1 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
4 * config/i386/i386.c (construct_container): Remove TFmode check
5 for X86_64_INTEGER_CLASS.
7 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
10 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
11 only when -Wpsabi is enabled.
13 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
16 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
17 passing arrays in registers are the same as for structs, so remove the
18 special case for them.
20 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
22 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
23 destination type, extract only the valid bits if the source type is not
24 integral and has a different mode.
26 2014-02-19 Richard Biener <rguenther@suse.de>
29 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
32 2014-02-19 Richard Biener <rguenther@suse.de>
35 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
36 (ipa_modify_call_arguments): Emit an argument load explicitely and
37 preserve virtual SSA form there and for the replacement call.
38 Do not update SSA form nor free dominance info.
40 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
42 * ipa.c (function_and_variable_visibility): Also clear WEAK
43 flag when disolving COMDAT_GROUP.
45 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
47 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
48 * ipa-prop.c (ipa_set_jf_known_type): Return early when
50 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
51 do more sanity checks.
52 (detect_type_change): Return true when giving up early.
53 (compute_complex_assign_jump_func): Fix type parameter of
55 (compute_complex_ancestor_jump_func): Likewise.
56 (update_jump_functions_after_inlining): Fix updating of
58 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
60 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
62 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
63 inline clones when edge disappears.
65 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
68 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
69 Split 64-bit moves into 2 patterns. Do not allow the use of
70 direct move for TDmode in little endian, since the decimal value
71 has little endian bytes within a word, but the 64-bit pieces are
72 ordered in a big endian fashion, and normal subreg's of TDmode are
74 (mov<mode>_64bit_dm): Likewise.
75 (movtd_64bit_nodm): Likewise.
77 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
79 PR tree-optimization/60174
80 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
81 statement of an SSA_NAME that occurs in an abnormal PHI node.
83 2014-02-18 Jakub Jelinek <jakub@redhat.com>
86 * final.c (SEEN_BB): Remove.
87 (SEEN_NOTE, SEEN_EMITTED): Renumber.
88 (final_scan_insn): Don't force_source_line on second
89 NOTE_INSN_BASIC_BLOCK.
91 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
94 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
95 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
96 (type_natural_mode): Warn ABI change when %zmm register is not
97 available for AVX512F vector value passing.
99 2014-02-18 Kai Tietz <ktietz@redhat.com>
102 * config/i386/i386.c (ix86_expand_prologue): Use value in
103 rax register as displacement when restoring %r10 or %rax.
104 Fix wrong offset when restoring both registers.
106 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
108 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
109 assertion with conditional return.
111 2014-02-18 Jakub Jelinek <jakub@redhat.com>
112 Uros Bizjak <ubizjak@gmail.com>
115 * config/i386/driver-i386.c (host_detect_local_cpu): If
116 YMM state is not saved by the OS, also clear has_f16c. Move
117 CPUID 0x80000001 handling before YMM state saving checking.
119 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
121 PR rtl-optimization/58960
122 * haifa-sched.c (alloc_global_sched_pressure_data): New,
123 factored out from ...
124 (sched_init): ... here.
125 (free_global_sched_pressure_data): New, factored out from ...
126 (sched_finish): ... here.
127 * sched-int.h (free_global_sched_pressure_data): Declare.
128 * sched-rgn.c (nr_regions_initial): New static global.
129 (haifa_find_rgns): Initialize it.
130 (schedule_region): Disable sched-pressure for the newly
133 2014-02-17 Richard Biener <rguenther@suse.de>
135 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
136 release SSA defs of pattern stmts.
138 2014-02-17 Richard Biener <rguenther@suse.de>
140 * tree-inline.c (expand_call_inline): Release the virtual
141 operand defined by the call we are about to inline.
143 2014-02-17 Richard Biener <rguenther@suse.de>
145 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
147 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
148 Ilya Tocar <ilya.tocar@intel.com>
150 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
151 arguments order in builtin.
152 (_mm512_permutexvar_epi64): Ditto.
153 (_mm512_mask_permutexvar_epi64): Ditto
154 (_mm512_maskz_permutexvar_epi32): Ditto
155 (_mm512_permutexvar_epi32): Ditto
156 (_mm512_mask_permutexvar_epi32): Ditto
158 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
160 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
161 (p8_vmrgow): Likewise.
163 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
165 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
168 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
171 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
172 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
173 into 64-bit and 32-bit moves. On 64-bit moves, add support for
174 using direct move instructions on ISA 2.07. Also adjust
175 instruction length for 64-bit.
176 (mov<mode>_64bit, TFmode/TDmode): Likewise.
177 (mov<mode>_32bit, TFmode/TDmode): Likewise.
179 2014-02-15 Alan Modra <amodra@gmail.com>
183 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
184 find_replacement on parts of insn rtl that might be reloaded.
186 2014-02-15 Richard Biener <rguenther@suse.de>
188 PR tree-optimization/60183
189 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
190 (tree_ssa_phiprop): Calculate and free post-dominators.
192 2014-02-14 Jeff Law <law@redhat.com>
194 PR rtl-optimization/60131
195 * ree.c (get_extended_src_reg): New function.
196 (combine_reaching_defs): Use it rather than assuming location of REG.
197 (find_and_remove_re): Verify first operand of extension is
198 a REG before adding the insns to the copy list.
200 2014-02-14 Roland McGrath <mcgrathr@google.com>
202 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
203 * configure: Regenerated.
204 * config.in: Regenerated.
205 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
206 instead of ASM_SHORT.
208 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
209 Richard Earnshaw <rearnsha@arm.com>
211 PR rtl-optimization/59535
212 * lra-constraints.c (process_alt_operands): Encourage alternative
213 when unassigned pseudo class is superset of the alternative class.
214 (inherit_reload_reg): Don't inherit when optimizing for code size.
215 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
216 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
217 modes not less than 4 for Thumb1.
219 2014-02-14 Kyle McMartin <kyle@redhat.com>
222 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
224 2014-02-14 Richard Biener <rguenther@suse.de>
226 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
227 (get_frame_arg): Drop the assert with langhook types_compatible_p.
228 Do not strip INDIRECT_REFs.
230 2014-02-14 Richard Biener <rguenther@suse.de>
233 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
234 DECL_FUNCTION_SPECIFIC_TARGET.
235 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
236 * tree-streamer-out.c (pack_ts_target_option): Remove.
237 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
238 (write_ts_function_decl_tree_pointers): Do not stream
239 DECL_FUNCTION_SPECIFIC_TARGET.
240 * tree-streamer-in.c (unpack_ts_target_option): Remove.
241 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
242 (lto_input_ts_function_decl_tree_pointers): Do not stream
243 DECL_FUNCTION_SPECIFIC_TARGET.
245 2014-02-14 Jakub Jelinek <jakub@redhat.com>
247 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
248 (get_initial_def_for_induction, vectorizable_induction): Ignore
249 debug stmts when looking for exit_phi.
250 (vectorizable_live_operation): Fix up condition.
252 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
254 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
255 nreverse() because it changes the content of original tree list.
257 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
259 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
260 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
262 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
264 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
265 GNU coding standards.
267 2014-02-13 Jakub Jelinek <jakub@redhat.com>
270 * dwarf2out.c (gen_subprogram_die): Don't call
271 add_calling_convention_attribute if subr_die is old_die.
273 2014-02-13 Sharad Singhai <singhai@google.com>
275 * doc/optinfo.texi: Fix order of nodes.
277 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
279 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
280 operands[2], not operands[3].
282 2014-02-13 Richard Biener <rguenther@suse.de>
285 * doc/install.texi (ISL): Update recommended version to 0.12.2,
286 mention the possibility of an in-tree build.
287 (CLooG): Update recommended version to 0.18.1, mention the
288 possibility of an in-tree build and clarify that the ISL
289 bundled with CLooG does not work.
291 2014-02-13 Jakub Jelinek <jakub@redhat.com>
294 * expr.c (compress_float_constant): If x is a hard register,
295 extend into a pseudo and then move to x.
297 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
299 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
300 caused by bad second argument to warning_at() with -mhotpatch and
301 nested functions (e.g. with gfortran).
303 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
305 * opts.c (option_name): Remove "enabled by default" rider.
307 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
309 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
311 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
312 Uros Bizjak <ubizjak@gmail.com>
315 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
316 * configure: Regenerated.
318 2014-02-12 Richard Biener <rguenther@suse.de>
320 * vec.c (vec_prefix::calculate_allocation): Move as
321 inline variant to vec.h.
322 (vec_prefix::calculate_allocation_1): New out-of-line version.
323 * vec.h (vec_prefix::calculate_allocation_1): Declare.
324 (vec_prefix::m_has_auto_buf): Rename to ...
325 (vec_prefix::m_using_auto_storage): ... this.
326 (vec_prefix::calculate_allocation): Inline the easy cases
327 and dispatch to calculate_allocation_1 which doesn't need the
329 (va_heap::reserve): Use gcc_checking_assert.
330 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
331 m_using_auto_storage.
332 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
334 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
335 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
336 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
338 2014-02-12 Richard Biener <rguenther@suse.de>
340 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
341 when we found a dependence.
343 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
345 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
347 (maybe_fold_stmt): ... into this new function.
348 * omp-low.c (lower_omp): Update comment.
350 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
353 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
356 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
358 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
359 identifiers in comments.
360 (cortexa53_extra_costs): Likewise.
361 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
362 (cortexa7_extra_costs): Likewise.
363 (cortexa12_extra_costs): Likewise.
364 (cortexa15_extra_costs): Likewise.
365 (v7m_extra_costs): Likewise.
367 2014-02-12 Richard Biener <rguenther@suse.de>
370 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
371 of posix_memalign being successful.
372 (lower_stmt): Restrict lowering of posix_memalign to when
373 -ftree-bit-ccp is enabled.
375 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
377 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
379 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
381 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
383 PR rtl-optimization/60116
384 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
385 other_insn once the combination has been validated.
387 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
390 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
392 * ipa-devirt.c: Include demangle.h
393 (odr_violation_reported): New static variable.
394 (add_type_duplicate): Update odr_violations.
395 (maybe_record_node): Add completep parameter; update it.
396 (record_target_from_binfo): Add COMPLETEP parameter;
398 (possible_polymorphic_call_targets_1): Likewise.
399 (struct polymorphic_call_target_d): Add nonconstruction_targets;
400 rename FINAL to COMPLETE.
401 (record_targets_from_bases): Sanity check we found the binfo;
402 fix COMPLETEP updating.
403 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
404 parameter, fix computing of COMPLETEP.
405 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
406 at LTO time do demangling.
407 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
408 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
410 (gimple_get_virt_method_for_binfo): Likewise.
411 * gimple-fold.h (gimple_get_virt_method_for_binfo,
412 gimple_get_virt_method_for_vtable): Update prototypes.
414 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
417 * genautomata.c (add_presence_absence): Fix typo with
418 {final_}presence_list.
420 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
423 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
424 for VSX/Altivec vectors that land in GPR registers.
426 2014-02-11 Richard Henderson <rth@redhat.com>
427 Jakub Jelinek <jakub@redhat.com>
430 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
431 around drhs if type conversion to lacc->type is not useless.
433 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
435 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
437 (cortex-a57.cortex-a53): Likewise.
438 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
440 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
442 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
445 2014-02-11 Renlin Li <Renlin.Li@arm.com>
447 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
448 add_options_for_arm_vfp3.
450 2014-02-11 Jeff Law <law@redhat.com>
453 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
454 object with an undesirable mode.
456 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
459 * config/i386/sol2-9.h: New file.
460 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
461 *-*-solaris2.9*): Use it.
