1 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
4 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
5 only when -Wpsabi is enabled.
7 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
10 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
11 passing arrays in registers are the same as for structs, so remove the
12 special case for them.
14 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
16 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
17 destination type, extract only the valid bits if the source type is not
18 integral and has a different mode.
20 2014-02-19 Richard Biener <rguenther@suse.de>
23 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
26 2014-02-19 Richard Biener <rguenther@suse.de>
29 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
30 (ipa_modify_call_arguments): Emit an argument load explicitely and
31 preserve virtual SSA form there and for the replacement call.
32 Do not update SSA form nor free dominance info.
34 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
36 * ipa.c (function_and_variable_visibility): Also clear WEAK
37 flag when disolving COMDAT_GROUP.
39 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
41 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
42 * ipa-prop.c (ipa_set_jf_known_type): Return early when
44 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
45 do more sanity checks.
46 (detect_type_change): Return true when giving up early.
47 (compute_complex_assign_jump_func): Fix type parameter of
49 (compute_complex_ancestor_jump_func): Likewise.
50 (update_jump_functions_after_inlining): Fix updating of
52 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
54 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
56 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
57 inline clones when edge disappears.
59 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
62 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
63 Split 64-bit moves into 2 patterns. Do not allow the use of
64 direct move for TDmode in little endian, since the decimal value
65 has little endian bytes within a word, but the 64-bit pieces are
66 ordered in a big endian fashion, and normal subreg's of TDmode are
68 (mov<mode>_64bit_dm): Likewise.
69 (movtd_64bit_nodm): Likewise.
71 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
73 PR tree-optimization/60174
74 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
75 statement of an SSA_NAME that occurs in an abnormal PHI node.
77 2014-02-18 Jakub Jelinek <jakub@redhat.com>
80 * final.c (SEEN_BB): Remove.
81 (SEEN_NOTE, SEEN_EMITTED): Renumber.
82 (final_scan_insn): Don't force_source_line on second
83 NOTE_INSN_BASIC_BLOCK.
85 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
88 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
89 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
90 (type_natural_mode): Warn ABI change when %zmm register is not
91 available for AVX512F vector value passing.
93 2014-02-18 Kai Tietz <ktietz@redhat.com>
96 * config/i386/i386.c (ix86_expand_prologue): Use value in
97 rax register as displacement when restoring %r10 or %rax.
98 Fix wrong offset when restoring both registers.
100 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
102 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
103 assertion with conditional return.
105 2014-02-18 Jakub Jelinek <jakub@redhat.com>
106 Uros Bizjak <ubizjak@gmail.com>
109 * config/i386/driver-i386.c (host_detect_local_cpu): If
110 YMM state is not saved by the OS, also clear has_f16c. Move
111 CPUID 0x80000001 handling before YMM state saving checking.
113 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
115 PR rtl-optimization/58960
116 * haifa-sched.c (alloc_global_sched_pressure_data): New,
117 factored out from ...
118 (sched_init): ... here.
119 (free_global_sched_pressure_data): New, factored out from ...
120 (sched_finish): ... here.
121 * sched-int.h (free_global_sched_pressure_data): Declare.
122 * sched-rgn.c (nr_regions_initial): New static global.
123 (haifa_find_rgns): Initialize it.
124 (schedule_region): Disable sched-pressure for the newly
127 2014-02-17 Richard Biener <rguenther@suse.de>
129 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
130 release SSA defs of pattern stmts.
132 2014-02-17 Richard Biener <rguenther@suse.de>
134 * tree-inline.c (expand_call_inline): Release the virtual
135 operand defined by the call we are about to inline.
137 2014-02-17 Richard Biener <rguenther@suse.de>
139 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
141 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
142 Ilya Tocar <ilya.tocar@intel.com>
144 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
145 arguments order in builtin.
146 (_mm512_permutexvar_epi64): Ditto.
147 (_mm512_mask_permutexvar_epi64): Ditto
148 (_mm512_maskz_permutexvar_epi32): Ditto
149 (_mm512_permutexvar_epi32): Ditto
150 (_mm512_mask_permutexvar_epi32): Ditto
152 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
154 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
155 (p8_vmrgow): Likewise.
157 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
159 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
162 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
165 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
166 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
167 into 64-bit and 32-bit moves. On 64-bit moves, add support for
168 using direct move instructions on ISA 2.07. Also adjust
169 instruction length for 64-bit.
170 (mov<mode>_64bit, TFmode/TDmode): Likewise.
171 (mov<mode>_32bit, TFmode/TDmode): Likewise.
173 2014-02-15 Alan Modra <amodra@gmail.com>
177 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
178 find_replacement on parts of insn rtl that might be reloaded.
180 2014-02-15 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/60183
183 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
184 (tree_ssa_phiprop): Calculate and free post-dominators.
186 2014-02-14 Jeff Law <law@redhat.com>
188 PR rtl-optimization/60131
189 * ree.c (get_extended_src_reg): New function.
190 (combine_reaching_defs): Use it rather than assuming location of REG.
191 (find_and_remove_re): Verify first operand of extension is
192 a REG before adding the insns to the copy list.
194 2014-02-14 Roland McGrath <mcgrathr@google.com>
196 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
197 * configure: Regenerated.
198 * config.in: Regenerated.
199 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
200 instead of ASM_SHORT.
202 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
203 Richard Earnshaw <rearnsha@arm.com>
205 PR rtl-optimization/59535
206 * lra-constraints.c (process_alt_operands): Encourage alternative
207 when unassigned pseudo class is superset of the alternative class.
208 (inherit_reload_reg): Don't inherit when optimizing for code size.
209 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
210 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
211 modes not less than 4 for Thumb1.
213 2014-02-14 Kyle McMartin <kyle@redhat.com>
216 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
218 2014-02-14 Richard Biener <rguenther@suse.de>
220 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
221 (get_frame_arg): Drop the assert with langhook types_compatible_p.
222 Do not strip INDIRECT_REFs.
224 2014-02-14 Richard Biener <rguenther@suse.de>
227 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
228 DECL_FUNCTION_SPECIFIC_TARGET.
229 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
230 * tree-streamer-out.c (pack_ts_target_option): Remove.
231 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
232 (write_ts_function_decl_tree_pointers): Do not stream
233 DECL_FUNCTION_SPECIFIC_TARGET.
234 * tree-streamer-in.c (unpack_ts_target_option): Remove.
235 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
236 (lto_input_ts_function_decl_tree_pointers): Do not stream
237 DECL_FUNCTION_SPECIFIC_TARGET.
239 2014-02-14 Jakub Jelinek <jakub@redhat.com>
241 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
242 (get_initial_def_for_induction, vectorizable_induction): Ignore
243 debug stmts when looking for exit_phi.
244 (vectorizable_live_operation): Fix up condition.
246 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
248 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
249 nreverse() because it changes the content of original tree list.
251 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
253 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
254 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
256 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
258 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
259 GNU coding standards.
261 2014-02-13 Jakub Jelinek <jakub@redhat.com>
264 * dwarf2out.c (gen_subprogram_die): Don't call
265 add_calling_convention_attribute if subr_die is old_die.
267 2014-02-13 Sharad Singhai <singhai@google.com>
269 * doc/optinfo.texi: Fix order of nodes.
271 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
273 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
274 operands[2], not operands[3].
276 2014-02-13 Richard Biener <rguenther@suse.de>
279 * doc/install.texi (ISL): Update recommended version to 0.12.2,
280 mention the possibility of an in-tree build.
281 (CLooG): Update recommended version to 0.18.1, mention the
282 possibility of an in-tree build and clarify that the ISL
283 bundled with CLooG does not work.
285 2014-02-13 Jakub Jelinek <jakub@redhat.com>
288 * expr.c (compress_float_constant): If x is a hard register,
289 extend into a pseudo and then move to x.
291 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
293 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
294 caused by bad second argument to warning_at() with -mhotpatch and
295 nested functions (e.g. with gfortran).
297 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
299 * opts.c (option_name): Remove "enabled by default" rider.
301 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
303 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
305 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
306 Uros Bizjak <ubizjak@gmail.com>
309 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
310 * configure: Regenerated.
312 2014-02-12 Richard Biener <rguenther@suse.de>
314 * vec.c (vec_prefix::calculate_allocation): Move as
315 inline variant to vec.h.
316 (vec_prefix::calculate_allocation_1): New out-of-line version.
317 * vec.h (vec_prefix::calculate_allocation_1): Declare.
318 (vec_prefix::m_has_auto_buf): Rename to ...
319 (vec_prefix::m_using_auto_storage): ... this.
320 (vec_prefix::calculate_allocation): Inline the easy cases
321 and dispatch to calculate_allocation_1 which doesn't need the
323 (va_heap::reserve): Use gcc_checking_assert.
324 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
325 m_using_auto_storage.
326 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
328 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
329 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
330 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
332 2014-02-12 Richard Biener <rguenther@suse.de>
334 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
335 when we found a dependence.
337 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
339 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
341 (maybe_fold_stmt): ... into this new function.
342 * omp-low.c (lower_omp): Update comment.
344 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
347 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
350 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
352 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
353 identifiers in comments.
354 (cortexa53_extra_costs): Likewise.
355 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
356 (cortexa7_extra_costs): Likewise.
357 (cortexa12_extra_costs): Likewise.
358 (cortexa15_extra_costs): Likewise.
359 (v7m_extra_costs): Likewise.
361 2014-02-12 Richard Biener <rguenther@suse.de>
364 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
365 of posix_memalign being successful.
366 (lower_stmt): Restrict lowering of posix_memalign to when
367 -ftree-bit-ccp is enabled.
369 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
371 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
373 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
375 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
377 PR rtl-optimization/60116
378 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
379 other_insn once the combination has been validated.
381 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
384 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
386 * ipa-devirt.c: Include demangle.h
387 (odr_violation_reported): New static variable.
388 (add_type_duplicate): Update odr_violations.
389 (maybe_record_node): Add completep parameter; update it.
390 (record_target_from_binfo): Add COMPLETEP parameter;
392 (possible_polymorphic_call_targets_1): Likewise.
393 (struct polymorphic_call_target_d): Add nonconstruction_targets;
394 rename FINAL to COMPLETE.
395 (record_targets_from_bases): Sanity check we found the binfo;
396 fix COMPLETEP updating.
397 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
398 parameter, fix computing of COMPLETEP.
399 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
400 at LTO time do demangling.
401 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
402 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
404 (gimple_get_virt_method_for_binfo): Likewise.
405 * gimple-fold.h (gimple_get_virt_method_for_binfo,
406 gimple_get_virt_method_for_vtable): Update prototypes.
408 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
411 * genautomata.c (add_presence_absence): Fix typo with
412 {final_}presence_list.
414 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
417 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
418 for VSX/Altivec vectors that land in GPR registers.
420 2014-02-11 Richard Henderson <rth@redhat.com>
421 Jakub Jelinek <jakub@redhat.com>
424 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
425 around drhs if type conversion to lacc->type is not useless.
427 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
429 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
431 (cortex-a57.cortex-a53): Likewise.
432 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
434 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
436 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
439 2014-02-11 Renlin Li <Renlin.Li@arm.com>
441 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
442 add_options_for_arm_vfp3.
444 2014-02-11 Jeff Law <law@redhat.com>
447 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
448 object with an undesirable mode.
450 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
453 * config/i386/sol2-9.h: New file.
454 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
455 *-*-solaris2.9*): Use it.
457 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
459 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
460 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
462 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
464 * config/microblaze/microblaze.c: Extend mcpu version format
466 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
468 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
470 2014-02-10 Richard Henderson <rth@redhat.com>
473 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
474 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
475 ms-abi vs -mno-accumulate-outgoing-args.
476 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
477 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
480 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
483 * cfgexpand.c (expand_asm_operands): Attach source location to
484 ASM_INPUT rtx objects.
485 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
487 2014-02-10 Nick Clifton <nickc@redhat.com>
489 * config/mn10300/mn10300.c (popcount): New function.
