gimple_call_builtin_p): New overload and adjust.
3200 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3201 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3202 (gimple_fold_stmt_to_constant_1): Likewise.
3203 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3205 2014-01-08 Richard Biener <rguenther@suse.de>
3208 * gimplify.c (gimplify_expr): Gimplify register-register type
3209 VIEW_CONVERT_EXPRs to separate stmts.
3211 2014-01-07 Jeff Law <law@redhat.com>
3214 * ree.c (combine_set_extension): Handle case where source
3215 and destination registers in an extension insn are different.
3216 (combine_reaching_defs): Allow source and destination registers
3217 in extension to be different under limited circumstances.
3218 (add_removable_extension): Remove restriction that the
3219 source and destination registers in the extension are the same.
3220 (find_and_remove_re): Emit a copy from the extension's
3221 destination to its source after the defining insn if
3222 the source and destination registers are different.
3225 * ifcvt.c (merge_if_block): If we are merging a block with more than
3226 one successor with a block with no successors, remove any BARRIER
3227 after the second block.
3229 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3231 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3233 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3236 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3237 for 14-bit register offsets when INT14_OK_STRICT is false.
3239 2014-01-07 Roland Stigge <stigge@antcom.de>
3240 Michael Meissner <meissner@linux.vnet.ibm.com>
3243 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3244 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3246 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3248 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3251 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3253 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3254 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3255 rtx is const0_rtx or not.
3257 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3260 * target-globals.c (save_target_globals): Remove this_fn_optab
3262 * toplev.c: Include optabs.h.
3263 (target_reinit): Temporarily restore the global options if another
3264 set of options are in force.
3266 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3268 PR rtl-optimization/58668
3269 * cfgcleanup.c (flow_find_cross_jump): Don't count
3270 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3271 to determine what is counted.
3272 (flow_find_head_matching_sequence): Use active_insn_p to determine
3274 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3276 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3277 determine what is counted.
3279 PR tree-optimization/59643
3280 * tree-predcom.c (split_data_refs_to_components): If one dr is
3281 read and one write, determine_offset fails and the write isn't
3282 in the bad component, just put the read into the bad component.
3284 2014-01-07 Mike Stump <mikestump@comcast.net>
3285 Jakub Jelinek <jakub@redhat.com>
3288 * tree-core.h (struct tree_optimization_option): Change optabs
3289 type from unsigned char * to void *.
3290 * optabs.c (init_tree_optimization_optabs): Adjust
3291 TREE_OPTIMIZATION_OPTABS initialization.
3293 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3296 * config/i386/i386.h (struct machine_function): Add
3297 no_drap_save_restore field.
3298 * config/i386/i386.c (ix86_save_reg): Use
3299 !cfun->machine->no_drap_save_restore instead of
3300 crtl->stack_realign_needed.
3301 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3302 this function clears frame_pointer_needed. Set
3303 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3304 and DRAP reg is needed.
3306 2014-01-06 Marek Polacek <polacek@redhat.com>
3309 * doc/implement-c.texi: Mention that other integer types are
3310 permitted as bit-field types in strictly conforming mode.
3312 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3314 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3317 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3319 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3321 2014-01-06 Martin Jambor <mjambor@suse.cz>
3324 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3326 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3328 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3332 * var-tracking.c (add_stores): Preserve the value of the source even if
3333 we don't record the store.
3335 2014-01-06 Terry Guo <terry.guo@arm.com>
3337 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3339 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3342 * config/darwin.c (darwin_function_section): Adjust return values to
3343 correspond to optimisation changes made in r206070.
3345 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3347 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3348 from prefetch_block tune setting.
3349 (nocona_cost): Correct size of prefetch block to 64.
3351 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3353 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3354 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3355 used to save the static chain register in the computation of the offset
3356 from which the FP registers need to be restored.
3358 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3360 PR tree-optimization/59519
3361 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3362 ICE if get_current_def (current_new_name) is already non-NULL, as long
3363 as it is a phi result of some other phi in *new_exit_bb that has
3366 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3367 or vmovdqu* for misaligned_operand.