463 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
465 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
466 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
468 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
470 * config/microblaze/microblaze.c: Extend mcpu version format
472 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
474 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
476 2014-02-10 Richard Henderson <rth@redhat.com>
479 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
480 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
481 ms-abi vs -mno-accumulate-outgoing-args.
482 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
483 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
486 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
489 * cfgexpand.c (expand_asm_operands): Attach source location to
490 ASM_INPUT rtx objects.
491 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
493 2014-02-10 Nick Clifton <nickc@redhat.com>
495 * config/mn10300/mn10300.c (popcount): New function.
496 (mn10300_expand_prologue): Include saved registers in stack usage
499 2014-02-10 Jeff Law <law@redhat.com>
502 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
503 when changing the SET_DEST of a prior insn to avoid an input reload.
505 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
507 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
508 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
509 -mcall-openbsd, or -mcall-linux.
510 (CC1_ENDIAN_BIG_SPEC): Remove.
511 (CC1_ENDIAN_LITTLE_SPEC): Remove.
512 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
513 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
514 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
515 and %cc1_endian_default.
516 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
518 2014-02-10 Richard Biener <rguenther@suse.de>
520 PR tree-optimization/60115
521 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
522 MEM_REF handling. Properly verify that the accesses are not
523 out of the objects bound.
525 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
527 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
530 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
532 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
533 proper constants and fix formatting.
534 (possible_polymorphic_call_targets): Fix formatting.
536 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
537 Ilya Tocar <ilya.tocar@intel.com>
539 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
540 (_mm512_loadu_epi32): Renamed into...
541 (_mm512_loadu_si512): This.
542 (_mm512_storeu_epi32): Renamed into...
543 (_mm512_storeu_si512): This.
544 (_mm512_maskz_ceil_ps): Removed.
545 (_mm512_maskz_ceil_pd): Ditto.
546 (_mm512_maskz_floor_ps): Ditto.
547 (_mm512_maskz_floor_pd): Ditto.
548 (_mm512_floor_round_ps): Ditto.
549 (_mm512_floor_round_pd): Ditto.
550 (_mm512_ceil_round_ps): Ditto.
551 (_mm512_ceil_round_pd): Ditto.
552 (_mm512_mask_floor_round_ps): Ditto.
553 (_mm512_mask_floor_round_pd): Ditto.
554 (_mm512_mask_ceil_round_ps): Ditto.
555 (_mm512_mask_ceil_round_pd): Ditto.
556 (_mm512_maskz_floor_round_ps): Ditto.
557 (_mm512_maskz_floor_round_pd): Ditto.
558 (_mm512_maskz_ceil_round_ps): Ditto.
559 (_mm512_maskz_ceil_round_pd): Ditto.
560 (_mm512_expand_pd): Ditto.
561 (_mm512_expand_ps): Ditto.
562 * config/i386/i386.c (ix86_builtins): Remove
563 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
565 * config/i386/predicates.md (const1256_operand): New.
566 (const_1_to_2_operand): Ditto.
567 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
568 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
569 (*avx512pf_gatherpf<mode>sf): Ditto.
570 (avx512pf_gatherpf<mode>df): Ditto.
571 (*avx512pf_gatherpf<mode>df_mask): Ditto.
572 (*avx512pf_gatherpf<mode>df): Ditto.
573 (avx512pf_scatterpf<mode>sf): Ditto.
574 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
575 (*avx512pf_scatterpf<mode>sf): Ditto.
576 (avx512pf_scatterpf<mode>df): Ditto.
577 (*avx512pf_scatterpf<mode>df_mask): Ditto.
578 (*avx512pf_scatterpf<mode>df): Ditto.
579 (avx512f_expand<mode>): Removed.
580 (<shift_insn><mode>3<mask_name>): Change predicate type.
582 2014-02-08 Jakub Jelinek <jakub@redhat.com>
584 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
585 not at the end of datarefs vector use ordered_remove to avoid
586 reordering datarefs vector.
589 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
590 mark local addressable non-static vars as GOVD_PRIVATE
591 instead of GOVD_LOCAL.
592 * omp-low.c (lower_omp_for): Move gimple_bind_vars
593 and BLOCK_VARS of gimple_bind_block to new_stmt rather
597 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
598 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
599 assume_aligned or alloc_align attributes.
600 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
601 arguments. Handle also assume_aligned and alloc_align attributes.
602 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
603 calls to functions with assume_aligned or alloc_align attributes.
604 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
606 2014-02-08 Terry Guo <terry.guo@arm.com>
608 * doc/invoke.texi: Document ARM -march=armv7e-m.
610 2014-02-08 Jakub Jelinek <jakub@redhat.com>
612 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
613 flag on __cilkrts_rethrow builtin.
616 * ipa-cp.c (determine_versionability): Fail at -O0
617 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
618 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
621 2014-02-04 Jakub Jelinek <jakub@redhat.com>
624 * tree-inline.c (copy_forbidden): Fail for
625 __attribute__((optimize (0))) functions.
627 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
629 * varpool.c: Include pointer-set.h.
630 (varpool_remove_unreferenced_decls): Variables in other partitions
631 will not be output; be however careful to not lose information
634 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
636 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
637 lookup in the vtable constructor.
639 2014-02-07 Jeff Law <law@redhat.com>
642 * config/m68k/m68k.md (ashldi_extsi): Turn into a
643 define_insn_and_split.
645 * ipa-inline.c (inline_small_functions): Fix typos.
647 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
649 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
650 (s390_can_use_return_insn): Declare.
651 * config/s390/s390.h (EPILOGUE_USES): Define.
652 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
654 (s390_chunkify_start): Handle return JUMP_LABELs.
655 (s390_early_mach): Emit a main_pool instruction on the entry edge.
656 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
657 (s390_can_use_return_insn): New functions.
658 (s390_fix_long_loop_prediction): Handle conditional returns.
659 (TARGET_SET_UP_BY_PROLOGUE): Define.
660 * config/s390/s390.md (ANY_RETURN): New code iterator.
661 (*creturn, *csimple_return, return, simple_return): New patterns.
663 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
665 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
666 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
667 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
668 REG_CFA_RESTORE list when deciding not to restore a register.
670 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
672 * config/s390/s390.c: Include tree-pass.h and context.h.
673 (s390_early_mach): New function, split out from...
674 (s390_emit_prologue): ...here.
675 (pass_data_s390_early_mach): New pass structure.
676 (pass_s390_early_mach): New class.
677 (s390_option_override): Create and register early_mach pass.
680 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
682 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
683 to match for the exit block.
685 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
687 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
688 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
689 Reject misaligned operands.
691 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
693 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
695 2014-02-07 Richard Biener <rguenther@suse.de>
698 * gimple-low.c (lower_builtin_posix_memalign): New function.
699 (lower_stmt): Call it to lower posix_memalign in a way
700 to make alignment info accessible.
702 2014-02-07 Jakub Jelinek <jakub@redhat.com>
705 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
706 __builtin_setjmp_receiver.
708 2014-02-07 Richard Biener <rguenther@suse.de>
711 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
712 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
713 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
714 Handle BUILT_IN_POSIX_MEMALIGN.
715 (find_func_clobbers): Likewise.
716 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
717 (call_may_clobber_ref_p_1): Likewise.
719 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
722 * ipa-devirt.c (record_target_from_binfo): Remove overactive
725 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
728 * lto-cgraph.c (lto_output_node): Use
729 symtab_get_symbol_partitioning_class.
730 (lto_output_varpool_node): likewise.
731 (symtab_get_symbol_partitioning_class): Move here from
733 * cgraph.h (symbol_partitioning_class): Likewise.
734 (symtab_get_symbol_partitioning_class): Declare.
736 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
738 * ggc.h (ggc_internal_cleared_alloc): New macro.
739 * vec.h (vec_safe_copy): Handle memory stats.
740 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
741 * target-globals.c (save_target_globals): Likewise.
743 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
746 * expr.c (emit_move_resolve_push): Export; be bit more selective
747 on when to clear alias set.
748 * expr.h (emit_move_resolve_push): Declare.
749 * function.h (struct function): Add tail_call_marked.
750 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
751 * config/i386/i386-protos.h (ix86_expand_push): Remove.
752 * config/i386/i386.md (TImode move expander): De not call
754 (FP push expanders): Preserve memory attributes.
755 * config/i386/sse.md (push<mode>1): Remove.
756 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
757 (ix86_expand_push): Remove.
758 * config/i386/mmx.md (push<mode>1): Remove.
760 2014-02-06 Jakub Jelinek <jakub@redhat.com>
762 PR rtl-optimization/60030
763 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
764 lopart with paradoxical subreg before shifting it up by hprec.
766 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
768 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
769 Remove extra newline at end of file.
770 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
771 (arm_issue_rate): Handle cortexa57.
772 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
773 (cortex-a57.cortex-a53): Likewise.
775 2014-02-06 Jakub Jelinek <jakub@redhat.com>
778 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
779 don't record in REG_FRAME_RELATED_EXPR registers not set in that
781 (arm_expand_prologue): Adjust all callers.
782 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
783 info, registers also at the lowest numbered registers side. Use
784 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
788 * var-tracking.c (adjust_mems): Before adding a SET to
789 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
791 2014-02-06 Alan Modra <amodra@gmail.com>
794 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
795 change SDmode to DDmode when lra_in_progress.
797 2014-02-06 Jakub Jelinek <jakub@redhat.com>
800 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
801 free_data_ref on the dr first, and before goto again also set dr
802 to the next dr. For simd_lane_access, free old datarefs[i] before
803 overwriting it. For get_vectype_for_scalar_type failure, don't
804 free_data_ref if simd_lane_access.
806 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
809 * tree.h (opts_for_fn): New inline function.
810 (opt_for_fn): Define.
811 * config/i386/i386.c (ix86_function_regparm): Use
812 opt_for_fn (decl, optimize) instead of optimize.
814 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
816 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
817 for SYMBOL_REF in large memory model.
819 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
821 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
823 (cortex-a57): Likewise.
824 (cortex-a57.cortex-a53): Likewise.
826 2014-02-06 Yury Gribov <y.gribov@samsung.com>
827 Kugan Vivekanandarajah <kuganv@linaro.org>
829 * config/arm/arm.c (arm_vector_alignment_reachable): Check
831 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
833 2014-02-06 Richard Biener <rguenther@suse.de>
835 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
836 set_loop_copy and initialize_original_copy_tables.
838 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
840 * config/aarch64/aarch64-simd.md
841 (aarch64_ashr_simddi): Change QI to SI.
843 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
844 Jakub Jelinek <jakub@redhat.com>
847 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
850 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
852 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
853 CODE_FOR_altivec_vpku[hw]um to
854 CODE_FOR_altivec_vpku[hw]um_direct.
855 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
856 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
857 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
858 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
860 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
862 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
863 generation for -maltivec=be.
864 (altivec_vsumsws): Simplify redundant test.
866 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
868 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
869 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
870 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
871 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
873 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
875 (altivec_vpks<VI_char>ss): Likewise.
876 (altivec_vpks<VI_char>us): Likewise.
877 (altivec_vpku<VI_char>us): Likewise.
878 (altivec_vpku<VI_char>um): Likewise.
879 (altivec_vpku<VI_char>um_direct): New (copy of
880 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
882 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
883 target is little endian and -maltivec=be is not specified.
884 (*altivec_vupkhs<VU_char>_direct): New (copy of
885 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
886 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
887 target is little endian and -maltivec=be is not specified.
888 (*altivec_vupkls<VU_char>_direct): New (copy of
889 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
890 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
891 little endian and -maltivec=be is not specified.
892 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
893 little endian and -maltivec=be is not specified.
895 2014-02-05 Richard Henderson <rth@redhat.com>
898 * combine-stack-adj.c: Revert r206943.