490 (mn10300_expand_prologue): Include saved registers in stack usage
493 2014-02-10 Jeff Law <law@redhat.com>
496 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
497 when changing the SET_DEST of a prior insn to avoid an input reload.
499 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
501 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
502 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
503 -mcall-openbsd, or -mcall-linux.
504 (CC1_ENDIAN_BIG_SPEC): Remove.
505 (CC1_ENDIAN_LITTLE_SPEC): Remove.
506 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
507 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
508 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
509 and %cc1_endian_default.
510 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
512 2014-02-10 Richard Biener <rguenther@suse.de>
514 PR tree-optimization/60115
515 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
516 MEM_REF handling. Properly verify that the accesses are not
517 out of the objects bound.
519 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
521 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
524 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
526 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
527 proper constants and fix formatting.
528 (possible_polymorphic_call_targets): Fix formatting.
530 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
531 Ilya Tocar <ilya.tocar@intel.com>
533 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
534 (_mm512_loadu_epi32): Renamed into...
535 (_mm512_loadu_si512): This.
536 (_mm512_storeu_epi32): Renamed into...
537 (_mm512_storeu_si512): This.
538 (_mm512_maskz_ceil_ps): Removed.
539 (_mm512_maskz_ceil_pd): Ditto.
540 (_mm512_maskz_floor_ps): Ditto.
541 (_mm512_maskz_floor_pd): Ditto.
542 (_mm512_floor_round_ps): Ditto.
543 (_mm512_floor_round_pd): Ditto.
544 (_mm512_ceil_round_ps): Ditto.
545 (_mm512_ceil_round_pd): Ditto.
546 (_mm512_mask_floor_round_ps): Ditto.
547 (_mm512_mask_floor_round_pd): Ditto.
548 (_mm512_mask_ceil_round_ps): Ditto.
549 (_mm512_mask_ceil_round_pd): Ditto.
550 (_mm512_maskz_floor_round_ps): Ditto.
551 (_mm512_maskz_floor_round_pd): Ditto.
552 (_mm512_maskz_ceil_round_ps): Ditto.
553 (_mm512_maskz_ceil_round_pd): Ditto.
554 (_mm512_expand_pd): Ditto.
555 (_mm512_expand_ps): Ditto.
556 * config/i386/i386.c (ix86_builtins): Remove
557 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
559 * config/i386/predicates.md (const1256_operand): New.
560 (const_1_to_2_operand): Ditto.
561 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
562 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
563 (*avx512pf_gatherpf<mode>sf): Ditto.
564 (avx512pf_gatherpf<mode>df): Ditto.
565 (*avx512pf_gatherpf<mode>df_mask): Ditto.
566 (*avx512pf_gatherpf<mode>df): Ditto.
567 (avx512pf_scatterpf<mode>sf): Ditto.
568 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
569 (*avx512pf_scatterpf<mode>sf): Ditto.
570 (avx512pf_scatterpf<mode>df): Ditto.
571 (*avx512pf_scatterpf<mode>df_mask): Ditto.
572 (*avx512pf_scatterpf<mode>df): Ditto.
573 (avx512f_expand<mode>): Removed.
574 (<shift_insn><mode>3<mask_name>): Change predicate type.
576 2014-02-08 Jakub Jelinek <jakub@redhat.com>
578 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
579 not at the end of datarefs vector use ordered_remove to avoid
580 reordering datarefs vector.
583 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
584 mark local addressable non-static vars as GOVD_PRIVATE
585 instead of GOVD_LOCAL.
586 * omp-low.c (lower_omp_for): Move gimple_bind_vars
587 and BLOCK_VARS of gimple_bind_block to new_stmt rather
591 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
592 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
593 assume_aligned or alloc_align attributes.
594 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
595 arguments. Handle also assume_aligned and alloc_align attributes.
596 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
597 calls to functions with assume_aligned or alloc_align attributes.
598 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
600 2014-02-08 Terry Guo <terry.guo@arm.com>
602 * doc/invoke.texi: Document ARM -march=armv7e-m.
604 2014-02-08 Jakub Jelinek <jakub@redhat.com>
606 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
607 flag on __cilkrts_rethrow builtin.
610 * ipa-cp.c (determine_versionability): Fail at -O0
611 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
612 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
615 2014-02-04 Jakub Jelinek <jakub@redhat.com>
618 * tree-inline.c (copy_forbidden): Fail for
619 __attribute__((optimize (0))) functions.
621 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
623 * varpool.c: Include pointer-set.h.
624 (varpool_remove_unreferenced_decls): Variables in other partitions
625 will not be output; be however careful to not lose information
628 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
630 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
631 lookup in the vtable constructor.
633 2014-02-07 Jeff Law <law@redhat.com>
636 * config/m68k/m68k.md (ashldi_extsi): Turn into a
637 define_insn_and_split.
639 * ipa-inline.c (inline_small_functions): Fix typos.
641 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
643 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
644 (s390_can_use_return_insn): Declare.
645 * config/s390/s390.h (EPILOGUE_USES): Define.
646 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
648 (s390_chunkify_start): Handle return JUMP_LABELs.
649 (s390_early_mach): Emit a main_pool instruction on the entry edge.
650 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
651 (s390_can_use_return_insn): New functions.
652 (s390_fix_long_loop_prediction): Handle conditional returns.
653 (TARGET_SET_UP_BY_PROLOGUE): Define.
654 * config/s390/s390.md (ANY_RETURN): New code iterator.
655 (*creturn, *csimple_return, return, simple_return): New patterns.
657 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
659 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
660 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
661 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
662 REG_CFA_RESTORE list when deciding not to restore a register.
664 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
666 * config/s390/s390.c: Include tree-pass.h and context.h.
667 (s390_early_mach): New function, split out from...
668 (s390_emit_prologue): ...here.
669 (pass_data_s390_early_mach): New pass structure.
670 (pass_s390_early_mach): New class.
671 (s390_option_override): Create and register early_mach pass.
674 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
676 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
677 to match for the exit block.
679 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
681 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
682 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
683 Reject misaligned operands.
685 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
687 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
689 2014-02-07 Richard Biener <rguenther@suse.de>
692 * gimple-low.c (lower_builtin_posix_memalign): New function.
693 (lower_stmt): Call it to lower posix_memalign in a way
694 to make alignment info accessible.
696 2014-02-07 Jakub Jelinek <jakub@redhat.com>
699 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
700 __builtin_setjmp_receiver.
702 2014-02-07 Richard Biener <rguenther@suse.de>
705 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
706 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
707 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
708 Handle BUILT_IN_POSIX_MEMALIGN.
709 (find_func_clobbers): Likewise.
710 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
711 (call_may_clobber_ref_p_1): Likewise.
713 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
716 * ipa-devirt.c (record_target_from_binfo): Remove overactive
719 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
722 * lto-cgraph.c (lto_output_node): Use
723 symtab_get_symbol_partitioning_class.
724 (lto_output_varpool_node): likewise.
725 (symtab_get_symbol_partitioning_class): Move here from
727 * cgraph.h (symbol_partitioning_class): Likewise.
728 (symtab_get_symbol_partitioning_class): Declare.
730 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
732 * ggc.h (ggc_internal_cleared_alloc): New macro.
733 * vec.h (vec_safe_copy): Handle memory stats.
734 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
735 * target-globals.c (save_target_globals): Likewise.
737 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
740 * expr.c (emit_move_resolve_push): Export; be bit more selective
741 on when to clear alias set.
742 * expr.h (emit_move_resolve_push): Declare.
743 * function.h (struct function): Add tail_call_marked.
744 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
745 * config/i386/i386-protos.h (ix86_expand_push): Remove.
746 * config/i386/i386.md (TImode move expander): De not call
748 (FP push expanders): Preserve memory attributes.
749 * config/i386/sse.md (push<mode>1): Remove.
750 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
751 (ix86_expand_push): Remove.
752 * config/i386/mmx.md (push<mode>1): Remove.
754 2014-02-06 Jakub Jelinek <jakub@redhat.com>
756 PR rtl-optimization/60030
757 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
758 lopart with paradoxical subreg before shifting it up by hprec.
760 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
762 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
763 Remove extra newline at end of file.
764 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
765 (arm_issue_rate): Handle cortexa57.
766 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
767 (cortex-a57.cortex-a53): Likewise.
769 2014-02-06 Jakub Jelinek <jakub@redhat.com>
772 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
773 don't record in REG_FRAME_RELATED_EXPR registers not set in that
775 (arm_expand_prologue): Adjust all callers.
776 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
777 info, registers also at the lowest numbered registers side. Use
778 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
782 * var-tracking.c (adjust_mems): Before adding a SET to
783 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
785 2014-02-06 Alan Modra <amodra@gmail.com>
788 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
789 change SDmode to DDmode when lra_in_progress.
791 2014-02-06 Jakub Jelinek <jakub@redhat.com>
794 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
795 free_data_ref on the dr first, and before goto again also set dr
796 to the next dr. For simd_lane_access, free old datarefs[i] before
797 overwriting it. For get_vectype_for_scalar_type failure, don't
798 free_data_ref if simd_lane_access.
800 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
803 * tree.h (opts_for_fn): New inline function.
804 (opt_for_fn): Define.
805 * config/i386/i386.c (ix86_function_regparm): Use
806 opt_for_fn (decl, optimize) instead of optimize.
808 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
810 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
811 for SYMBOL_REF in large memory model.
813 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
815 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
817 (cortex-a57): Likewise.
818 (cortex-a57.cortex-a53): Likewise.
820 2014-02-06 Yury Gribov <y.gribov@samsung.com>
821 Kugan Vivekanandarajah <kuganv@linaro.org>
823 * config/arm/arm.c (arm_vector_alignment_reachable): Check
825 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
827 2014-02-06 Richard Biener <rguenther@suse.de>
829 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
830 set_loop_copy and initialize_original_copy_tables.
832 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
834 * config/aarch64/aarch64-simd.md
835 (aarch64_ashr_simddi): Change QI to SI.
837 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
838 Jakub Jelinek <jakub@redhat.com>
841 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
844 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
846 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
847 CODE_FOR_altivec_vpku[hw]um to
848 CODE_FOR_altivec_vpku[hw]um_direct.
849 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
850 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
851 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
852 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
854 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
856 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
857 generation for -maltivec=be.
858 (altivec_vsumsws): Simplify redundant test.
860 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
862 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
863 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
864 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
865 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
867 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
869 (altivec_vpks<VI_char>ss): Likewise.
870 (altivec_vpks<VI_char>us): Likewise.
871 (altivec_vpku<VI_char>us): Likewise.
872 (altivec_vpku<VI_char>um): Likewise.
873 (altivec_vpku<VI_char>um_direct): New (copy of
874 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
876 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
877 target is little endian and -maltivec=be is not specified.
878 (*altivec_vupkhs<VU_char>_direct): New (copy of
879 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
880 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
881 target is little endian and -maltivec=be is not specified.
882 (*altivec_vupkls<VU_char>_direct): New (copy of
883 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
884 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
885 little endian and -maltivec=be is not specified.
886 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
887 little endian and -maltivec=be is not specified.
889 2014-02-05 Richard Henderson <rth@redhat.com>
892 * combine-stack-adj.c: Revert r206943.
893 * sched-int.h (struct deps_desc): Add last_args_size.
894 * sched-deps.c (init_deps): Initialize it.
895 (sched_analyze_insn): Add OUTPUT dependencies between insns that
896 contain REG_ARGS_SIZE notes.
898 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
900 * lto-cgraph.c (asm_nodes_output): Make global.
901 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
902 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
903 (driver_handle_option): Handle OPT_fwpa.
905 2014-02-05 Jakub Jelinek <jakub@redhat.com>
908 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
909 a comment typo and formatting issue. If odr_hash hasn't been
910 created, return vNULL and set *completep to false.
913 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
914 bb with no successors.
916 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
919 * doc/invoke.texi (-march): Clarify documentation for ARM.
923 2014-02-05 Richard Biener <rguenther@suse.de>
925 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
926 when not vectorizing because of too many alias checks.
927 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
928 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
930 2014-02-05 Nick Clifton <nickc@redhat.com>
932 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
933 accept extended registers in any mode when compiling for the MN10300.