3368 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3369 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3370 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3371 aligned_mem for AVX512F masked aligned load and store builtins and for
3374 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3376 PR tree-optimization/59651
3377 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3378 Address range for negative step should be added by TYPE_SIZE_UNIT.
3380 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3382 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3383 overlapping registers also for registers other than the stack pointer.
3385 2014-01-03 Marek Polacek <polacek@redhat.com>
3388 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3391 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3394 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3397 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3398 (push splitter): Use <P:MODE_SIZE> instead of
3399 GET_MODE_SIZE (<P:MODE>mode).
3400 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3401 (mov -1, reg peephole2): Likewise.
3402 * config/i386/sse.md (*mov<mode>_internal,
3403 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3404 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3405 *<code><mode>3, *andnot<mode>3<mask_name>,
3406 <mask_codefor><code><mode>3<mask_name>): Likewise.
3407 * config/i386/subst.md (mask_mode512bit_condition,
3408 sd_mask_mode512bit_condition): Likewise.
3410 2014-01-02 Xinliang David Li <davidxl@google.com>
3412 PR tree-optimization/59303
3413 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3414 (dump_predicates): Better output format.
3415 (pred_equal_p): New function.
3416 (is_neq_relop_p): Ditto.
3417 (is_neq_zero_form_p): Ditto.
3418 (pred_expr_equal_p): Ditto.
3419 (pred_neg_p): Ditto.
3420 (simplify_pred): Ditto.
3421 (simplify_preds_2): Ditto.
3422 (simplify_preds_3): Ditto.
3423 (simplify_preds_4): Ditto.
3424 (simplify_preds): Ditto.
3426 (push_to_worklist): Ditto.
3427 (get_pred_info_from_cmp): Ditto.
3428 (is_degenerated_phi): Ditto.
3429 (normalize_one_pred_1): Ditto.
3430 (normalize_one_pred): Ditto.
3431 (normalize_one_pred_chain): Ditto.
3432 (normalize_preds): Ditto.
3433 (normalize_cond_1): Remove function.
3434 (normalize_cond): Ditto.
3435 (is_gcond_subset_of): Ditto.
3436 (is_subset_of_any): Ditto.
3437 (is_or_set_subset_of): Ditto.
3438 (is_and_set_subset_of): Ditto.
3439 (is_norm_cond_subset_of): Ditto.
3440 (pred_chain_length_cmp): Ditto.
3441 (convert_control_dep_chain_into_preds): Type change.
3442 (find_predicates): Ditto.
3443 (find_def_preds): Ditto.
3444 (destroy_predicates_vecs): Ditto.
3445 (find_matching_predicates_in_rest_chains): Ditto.
3446 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3447 (is_pred_expr_subset): Ditto.
3448 (is_pred_chain_subset_of): Ditto.
3449 (is_included_in): Ditto.
3450 (is_superset_of): Ditto.
3452 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3454 Update copyright years.
3456 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3458 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3459 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3460 config/arc/arc.md, config/arc/arc.opt,
3461 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3462 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3463 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3464 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3465 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3466 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3467 vtable-verify.c, vtable-verify.h: Use the standard form for the
3470 2014-01-02 Tobias Burnus <burnus@net-b.de>
3472 * gcc.c (process_command): Update copyright notice dates.
3473 * gcov-dump.c: Ditto.
3475 * doc/cpp.texi: Bump @copying's copyright year.
3476 * doc/cppinternals.texi: Ditto.
3477 * doc/gcc.texi: Ditto.
3478 * doc/gccint.texi: Ditto.
3479 * doc/gcov.texi: Ditto.
3480 * doc/install.texi: Ditto.
3481 * doc/invoke.texi: Ditto.
3483 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3485 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3487 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3489 * config/i386/sse.md (*mov<mode>_internal): Guard
3490 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3492 PR rtl-optimization/59647
3493 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3494 new_rtx into UNSIGNED_FLOAT rtxes.
3496 Copyright (C) 2014 Free Software Foundation, Inc.
3498 Copying and distribution of this file, with or without modification,
3499 are permitted in any medium without royalty provided the copyright
3500 notice and this notice are preserved.