899 * sched-int.h (struct deps_desc): Add last_args_size.
900 * sched-deps.c (init_deps): Initialize it.
901 (sched_analyze_insn): Add OUTPUT dependencies between insns that
902 contain REG_ARGS_SIZE notes.
904 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
906 * lto-cgraph.c (asm_nodes_output): Make global.
907 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
908 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
909 (driver_handle_option): Handle OPT_fwpa.
911 2014-02-05 Jakub Jelinek <jakub@redhat.com>
914 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
915 a comment typo and formatting issue. If odr_hash hasn't been
916 created, return vNULL and set *completep to false.
919 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
920 bb with no successors.
922 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
925 * doc/invoke.texi (-march): Clarify documentation for ARM.
929 2014-02-05 Richard Biener <rguenther@suse.de>
931 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
932 when not vectorizing because of too many alias checks.
933 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
934 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
936 2014-02-05 Nick Clifton <nickc@redhat.com>
938 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
939 accept extended registers in any mode when compiling for the MN10300.
941 2014-02-05 Yury Gribov <y.gribov@samsung.com>
943 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
944 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
945 sanitization attributes.
946 (can_inline_edge_p): Likewise.
947 (sanitize_attrs_match_for_inline_p): New function.
949 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
951 * ipa-prop.c (detect_type_change): Shor circuit testing of
952 type changes on THIS pointer.
954 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
957 * config/pa/pa.c (legitimize_tls_address): Return original address
958 if not passed a SYMBOL_REF rtx.
959 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
961 (pa_emit_move_sequence): Simplify TLS source operands.
962 (pa_legitimate_constant_p): Reject all TLS constants.
963 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
964 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
966 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
968 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
969 groups when we know they are controlled by LTO.
970 * varasm.c (default_binds_local_p_1): If object is in other partition,
971 it will be resolved locally.
973 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
975 * config/host-linux.c (linux_gt_pch_use_address): Don't
976 use SSIZE_MAX because it is not always defined.
978 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
981 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
982 threshold for pseudo splitting.
983 (update_ebb_live_info): Process call argument hard registers and
984 hard registers from insn definition too.
985 (max_small_class_regs_num): New constant.
986 (inherit_in_ebb): Update live hard regs through EBBs. Update
987 reloads_num only for small register classes. Don't split for
990 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
993 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
996 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
998 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1001 2014-02-04 Marek Polacek <polacek@redhat.com>
1003 * gdbinit.in (pel): Define.
1005 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1007 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1010 2014-02-04 Richard Biener <rguenther@suse.de>
1013 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1014 in function context local.
1015 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1016 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1017 similar to LTO_imported_decl_ref.
1019 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1021 PR tree-optimization/60002
1022 * cgraphclones.c (build_function_decl_skip_args): Clear
1025 PR tree-optimization/60023
1026 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1027 false to gsi_replace.
1028 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1029 has been in some EH region and vec_stmt could throw, add
1030 vec_stmt into the same EH region.
1031 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1032 has no lhs, ignore it.
1033 * internal-fn.c (expand_MASK_LOAD): Likewise.
1036 * tree-inline.c (copy_forbidden): Fail for
1037 __attribute__((optimize (0))) functions.
1040 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1041 is set, copy one less argument.
1042 (expand_simd_clones): Don't subtract clone_info->inbranch
1043 from simd_clone_struct_alloc argument.
1045 PR rtl-optimization/57915
1046 * recog.c (simplify_while_replacing): If all unary/binary/relational
1047 operation arguments are constant, attempt to simplify those.
1050 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1051 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1053 2014-02-04 Richard Biener <rguenther@suse.de>
1055 PR tree-optimization/60012
1056 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1057 TBAA disambiguation to all DDRs.
1059 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1062 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1063 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1065 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1068 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1070 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1072 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1073 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1075 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1078 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1079 to figure out targets of polymorphic calls with known decl.
1080 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1081 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1082 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1083 (get_polymorphic_call_info): ... here.
1084 (get_polymorphic_call_info_from_invariant): New function.
1086 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1088 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1089 lookup via vtable pointer; check for type consistency
1090 and turn inconsitent facts into UNREACHABLE.
1091 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1092 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1093 type inconsistent querries; return UNREACHABLE instead.
1095 2014-02-03 Richard Henderson <rth@twiddle.net>
1098 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1099 already processed this node.
1100 (normalize_one_pred_1): Pass along mark_set.
1101 (normalize_one_pred): Create and destroy a pointer_set_t.
1102 (normalize_one_pred_chain): Likewise.
1104 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1106 PR gcov-profile/58602
1107 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1109 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1112 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1113 -fno-devirtualize; try to devirtualize by the knowledge of
1114 virtual table pointer given by aggregate propagation.
1115 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1116 (ipa_print_node_jump_functions): Dump also offset that
1117 is relevant for polymorphic calls.
1118 (determine_known_aggregate_parts): Add arg_type parameter; use it
1119 instead of determining the type from pointer type.
1120 (ipa_compute_jump_functions_for_edge): Update call of
1121 determine_known_aggregate_parts.
1122 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1123 (gimple_get_virt_method_for_binfo): ... here; simplify using
1124 vtable_pointer_value_to_vtable.
1125 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1126 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1127 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1128 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1130 (vtable_pointer_value_to_binfo): ... here.
1131 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1133 2014-02-03 Teresa Johnson <tejohnson@google.com>
1135 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1136 redef of outer loop index variable.
1138 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1142 * doc/extend.texi (Function Attributes): Typo.
1144 2014-02-03 Cong Hou <congh@google.com>
1146 PR tree-optimization/60000
1147 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1148 if the vectorized statement is a store. A store statement can only
1149 appear at the end of pattern statements.
1151 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1153 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1154 (ix86_option_override_internal): Default long double to 64-bit for
1155 32-bit Bionic and to 128-bit for 64-bit Bionic.
1157 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1158 TARGET_LONG_DOUBLE_128 is true.
1159 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1161 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1162 (mlong-double-64): Negate -mlong-double-128.
1163 (mlong-double-128): New option.
1165 * config/i386/i386-c.c (ix86_target_macros): Define
1166 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1168 * doc/invoke.texi: Document -mlong-double-128.
1170 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1172 PR rtl-optimization/60024
1173 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1175 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1177 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1179 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1181 PR rtl-optimization/57662
1182 * sel-sched.c (code_motion_path_driver): Do not mark already not
1183 existing blocks in the visiting bitmap.
1185 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1187 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1188 on the insn being emitted.
1190 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1191 Will Deacon <will.deacon@arm.com>
1193 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1195 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1197 * config/arm/arm-tables.opt: Regenerate.
1199 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1201 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1202 for vector types other than V16QImode.
1203 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1204 define_expand, and call altivec_expand_vec_perm_le when producing
1205 code with little endian element order.
1206 (*altivec_vperm_<mode>_internal): New insn having previous
1207 behavior of altivec_vperm_<mode>.
1208 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1209 altivec_expand_vec_perm_le when producing code with little endian
1211 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1212 behavior of altivec_vperm_<mode>_uns.
1214 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1216 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1217 (altivec_vsumsws): Add handling for -maltivec=be with a little
1219 (altivec_vsumsws_direct): New.
1220 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1221 gen_altivec_vsumsws.
1223 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1225 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1226 vtable_pointer_value_to_binfo): New functions.
1227 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1228 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1230 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1232 * config/nios2/nios2.md (load_got_register): Initialize GOT
1233 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1234 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1236 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1238 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1239 preserverd by passthrough, do not propagate the type.
1241 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1243 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1244 (mips_atomic_assign_expand_fenv): New function.
1245 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1247 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1249 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1250 (__builtin_mips_set_fcsr): Likewise.
1251 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1252 MIPS_USI_FTYPE_VOID.
1253 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1254 (mips16_expand_set_fcsr): Likewise.
1255 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1256 (mips16_set_fcsr_stub): Likewise.
1257 (mips16_get_fcsr_one_only_stub): New class.
1258 (mips16_set_fcsr_one_only_stub): Likewise.
1259 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1260 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1261 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1262 (hard_float): New availability predicate.
1263 (mips_builtins): Add get_fcsr and set_fcsr.
1264 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1265 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1266 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1267 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1268 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1271 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1273 * config/mips/mips.c (mips_one_only_stub): New class.
1274 (mips_need_mips16_rdhwr_p): Replace with...
1275 (mips16_rdhwr_stub): ...this new variable.
1276 (mips16_stub_call_address): New function.
1277 (mips16_rdhwr_one_only_stub): New class.
1278 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1279 (mips_output_mips16_rdhwr): Delete.
1280 (mips_finish_stub): New function.
1281 (mips_code_end): Use it to handle rdhwr stubs.
1283 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1286 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1287 when calculating size of integer atomic types.
1289 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1291 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1293 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1295 PR tree-optimization/60003
1296 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1297 * profile.c (branch_prob): Use gimple_call_builtin_p
1298 to check for BUILT_IN_SETJMP_RECEIVER.
1299 * tree-inline.c (copy_bb): Call notice_special_calls.
1301 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1304 * lra-constraints.c (process_alt_operands): Update reload_sum only
1307 2014-01-31 Richard Henderson <rth@redhat.com>
1310 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1311 until after else_eh is processed.
1313 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1315 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1316 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1317 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1318 in smmintrin.h, remove them.
1319 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1320 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1321 * config/i386/i386.md (ROUND_SAE): Fix value.
1322 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1323 (const48_operand): New.
1324 * config/i386/subst.md (round), (round_expand): Use
1325 const_4_or_8_to_11_operand.
1326 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1328 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1330 * config/i386/constraints.md (Yk): Swap meaning with k.
1331 * config/i386/i386.md (movhi_internal): Change Yk to k.
1332 (movqi_internal): Ditto.
1333 (*k<logic><mode>): Ditto.
1336 (kandn<mode>): Ditto.
1337 (*<code>hi_1): Ditto.
1338 (*<code>qi_1): Ditto.
1339 (kxnor<mode>): Ditto.
1340 (kortestzhi): Ditto.
1341 (kortestchi): Ditto.
1343 (*one_cmplhi2_1): Ditto.
1344 (*one_cmplqi2_1): Ditto.
1345 * config/i386/sse.md (): Change k to Yk.
1346 (avx512f_load<mode>_mask): Ditto.
1347 (avx512f_blendm<mode>): Ditto.
1348 (avx512f_store<mode>_mask): Ditto.
1349 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1350 (avx512f_storedqu<mode>_mask): Ditto.
1351 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1353 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1354 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1355 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1356 (avx512f_maskcmp<mode>3): Ditto.
1357 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1358 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1359 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1360 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1361 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1362 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1363 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1364 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1365 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1366 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1367 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1368 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1369 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1370 (vec_extract_lo_<mode>_maskm): Ditto.
1371 (vec_extract_hi_<mode>_maskm): Ditto.
1372 (avx512f_vternlog<mode>_mask): Ditto.
1373 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1374 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1375 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1376 (avx512f_<code>v8div16qi2_mask): Ditto.
1377 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1378 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1379 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1380 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1381 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1382 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1383 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1384 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1385 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1386 (avx512cd_maskb_vec_dupv8di): Ditto.
1387 (avx512cd_maskw_vec_dupv16si): Ditto.
1388 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1389 (avx512f_vpermi2var<mode>3_mask): Ditto.
1390 (avx512f_vpermi2var<mode>3_mask): Ditto.
1391 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1392 (*avx512f_gathersi<mode>): Ditto.
1393 (*avx512f_gathersi<mode>_2): Ditto.
1394 (*avx512f_gatherdi<mode>): Ditto.