935 2014-02-05 Yury Gribov <y.gribov@samsung.com>
937 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
938 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
939 sanitization attributes.
940 (can_inline_edge_p): Likewise.
941 (sanitize_attrs_match_for_inline_p): New function.
943 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
945 * ipa-prop.c (detect_type_change): Shor circuit testing of
946 type changes on THIS pointer.
948 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
951 * config/pa/pa.c (legitimize_tls_address): Return original address
952 if not passed a SYMBOL_REF rtx.
953 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
955 (pa_emit_move_sequence): Simplify TLS source operands.
956 (pa_legitimate_constant_p): Reject all TLS constants.
957 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
958 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
960 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
962 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
963 groups when we know they are controlled by LTO.
964 * varasm.c (default_binds_local_p_1): If object is in other partition,
965 it will be resolved locally.
967 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
969 * config/host-linux.c (linux_gt_pch_use_address): Don't
970 use SSIZE_MAX because it is not always defined.
972 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
975 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
976 threshold for pseudo splitting.
977 (update_ebb_live_info): Process call argument hard registers and
978 hard registers from insn definition too.
979 (max_small_class_regs_num): New constant.
980 (inherit_in_ebb): Update live hard regs through EBBs. Update
981 reloads_num only for small register classes. Don't split for
984 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
987 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
990 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
992 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
995 2014-02-04 Marek Polacek <polacek@redhat.com>
997 * gdbinit.in (pel): Define.
999 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1001 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1004 2014-02-04 Richard Biener <rguenther@suse.de>
1007 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1008 in function context local.
1009 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1010 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1011 similar to LTO_imported_decl_ref.
1013 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1015 PR tree-optimization/60002
1016 * cgraphclones.c (build_function_decl_skip_args): Clear
1019 PR tree-optimization/60023
1020 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1021 false to gsi_replace.
1022 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1023 has been in some EH region and vec_stmt could throw, add
1024 vec_stmt into the same EH region.
1025 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1026 has no lhs, ignore it.
1027 * internal-fn.c (expand_MASK_LOAD): Likewise.
1030 * tree-inline.c (copy_forbidden): Fail for
1031 __attribute__((optimize (0))) functions.
1034 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1035 is set, copy one less argument.
1036 (expand_simd_clones): Don't subtract clone_info->inbranch
1037 from simd_clone_struct_alloc argument.
1039 PR rtl-optimization/57915
1040 * recog.c (simplify_while_replacing): If all unary/binary/relational
1041 operation arguments are constant, attempt to simplify those.
1044 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1045 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1047 2014-02-04 Richard Biener <rguenther@suse.de>
1049 PR tree-optimization/60012
1050 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1051 TBAA disambiguation to all DDRs.
1053 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1056 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1057 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1059 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1062 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1064 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1066 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1067 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1069 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1072 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1073 to figure out targets of polymorphic calls with known decl.
1074 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1075 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1076 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1077 (get_polymorphic_call_info): ... here.
1078 (get_polymorphic_call_info_from_invariant): New function.
1080 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1082 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1083 lookup via vtable pointer; check for type consistency
1084 and turn inconsitent facts into UNREACHABLE.
1085 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1086 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1087 type inconsistent querries; return UNREACHABLE instead.
1089 2014-02-03 Richard Henderson <rth@twiddle.net>
1092 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1093 already processed this node.
1094 (normalize_one_pred_1): Pass along mark_set.
1095 (normalize_one_pred): Create and destroy a pointer_set_t.
1096 (normalize_one_pred_chain): Likewise.
1098 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1100 PR gcov-profile/58602
1101 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1103 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1106 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1107 -fno-devirtualize; try to devirtualize by the knowledge of
1108 virtual table pointer given by aggregate propagation.
1109 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1110 (ipa_print_node_jump_functions): Dump also offset that
1111 is relevant for polymorphic calls.
1112 (determine_known_aggregate_parts): Add arg_type parameter; use it
1113 instead of determining the type from pointer type.
1114 (ipa_compute_jump_functions_for_edge): Update call of
1115 determine_known_aggregate_parts.
1116 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1117 (gimple_get_virt_method_for_binfo): ... here; simplify using
1118 vtable_pointer_value_to_vtable.
1119 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1120 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1121 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1122 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1124 (vtable_pointer_value_to_binfo): ... here.
1125 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1127 2014-02-03 Teresa Johnson <tejohnson@google.com>
1129 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1130 redef of outer loop index variable.
1132 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1136 * doc/extend.texi (Function Attributes): Typo.
1138 2014-02-03 Cong Hou <congh@google.com>
1140 PR tree-optimization/60000
1141 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1142 if the vectorized statement is a store. A store statement can only
1143 appear at the end of pattern statements.
1145 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1147 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1148 (ix86_option_override_internal): Default long double to 64-bit for
1149 32-bit Bionic and to 128-bit for 64-bit Bionic.
1151 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1152 TARGET_LONG_DOUBLE_128 is true.
1153 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1155 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1156 (mlong-double-64): Negate -mlong-double-128.
1157 (mlong-double-128): New option.
1159 * config/i386/i386-c.c (ix86_target_macros): Define
1160 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1162 * doc/invoke.texi: Document -mlong-double-128.
1164 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1166 PR rtl-optimization/60024
1167 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1169 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1171 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1173 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1175 PR rtl-optimization/57662
1176 * sel-sched.c (code_motion_path_driver): Do not mark already not
1177 existing blocks in the visiting bitmap.
1179 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1181 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1182 on the insn being emitted.
1184 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1185 Will Deacon <will.deacon@arm.com>
1187 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1189 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1191 * config/arm/arm-tables.opt: Regenerate.
1193 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1195 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1196 for vector types other than V16QImode.
1197 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1198 define_expand, and call altivec_expand_vec_perm_le when producing
1199 code with little endian element order.
1200 (*altivec_vperm_<mode>_internal): New insn having previous
1201 behavior of altivec_vperm_<mode>.
1202 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1203 altivec_expand_vec_perm_le when producing code with little endian
1205 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1206 behavior of altivec_vperm_<mode>_uns.
1208 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1210 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1211 (altivec_vsumsws): Add handling for -maltivec=be with a little
1213 (altivec_vsumsws_direct): New.
1214 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1215 gen_altivec_vsumsws.
1217 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1219 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1220 vtable_pointer_value_to_binfo): New functions.
1221 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1222 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1224 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1226 * config/nios2/nios2.md (load_got_register): Initialize GOT
1227 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1228 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1230 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1232 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1233 preserverd by passthrough, do not propagate the type.
1235 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1237 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1238 (mips_atomic_assign_expand_fenv): New function.
1239 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1241 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1243 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1244 (__builtin_mips_set_fcsr): Likewise.
1245 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1246 MIPS_USI_FTYPE_VOID.
1247 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1248 (mips16_expand_set_fcsr): Likewise.
1249 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1250 (mips16_set_fcsr_stub): Likewise.
1251 (mips16_get_fcsr_one_only_stub): New class.
1252 (mips16_set_fcsr_one_only_stub): Likewise.
1253 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1254 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1255 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1256 (hard_float): New availability predicate.
1257 (mips_builtins): Add get_fcsr and set_fcsr.
1258 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1259 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1260 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1261 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1262 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1265 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1267 * config/mips/mips.c (mips_one_only_stub): New class.
1268 (mips_need_mips16_rdhwr_p): Replace with...
1269 (mips16_rdhwr_stub): ...this new variable.
1270 (mips16_stub_call_address): New function.
1271 (mips16_rdhwr_one_only_stub): New class.
1272 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1273 (mips_output_mips16_rdhwr): Delete.
1274 (mips_finish_stub): New function.
1275 (mips_code_end): Use it to handle rdhwr stubs.
1277 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1280 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1281 when calculating size of integer atomic types.
1283 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1285 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1287 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1289 PR tree-optimization/60003
1290 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1291 * profile.c (branch_prob): Use gimple_call_builtin_p
1292 to check for BUILT_IN_SETJMP_RECEIVER.
1293 * tree-inline.c (copy_bb): Call notice_special_calls.
1295 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1298 * lra-constraints.c (process_alt_operands): Update reload_sum only
1301 2014-01-31 Richard Henderson <rth@redhat.com>
1304 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1305 until after else_eh is processed.
1307 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1309 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1310 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1311 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1312 in smmintrin.h, remove them.
1313 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1314 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1315 * config/i386/i386.md (ROUND_SAE): Fix value.
1316 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1317 (const48_operand): New.
1318 * config/i386/subst.md (round), (round_expand): Use
1319 const_4_or_8_to_11_operand.
1320 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1322 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1324 * config/i386/constraints.md (Yk): Swap meaning with k.
1325 * config/i386/i386.md (movhi_internal): Change Yk to k.
1326 (movqi_internal): Ditto.
1327 (*k<logic><mode>): Ditto.
1330 (kandn<mode>): Ditto.
1331 (*<code>hi_1): Ditto.
1332 (*<code>qi_1): Ditto.
1333 (kxnor<mode>): Ditto.
1334 (kortestzhi): Ditto.
1335 (kortestchi): Ditto.
1337 (*one_cmplhi2_1): Ditto.
1338 (*one_cmplqi2_1): Ditto.
1339 * config/i386/sse.md (): Change k to Yk.
1340 (avx512f_load<mode>_mask): Ditto.
1341 (avx512f_blendm<mode>): Ditto.
1342 (avx512f_store<mode>_mask): Ditto.
1343 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1344 (avx512f_storedqu<mode>_mask): Ditto.
1345 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1347 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1348 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1349 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1350 (avx512f_maskcmp<mode>3): Ditto.
1351 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1352 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1353 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1354 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1355 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1356 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1357 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1358 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1359 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1360 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1361 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1362 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1363 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1364 (vec_extract_lo_<mode>_maskm): Ditto.
1365 (vec_extract_hi_<mode>_maskm): Ditto.
1366 (avx512f_vternlog<mode>_mask): Ditto.
1367 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1368 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1369 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1370 (avx512f_<code>v8div16qi2_mask): Ditto.
1371 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1372 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1373 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1374 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1375 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1376 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1377 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1378 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1379 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1380 (avx512cd_maskb_vec_dupv8di): Ditto.
1381 (avx512cd_maskw_vec_dupv16si): Ditto.
1382 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1383 (avx512f_vpermi2var<mode>3_mask): Ditto.
1384 (avx512f_vpermi2var<mode>3_mask): Ditto.
1385 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1386 (*avx512f_gathersi<mode>): Ditto.
1387 (*avx512f_gathersi<mode>_2): Ditto.
1388 (*avx512f_gatherdi<mode>): Ditto.
1389 (*avx512f_gatherdi<mode>_2): Ditto.
1390 (*avx512f_scattersi<mode>): Ditto.
1391 (*avx512f_scatterdi<mode>): Ditto.
1392 (avx512f_compress<mode>_mask): Ditto.
1393 (avx512f_compressstore<mode>_mask): Ditto.
1394 (avx512f_expand<mode>_mask): Ditto.
1395 * config/i386/subst.md (mask): Change k to Yk.
1396 (mask_scalar_merge): Ditto.
1399 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1401 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1403 2014-01-31 Richard Biener <rguenther@suse.de>
1406 * builtins.c (fold_builtin_memory_op): Make sure to not
1407 use a floating-point mode or a boolean or enumeral type for
1410 2014-01-30 DJ Delorie <dj@redhat.com>
1412 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1413 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1414 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1415 whenever main() has an epilogue.
1417 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1419 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1420 unused variable "field".
1421 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1422 (vsx_mergeh_<mode>): Likewise.
1423 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1424 (altivec_vmrghh): Likewise.
1425 (altivec_vmrghw): Likewise.
1426 (altivec_vmrglb): Likewise.
1427 (altivec_vmrglh): Likewise.
1428 (altivec_vmrglw): Likewise.
1429 (altivec_vspltb): Add missing uses.
1430 (altivec_vsplth): Likewise.
1431 (altivec_vspltw): Likewise.
1432 (altivec_vspltsf): Likewise.
1434 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1437 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1438 frame related instructions.
1440 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1442 PR rtl-optimization/59959
1443 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1444 any reload of register whose subreg is invalid.
1446 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1448 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1449 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1450 Add missing return type - void.
1452 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1454 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1455 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1456 remove element index adjustment for endian (now handled in vsx.md
1458 (altivec_expand_vec_perm_const): Use
1459 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1460 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1461 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1462 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1463 define_expand and a new define_insn *altivec_vspltb_internal;
1464 adjust for -maltivec=be on a little endian target.
1465 (altivec_vspltb_direct): New.
1466 (altivec_vsplth): Divide into a define_expand and a new
1467 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1468 little endian target.
1469 (altivec_vsplth_direct): New.
1470 (altivec_vspltw): Divide into a define_expand and a new
1471 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1472 little endian target.
1473 (altivec_vspltw_direct): New.
1474 (altivec_vspltsf): Divide into a define_expand and a new
1475 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1476 a little endian target.
1478 2014-01-30 Richard Biener <rguenther@suse.de>
1480 PR tree-optimization/59993
1481 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1482 can propagate form the earlier stmt and avoid the transform
1483 when the intermediate result is needed.
1485 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1487 * README.Portability: Fix typo.
1489 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1491 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1492 comparison_operator with ordered_comparison_operator.
1494 2014-01-30 Nick Clifton <nickc@redhat.com>
1496 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1497 Rename to mn10300_store_multiple_regs.
1498 * config/mn10300/mn10300.c: Likewise.
1499 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1500 store_multiple_regs.
1501 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1502 Call mn10300_store_multiple_regs.
1504 2014-01-30 Nick Clifton <nickc@redhat.com>
1505 DJ Delorie <dj@redhat.com>
1507 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1508 %fp 2 to keep registers after it properly word-aligned.
1509 (rl78_alloc_physical_registers_umul): Handle the case where both
1510 input operands are the same.
1512 2014-01-30 Richard Biener <rguenther@suse.de>
1514 PR tree-optimization/59903
1515 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1518 2014-01-30 Jason Merrill <jason@redhat.com>
1521 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1524 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1526 2014-01-30 Richard Biener <rguenther@suse.de>
1528 PR tree-optimization/59951
1529 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1531 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1534 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1535 SFmode to DFmode case.
1537 2014-01-29 DJ Delorie <dj@redhat.com>
1539 * config/msp430/msp430.opt (-minrt): New.
1540 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1542 (ENDFILE_SPEC): Likewise.
1544 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1546 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1547 (estimate_function_body_sizes): Use it.
1549 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1552 * dwarf2out.c (is_cxx_auto): New.
1553 (is_base_type): Use it.
1554 (gen_type_die_with_usage): Likewise.
1556 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1558 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1559 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1560 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1561 -maltivec=be with LE targets.
1562 (vsx_mergeh_<mode>): Likewise.
1563 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1564 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1565 (altivec_vmrghb): Replace with define_expand and new
1566 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1567 (altivec_vmrghb_direct): New define_insn.
1568 (altivec_vmrghh): Replace with define_expand and new
1569 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1570 (altivec_vmrghh_direct): New define_insn.
1571 (altivec_vmrghw): Replace with define_expand and new
1572 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1573 (altivec_vmrghw_direct): New define_insn.
1574 (*altivec_vmrghsf): Adjust for endianness.
1575 (altivec_vmrglb): Replace with define_expand and new
1576 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1577 (altivec_vmrglb_direct): New define_insn.
1578 (altivec_vmrglh): Replace with define_expand and new
1579 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1580 (altivec_vmrglh_direct): New define_insn.
1581 (altivec_vmrglw): Replace with define_expand and new
1582 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1583 (altivec_vmrglw_direct): New define_insn.
1584 (*altivec_vmrglsf): Adjust for endianness.
1585 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1586 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1587 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1588 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1589 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1590 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1591 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1592 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1594 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1596 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1597 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1600 2014-01-29 Richard Biener <rguenther@suse.de>
1602 PR tree-optimization/58742
1603 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1604 associate_pointerplus_align.
1605 (associate_pointerplus_diff): New function.
1606 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1607 and associate_pointerplus_diff.
1609 2014-01-29 Richard Biener <rguenther@suse.de>
1611 * lto-streamer.h (LTO_major_version): Bump to 3.
1612 (LTO_minor_version): Reset to 0.
1614 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1616 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1617 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1618 (arm_file_start): Generate correct asm header for armv7ve.
1619 * config/arm/bpabi.h: Add multilib support for armv7ve.
1620 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1621 and cortex-a15 to armv7ve.
1622 * config/arm/t-aprofile: Add multilib support for armv7ve.
1623 * doc/invoke.texi: Document -march=armv7ve.
1625 2014-01-29 Richard Biener <rguenther@suse.de>
1627 PR tree-optimization/58742
1628 * tree-ssa-forwprop.c (associate_plusminus): Return true
1629 if we changed sth, defer EH cleanup to ...
1630 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1631 (simplify_mult): New function.
1633 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1636 PR tree-optimization/59920
1637 * tree.c (build_common_builtin_nodes): Remove
1638 __builtin_setjmp_dispatcher initialization.
1639 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1640 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1641 instead of gsi_after_labels + manually skipping debug stmts.
1642 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1643 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1644 * tree-inline.c (copy_edges_for_bb): Remove
1645 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1646 argument. Ignore computed_goto_p stmts. Don't call
1647 make_abnormal_goto_edges. If a call might need abnormal edges
1648 for non-local gotos, see if it already has an edge to
1649 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1650 with true argument, don't do anything then, otherwise add
1651 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1652 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1654 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1655 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1656 (lower_stmt): Don't set data->calls_builtin_setjmp.
1657 (lower_builtin_setjmp): Adjust comment.
1658 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1659 * tree-cfg.c (found_computed_goto): Remove.
1660 (factor_computed_gotos): Remove.
1661 (make_goto_expr_edges): Return bool, true for computed gotos.
1662 Don't call make_abnormal_goto_edges.
1663 (build_gimple_cfg): Don't set found_computed_goto, don't call
1664 factor_computed_gotos.
1665 (computed_goto_p): No longer static.
1666 (make_blocks): Don't set found_computed_goto.
1667 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1668 (make_edges): If make_goto_expr_edges returns true, push bb
1669 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1670 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1671 vector. Record mapping between bbs and OpenMP regions if there
1672 are any, adjust make_gimple_omp_edges caller. Call
1673 handle_abnormal_edges.
1674 (make_abnormal_goto_edges): Remove.
1675 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1676 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1677 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1678 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1679 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1680 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1681 filling *region also set *region_idx to (*region)->entry->index.
1684 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1685 For REGs set ORIGINAL_REGNO.
1687 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1689 * doc/md.texi: Mention that a target shouldn't implement
1690 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1693 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1695 PR tree-optimization/59594
1696 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1697 a copy of the datarefs vector rather than the vector itself.
1699 2014-01-28 Jason Merrill <jason@redhat.com>
1702 * dwarf2out.c (auto_die): New static.
1703 (gen_type_die_with_usage): Handle C++1y 'auto'.
1704 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1707 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1710 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1711 (SPEC_X32): Likewise.
1712 (SPEC_64): Likewise.
1713 * config/i386/i386.c (ix86_option_override_internal): Turn off
1714 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1716 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1717 * config/i386/i386.h (TARGET_16BIT): New macro.
1718 (TARGET_16BIT_P): Likewise.
1719 * config/i386/i386.opt: Add m16.
1720 * doc/invoke.texi: Document -m16.
1722 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1724 PR preprocessor/59935
1725 * input.c (location_get_source_line): Bail out on when line number
1726 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1728 2014-01-28 Richard Biener <rguenther@suse.de>
1730 PR tree-optimization/58742
1731 * tree-ssa-forwprop.c (associate_plusminus): Handle
1732 pointer subtraction of the form (T)(P + A) - (T)P.
1734 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1736 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1739 2014-01-28 Richard Biener <rguenther@suse.de>
1742 2014-01-28 Richard Biener <rguenther@suse.de>
1744 PR rtl-optimization/45364
1745 PR rtl-optimization/59890
1746 * var-tracking.c (local_get_addr_clear_given_value): Handle
1747 already cleared slot.
1748 (val_reset): Handle not allocated local_get_addr_cache.
1749 (vt_find_locations): Use post-order on the inverted CFG.
1751 2014-01-28 Richard Biener <rguenther@suse.de>
1753 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1755 2014-01-28 Richard Biener <rguenther@suse.de>
1757 PR rtl-optimization/45364
1758 PR rtl-optimization/59890
1759 * var-tracking.c (local_get_addr_clear_given_value): Handle
1760 already cleared slot.
1761 (val_reset): Handle not allocated local_get_addr_cache.
1762 (vt_find_locations): Use post-order on the inverted CFG.
1764 2014-01-28 Alan Modra <amodra@gmail.com>
1766 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1767 * configure.ac <recursive call for build != host>: Define
1768 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1769 and LD_FOR_BUILD too.
1770 * configure: Regenerate.
1772 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1774 * config/i386/i386.c (get_builtin_code_for_version): Separate
1775 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1776 Broadwell from Haswell.
1778 2014-01-27 Steve Ellcey <sellcey@mips.com>
1780 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1781 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1782 * config/mips/mips.c (mips_option_override): Change setting
1784 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1785 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1786 Change from Mask to Var.
1788 2014-01-27 Jeff Law <law@redhat.com>
1790 * ipa-inline.c (inline_small_functions): Fix typo.
1792 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1794 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1795 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1796 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1797 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1798 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1799 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1800 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1801 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1802 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1803 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1804 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1805 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1806 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1807 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1808 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1809 (_mm512_storeu_epi64): Ditto.
1810 (_mm512_cmpge_epi32_mask): Ditto.
1811 (_mm512_cmpge_epu32_mask): Ditto.
1812 (_mm512_cmpge_epi64_mask): Ditto.
1813 (_mm512_cmpge_epu64_mask): Ditto.
1814 (_mm512_cmple_epi32_mask): Ditto.
1815 (_mm512_cmple_epu32_mask): Ditto.
1816 (_mm512_cmple_epi64_mask): Ditto.
1817 (_mm512_cmple_epu64_mask): Ditto.
1818 (_mm512_cmplt_epi32_mask): Ditto.
1819 (_mm512_cmplt_epu32_mask): Ditto.
1820 (_mm512_cmplt_epi64_mask): Ditto.
1821 (_mm512_cmplt_epu64_mask): Ditto.
1822 (_mm512_cmpneq_epi32_mask): Ditto.
1823 (_mm512_cmpneq_epu32_mask): Ditto.
1824 (_mm512_cmpneq_epi64_mask): Ditto.
1825 (_mm512_cmpneq_epu64_mask): Ditto.
1826 (_mm512_expand_pd): Ditto.
1827 (_mm512_expand_ps): Ditto.
1828 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1829 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1830 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1831 * config/i386/i386.c (ix86_builtins): Add
1832 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1833 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1834 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1835 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1836 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1837 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1838 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1839 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1840 IX86_BUILTIN_PMOVUSQW512_MEM.
1841 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1842 __builtin_ia32_pmovsqd512mem_mask,
1843 __builtin_ia32_pmovqd512mem_mask,
1844 __builtin_ia32_pmovusqw512mem_mask,
1845 __builtin_ia32_pmovsqw512mem_mask,
1846 __builtin_ia32_pmovqw512mem_mask,
1847 __builtin_ia32_pmovusdw512mem_mask,
1848 __builtin_ia32_pmovsdw512mem_mask,
1849 __builtin_ia32_pmovdw512mem_mask,
1850 __builtin_ia32_pmovqb512mem_mask,
1851 __builtin_ia32_pmovusqb512mem_mask,
1852 __builtin_ia32_pmovsqb512mem_mask,
1853 __builtin_ia32_pmovusdb512mem_mask,
1854 __builtin_ia32_pmovsdb512mem_mask,
1855 __builtin_ia32_pmovdb512mem_mask.