1395 (*avx512f_gatherdi<mode>_2): Ditto.
1396 (*avx512f_scattersi<mode>): Ditto.
1397 (*avx512f_scatterdi<mode>): Ditto.
1398 (avx512f_compress<mode>_mask): Ditto.
1399 (avx512f_compressstore<mode>_mask): Ditto.
1400 (avx512f_expand<mode>_mask): Ditto.
1401 * config/i386/subst.md (mask): Change k to Yk.
1402 (mask_scalar_merge): Ditto.
1405 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1407 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1409 2014-01-31 Richard Biener <rguenther@suse.de>
1412 * builtins.c (fold_builtin_memory_op): Make sure to not
1413 use a floating-point mode or a boolean or enumeral type for
1416 2014-01-30 DJ Delorie <dj@redhat.com>
1418 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1419 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1420 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1421 whenever main() has an epilogue.
1423 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1425 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1426 unused variable "field".
1427 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1428 (vsx_mergeh_<mode>): Likewise.
1429 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1430 (altivec_vmrghh): Likewise.
1431 (altivec_vmrghw): Likewise.
1432 (altivec_vmrglb): Likewise.
1433 (altivec_vmrglh): Likewise.
1434 (altivec_vmrglw): Likewise.
1435 (altivec_vspltb): Add missing uses.
1436 (altivec_vsplth): Likewise.
1437 (altivec_vspltw): Likewise.
1438 (altivec_vspltsf): Likewise.
1440 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1443 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1444 frame related instructions.
1446 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1448 PR rtl-optimization/59959
1449 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1450 any reload of register whose subreg is invalid.
1452 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1454 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1455 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1456 Add missing return type - void.
1458 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1460 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1461 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1462 remove element index adjustment for endian (now handled in vsx.md
1464 (altivec_expand_vec_perm_const): Use
1465 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1466 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1467 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1468 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1469 define_expand and a new define_insn *altivec_vspltb_internal;
1470 adjust for -maltivec=be on a little endian target.
1471 (altivec_vspltb_direct): New.
1472 (altivec_vsplth): Divide into a define_expand and a new
1473 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1474 little endian target.
1475 (altivec_vsplth_direct): New.
1476 (altivec_vspltw): Divide into a define_expand and a new
1477 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1478 little endian target.
1479 (altivec_vspltw_direct): New.
1480 (altivec_vspltsf): Divide into a define_expand and a new
1481 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1482 a little endian target.
1484 2014-01-30 Richard Biener <rguenther@suse.de>
1486 PR tree-optimization/59993
1487 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1488 can propagate form the earlier stmt and avoid the transform
1489 when the intermediate result is needed.
1491 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1493 * README.Portability: Fix typo.
1495 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1497 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1498 comparison_operator with ordered_comparison_operator.
1500 2014-01-30 Nick Clifton <nickc@redhat.com>
1502 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1503 Rename to mn10300_store_multiple_regs.
1504 * config/mn10300/mn10300.c: Likewise.
1505 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1506 store_multiple_regs.
1507 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1508 Call mn10300_store_multiple_regs.
1510 2014-01-30 Nick Clifton <nickc@redhat.com>
1511 DJ Delorie <dj@redhat.com>
1513 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1514 %fp 2 to keep registers after it properly word-aligned.
1515 (rl78_alloc_physical_registers_umul): Handle the case where both
1516 input operands are the same.
1518 2014-01-30 Richard Biener <rguenther@suse.de>
1520 PR tree-optimization/59903
1521 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1524 2014-01-30 Jason Merrill <jason@redhat.com>
1527 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1530 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1532 2014-01-30 Richard Biener <rguenther@suse.de>
1534 PR tree-optimization/59951
1535 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1537 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1540 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1541 SFmode to DFmode case.
1543 2014-01-29 DJ Delorie <dj@redhat.com>
1545 * config/msp430/msp430.opt (-minrt): New.
1546 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1548 (ENDFILE_SPEC): Likewise.
1550 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1552 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1553 (estimate_function_body_sizes): Use it.
1555 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1558 * dwarf2out.c (is_cxx_auto): New.
1559 (is_base_type): Use it.
1560 (gen_type_die_with_usage): Likewise.
1562 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1564 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1565 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1566 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1567 -maltivec=be with LE targets.
1568 (vsx_mergeh_<mode>): Likewise.
1569 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1570 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1571 (altivec_vmrghb): Replace with define_expand and new
1572 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1573 (altivec_vmrghb_direct): New define_insn.
1574 (altivec_vmrghh): Replace with define_expand and new
1575 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1576 (altivec_vmrghh_direct): New define_insn.
1577 (altivec_vmrghw): Replace with define_expand and new
1578 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1579 (altivec_vmrghw_direct): New define_insn.
1580 (*altivec_vmrghsf): Adjust for endianness.
1581 (altivec_vmrglb): Replace with define_expand and new
1582 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1583 (altivec_vmrglb_direct): New define_insn.
1584 (altivec_vmrglh): Replace with define_expand and new
1585 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1586 (altivec_vmrglh_direct): New define_insn.
1587 (altivec_vmrglw): Replace with define_expand and new
1588 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1589 (altivec_vmrglw_direct): New define_insn.
1590 (*altivec_vmrglsf): Adjust for endianness.
1591 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1592 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1593 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1594 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1595 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1596 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1597 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1598 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1600 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1602 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1603 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1606 2014-01-29 Richard Biener <rguenther@suse.de>
1608 PR tree-optimization/58742
1609 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1610 associate_pointerplus_align.
1611 (associate_pointerplus_diff): New function.
1612 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1613 and associate_pointerplus_diff.
1615 2014-01-29 Richard Biener <rguenther@suse.de>
1617 * lto-streamer.h (LTO_major_version): Bump to 3.
1618 (LTO_minor_version): Reset to 0.
1620 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1622 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1623 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1624 (arm_file_start): Generate correct asm header for armv7ve.
1625 * config/arm/bpabi.h: Add multilib support for armv7ve.
1626 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1627 and cortex-a15 to armv7ve.
1628 * config/arm/t-aprofile: Add multilib support for armv7ve.
1629 * doc/invoke.texi: Document -march=armv7ve.
1631 2014-01-29 Richard Biener <rguenther@suse.de>
1633 PR tree-optimization/58742
1634 * tree-ssa-forwprop.c (associate_plusminus): Return true
1635 if we changed sth, defer EH cleanup to ...
1636 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1637 (simplify_mult): New function.
1639 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1642 PR tree-optimization/59920
1643 * tree.c (build_common_builtin_nodes): Remove
1644 __builtin_setjmp_dispatcher initialization.
1645 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1646 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1647 instead of gsi_after_labels + manually skipping debug stmts.
1648 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1649 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1650 * tree-inline.c (copy_edges_for_bb): Remove
1651 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1652 argument. Ignore computed_goto_p stmts. Don't call
1653 make_abnormal_goto_edges. If a call might need abnormal edges
1654 for non-local gotos, see if it already has an edge to
1655 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1656 with true argument, don't do anything then, otherwise add
1657 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1658 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1660 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1661 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1662 (lower_stmt): Don't set data->calls_builtin_setjmp.
1663 (lower_builtin_setjmp): Adjust comment.
1664 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1665 * tree-cfg.c (found_computed_goto): Remove.
1666 (factor_computed_gotos): Remove.
1667 (make_goto_expr_edges): Return bool, true for computed gotos.
1668 Don't call make_abnormal_goto_edges.
1669 (build_gimple_cfg): Don't set found_computed_goto, don't call
1670 factor_computed_gotos.
1671 (computed_goto_p): No longer static.
1672 (make_blocks): Don't set found_computed_goto.
1673 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1674 (make_edges): If make_goto_expr_edges returns true, push bb
1675 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1676 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1677 vector. Record mapping between bbs and OpenMP regions if there
1678 are any, adjust make_gimple_omp_edges caller. Call
1679 handle_abnormal_edges.
1680 (make_abnormal_goto_edges): Remove.
1681 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1682 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1683 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1684 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1685 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1686 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1687 filling *region also set *region_idx to (*region)->entry->index.
1690 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1691 For REGs set ORIGINAL_REGNO.
1693 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1695 * doc/md.texi: Mention that a target shouldn't implement
1696 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1699 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1701 PR tree-optimization/59594
1702 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1703 a copy of the datarefs vector rather than the vector itself.
1705 2014-01-28 Jason Merrill <jason@redhat.com>
1708 * dwarf2out.c (auto_die): New static.
1709 (gen_type_die_with_usage): Handle C++1y 'auto'.
1710 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1713 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1716 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1717 (SPEC_X32): Likewise.
1718 (SPEC_64): Likewise.
1719 * config/i386/i386.c (ix86_option_override_internal): Turn off
1720 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1722 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1723 * config/i386/i386.h (TARGET_16BIT): New macro.
1724 (TARGET_16BIT_P): Likewise.
1725 * config/i386/i386.opt: Add m16.
1726 * doc/invoke.texi: Document -m16.
1728 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1730 PR preprocessor/59935
1731 * input.c (location_get_source_line): Bail out on when line number
1732 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1734 2014-01-28 Richard Biener <rguenther@suse.de>
1736 PR tree-optimization/58742
1737 * tree-ssa-forwprop.c (associate_plusminus): Handle
1738 pointer subtraction of the form (T)(P + A) - (T)P.
1740 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1742 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1745 2014-01-28 Richard Biener <rguenther@suse.de>
1748 2014-01-28 Richard Biener <rguenther@suse.de>
1750 PR rtl-optimization/45364
1751 PR rtl-optimization/59890
1752 * var-tracking.c (local_get_addr_clear_given_value): Handle
1753 already cleared slot.
1754 (val_reset): Handle not allocated local_get_addr_cache.
1755 (vt_find_locations): Use post-order on the inverted CFG.
1757 2014-01-28 Richard Biener <rguenther@suse.de>
1759 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1761 2014-01-28 Richard Biener <rguenther@suse.de>
1763 PR rtl-optimization/45364
1764 PR rtl-optimization/59890
1765 * var-tracking.c (local_get_addr_clear_given_value): Handle
1766 already cleared slot.
1767 (val_reset): Handle not allocated local_get_addr_cache.
1768 (vt_find_locations): Use post-order on the inverted CFG.
1770 2014-01-28 Alan Modra <amodra@gmail.com>
1772 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1773 * configure.ac <recursive call for build != host>: Define
1774 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1775 and LD_FOR_BUILD too.
1776 * configure: Regenerate.
1778 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1780 * config/i386/i386.c (get_builtin_code_for_version): Separate
1781 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1782 Broadwell from Haswell.
1784 2014-01-27 Steve Ellcey <sellcey@mips.com>
1786 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1787 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1788 * config/mips/mips.c (mips_option_override): Change setting
1790 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1791 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1792 Change from Mask to Var.
1794 2014-01-27 Jeff Law <law@redhat.com>
1796 * ipa-inline.c (inline_small_functions): Fix typo.
1798 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1800 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1801 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1802 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1803 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1804 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1805 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1806 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1807 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1808 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1809 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1810 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1811 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1812 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1813 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1814 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1815 (_mm512_storeu_epi64): Ditto.
1816 (_mm512_cmpge_epi32_mask): Ditto.
1817 (_mm512_cmpge_epu32_mask): Ditto.
1818 (_mm512_cmpge_epi64_mask): Ditto.
1819 (_mm512_cmpge_epu64_mask): Ditto.
1820 (_mm512_cmple_epi32_mask): Ditto.
1821 (_mm512_cmple_epu32_mask): Ditto.
1822 (_mm512_cmple_epi64_mask): Ditto.
1823 (_mm512_cmple_epu64_mask): Ditto.