1856 (bdesc_args): Add __builtin_ia32_expanddf512,
1857 __builtin_ia32_expandsf512.
1858 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1859 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1860 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1861 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1862 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1863 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1864 (avx512f_<code>v8div16qi2_mask_store): This.
1865 (avx512f_expand<mode>): New.
1867 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1869 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1871 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1872 (_mm512_prefetch_i32scatter_pd): Ditto.
1873 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1874 (_mm512_prefetch_i64scatter_pd): Ditto.
1875 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1876 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1877 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1878 (_mm512_prefetch_i32scatter_ps): Ditto.
1879 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1880 (_mm512_prefetch_i64scatter_ps): Ditto.
1881 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1882 * config/i386/i386-builtin-types.def: Define
1883 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1884 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1885 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1886 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1887 IX86_BUILTIN_SCATTERPFQPD.
1888 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1889 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1890 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1891 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1892 __builtin_ia32_scatterpfqps.
1893 (ix86_expand_builtin): Expand new built-ins.
1894 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1895 fix memory access data type.
1896 (*avx512pf_gatherpf<mode>_mask): Ditto.
1897 (*avx512pf_gatherpf<mode>): Ditto.
1898 (avx512pf_scatterpf<mode>): Ditto.
1899 (*avx512pf_scatterpf<mode>_mask): Ditto.
1900 (*avx512pf_scatterpf<mode>): Ditto.
1901 (GATHER_SCATTER_SF_MEM_MODE): New.
1902 (avx512pf_gatherpf<mode>df): Ditto.
1903 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1904 (*avx512pf_scatterpf<mode>df): Ditto.
1906 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1909 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1910 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1913 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1915 * common/config/arm/arm-common.c
1916 (arm_rewrite_mcpu): Handle multiple names.
1918 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1920 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1922 * gimple-builder.h (create_gimple_tmp): Delete.
1924 2014-01-27 Christian Bruel <christian.bruel@st.com>
1926 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1929 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1931 * config/pa/pa.md (call): Generate indirect long calls to non-local
1932 functions when outputing 32-bit code.
1933 (call_value): Likewise except for special call to buggy powf function.
1935 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1936 portable runtime and PIC indirect calls.
1937 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1938 and PIC call sequences. Use ldo instead of blr to set return register
1939 in PIC call sequence.
1941 2014-01-25 Walter Lee <walt@tilera.com>
1943 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1944 avoid clobbering a live register.
1946 2014-01-25 Walter Lee <walt@tilera.com>
1948 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1949 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1950 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1951 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1953 2014-01-25 Walter Lee <walt@tilera.com>
1955 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1956 arguments on even registers.
1957 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1959 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1960 (BIGGEST_ALIGNMENT): Ditto.
1961 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1963 2014-01-25 Walter Lee <walt@tilera.com>
1965 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1966 insns before bundling.
1967 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1969 2014-01-25 Walter Lee <walt@tilera.com>
1971 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1972 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1973 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1975 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1977 * config/mips/constraints.md (kl): Delete.
1978 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1979 define expands, using...
1980 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1981 instructions for MIPS16.
1982 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1983 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1985 2014-01-25 Walter Lee <walt@tilera.com>
1987 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1991 2014-01-25 Walter Lee <walt@tilera.com>
1993 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1994 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1996 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1998 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2001 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2003 * print-rtl.c (in_call_function_usage): New var.
2004 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2005 EXPR_LIST mode as mode and not as reg note name.
2008 * cfgloopmanip.c (copy_loop_info): If
2009 loop->warned_aggressive_loop_optimizations, make sure
2010 the flag is set in target loop too.
2012 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2014 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2016 * builtins.def: Likewise.
2017 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2018 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2019 * ira.c (ira_setup_eliminable_regset): Likewise.
2020 * omp-low.c (gate_expand_omp): Likewise.
2021 (execute_lower_omp): Likewise.
2022 (diagnose_sb_0): Likewise.
2023 (gate_diagnose_omp_blocks): Likewise.
2024 (simd_clone_clauses_extract): Likewise.
2027 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2029 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2030 correction for little endian...
2031 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2034 2014-01-24 Jeff Law <law@redhat.com>
2036 PR tree-optimization/59919
2037 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2038 for non-returning calls.
2040 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2042 * common/config/aarch64/aarch64-common.c
2043 (aarch64_rewrite_mcpu): Handle multiple names.
2044 * config/aarch64/aarch64.h
2045 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2047 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2049 * input.c (add_file_to_cache_tab): Handle the case where fopen
2052 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2055 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2056 from push operand if code of push isn't PRE_DEC.
2058 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2061 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2062 -mquad-memory-atomic. Update -mquad-memory documentation to say
2063 it is only used for non-atomic loads/stores.
2065 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2066 -mquad-memory or -mquad-memory-atomic switches.
2068 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2069 -mquad-memory-atomic to ISA 2.07 support.
2071 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2072 to separate support of normal quad word memory operations (ldq, stq)
2073 from the atomic quad word memory operations.
2075 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2076 support to separate non-atomic quad word operations from atomic
2077 quad word operations. Disable non-atomic quad word operations in
2078 little endian mode so that we don't have to swap words after the
2079 load and before the store.
2080 (quad_load_store_p): Add comment about atomic quad word support.
2081 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2082 options printed with -mdebug=reg.
2084 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2085 -mquad-memory-atomic as the test for whether we have quad word
2086 atomic instructions.
2087 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2088 or -mp8-vector are used, allow byte/half-word atomic operations.
2090 * config/rs6000/sync.md (load_lockedti): Insure that the address
2091 is a proper indexed or indirect address for the lqarx instruction.
2092 On little endian systems, swap the hi/lo registers after the lqarx
2094 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2095 insure the address is valid for the lqarx instruction.
2096 (store_conditionalti): Insure that the address is a proper indexed
2097 or indirect address for the stqcrx. instruction. On little endian
2098 systems, swap the hi/lo registers before doing the stqcrx.
2100 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2101 insure the address is valid for the stqcrx. instruction.
2103 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2104 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2105 type of quad memory support is available.
2107 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2110 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2111 there is a danger of looping.
2113 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2115 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2116 force flag_ira_loop_pressure if set via command line.
2118 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2120 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2121 (ashr_simd): New builtin handling DI mode.
2122 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2123 (aarch64_sshr_simddi): New match pattern.
2124 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2125 (vshrd_n_s64): Likewise.
2126 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2128 2014-01-23 Nick Clifton <nickc@redhat.com>
2130 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2131 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2132 favour of mcu specific scripts.
2133 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2136 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2137 Alex Velenko <Alex.Velenko@arm.com>
2139 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2140 (vaddv_s16): Likewise.
2141 (vaddv_s32): Likewise.
2142 (vaddv_u8): Likewise.
2143 (vaddv_u16): Likewise.
2144 (vaddv_u32): Likewise.
2145 (vaddvq_s8): Likewise.
2146 (vaddvq_s16): Likewise.
2147 (vaddvq_s32): Likewise.
2148 (vaddvq_s64): Likewise.
2149 (vaddvq_u8): Likewise.
2150 (vaddvq_u16): Likewise.
2151 (vaddvq_u32): Likewise.
2152 (vaddvq_u64): Likewise.
2153 (vaddv_f32): Likewise.
2154 (vaddvq_f32): Likewise.
2155 (vaddvq_f64): Likewise.
2156 (vmaxv_f32): Likewise.
2157 (vmaxv_s8): Likewise.
2158 (vmaxv_s16): Likewise.
2159 (vmaxv_s32): Likewise.
2160 (vmaxv_u8): Likewise.
2161 (vmaxv_u16): Likewise.
2162 (vmaxv_u32): Likewise.
2163 (vmaxvq_f32): Likewise.
2164 (vmaxvq_f64): Likewise.
2165 (vmaxvq_s8): Likewise.
2166 (vmaxvq_s16): Likewise.
2167 (vmaxvq_s32): Likewise.
2168 (vmaxvq_u8): Likewise.
2169 (vmaxvq_u16): Likewise.
2170 (vmaxvq_u32): Likewise.
2171 (vmaxnmv_f32): Likewise.
2172 (vmaxnmvq_f32): Likewise.
2173 (vmaxnmvq_f64): Likewise.
2174 (vminv_f32): Likewise.
2175 (vminv_s8): Likewise.
2176 (vminv_s16): Likewise.
2177 (vminv_s32): Likewise.
2178 (vminv_u8): Likewise.
2179 (vminv_u16): Likewise.
2180 (vminv_u32): Likewise.
2181 (vminvq_f32): Likewise.
2182 (vminvq_f64): Likewise.
2183 (vminvq_s8): Likewise.
2184 (vminvq_s16): Likewise.
2185 (vminvq_s32): Likewise.
2186 (vminvq_u8): Likewise.
2187 (vminvq_u16): Likewise.
2188 (vminvq_u32): Likewise.
2189 (vminnmv_f32): Likewise.
2190 (vminnmvq_f32): Likewise.
2191 (vminnmvq_f64): Likewise.
2193 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2195 * config/aarch64/aarch64-simd.md
2196 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2197 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2198 (*aarch64_mul3_elt<mode>): Likewise.
2199 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2200 (*aarch64_mul3_elt_to_64v2df): Likewise.
2201 (*aarch64_mla_elt<mode>): Likewise.
2202 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2203 (*aarch64_mls_elt<mode>): Likewise.
2204 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2205 (*aarch64_fma4_elt<mode>): Likewise.
2206 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2207 (*aarch64_fma4_elt_to_64v2df): Likewise.
2208 (*aarch64_fnma4_elt<mode>): Likewise.
2209 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2210 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2211 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2212 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2213 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2214 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2215 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2216 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2217 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2219 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2221 * config/aarch64/aarch64-simd.md
2222 (aarch64_be_checked_get_lane<mode>): New define_expand.
2223 * config/aarch64/aarch64-simd-builtins.def
2224 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2225 New builtin definition.
2226 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2227 Use new safe be builtin.
2229 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2231 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2233 (aarch64_be_st1<mode>): Likewise.
2234 (aarch_ld1<VALL:mode>): Define_expand modified.
2235 (aarch_st1<VALL:mode>): Likewise.
2236 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2237 (UNSPEC_ST1): Likewise.
2239 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2241 * config/microblaze/microblaze.md: Add trap insn and attribute
2243 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2245 PR preprocessor/58580
2246 * input.h (location_get_source_line): Take an additional line_size
2248 (void diagnostics_file_cache_fini): Declare new function.
2249 * input.c (struct fcache): New type.
2250 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2251 New static constants.
2252 (diagnostic_file_cache_init, total_lines_num)
2253 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2254 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2255 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2256 (get_next_line, read_next_line, goto_next_line, read_line_num):
2257 New static function definitions.
2258 (diagnostic_file_cache_fini): New function.
2259 (location_get_source_line): Take an additional output line_len
2260 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2262 * diagnostic.c (diagnostic_finish): Call
2263 diagnostic_file_cache_fini.
2264 (adjust_line): Take an additional input parameter for the length
2265 of the line, rather than calculating it with strlen.
2266 (diagnostic_show_locus): Adjust the use of
2267 location_get_source_line and adjust_line with respect to their new
2268 signature. While displaying a line now, do not stop at the first
2269 null byte. Rather, display the zero byte as a space and keep
2270 going until we reach the size of the line.
2271 * Makefile.in: Add vec.o to OBJS-libcommon
2273 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2274 Ilya Tocar <ilya.tocar@intel.com>
2276 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2277 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2278 (__builtin_ia32_kmov16): Ditto.
2279 * config/i386/i386.md (UNSPEC_KMOV): New.
2282 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2284 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2285 (_mm512_storeu_si512): Ditto.
2287 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2290 * rtl.h (get_referenced_operands): Declare.
2291 * recog.c (get_referenced_operands): New function.
2292 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2293 operands have been referenced when recording LO_SUM references.
2295 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2297 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2299 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2301 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2302 Enable for generic and recent AMD targets.
2304 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2306 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2307 ARG_SIZE note when adjustment was eliminated.