1824 (_mm512_cmplt_epi32_mask): Ditto.
1825 (_mm512_cmplt_epu32_mask): Ditto.
1826 (_mm512_cmplt_epi64_mask): Ditto.
1827 (_mm512_cmplt_epu64_mask): Ditto.
1828 (_mm512_cmpneq_epi32_mask): Ditto.
1829 (_mm512_cmpneq_epu32_mask): Ditto.
1830 (_mm512_cmpneq_epi64_mask): Ditto.
1831 (_mm512_cmpneq_epu64_mask): Ditto.
1832 (_mm512_expand_pd): Ditto.
1833 (_mm512_expand_ps): Ditto.
1834 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1835 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1836 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1837 * config/i386/i386.c (ix86_builtins): Add
1838 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1839 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1840 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1841 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1842 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1843 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1844 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1845 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1846 IX86_BUILTIN_PMOVUSQW512_MEM.
1847 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1848 __builtin_ia32_pmovsqd512mem_mask,
1849 __builtin_ia32_pmovqd512mem_mask,
1850 __builtin_ia32_pmovusqw512mem_mask,
1851 __builtin_ia32_pmovsqw512mem_mask,
1852 __builtin_ia32_pmovqw512mem_mask,
1853 __builtin_ia32_pmovusdw512mem_mask,
1854 __builtin_ia32_pmovsdw512mem_mask,
1855 __builtin_ia32_pmovdw512mem_mask,
1856 __builtin_ia32_pmovqb512mem_mask,
1857 __builtin_ia32_pmovusqb512mem_mask,
1858 __builtin_ia32_pmovsqb512mem_mask,
1859 __builtin_ia32_pmovusdb512mem_mask,
1860 __builtin_ia32_pmovsdb512mem_mask,
1861 __builtin_ia32_pmovdb512mem_mask.
1862 (bdesc_args): Add __builtin_ia32_expanddf512,
1863 __builtin_ia32_expandsf512.
1864 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1865 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1866 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1867 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1868 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1869 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1870 (avx512f_<code>v8div16qi2_mask_store): This.
1871 (avx512f_expand<mode>): New.
1873 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1875 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1877 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1878 (_mm512_prefetch_i32scatter_pd): Ditto.
1879 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1880 (_mm512_prefetch_i64scatter_pd): Ditto.
1881 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1882 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1883 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1884 (_mm512_prefetch_i32scatter_ps): Ditto.
1885 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1886 (_mm512_prefetch_i64scatter_ps): Ditto.
1887 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1888 * config/i386/i386-builtin-types.def: Define
1889 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1890 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1891 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1892 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1893 IX86_BUILTIN_SCATTERPFQPD.
1894 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1895 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1896 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1897 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1898 __builtin_ia32_scatterpfqps.
1899 (ix86_expand_builtin): Expand new built-ins.
1900 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1901 fix memory access data type.
1902 (*avx512pf_gatherpf<mode>_mask): Ditto.
1903 (*avx512pf_gatherpf<mode>): Ditto.
1904 (avx512pf_scatterpf<mode>): Ditto.
1905 (*avx512pf_scatterpf<mode>_mask): Ditto.
1906 (*avx512pf_scatterpf<mode>): Ditto.
1907 (GATHER_SCATTER_SF_MEM_MODE): New.
1908 (avx512pf_gatherpf<mode>df): Ditto.
1909 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1910 (*avx512pf_scatterpf<mode>df): Ditto.
1912 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1915 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1916 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1919 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1921 * common/config/arm/arm-common.c
1922 (arm_rewrite_mcpu): Handle multiple names.
1924 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1926 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1928 * gimple-builder.h (create_gimple_tmp): Delete.
1930 2014-01-27 Christian Bruel <christian.bruel@st.com>
1932 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1935 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1937 * config/pa/pa.md (call): Generate indirect long calls to non-local
1938 functions when outputing 32-bit code.
1939 (call_value): Likewise except for special call to buggy powf function.
1941 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1942 portable runtime and PIC indirect calls.
1943 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1944 and PIC call sequences. Use ldo instead of blr to set return register
1945 in PIC call sequence.
1947 2014-01-25 Walter Lee <walt@tilera.com>
1949 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1950 avoid clobbering a live register.
1952 2014-01-25 Walter Lee <walt@tilera.com>
1954 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1955 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1956 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1957 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1959 2014-01-25 Walter Lee <walt@tilera.com>
1961 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1962 arguments on even registers.
1963 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1965 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1966 (BIGGEST_ALIGNMENT): Ditto.
1967 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1969 2014-01-25 Walter Lee <walt@tilera.com>
1971 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1972 insns before bundling.
1973 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1975 2014-01-25 Walter Lee <walt@tilera.com>
1977 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1978 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1979 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1981 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1983 * config/mips/constraints.md (kl): Delete.
1984 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1985 define expands, using...
1986 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1987 instructions for MIPS16.
1988 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1989 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1991 2014-01-25 Walter Lee <walt@tilera.com>
1993 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1997 2014-01-25 Walter Lee <walt@tilera.com>
1999 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2000 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2002 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2004 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2007 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2009 * print-rtl.c (in_call_function_usage): New var.
2010 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2011 EXPR_LIST mode as mode and not as reg note name.
2014 * cfgloopmanip.c (copy_loop_info): If
2015 loop->warned_aggressive_loop_optimizations, make sure
2016 the flag is set in target loop too.
2018 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2020 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2022 * builtins.def: Likewise.
2023 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2024 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2025 * ira.c (ira_setup_eliminable_regset): Likewise.
2026 * omp-low.c (gate_expand_omp): Likewise.
2027 (execute_lower_omp): Likewise.
2028 (diagnose_sb_0): Likewise.
2029 (gate_diagnose_omp_blocks): Likewise.
2030 (simd_clone_clauses_extract): Likewise.
2033 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2035 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2036 correction for little endian...
2037 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2040 2014-01-24 Jeff Law <law@redhat.com>
2042 PR tree-optimization/59919
2043 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2044 for non-returning calls.
2046 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2048 * common/config/aarch64/aarch64-common.c
2049 (aarch64_rewrite_mcpu): Handle multiple names.
2050 * config/aarch64/aarch64.h
2051 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2053 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2055 * input.c (add_file_to_cache_tab): Handle the case where fopen
2058 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2061 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2062 from push operand if code of push isn't PRE_DEC.
2064 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2067 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2068 -mquad-memory-atomic. Update -mquad-memory documentation to say
2069 it is only used for non-atomic loads/stores.
2071 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2072 -mquad-memory or -mquad-memory-atomic switches.
2074 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2075 -mquad-memory-atomic to ISA 2.07 support.
2077 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2078 to separate support of normal quad word memory operations (ldq, stq)
2079 from the atomic quad word memory operations.
2081 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2082 support to separate non-atomic quad word operations from atomic
2083 quad word operations. Disable non-atomic quad word operations in
2084 little endian mode so that we don't have to swap words after the
2085 load and before the store.
2086 (quad_load_store_p): Add comment about atomic quad word support.
2087 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2088 options printed with -mdebug=reg.
2090 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2091 -mquad-memory-atomic as the test for whether we have quad word
2092 atomic instructions.
2093 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2094 or -mp8-vector are used, allow byte/half-word atomic operations.
2096 * config/rs6000/sync.md (load_lockedti): Insure that the address
2097 is a proper indexed or indirect address for the lqarx instruction.
2098 On little endian systems, swap the hi/lo registers after the lqarx
2100 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2101 insure the address is valid for the lqarx instruction.
2102 (store_conditionalti): Insure that the address is a proper indexed
2103 or indirect address for the stqcrx. instruction. On little endian
2104 systems, swap the hi/lo registers before doing the stqcrx.
2106 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2107 insure the address is valid for the stqcrx. instruction.
2109 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2110 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2111 type of quad memory support is available.
2113 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2116 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2117 there is a danger of looping.
2119 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2121 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2122 force flag_ira_loop_pressure if set via command line.
2124 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2126 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2127 (ashr_simd): New builtin handling DI mode.
2128 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2129 (aarch64_sshr_simddi): New match pattern.
2130 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2131 (vshrd_n_s64): Likewise.
2132 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2134 2014-01-23 Nick Clifton <nickc@redhat.com>
2136 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2137 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2138 favour of mcu specific scripts.
2139 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2142 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2143 Alex Velenko <Alex.Velenko@arm.com>
2145 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2146 (vaddv_s16): Likewise.
2147 (vaddv_s32): Likewise.
2148 (vaddv_u8): Likewise.
2149 (vaddv_u16): Likewise.
2150 (vaddv_u32): Likewise.
2151 (vaddvq_s8): Likewise.
2152 (vaddvq_s16): Likewise.
2153 (vaddvq_s32): Likewise.
2154 (vaddvq_s64): Likewise.
2155 (vaddvq_u8): Likewise.
2156 (vaddvq_u16): Likewise.
2157 (vaddvq_u32): Likewise.
2158 (vaddvq_u64): Likewise.
2159 (vaddv_f32): Likewise.
2160 (vaddvq_f32): Likewise.
2161 (vaddvq_f64): Likewise.
2162 (vmaxv_f32): Likewise.
2163 (vmaxv_s8): Likewise.
2164 (vmaxv_s16): Likewise.
2165 (vmaxv_s32): Likewise.
2166 (vmaxv_u8): Likewise.
2167 (vmaxv_u16): Likewise.
2168 (vmaxv_u32): Likewise.
2169 (vmaxvq_f32): Likewise.
2170 (vmaxvq_f64): Likewise.
2171 (vmaxvq_s8): Likewise.
2172 (vmaxvq_s16): Likewise.
2173 (vmaxvq_s32): Likewise.
2174 (vmaxvq_u8): Likewise.
2175 (vmaxvq_u16): Likewise.
2176 (vmaxvq_u32): Likewise.
2177 (vmaxnmv_f32): Likewise.
2178 (vmaxnmvq_f32): Likewise.
2179 (vmaxnmvq_f64): Likewise.
2180 (vminv_f32): Likewise.
2181 (vminv_s8): Likewise.
2182 (vminv_s16): Likewise.
2183 (vminv_s32): Likewise.
2184 (vminv_u8): Likewise.
2185 (vminv_u16): Likewise.
2186 (vminv_u32): Likewise.
2187 (vminvq_f32): Likewise.
2188 (vminvq_f64): Likewise.
2189 (vminvq_s8): Likewise.
2190 (vminvq_s16): Likewise.
2191 (vminvq_s32): Likewise.
2192 (vminvq_u8): Likewise.
2193 (vminvq_u16): Likewise.
2194 (vminvq_u32): Likewise.
2195 (vminnmv_f32): Likewise.
2196 (vminnmvq_f32): Likewise.
2197 (vminnmvq_f64): Likewise.
2199 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2201 * config/aarch64/aarch64-simd.md
2202 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2203 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2204 (*aarch64_mul3_elt<mode>): Likewise.
2205 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2206 (*aarch64_mul3_elt_to_64v2df): Likewise.
2207 (*aarch64_mla_elt<mode>): Likewise.
2208 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2209 (*aarch64_mls_elt<mode>): Likewise.
2210 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2211 (*aarch64_fma4_elt<mode>): Likewise.
2212 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2213 (*aarch64_fma4_elt_to_64v2df): Likewise.
2214 (*aarch64_fnma4_elt<mode>): Likewise.
2215 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2216 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2217 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2218 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2219 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2220 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2221 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2222 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2223 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2225 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2227 * config/aarch64/aarch64-simd.md
2228 (aarch64_be_checked_get_lane<mode>): New define_expand.
2229 * config/aarch64/aarch64-simd-builtins.def
2230 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2231 New builtin definition.
2232 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2233 Use new safe be builtin.