2309 2014-01-22 Jeff Law <law@redhat.com>
2311 PR tree-optimization/59597
2312 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2313 in file. Accept new argument REGISTERING and use it to modify
2314 dump output appropriately.
2315 (register_jump_thread): Corresponding changes.
2316 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2317 thread paths involving joiner blocks. Add code to dump cancelled
2318 jump threading paths.
2320 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2322 PR rtl-optimization/59477
2323 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2324 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2326 2014-01-22 Tom Tromey <tromey@redhat.com>
2328 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2330 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2332 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2334 PR rtl-optimization/59896
2335 * lra-constraints.c (process_alt_operands): Check unused note for
2336 matched operands of insn with no output reloads.
2338 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2340 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2341 (mips_move_from_gpr_cost): Likewise.
2343 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2345 PR rtl-optimization/59858
2346 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2347 ira_class_hard_regs_num.
2348 (process_alt_operands): Increase reject for dying matched operand.
2350 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2353 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2354 smaller than size, perform several stores or loads and stores
2355 at dst + count - size to store or copy all of size bytes, rather
2356 than just last modesize bytes.
2358 2014-01-20 DJ Delorie <dj@redhat.com>
2360 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2361 that CLOBBERs are REGs before propogating their values.
2363 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2366 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2367 (cgraph_inline_failed_type): New function.
2368 * cgraph.h (DEFCIFCODE): Add type.
2369 (cgraph_inline_failed_type_t): New enum.
2370 (cgraph_inline_failed_type): New prototype.
2371 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2372 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2373 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2374 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2375 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2376 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2377 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2378 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2379 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2380 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2381 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2382 OPTIMIZATION_MISMATCH.
2383 * tree-inline.c (expand_call_inline): Emit errors during
2384 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2386 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2389 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2390 mode attribute in insn output.
2392 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2394 * output.h (output_constant): Delete.
2395 * varasm.c (output_constant): Make private.
2397 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2399 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2401 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2404 * tree.h (fold_builtin_strcat): New prototype.
2405 * builtins.c (fold_builtin_strcat): No longer static. Add len
2406 argument, if non-NULL, don't call c_strlen. Optimize
2407 directly into __builtin_memcpy instead of __builtin_strcpy.
2408 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2409 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2411 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2413 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2414 for SImode_address_operand operands, having only a REG argument.
2416 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2418 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2419 loader name using mbig-endian.
2420 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2422 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2424 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2428 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2430 * config/aarch64/aarch64-protos.h
2431 (aarch64_cannot_change_mode_class_ptr): Declare.
2432 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2433 aarch64_cannot_change_mode_class_ptr): New.
2434 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2435 backend hook aarch64_cannot_change_mode_class.
2437 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2439 * common/config/aarch64/aarch64-common.c
2440 (aarch64_handle_option): Don't handle any option order logic here.
2441 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2442 selected_cpu, warn on architecture version mismatch.
2443 (aarch64_override_options): Fix parsing order for option strings.
2445 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2446 Iain Sandoe <iain@codesourcery.com>
2449 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2450 warning. Amend comment to reflect current functionality.
2452 2014-01-20 Richard Biener <rguenther@suse.de>
2455 * builtins.c (fold_builtin_strcat): Remove case better handled
2456 by tree-ssa-strlen.c.
2458 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2460 * config/aarch64/aarch64.opt
2461 (mcpu, march, mtune): Make case-insensitive.
2463 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2466 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2467 if operands[1] is a REG or ZERO_EXTEND of a REG.
2469 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2471 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2473 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2475 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2476 long non-pic millicode calls.
2478 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2480 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2482 2014-01-19 Kito Cheng <kito@0xlab.org>
2484 * builtins.c (expand_movstr): Check movstr expand done or fail.
2486 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2487 H.J. Lu <hongjiu.lu@intel.com>
2490 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2491 to DImode for zero-extended addresses.
2493 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2495 PR rtl-optimization/57763
2496 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2497 on the new indirect jump_insn and increment LABEL_NUSES (label).
2499 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2503 * config.gcc (x86_archs): New variable.
2504 (x86_64_archs): Likewise.
2505 (x86_cpus): Likewise.
2506 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2507 --with-arch/--with-cpu= options.
2508 Support --with-arch=/--with-cpu={nehalem,westmere,
2509 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2511 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2513 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2514 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2516 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2518 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2520 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2523 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2524 clear cpp_get_options (parse_in)->warn_unused_macros for
2525 ix86_target_macros_internal with cpp_define.
2527 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2529 * jump.c (delete_related_insns): Keep (use (insn))s.
2530 * reorg.c (redundant_insn): Check for barriers too.
2532 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2534 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2536 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2538 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2539 call to $$dyncall when TARGET_LONG_CALLS is true.
2541 2014-01-17 Jeff Law <law@redhat.com>
2543 * ree.c (combine_set_extension): Temporarily disable test for
2544 changing number of hard registers.
2546 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2549 * ipa-inline-analysis.c (inline_free_summary):
2550 Do not free summary of aliases.
2552 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2555 * gimplify.c (gimplify_expr): Use create_tmp_var
2556 instead of create_tmp_var_raw. If cond doesn't have
2557 integral type, don't add the IFN_ANNOTATE builtin at all.
2559 2014-01-17 Martin Jambor <mjambor@suse.cz>
2562 * ipa-cp.c (prev_edge_clone): New variable.
2563 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2564 Also resize prev_edge_clone vector.
2565 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2566 (ipcp_edge_removal_hook): New function.
2567 (ipcp_driver): Register ipcp_edge_removal_hook.
2569 2014-01-17 Andrew Pinski <apinski@cavium.com>
2570 Steve Ellcey <sellcey@mips.com>
2573 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2576 2014-01-17 Jeff Law <law@redhat.com>
2579 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2580 so that pass_ccp runs first.
2582 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2584 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2585 (ix86_adjust_cost): Use !TARGET_XXX.
2586 (do_reorder_for_imul): Likewise.
2587 (swap_top_of_ready_list): Likewise.
2588 (ix86_sched_reorder): Likewise.
2590 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2592 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2593 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2594 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2595 (intel_memset): New. Duplicate slm_memset.
2596 (intel_cost): New. Duplicate slm_cost.
2597 (m_INTEL): New macro.
2598 (processor_target_table): Add "intel".
2599 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2600 with PROCESSOR_INTEL for "intel".
2601 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2602 PROCESSOR_SILVERMONT.
2603 (ix86_issue_rate): Likewise.
2604 (ix86_adjust_cost): Likewise.
2605 (ia32_multipass_dfa_lookahead): Likewise.
2606 (swap_top_of_ready_list): Likewise.
2607 (ix86_sched_reorder): Likewise.
2608 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2609 instead of TARGET_OPT_AGU.
2610 * config/i386/i386.h (TARGET_INTEL): New.
2611 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2612 (processor_type): Add PROCESSOR_INTEL.
2613 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2614 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2616 2014-01-17 Marek Polacek <polacek@redhat.com>
2619 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2622 2014-01-17 Richard Biener <rguenther@suse.de>
2624 PR tree-optimization/46590
2625 * opts.c (default_options_table): Add entries for
2626 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2627 all enabled at -O1 but not for -Og.
2628 * common.opt (fbranch-count-reg): Remove Init(1).
2629 (fmove-loop-invariants): Likewise.
2630 (ftree-pta): Likewise.
2632 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2634 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2635 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2636 decls to at least the GCC 4.8 used alignments.
2639 * tree-nested.c (convert_nonlocal_reference_stmt,
2640 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2641 of GIMPLE_BIND stmts, adjust associated decls.
2643 2014-01-17 Richard Biener <rguenther@suse.de>
2645 PR tree-optimization/46590
2646 * vec.h (vec<>::bseach): New member function implementing
2647 binary search according to C89 bsearch.
2648 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2649 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2650 bitmap pointer again. Make accesses_in_loop a flat array.
2651 (mem_ref_obstack): New global.
2652 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2653 (mark_ref_stored): Likewise.
2654 (ref_indep_loop_p_2): Likewise.
2655 (set_ref_stored_in_loop): New helper function.
2656 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2657 (memref_free): Adjust.
2658 (record_mem_ref_loc): Simplify.
2659 (gather_mem_refs_stmt): Adjust.
2660 (sort_locs_in_loop_postorder_cmp): New function.
2661 (analyze_memory_references): Sort accesses_in_loop after
2662 loop postorder number.
2663 (find_ref_loc_in_loop_cmp): New function.
2664 (for_all_locs_in_loop): Find relevant cluster of locs in
2665 accesses_in_loop and iterate without recursion.
2666 (execute_sm): Avoid uninit warning.
2667 (struct ref_always_accessed): Simplify.
2668 (ref_always_accessed::operator ()): Likewise.
2669 (ref_always_accessed_p): Likewise.
2670 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2671 loop postorder numbers here.
2672 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2675 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2678 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2679 on decls for which assemble_alias has been called.
2681 2014-01-17 Nick Clifton <nickc@redhat.com>
2683 * config/msp430/msp430.opt: (mcpu): New option.
2684 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2685 (msp430_option_override): Parse target_cpu. If the MCU name
2686 matches a generic string, clear target_mcu.
2687 (msp430_attr): Allow numeric interrupt values up to 63.
2688 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2689 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2691 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2693 * config/msp430/msp430.md (popm): Use %J rather than %I.
2694 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2695 (addhi_cy_i): Use immediate_operand for operand 2.
2696 * doc/invoke.texi: Document -mcpu option.
2698 2014-01-17 Richard Biener <rguenther@suse.de>
2700 PR rtl-optimization/38518
2701 * df.h (df_analyze_loop): Declare.
2702 * df-core.c: Include cfgloop.h.
2703 (df_analyze_1): Split out main part of df_analyze.
2704 (df_analyze): Adjust.
2705 (loop_inverted_post_order_compute): New function.
2706 (loop_post_order_compute): Likewise.
2707 (df_analyze_loop): New function avoiding whole-function
2709 * loop-invariant.c (find_defs): Use df_analyze_loop.
2710 (find_invariants): Adjust.
2711 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2713 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2715 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2716 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2718 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2720 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2721 traversal when removing references.
2723 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2726 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2728 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2731 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2732 pushing a reload for an autoinc when we had previously reloaded an
2733 inner part of the address.
2735 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2737 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2739 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2740 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2741 when not giving up or versioning for alias only because of
2743 (vect_analyze_data_ref_dependences): Set to true.
2744 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2746 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2747 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2751 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2754 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2755 operand 0 predicate for gathers, use a new pseudo as subtarget.
2757 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2760 * lra-constraints.c (process_alt_operands): Add printing debug info.
2761 Check absence of input/output reloads for matched operands too.
2763 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2765 PR rtl-optimization/59835
2766 * ira.c (ira_init_register_move_cost): Increase cost for
2769 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2771 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2773 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2776 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2777 non-register objects. Use gen_(high/low)part more consistently.
2780 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2783 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2784 endian support, remove tests for WORDS_BIG_ENDIAN.
2785 (p8_mfvsrd_3_<mode>): Likewise.
2786 (reload_gpr_from_vsx<mode>): Likewise.
2787 (reload_gpr_from_vsxsf): Likewise.
2788 (p8_mfvsrd_4_disf): Likewise.
2790 2014-01-16 Richard Biener <rguenther@suse.de>
2792 PR rtl-optimization/46590
2793 * lcm.c (compute_antinout_edge): Use postorder iteration.
2794 (compute_laterin): Use inverted postorder iteration.
2796 2014-01-16 Nick Clifton <nickc@redhat.com>
2799 * varasm.c (output_constant): Return the number of bytes actually
2801 (output_constructor_array_range): Update the field size with the
2802 number of bytes emitted by output_constant.
2803 (output_constructor_regular_field): Likewise. Also do not
2804 complain if the total number of bytes emitted is now greater
2805 than the expected fieldpos.
2806 * output.h (output_constant): Update prototype and descriptive comment.
2808 2014-01-16 Marek Polacek <polacek@redhat.com>
2811 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2812 it is error_mark_node.
2814 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2816 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2817 VALID_AVX256_REG_OR_OI_MODE.