2235 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2237 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2239 (aarch64_be_st1<mode>): Likewise.
2240 (aarch_ld1<VALL:mode>): Define_expand modified.
2241 (aarch_st1<VALL:mode>): Likewise.
2242 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2243 (UNSPEC_ST1): Likewise.
2245 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2247 * config/microblaze/microblaze.md: Add trap insn and attribute
2249 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2251 PR preprocessor/58580
2252 * input.h (location_get_source_line): Take an additional line_size
2254 (void diagnostics_file_cache_fini): Declare new function.
2255 * input.c (struct fcache): New type.
2256 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2257 New static constants.
2258 (diagnostic_file_cache_init, total_lines_num)
2259 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2260 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2261 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2262 (get_next_line, read_next_line, goto_next_line, read_line_num):
2263 New static function definitions.
2264 (diagnostic_file_cache_fini): New function.
2265 (location_get_source_line): Take an additional output line_len
2266 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2268 * diagnostic.c (diagnostic_finish): Call
2269 diagnostic_file_cache_fini.
2270 (adjust_line): Take an additional input parameter for the length
2271 of the line, rather than calculating it with strlen.
2272 (diagnostic_show_locus): Adjust the use of
2273 location_get_source_line and adjust_line with respect to their new
2274 signature. While displaying a line now, do not stop at the first
2275 null byte. Rather, display the zero byte as a space and keep
2276 going until we reach the size of the line.
2277 * Makefile.in: Add vec.o to OBJS-libcommon
2279 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2280 Ilya Tocar <ilya.tocar@intel.com>
2282 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2283 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2284 (__builtin_ia32_kmov16): Ditto.
2285 * config/i386/i386.md (UNSPEC_KMOV): New.
2288 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2290 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2291 (_mm512_storeu_si512): Ditto.
2293 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2296 * rtl.h (get_referenced_operands): Declare.
2297 * recog.c (get_referenced_operands): New function.
2298 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2299 operands have been referenced when recording LO_SUM references.
2301 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2303 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2305 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2307 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2308 Enable for generic and recent AMD targets.
2310 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2312 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2313 ARG_SIZE note when adjustment was eliminated.
2315 2014-01-22 Jeff Law <law@redhat.com>
2317 PR tree-optimization/59597
2318 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2319 in file. Accept new argument REGISTERING and use it to modify
2320 dump output appropriately.
2321 (register_jump_thread): Corresponding changes.
2322 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2323 thread paths involving joiner blocks. Add code to dump cancelled
2324 jump threading paths.
2326 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2328 PR rtl-optimization/59477
2329 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2330 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2332 2014-01-22 Tom Tromey <tromey@redhat.com>
2334 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2336 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2338 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2340 PR rtl-optimization/59896
2341 * lra-constraints.c (process_alt_operands): Check unused note for
2342 matched operands of insn with no output reloads.
2344 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2346 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2347 (mips_move_from_gpr_cost): Likewise.
2349 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2351 PR rtl-optimization/59858
2352 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2353 ira_class_hard_regs_num.
2354 (process_alt_operands): Increase reject for dying matched operand.
2356 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2359 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2360 smaller than size, perform several stores or loads and stores
2361 at dst + count - size to store or copy all of size bytes, rather
2362 than just last modesize bytes.
2364 2014-01-20 DJ Delorie <dj@redhat.com>
2366 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2367 that CLOBBERs are REGs before propogating their values.
2369 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2372 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2373 (cgraph_inline_failed_type): New function.
2374 * cgraph.h (DEFCIFCODE): Add type.
2375 (cgraph_inline_failed_type_t): New enum.
2376 (cgraph_inline_failed_type): New prototype.
2377 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2378 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2379 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2380 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2381 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2382 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2383 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2384 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2385 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2386 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2387 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2388 OPTIMIZATION_MISMATCH.
2389 * tree-inline.c (expand_call_inline): Emit errors during
2390 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2392 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2395 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2396 mode attribute in insn output.
2398 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2400 * output.h (output_constant): Delete.
2401 * varasm.c (output_constant): Make private.
2403 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2405 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2407 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2410 * tree.h (fold_builtin_strcat): New prototype.
2411 * builtins.c (fold_builtin_strcat): No longer static. Add len
2412 argument, if non-NULL, don't call c_strlen. Optimize
2413 directly into __builtin_memcpy instead of __builtin_strcpy.
2414 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2415 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2417 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2419 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2420 for SImode_address_operand operands, having only a REG argument.
2422 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2424 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2425 loader name using mbig-endian.
2426 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2428 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2430 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2434 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2436 * config/aarch64/aarch64-protos.h
2437 (aarch64_cannot_change_mode_class_ptr): Declare.
2438 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2439 aarch64_cannot_change_mode_class_ptr): New.
2440 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2441 backend hook aarch64_cannot_change_mode_class.
2443 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2445 * common/config/aarch64/aarch64-common.c
2446 (aarch64_handle_option): Don't handle any option order logic here.
2447 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2448 selected_cpu, warn on architecture version mismatch.
2449 (aarch64_override_options): Fix parsing order for option strings.
2451 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2452 Iain Sandoe <iain@codesourcery.com>
2455 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2456 warning. Amend comment to reflect current functionality.
2458 2014-01-20 Richard Biener <rguenther@suse.de>
2461 * builtins.c (fold_builtin_strcat): Remove case better handled
2462 by tree-ssa-strlen.c.
2464 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2466 * config/aarch64/aarch64.opt
2467 (mcpu, march, mtune): Make case-insensitive.
2469 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2472 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2473 if operands[1] is a REG or ZERO_EXTEND of a REG.
2475 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2477 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2479 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2481 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2482 long non-pic millicode calls.
2484 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2486 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2488 2014-01-19 Kito Cheng <kito@0xlab.org>
2490 * builtins.c (expand_movstr): Check movstr expand done or fail.
2492 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2493 H.J. Lu <hongjiu.lu@intel.com>
2496 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2497 to DImode for zero-extended addresses.
2499 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2501 PR rtl-optimization/57763
2502 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2503 on the new indirect jump_insn and increment LABEL_NUSES (label).
2505 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2509 * config.gcc (x86_archs): New variable.
2510 (x86_64_archs): Likewise.
2511 (x86_cpus): Likewise.
2512 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2513 --with-arch/--with-cpu= options.
2514 Support --with-arch=/--with-cpu={nehalem,westmere,
2515 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2517 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2519 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2520 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2522 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2524 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2526 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2529 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2530 clear cpp_get_options (parse_in)->warn_unused_macros for
2531 ix86_target_macros_internal with cpp_define.
2533 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2535 * jump.c (delete_related_insns): Keep (use (insn))s.
2536 * reorg.c (redundant_insn): Check for barriers too.
2538 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2540 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2542 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2544 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2545 call to $$dyncall when TARGET_LONG_CALLS is true.
2547 2014-01-17 Jeff Law <law@redhat.com>
2549 * ree.c (combine_set_extension): Temporarily disable test for
2550 changing number of hard registers.
2552 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2555 * ipa-inline-analysis.c (inline_free_summary):
2556 Do not free summary of aliases.
2558 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2561 * gimplify.c (gimplify_expr): Use create_tmp_var
2562 instead of create_tmp_var_raw. If cond doesn't have
2563 integral type, don't add the IFN_ANNOTATE builtin at all.
2565 2014-01-17 Martin Jambor <mjambor@suse.cz>
2568 * ipa-cp.c (prev_edge_clone): New variable.
2569 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2570 Also resize prev_edge_clone vector.
2571 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2572 (ipcp_edge_removal_hook): New function.
2573 (ipcp_driver): Register ipcp_edge_removal_hook.
2575 2014-01-17 Andrew Pinski <apinski@cavium.com>
2576 Steve Ellcey <sellcey@mips.com>
2579 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2582 2014-01-17 Jeff Law <law@redhat.com>
2585 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2586 so that pass_ccp runs first.
2588 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2590 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2591 (ix86_adjust_cost): Use !TARGET_XXX.
2592 (do_reorder_for_imul): Likewise.
2593 (swap_top_of_ready_list): Likewise.
2594 (ix86_sched_reorder): Likewise.
2596 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2598 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2599 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2600 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2601 (intel_memset): New. Duplicate slm_memset.
2602 (intel_cost): New. Duplicate slm_cost.
2603 (m_INTEL): New macro.
2604 (processor_target_table): Add "intel".
2605 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2606 with PROCESSOR_INTEL for "intel".
2607 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2608 PROCESSOR_SILVERMONT.
2609 (ix86_issue_rate): Likewise.
2610 (ix86_adjust_cost): Likewise.
2611 (ia32_multipass_dfa_lookahead): Likewise.
2612 (swap_top_of_ready_list): Likewise.
2613 (ix86_sched_reorder): Likewise.
2614 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2615 instead of TARGET_OPT_AGU.
2616 * config/i386/i386.h (TARGET_INTEL): New.
2617 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2618 (processor_type): Add PROCESSOR_INTEL.
2619 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2620 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2622 2014-01-17 Marek Polacek <polacek@redhat.com>
2625 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2628 2014-01-17 Richard Biener <rguenther@suse.de>
2630 PR tree-optimization/46590
2631 * opts.c (default_options_table): Add entries for
2632 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2633 all enabled at -O1 but not for -Og.
2634 * common.opt (fbranch-count-reg): Remove Init(1).
2635 (fmove-loop-invariants): Likewise.
2636 (ftree-pta): Likewise.
2638 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2640 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2641 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2642 decls to at least the GCC 4.8 used alignments.
2645 * tree-nested.c (convert_nonlocal_reference_stmt,
2646 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2647 of GIMPLE_BIND stmts, adjust associated decls.
2649 2014-01-17 Richard Biener <rguenther@suse.de>
2651 PR tree-optimization/46590
2652 * vec.h (vec<>::bseach): New member function implementing
2653 binary search according to C89 bsearch.
2654 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2655 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2656 bitmap pointer again. Make accesses_in_loop a flat array.
2657 (mem_ref_obstack): New global.
2658 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2659 (mark_ref_stored): Likewise.
2660 (ref_indep_loop_p_2): Likewise.
2661 (set_ref_stored_in_loop): New helper function.
2662 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2663 (memref_free): Adjust.
2664 (record_mem_ref_loc): Simplify.
2665 (gather_mem_refs_stmt): Adjust.
2666 (sort_locs_in_loop_postorder_cmp): New function.
2667 (analyze_memory_references): Sort accesses_in_loop after
2668 loop postorder number.
2669 (find_ref_loc_in_loop_cmp): New function.
2670 (for_all_locs_in_loop): Find relevant cluster of locs in
2671 accesses_in_loop and iterate without recursion.
2672 (execute_sm): Avoid uninit warning.
2673 (struct ref_always_accessed): Simplify.
2674 (ref_always_accessed::operator ()): Likewise.
2675 (ref_always_accessed_p): Likewise.
2676 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2677 loop postorder numbers here.
2678 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2681 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2684 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2685 on decls for which assemble_alias has been called.
2687 2014-01-17 Nick Clifton <nickc@redhat.com>
2689 * config/msp430/msp430.opt: (mcpu): New option.
2690 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2691 (msp430_option_override): Parse target_cpu. If the MCU name
2692 matches a generic string, clear target_mcu.
2693 (msp430_attr): Allow numeric interrupt values up to 63.
2694 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2695 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2697 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2699 * config/msp430/msp430.md (popm): Use %J rather than %I.
2700 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2701 (addhi_cy_i): Use immediate_operand for operand 2.
2702 * doc/invoke.texi: Document -mcpu option.
2704 2014-01-17 Richard Biener <rguenther@suse.de>
2706 PR rtl-optimization/38518
2707 * df.h (df_analyze_loop): Declare.
2708 * df-core.c: Include cfgloop.h.
2709 (df_analyze_1): Split out main part of df_analyze.
2710 (df_analyze): Adjust.
2711 (loop_inverted_post_order_compute): New function.
2712 (loop_post_order_compute): Likewise.
2713 (df_analyze_loop): New function avoiding whole-function
2715 * loop-invariant.c (find_defs): Use df_analyze_loop.
2716 (find_invariants): Adjust.
2717 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2719 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2721 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2722 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2724 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2726 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2727 traversal when removing references.
2729 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2732 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2734 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2737 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2738 pushing a reload for an autoinc when we had previously reloaded an
2739 inner part of the address.
2741 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2743 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2745 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2746 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2747 when not giving up or versioning for alias only because of
2749 (vect_analyze_data_ref_dependences): Set to true.
2750 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2752 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2753 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2757 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2760 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2761 operand 0 predicate for gathers, use a new pseudo as subtarget.
2763 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2766 * lra-constraints.c (process_alt_operands): Add printing debug info.
2767 Check absence of input/output reloads for matched operands too.
2769 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2771 PR rtl-optimization/59835
2772 * ira.c (ira_init_register_move_cost): Increase cost for
2775 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2777 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2779 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2782 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2783 non-register objects. Use gen_(high/low)part more consistently.
2786 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2789 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2790 endian support, remove tests for WORDS_BIG_ENDIAN.
2791 (p8_mfvsrd_3_<mode>): Likewise.
2792 (reload_gpr_from_vsx<mode>): Likewise.
2793 (reload_gpr_from_vsxsf): Likewise.
2794 (p8_mfvsrd_4_disf): Likewise.
2796 2014-01-16 Richard Biener <rguenther@suse.de>
2798 PR rtl-optimization/46590
2799 * lcm.c (compute_antinout_edge): Use postorder iteration.
2800 (compute_laterin): Use inverted postorder iteration.
2802 2014-01-16 Nick Clifton <nickc@redhat.com>
2805 * varasm.c (output_constant): Return the number of bytes actually
2807 (output_constructor_array_range): Update the field size with the
2808 number of bytes emitted by output_constant.
2809 (output_constructor_regular_field): Likewise. Also do not
2810 complain if the total number of bytes emitted is now greater
2811 than the expected fieldpos.
2812 * output.h (output_constant): Update prototype and descriptive comment.
2814 2014-01-16 Marek Polacek <polacek@redhat.com>
2817 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2818 it is error_mark_node.
2820 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2822 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2823 VALID_AVX256_REG_OR_OI_MODE.
2825 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2827 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2828 current procedure should be profiled.
2830 2014-01-15 Andrew Pinski <apinski@cavium.com>
2832 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2833 of moving from/to the STACK_REG register class.
2835 2014-01-15 Richard Henderson <rth@redhat.com>
2838 * reginfo.c (global_regs_decl): Globalize.
2839 * rtl.h (global_regs_decl): Declare.
2840 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2841 reserved via global_regs.
2843 2014-01-15 Teresa Johnson <tejohnson@google.com>
2845 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2847 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2849 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2850 and vmulosh rather than call gen_vec_widen_smult_*.
2851 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2852 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2853 (vec_widen_smult_even_v16qi): Likewise.
2854 (vec_widen_umult_even_v8hi): Likewise.
2855 (vec_widen_smult_even_v8hi): Likewise.
2856 (vec_widen_umult_odd_v16qi): Likewise.
2857 (vec_widen_smult_odd_v16qi): Likewise.
2858 (vec_widen_umult_odd_v8hi): Likewise.
2859 (vec_widen_smult_odd_v8hi): Likewise.
2860 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2861 vmuloub rather than call gen_vec_widen_umult_*.
2862 (vec_widen_umult_lo_v16qi): Likewise.
2863 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2864 vmulosb rather than call gen_vec_widen_smult_*.
2865 (vec_widen_smult_lo_v16qi): Likewise.
2866 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2867 rather than call gen_vec_widen_umult_*.
2868 (vec_widen_umult_lo_v8hi): Likewise.
2869 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2870 rather than call gen_vec_widen_smult_*.
2871 (vec_widen_smult_lo_v8hi): Likewise.
2873 2014-01-15 Jeff Law <law@redhat.com>
2875 PR tree-optimization/59747
2876 * ree.c (find_and_remove_re): Properly handle case where a second
2877 eliminated extension requires widening a copy created for elimination
2878 of a prior extension.
2879 (combine_set_extension): Ensure that the number of hard regs needed
2880 for a destination register does not change when we widen it.
2882 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2884 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2885 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2886 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2887 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2888 (avr-*-rtems*): Likewise.
2889 (bfin*-rtems*): Likewise.
2890 (moxie-*-rtems*): Likewise.
2891 (h8300-*-rtems*): Likewise.
2892 (i[34567]86-*-rtems*): Likewise.
2893 (lm32-*-rtems*): Likewise.
2894 (m32r-*-rtems*): Likewise.
2895 (m68k-*-rtems*): Likewise.
2896 (microblaze*-*-rtems*): Likewise.
2897 (mips*-*-rtems*): Likewise.
2898 (powerpc-*-rtems*): Likewise.
2899 (sh-*-rtems*): Likewise.
2900 (sparc-*-rtems*): Likewise.
2901 (sparc64-*-rtems*): Likewise.
2902 (v850-*-rtems*): Likewise.
2903 (m32c-*-rtems*): Likewise.
2905 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2907 PR rtl-optimization/59511
2908 * ira.c (ira_init_register_move_cost): Use memory costs for some
2909 cases of register move cost calculations.
2910 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2911 instead of BB frequency.
2912 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2913 * lra-assigns.c (find_hard_regno_for): Ditto.
2915 2014-01-15 Richard Biener <rguenther@suse.de>
2917 PR tree-optimization/59822
2918 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2919 (vectorizable_load): Use it to hoist defs of uses of invariant
2920 loads out of the loop.
2922 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2923 Kugan Vivekanandarajah <kuganv@linaro.org>
2926 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2929 2014-01-15 Richard Biener <rguenther@suse.de>
2931 PR rtl-optimization/59802
2932 * lcm.c (compute_available): Use inverted postorder to seed
2933 the initial worklist.
2935 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2938 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2939 ADDR_REGS for invalid symrefs in non-PIC code.
2941 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2944 * builtins.c (determine_block_size): Initialize *probable_max_size
2945 even if len_rtx is CONST_INT.
2947 2014-01-14 Andrew Pinski <apinski@cavium.com>
2949 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2950 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2951 (cortexa53_tunings): Likewise.
2952 (aarch64_sched_issue_rate): New function.
2953 (TARGET_SCHED_ISSUE_RATE): Define.
2955 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2957 * ira-costs.c (find_costs_and_classes): Add missed
2958 ira_init_register_move_cost_if_necessary.
2960 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2963 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2965 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2968 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2969 to indicate if type is used for function return value. Warn ABI
2970 change if the vector mode isn't available for function return value.
2971 (ix86_function_arg_advance): Pass false to type_natural_mode.
2972 (ix86_function_arg): Likewise.
2973 (ix86_gimplify_va_arg): Likewise.
2974 (function_arg_32): Don't warn ABI change.
2975 (ix86_function_value): Pass true to type_natural_mode.
2976 (ix86_return_in_memory): Likewise.
2977 (ix86_struct_value_rtx): Removed.
2978 (TARGET_STRUCT_VALUE_RTX): Likewise.
2980 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2982 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2983 converting a conditional jump into a conditional return.
2985 2014-01-14 Richard Biener <rguenther@suse.de>
2987 PR tree-optimization/58921
2988 PR tree-optimization/59006
2989 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2990 hoisting invariant stmts.
2991 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2992 invariant loads on the preheader edge if possible.
2994 2014-01-14 Joey Ye <joey.ye@arm.com>
2996 * doc/plugin.texi (Building GCC plugins): Update to C++.
2998 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3000 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3001 (_mm_rcp28_round_ss): Ditto.
3002 (_mm_rsqrt28_round_sd): Ditto.
3003 (_mm_rsqrt28_round_ss): Ditto.
3004 (_mm_rcp28_sd): Ditto.
3005 (_mm_rcp28_ss): Ditto.
3006 (_mm_rsqrt28_sd): Ditto.
3007 (_mm_rsqrt28_ss): Ditto.
3008 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3009 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3010 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3011 (IX86_BUILTIN_RCP28SD): Ditto.
3012 (IX86_BUILTIN_RCP28SS): Ditto.
3013 (IX86_BUILTIN_RSQRT28SD): Ditto.
3014 (IX86_BUILTIN_RSQRT28SS): Ditto.
3015 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3016 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3017 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3018 (ix86_expand_special_args_builtin): Expand new FTYPE.
3019 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3020 (srcp14<mode>): Make insn unary.
3021 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3022 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3023 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3024 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3025 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3026 Fix rounding: make it SAE only.
3027 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3029 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3031 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3032 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3033 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3034 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3035 (round_saeonly_mask_scalar_operand4): Ditto.
3036 (round_saeonly_mask_scalar_op3): Ditto.
3037 (round_saeonly_mask_scalar_op4): Ditto.
3039 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3041 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3042 Implement -maltivec=be for vec_insert and vec_extract.
3044 2014-01-10 DJ Delorie <dj@redhat.com>
3046 * config/msp430/msp430.md (call_internal): Don't allow memory
3047 references with SP as the base register.
3048 (call_value_internal): Likewise.
3049 * config/msp430/constraints.md (Yc): New. For memory references
3050 that don't use SP as a base register.
3052 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3053 "an integer without a # prefix"
3054 * config/msp430/msp430.md (epilogue_helper): Use it.
3056 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3059 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3060 AVX512F gather builtins.
3061 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3062 on gather decls with INTEGER_TYPE masktype.
3063 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3064 directly into the builtin rather than hoisting it before loop.
3066 PR tree-optimization/59387
3067 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3068 (scev_const_prop): If folded_casts and type has undefined overflow,
3069 use force_gimple_operand instead of force_gimple_operand_gsi and
3070 for each added stmt if it is assign with
3071 arith_code_with_undefined_signed_overflow, call
3072 rewrite_to_defined_overflow.
3073 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3074 gimple-fold.h instead.
3075 (arith_code_with_undefined_signed_overflow,
3076 rewrite_to_defined_overflow): Moved to ...
3077 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3078 rewrite_to_defined_overflow): ... here. No longer static.
3079 Include gimplify-me.h.
3080 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3081 rewrite_to_defined_overflow): New prototypes.
3083 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3085 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3087 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3089 * builtins.c (get_object_alignment_2): Minor tweak.
3090 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3092 2014-01-13 Christian Bruel <christian.bruel@st.com>
3094 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3095 optimized non constant lengths.
3097 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3100 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3101 load as __atomic_load_N if possible.
3103 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3105 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3107 (rs6000_expand_builtin): Adjust call.
3109 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3112 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3113 * config/rs6000/rs6000.c: Include target-globals.h.
3114 (rs6000_set_current_function): Instead of doing target_reinit
3115 unconditionally, use save_target_globals_default_opts and
3116 restore_target_globals.
3118 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3120 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3121 (rs6000_expand_builtin): Handle mffs and mtfsf.
3122 (rs6000_init_builtins): Define mffs and mtfsf.
3123 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3124 (rs6000_mffs): New pattern.
3125 (rs6000_mtfsf): New pattern.
3127 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3129 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3130 Start narrowing with START. Apply candidate-use pair
3131 and check overall cost in narrowing.
3132 (iv_ca_prune): Pass new argument.