2819 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2821 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2822 current procedure should be profiled.
2824 2014-01-15 Andrew Pinski <apinski@cavium.com>
2826 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2827 of moving from/to the STACK_REG register class.
2829 2014-01-15 Richard Henderson <rth@redhat.com>
2832 * reginfo.c (global_regs_decl): Globalize.
2833 * rtl.h (global_regs_decl): Declare.
2834 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2835 reserved via global_regs.
2837 2014-01-15 Teresa Johnson <tejohnson@google.com>
2839 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2841 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2843 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2844 and vmulosh rather than call gen_vec_widen_smult_*.
2845 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2846 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2847 (vec_widen_smult_even_v16qi): Likewise.
2848 (vec_widen_umult_even_v8hi): Likewise.
2849 (vec_widen_smult_even_v8hi): Likewise.
2850 (vec_widen_umult_odd_v16qi): Likewise.
2851 (vec_widen_smult_odd_v16qi): Likewise.
2852 (vec_widen_umult_odd_v8hi): Likewise.
2853 (vec_widen_smult_odd_v8hi): Likewise.
2854 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2855 vmuloub rather than call gen_vec_widen_umult_*.
2856 (vec_widen_umult_lo_v16qi): Likewise.
2857 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2858 vmulosb rather than call gen_vec_widen_smult_*.
2859 (vec_widen_smult_lo_v16qi): Likewise.
2860 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2861 rather than call gen_vec_widen_umult_*.
2862 (vec_widen_umult_lo_v8hi): Likewise.
2863 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2864 rather than call gen_vec_widen_smult_*.
2865 (vec_widen_smult_lo_v8hi): Likewise.
2867 2014-01-15 Jeff Law <law@redhat.com>
2869 PR tree-optimization/59747
2870 * ree.c (find_and_remove_re): Properly handle case where a second
2871 eliminated extension requires widening a copy created for elimination
2872 of a prior extension.
2873 (combine_set_extension): Ensure that the number of hard regs needed
2874 for a destination register does not change when we widen it.
2876 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2878 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2879 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2880 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2881 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2882 (avr-*-rtems*): Likewise.
2883 (bfin*-rtems*): Likewise.
2884 (moxie-*-rtems*): Likewise.
2885 (h8300-*-rtems*): Likewise.
2886 (i[34567]86-*-rtems*): Likewise.
2887 (lm32-*-rtems*): Likewise.
2888 (m32r-*-rtems*): Likewise.
2889 (m68k-*-rtems*): Likewise.
2890 (microblaze*-*-rtems*): Likewise.
2891 (mips*-*-rtems*): Likewise.
2892 (powerpc-*-rtems*): Likewise.
2893 (sh-*-rtems*): Likewise.
2894 (sparc-*-rtems*): Likewise.
2895 (sparc64-*-rtems*): Likewise.
2896 (v850-*-rtems*): Likewise.
2897 (m32c-*-rtems*): Likewise.
2899 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2901 PR rtl-optimization/59511
2902 * ira.c (ira_init_register_move_cost): Use memory costs for some
2903 cases of register move cost calculations.
2904 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2905 instead of BB frequency.
2906 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2907 * lra-assigns.c (find_hard_regno_for): Ditto.
2909 2014-01-15 Richard Biener <rguenther@suse.de>
2911 PR tree-optimization/59822
2912 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2913 (vectorizable_load): Use it to hoist defs of uses of invariant
2914 loads out of the loop.
2916 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2917 Kugan Vivekanandarajah <kuganv@linaro.org>
2920 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2923 2014-01-15 Richard Biener <rguenther@suse.de>
2925 PR rtl-optimization/59802
2926 * lcm.c (compute_available): Use inverted postorder to seed
2927 the initial worklist.
2929 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2932 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2933 ADDR_REGS for invalid symrefs in non-PIC code.
2935 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2938 * builtins.c (determine_block_size): Initialize *probable_max_size
2939 even if len_rtx is CONST_INT.
2941 2014-01-14 Andrew Pinski <apinski@cavium.com>
2943 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2944 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2945 (cortexa53_tunings): Likewise.
2946 (aarch64_sched_issue_rate): New function.
2947 (TARGET_SCHED_ISSUE_RATE): Define.
2949 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2951 * ira-costs.c (find_costs_and_classes): Add missed
2952 ira_init_register_move_cost_if_necessary.
2954 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2957 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2959 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2962 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2963 to indicate if type is used for function return value. Warn ABI
2964 change if the vector mode isn't available for function return value.
2965 (ix86_function_arg_advance): Pass false to type_natural_mode.
2966 (ix86_function_arg): Likewise.
2967 (ix86_gimplify_va_arg): Likewise.
2968 (function_arg_32): Don't warn ABI change.
2969 (ix86_function_value): Pass true to type_natural_mode.
2970 (ix86_return_in_memory): Likewise.
2971 (ix86_struct_value_rtx): Removed.
2972 (TARGET_STRUCT_VALUE_RTX): Likewise.
2974 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2976 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2977 converting a conditional jump into a conditional return.
2979 2014-01-14 Richard Biener <rguenther@suse.de>
2981 PR tree-optimization/58921
2982 PR tree-optimization/59006
2983 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2984 hoisting invariant stmts.
2985 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2986 invariant loads on the preheader edge if possible.
2988 2014-01-14 Joey Ye <joey.ye@arm.com>
2990 * doc/plugin.texi (Building GCC plugins): Update to C++.
2992 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2994 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2995 (_mm_rcp28_round_ss): Ditto.
2996 (_mm_rsqrt28_round_sd): Ditto.
2997 (_mm_rsqrt28_round_ss): Ditto.
2998 (_mm_rcp28_sd): Ditto.
2999 (_mm_rcp28_ss): Ditto.
3000 (_mm_rsqrt28_sd): Ditto.
3001 (_mm_rsqrt28_ss): Ditto.
3002 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3003 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3004 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3005 (IX86_BUILTIN_RCP28SD): Ditto.
3006 (IX86_BUILTIN_RCP28SS): Ditto.
3007 (IX86_BUILTIN_RSQRT28SD): Ditto.
3008 (IX86_BUILTIN_RSQRT28SS): Ditto.
3009 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3010 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3011 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3012 (ix86_expand_special_args_builtin): Expand new FTYPE.
3013 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3014 (srcp14<mode>): Make insn unary.
3015 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3016 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3017 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3018 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3019 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3020 Fix rounding: make it SAE only.
3021 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3023 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3025 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3026 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3027 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3028 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3029 (round_saeonly_mask_scalar_operand4): Ditto.
3030 (round_saeonly_mask_scalar_op3): Ditto.
3031 (round_saeonly_mask_scalar_op4): Ditto.
3033 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3035 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3036 Implement -maltivec=be for vec_insert and vec_extract.
3038 2014-01-10 DJ Delorie <dj@redhat.com>
3040 * config/msp430/msp430.md (call_internal): Don't allow memory
3041 references with SP as the base register.
3042 (call_value_internal): Likewise.
3043 * config/msp430/constraints.md (Yc): New. For memory references
3044 that don't use SP as a base register.
3046 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3047 "an integer without a # prefix"
3048 * config/msp430/msp430.md (epilogue_helper): Use it.
3050 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3053 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3054 AVX512F gather builtins.
3055 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3056 on gather decls with INTEGER_TYPE masktype.
3057 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3058 directly into the builtin rather than hoisting it before loop.
3060 PR tree-optimization/59387
3061 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3062 (scev_const_prop): If folded_casts and type has undefined overflow,
3063 use force_gimple_operand instead of force_gimple_operand_gsi and
3064 for each added stmt if it is assign with
3065 arith_code_with_undefined_signed_overflow, call
3066 rewrite_to_defined_overflow.
3067 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3068 gimple-fold.h instead.
3069 (arith_code_with_undefined_signed_overflow,
3070 rewrite_to_defined_overflow): Moved to ...
3071 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3072 rewrite_to_defined_overflow): ... here. No longer static.
3073 Include gimplify-me.h.
3074 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3075 rewrite_to_defined_overflow): New prototypes.
3077 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3079 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3081 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3083 * builtins.c (get_object_alignment_2): Minor tweak.
3084 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3086 2014-01-13 Christian Bruel <christian.bruel@st.com>
3088 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3089 optimized non constant lengths.
3091 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3094 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3095 load as __atomic_load_N if possible.
3097 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3099 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3101 (rs6000_expand_builtin): Adjust call.
3103 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3106 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3107 * config/rs6000/rs6000.c: Include target-globals.h.
3108 (rs6000_set_current_function): Instead of doing target_reinit
3109 unconditionally, use save_target_globals_default_opts and
3110 restore_target_globals.
3112 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3114 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3115 (rs6000_expand_builtin): Handle mffs and mtfsf.
3116 (rs6000_init_builtins): Define mffs and mtfsf.
3117 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3118 (rs6000_mffs): New pattern.
3119 (rs6000_mtfsf): New pattern.
3121 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3123 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3124 Start narrowing with START. Apply candidate-use pair
3125 and check overall cost in narrowing.
3126 (iv_ca_prune): Pass new argument.
3128 2014-01-10 Jeff Law <law@redhat.com>
3131 * ree.c (combine_reaching_defs): Ensure the defining statement
3132 occurs before the extension when optimizing extensions with
3133 different source and destination hard registers.
3135 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3138 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3139 vtables into the type inheritance graph.
3141 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3143 PR rtl-optimization/59754
3144 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3145 modes in the REGNO != REGNO case.
3147 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3149 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3151 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3153 PR tree-optimization/59745
3154 * tree-predcom.c (tree_predictive_commoning_loop): Call
3155 free_affine_expand_cache if giving up because components is NULL.
3157 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3158 GC in payload of target_globals struct instead of allocating them on
3159 the heap and the larger structs separately using GC.
3160 * target-globals.h (struct target_globals): Make regs, hard_regs,
3161 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3162 of GTY((skip)) and change type to void *.
3163 (reset_target_globals): Cast loads from those fields to corresponding
3166 2014-01-10 Steve Ellcey <sellcey@mips.com>
3169 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3170 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3171 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3173 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3176 * aarch64-modes.def (CC_Zmode): New flags mode.
3177 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3178 represents an equality.
3179 (aarch64_get_condition_code): Handle CC_Zmode.
3180 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3182 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3184 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3185 extraction in good case.
3187 2014-01-10 Richard Biener <rguenther@suse.de>
3189 PR tree-optimization/59374
3190 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3191 checking after SLP discovery. Mark stmts not participating
3192 in any SLP instance properly.
3194 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3196 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3197 when handling a SET rtx.
3199 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3201 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3202 (cortex-a57): Likewise.
3203 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3205 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3207 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3208 non-iwmmxt builtins.
3210 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3214 * ipa-devirt.c record_target_from_binfo): Take as argument
3215 stack of binfos and lookup matching one for virtual inheritance.
3216 (possible_polymorphic_call_targets_1): Update.
3218 2014-01-10 Huacai Chen <chenhc@lemote.com>
3220 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3221 kernel strings for Loongson-2E/2F/3A.
3223 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3226 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3227 is_gimple_call before calling gimple_call_internal_p.
3229 2014-01-09 Steve Ellcey <sellcey@mips.com>
3231 * Makefile.in (TREE_FLOW_H): Remove.
3232 (TREE_SSA_H): Add file names from tree-flow.h.
3233 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3234 * tree.h: Remove tree-flow.h reference.
3235 * hash-table.h: Remove tree-flow.h reference.
3236 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3237 reference with tree-ssa-loop.h.
3239 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3241 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3242 default element-order behavior for -maltivec.
3243 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3244 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3245 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3246 when targeting big endian, at least for now.
3247 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3249 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3252 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3253 var satisfies use_register_for_decl, just take into account type
3254 alignment, rather than decl alignment.
3256 PR tree-optimization/59622
3257 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3258 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3259 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3260 Don't devirtualize for inplace at all. For targets.length () == 1,
3261 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3263 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3265 * config/i386/i386.md (cpu): Remove the unused btver1.
3267 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3269 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3271 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3274 * tree-core.h (struct target_globals): New forward declaration.
3275 (struct tree_target_option): Add globals field.
3276 * tree.h (TREE_TARGET_GLOBALS): Define.
3277 (prepare_target_option_nodes_for_pch): New prototype.
3278 * target-globals.h (struct target_globals): Define even if
3280 * tree.c (prepare_target_option_node_for_pch,
3281 prepare_target_option_nodes_for_pch): New functions.
3282 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3283 * config/i386/i386.c: Include target-globals.h.
3284 (ix86_set_current_function): Instead of doing target_reinit
3285 unconditionally, use save_target_globals_default_opts and
3286 restore_target_globals.
3288 2014-01-09 Richard Biener <rguenther@suse.de>
3290 PR tree-optimization/59715
3291 * tree-cfg.h (split_critical_edges): Declare.
3292 * tree-cfg.c (split_critical_edges): Export.
3293 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3295 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3297 * cfgexpand.c (expand_stack_vars): Optionally disable
3298 asan stack protection.
3299 (expand_used_vars): Likewise.
3300 (partition_stack_vars): Likewise.
3301 * asan.c (asan_emit_stack_protection): Optionally disable
3302 after return stack usage.
3303 (instrument_derefs): Optionally disable memory access instrumentation.
3304 (instrument_builtin_call): Likewise.
3305 (instrument_strlen_call): Likewise.
3306 (asan_protect_global): Optionally disable global variables protection.
3307 * doc/invoke.texi: Added doc for new options.
3308 * params.def: Added new options.
3309 * params.h: Likewise.
3311 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3313 PR rtl-optimization/59724
3314 * ifcvt.c (cond_exec_process_if_block): Don't call
3315 flow_find_head_matching_sequence with 0 longest_match.
3316 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3317 non-active insns if !stop_after.
3318 (try_head_merge_bb): Revert 2014-01-07 changes.
3320 2014-01-08 Jeff Law <law@redhat.com>
3322 * ree.c (get_sub_rtx): New function, extracted from...
3323 (merge_def_and_ext): Here.
3324 (combine_reaching_defs): Use get_sub_rtx.
3326 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3328 * cgraph.h (varpool_variable_node): Do not choke on null node.
3330 2014-01-08 Catherine Moore <clm@codesourcery.com>
3332 * config/mips/mips.md (simple_return): Attempt to use JRC
3334 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3336 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3338 PR rtl-optimization/59137
3339 * reorg.c (steal_delay_list_from_target): Call update_block for
3341 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3343 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3345 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3346 two duplicate entries.
3348 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3351 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3353 * config/mips/mips.c (mips_truncated_op_cost): New function.
3354 (mips_rtx_costs): Adjust test for BADDU.
3355 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3357 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3359 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3360 (*baddu_si): ...this new pattern.
3362 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3365 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3367 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3370 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3372 (expand_expr, expand_normal): Adjust.
3373 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3374 inner_reference_p. Use inner_reference_p to expand inner references.
3375 (store_expr): Adjust.
3376 * cfgexpand.c (expand_call_stmt): Adjust.
3378 2014-01-08 Rong Xu <xur@google.com>
3380 * gcov-io.c (gcov_var): Move from gcov-io.h.
3381 (gcov_position): Ditto.
3382 (gcov_is_error): Ditto.
3383 (gcov_rewrite): Ditto.
3384 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3385 only part to libgcc/libgcov.h.
3387 2014-01-08 Marek Polacek <polacek@redhat.com>
3390 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3392 2014-01-08 Marek Polacek <polacek@redhat.com>
3395 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3397 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3399 PR rtl-optimization/59649
3400 * stor-layout.c (get_mode_bounds): For BImode return
3401 0 and STORE_FLAG_VALUE.
3403 2014-01-08 Richard Biener <rguenther@suse.de>
3406 * gimple.h (is_gimple_builtin_call): Remove.
3407 (gimple_builtin_call_types_compatible_p): New.
3408 (gimple_call_builtin_p): New overload.
3409 * gimple.c (is_gimple_builtin_call): Remove.
3410 (validate_call): Rename to ...
3411 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3413 (validate_type): New static function.
3414 (gimple_call_builtin_p): New overload and adjust.
3415 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3416 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3417 (gimple_fold_stmt_to_constant_1): Likewise.
3418 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3420 2014-01-08 Richard Biener <rguenther@suse.de>
3423 * gimplify.c (gimplify_expr): Gimplify register-register type
3424 VIEW_CONVERT_EXPRs to separate stmts.
3426 2014-01-07 Jeff Law <law@redhat.com>
3429 * ree.c (combine_set_extension): Handle case where source
3430 and destination registers in an extension insn are different.
3431 (combine_reaching_defs): Allow source and destination registers
3432 in extension to be different under limited circumstances.
3433 (add_removable_extension): Remove restriction that the
3434 source and destination registers in the extension are the same.
3435 (find_and_remove_re): Emit a copy from the extension's
3436 destination to its source after the defining insn if
3437 the source and destination registers are different.
3440 * ifcvt.c (merge_if_block): If we are merging a block with more than
3441 one successor with a block with no successors, remove any BARRIER
3442 after the second block.
3444 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3446 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3448 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3451 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3452 for 14-bit register offsets when INT14_OK_STRICT is false.
3454 2014-01-07 Roland Stigge <stigge@antcom.de>
3455 Michael Meissner <meissner@linux.vnet.ibm.com>
3458 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3459 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3461 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3463 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3466 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3468 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3469 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3470 rtx is const0_rtx or not.
3472 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3475 * target-globals.c (save_target_globals): Remove this_fn_optab
3477 * toplev.c: Include optabs.h.
3478 (target_reinit): Temporarily restore the global options if another
3479 set of options are in force.
3481 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3483 PR rtl-optimization/58668
3484 * cfgcleanup.c (flow_find_cross_jump): Don't count
3485 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3486 to determine what is counted.
3487 (flow_find_head_matching_sequence): Use active_insn_p to determine
3489 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3491 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3492 determine what is counted.
3494 PR tree-optimization/59643
3495 * tree-predcom.c (split_data_refs_to_components): If one dr is
3496 read and one write, determine_offset fails and the write isn't
3497 in the bad component, just put the read into the bad component.
3499 2014-01-07 Mike Stump <mikestump@comcast.net>
3500 Jakub Jelinek <jakub@redhat.com>
3503 * tree-core.h (struct tree_optimization_option): Change optabs
3504 type from unsigned char * to void *.
3505 * optabs.c (init_tree_optimization_optabs): Adjust
3506 TREE_OPTIMIZATION_OPTABS initialization.
3508 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3511 * config/i386/i386.h (struct machine_function): Add
3512 no_drap_save_restore field.
3513 * config/i386/i386.c (ix86_save_reg): Use
3514 !cfun->machine->no_drap_save_restore instead of
3515 crtl->stack_realign_needed.
3516 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3517 this function clears frame_pointer_needed. Set
3518 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3519 and DRAP reg is needed.
3521 2014-01-06 Marek Polacek <polacek@redhat.com>
3524 * doc/implement-c.texi: Mention that other integer types are
3525 permitted as bit-field types in strictly conforming mode.
3527 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3529 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3532 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3534 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3536 2014-01-06 Martin Jambor <mjambor@suse.cz>
3539 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3541 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3543 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3547 * var-tracking.c (add_stores): Preserve the value of the source even if
3548 we don't record the store.
3550 2014-01-06 Terry Guo <terry.guo@arm.com>
3552 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3554 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3557 * config/darwin.c (darwin_function_section): Adjust return values to
3558 correspond to optimisation changes made in r206070.
3560 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3562 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3563 from prefetch_block tune setting.
3564 (nocona_cost): Correct size of prefetch block to 64.
3566 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3568 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3569 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3570 used to save the static chain register in the computation of the offset
3571 from which the FP registers need to be restored.
3573 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3575 PR tree-optimization/59519
3576 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3577 ICE if get_current_def (current_new_name) is already non-NULL, as long
3578 as it is a phi result of some other phi in *new_exit_bb that has
3581 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3582 or vmovdqu* for misaligned_operand.
3583 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3584 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3585 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3586 aligned_mem for AVX512F masked aligned load and store builtins and for
3589 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3591 PR tree-optimization/59651
3592 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3593 Address range for negative step should be added by TYPE_SIZE_UNIT.
3595 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3597 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3598 overlapping registers also for registers other than the stack pointer.
3600 2014-01-03 Marek Polacek <polacek@redhat.com>
3603 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3606 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3609 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3612 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3613 (push splitter): Use <P:MODE_SIZE> instead of
3614 GET_MODE_SIZE (<P:MODE>mode).
3615 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3616 (mov -1, reg peephole2): Likewise.
3617 * config/i386/sse.md (*mov<mode>_internal,
3618 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3619 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3620 *<code><mode>3, *andnot<mode>3<mask_name>,
3621 <mask_codefor><code><mode>3<mask_name>): Likewise.
3622 * config/i386/subst.md (mask_mode512bit_condition,
3623 sd_mask_mode512bit_condition): Likewise.
3625 2014-01-02 Xinliang David Li <davidxl@google.com>
3627 PR tree-optimization/59303
3628 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3629 (dump_predicates): Better output format.
3630 (pred_equal_p): New function.
3631 (is_neq_relop_p): Ditto.
3632 (is_neq_zero_form_p): Ditto.
3633 (pred_expr_equal_p): Ditto.
3634 (pred_neg_p): Ditto.
3635 (simplify_pred): Ditto.
3636 (simplify_preds_2): Ditto.
3637 (simplify_preds_3): Ditto.
3638 (simplify_preds_4): Ditto.
3639 (simplify_preds): Ditto.
3641 (push_to_worklist): Ditto.
3642 (get_pred_info_from_cmp): Ditto.
3643 (is_degenerated_phi): Ditto.
3644 (normalize_one_pred_1): Ditto.
3645 (normalize_one_pred): Ditto.
3646 (normalize_one_pred_chain): Ditto.
3647 (normalize_preds): Ditto.
3648 (normalize_cond_1): Remove function.
3649 (normalize_cond): Ditto.
3650 (is_gcond_subset_of): Ditto.
3651 (is_subset_of_any): Ditto.
3652 (is_or_set_subset_of): Ditto.
3653 (is_and_set_subset_of): Ditto.
3654 (is_norm_cond_subset_of): Ditto.
3655 (pred_chain_length_cmp): Ditto.
3656 (convert_control_dep_chain_into_preds): Type change.
3657 (find_predicates): Ditto.
3658 (find_def_preds): Ditto.
3659 (destroy_predicates_vecs): Ditto.
3660 (find_matching_predicates_in_rest_chains): Ditto.
3661 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3662 (is_pred_expr_subset): Ditto.
3663 (is_pred_chain_subset_of): Ditto.
3664 (is_included_in): Ditto.
3665 (is_superset_of): Ditto.
3667 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3669 Update copyright years.
3671 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3673 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3674 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3675 config/arc/arc.md, config/arc/arc.opt,
3676 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3677 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3678 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3679 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3680 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3681 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3682 vtable-verify.c, vtable-verify.h: Use the standard form for the
3685 2014-01-02 Tobias Burnus <burnus@net-b.de>
3687 * gcc.c (process_command): Update copyright notice dates.
3688 * gcov-dump.c: Ditto.
3690 * doc/cpp.texi: Bump @copying's copyright year.
3691 * doc/cppinternals.texi: Ditto.
3692 * doc/gcc.texi: Ditto.
3693 * doc/gccint.texi: Ditto.
3694 * doc/gcov.texi: Ditto.
3695 * doc/install.texi: Ditto.
3696 * doc/invoke.texi: Ditto.
3698 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3700 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3702 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3704 * config/i386/sse.md (*mov<mode>_internal): Guard
3705 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3707 PR rtl-optimization/59647
3708 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3709 new_rtx into UNSIGNED_FLOAT rtxes.
3711 Copyright (C) 2014 Free Software Foundation, Inc.
3713 Copying and distribution of this file, with or without modification,
3714 are permitted in any medium without royalty provided the copyright
3715 notice and this notice are preserved.