3134 2014-01-10 Jeff Law <law@redhat.com>
3137 * ree.c (combine_reaching_defs): Ensure the defining statement
3138 occurs before the extension when optimizing extensions with
3139 different source and destination hard registers.
3141 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3144 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3145 vtables into the type inheritance graph.
3147 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3149 PR rtl-optimization/59754
3150 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3151 modes in the REGNO != REGNO case.
3153 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3155 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3157 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3159 PR tree-optimization/59745
3160 * tree-predcom.c (tree_predictive_commoning_loop): Call
3161 free_affine_expand_cache if giving up because components is NULL.
3163 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3164 GC in payload of target_globals struct instead of allocating them on
3165 the heap and the larger structs separately using GC.
3166 * target-globals.h (struct target_globals): Make regs, hard_regs,
3167 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3168 of GTY((skip)) and change type to void *.
3169 (reset_target_globals): Cast loads from those fields to corresponding
3172 2014-01-10 Steve Ellcey <sellcey@mips.com>
3175 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3176 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3177 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3179 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3182 * aarch64-modes.def (CC_Zmode): New flags mode.
3183 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3184 represents an equality.
3185 (aarch64_get_condition_code): Handle CC_Zmode.
3186 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3188 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3190 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3191 extraction in good case.
3193 2014-01-10 Richard Biener <rguenther@suse.de>
3195 PR tree-optimization/59374
3196 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3197 checking after SLP discovery. Mark stmts not participating
3198 in any SLP instance properly.
3200 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3202 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3203 when handling a SET rtx.
3205 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3207 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3208 (cortex-a57): Likewise.
3209 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3211 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3213 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3214 non-iwmmxt builtins.
3216 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3220 * ipa-devirt.c record_target_from_binfo): Take as argument
3221 stack of binfos and lookup matching one for virtual inheritance.
3222 (possible_polymorphic_call_targets_1): Update.
3224 2014-01-10 Huacai Chen <chenhc@lemote.com>
3226 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3227 kernel strings for Loongson-2E/2F/3A.
3229 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3232 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3233 is_gimple_call before calling gimple_call_internal_p.
3235 2014-01-09 Steve Ellcey <sellcey@mips.com>
3237 * Makefile.in (TREE_FLOW_H): Remove.
3238 (TREE_SSA_H): Add file names from tree-flow.h.
3239 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3240 * tree.h: Remove tree-flow.h reference.
3241 * hash-table.h: Remove tree-flow.h reference.
3242 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3243 reference with tree-ssa-loop.h.
3245 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3247 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3248 default element-order behavior for -maltivec.
3249 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3251 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3252 when targeting big endian, at least for now.
3253 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3255 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3258 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3259 var satisfies use_register_for_decl, just take into account type
3260 alignment, rather than decl alignment.
3262 PR tree-optimization/59622
3263 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3264 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3265 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3266 Don't devirtualize for inplace at all. For targets.length () == 1,
3267 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3269 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3271 * config/i386/i386.md (cpu): Remove the unused btver1.
3273 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3275 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3277 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3280 * tree-core.h (struct target_globals): New forward declaration.
3281 (struct tree_target_option): Add globals field.
3282 * tree.h (TREE_TARGET_GLOBALS): Define.
3283 (prepare_target_option_nodes_for_pch): New prototype.
3284 * target-globals.h (struct target_globals): Define even if
3286 * tree.c (prepare_target_option_node_for_pch,
3287 prepare_target_option_nodes_for_pch): New functions.
3288 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3289 * config/i386/i386.c: Include target-globals.h.
3290 (ix86_set_current_function): Instead of doing target_reinit
3291 unconditionally, use save_target_globals_default_opts and
3292 restore_target_globals.
3294 2014-01-09 Richard Biener <rguenther@suse.de>
3296 PR tree-optimization/59715
3297 * tree-cfg.h (split_critical_edges): Declare.
3298 * tree-cfg.c (split_critical_edges): Export.
3299 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3301 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3303 * cfgexpand.c (expand_stack_vars): Optionally disable
3304 asan stack protection.
3305 (expand_used_vars): Likewise.
3306 (partition_stack_vars): Likewise.
3307 * asan.c (asan_emit_stack_protection): Optionally disable
3308 after return stack usage.
3309 (instrument_derefs): Optionally disable memory access instrumentation.
3310 (instrument_builtin_call): Likewise.
3311 (instrument_strlen_call): Likewise.
3312 (asan_protect_global): Optionally disable global variables protection.
3313 * doc/invoke.texi: Added doc for new options.
3314 * params.def: Added new options.
3315 * params.h: Likewise.
3317 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3319 PR rtl-optimization/59724
3320 * ifcvt.c (cond_exec_process_if_block): Don't call
3321 flow_find_head_matching_sequence with 0 longest_match.
3322 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3323 non-active insns if !stop_after.
3324 (try_head_merge_bb): Revert 2014-01-07 changes.
3326 2014-01-08 Jeff Law <law@redhat.com>
3328 * ree.c (get_sub_rtx): New function, extracted from...
3329 (merge_def_and_ext): Here.
3330 (combine_reaching_defs): Use get_sub_rtx.
3332 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3334 * cgraph.h (varpool_variable_node): Do not choke on null node.
3336 2014-01-08 Catherine Moore <clm@codesourcery.com>
3338 * config/mips/mips.md (simple_return): Attempt to use JRC
3340 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3342 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3344 PR rtl-optimization/59137
3345 * reorg.c (steal_delay_list_from_target): Call update_block for
3347 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3349 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3351 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3352 two duplicate entries.
3354 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3357 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3359 * config/mips/mips.c (mips_truncated_op_cost): New function.
3360 (mips_rtx_costs): Adjust test for BADDU.
3361 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3363 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3365 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3366 (*baddu_si): ...this new pattern.
3368 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3371 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3373 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3376 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3378 (expand_expr, expand_normal): Adjust.
3379 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3380 inner_reference_p. Use inner_reference_p to expand inner references.
3381 (store_expr): Adjust.
3382 * cfgexpand.c (expand_call_stmt): Adjust.
3384 2014-01-08 Rong Xu <xur@google.com>
3386 * gcov-io.c (gcov_var): Move from gcov-io.h.
3387 (gcov_position): Ditto.
3388 (gcov_is_error): Ditto.
3389 (gcov_rewrite): Ditto.
3390 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3391 only part to libgcc/libgcov.h.
3393 2014-01-08 Marek Polacek <polacek@redhat.com>
3396 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3398 2014-01-08 Marek Polacek <polacek@redhat.com>
3401 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3403 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3405 PR rtl-optimization/59649
3406 * stor-layout.c (get_mode_bounds): For BImode return
3407 0 and STORE_FLAG_VALUE.
3409 2014-01-08 Richard Biener <rguenther@suse.de>
3412 * gimple.h (is_gimple_builtin_call): Remove.
3413 (gimple_builtin_call_types_compatible_p): New.
3414 (gimple_call_builtin_p): New overload.
3415 * gimple.c (is_gimple_builtin_call): Remove.
3416 (validate_call): Rename to ...
3417 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3419 (validate_type): New static function.
3420 (gimple_call_builtin_p): New overload and adjust.
3421 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3422 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3423 (gimple_fold_stmt_to_constant_1): Likewise.
3424 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3426 2014-01-08 Richard Biener <rguenther@suse.de>
3429 * gimplify.c (gimplify_expr): Gimplify register-register type
3430 VIEW_CONVERT_EXPRs to separate stmts.
3432 2014-01-07 Jeff Law <law@redhat.com>
3435 * ree.c (combine_set_extension): Handle case where source
3436 and destination registers in an extension insn are different.
3437 (combine_reaching_defs): Allow source and destination registers
3438 in extension to be different under limited circumstances.
3439 (add_removable_extension): Remove restriction that the
3440 source and destination registers in the extension are the same.
3441 (find_and_remove_re): Emit a copy from the extension's
3442 destination to its source after the defining insn if
3443 the source and destination registers are different.
3446 * ifcvt.c (merge_if_block): If we are merging a block with more than
3447 one successor with a block with no successors, remove any BARRIER
3448 after the second block.
3450 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3452 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3454 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3457 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3458 for 14-bit register offsets when INT14_OK_STRICT is false.
3460 2014-01-07 Roland Stigge <stigge@antcom.de>
3461 Michael Meissner <meissner@linux.vnet.ibm.com>
3464 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3465 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3467 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3469 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3472 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3474 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3475 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3476 rtx is const0_rtx or not.
3478 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3481 * target-globals.c (save_target_globals): Remove this_fn_optab
3483 * toplev.c: Include optabs.h.
3484 (target_reinit): Temporarily restore the global options if another
3485 set of options are in force.
3487 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3489 PR rtl-optimization/58668
3490 * cfgcleanup.c (flow_find_cross_jump): Don't count
3491 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3492 to determine what is counted.
3493 (flow_find_head_matching_sequence): Use active_insn_p to determine
3495 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3497 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3498 determine what is counted.
3500 PR tree-optimization/59643
3501 * tree-predcom.c (split_data_refs_to_components): If one dr is
3502 read and one write, determine_offset fails and the write isn't
3503 in the bad component, just put the read into the bad component.
3505 2014-01-07 Mike Stump <mikestump@comcast.net>
3506 Jakub Jelinek <jakub@redhat.com>
3509 * tree-core.h (struct tree_optimization_option): Change optabs
3510 type from unsigned char * to void *.
3511 * optabs.c (init_tree_optimization_optabs): Adjust
3512 TREE_OPTIMIZATION_OPTABS initialization.
3514 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3517 * config/i386/i386.h (struct machine_function): Add
3518 no_drap_save_restore field.
3519 * config/i386/i386.c (ix86_save_reg): Use
3520 !cfun->machine->no_drap_save_restore instead of
3521 crtl->stack_realign_needed.
3522 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3523 this function clears frame_pointer_needed. Set
3524 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3525 and DRAP reg is needed.
3527 2014-01-06 Marek Polacek <polacek@redhat.com>
3530 * doc/implement-c.texi: Mention that other integer types are
3531 permitted as bit-field types in strictly conforming mode.
3533 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3535 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3538 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3540 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3542 2014-01-06 Martin Jambor <mjambor@suse.cz>
3545 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3547 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3549 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3553 * var-tracking.c (add_stores): Preserve the value of the source even if
3554 we don't record the store.
3556 2014-01-06 Terry Guo <terry.guo@arm.com>
3558 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3560 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3563 * config/darwin.c (darwin_function_section): Adjust return values to
3564 correspond to optimisation changes made in r206070.
3566 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3568 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3569 from prefetch_block tune setting.
3570 (nocona_cost): Correct size of prefetch block to 64.
3572 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3574 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3575 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3576 used to save the static chain register in the computation of the offset
3577 from which the FP registers need to be restored.
3579 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3581 PR tree-optimization/59519
3582 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3583 ICE if get_current_def (current_new_name) is already non-NULL, as long
3584 as it is a phi result of some other phi in *new_exit_bb that has
3587 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3588 or vmovdqu* for misaligned_operand.
3589 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3590 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3591 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3592 aligned_mem for AVX512F masked aligned load and store builtins and for
3595 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3597 PR tree-optimization/59651
3598 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3599 Address range for negative step should be added by TYPE_SIZE_UNIT.
3601 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3603 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3604 overlapping registers also for registers other than the stack pointer.
3606 2014-01-03 Marek Polacek <polacek@redhat.com>
3609 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3612 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3615 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3618 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3619 (push splitter): Use <P:MODE_SIZE> instead of
3620 GET_MODE_SIZE (<P:MODE>mode).
3621 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3622 (mov -1, reg peephole2): Likewise.
3623 * config/i386/sse.md (*mov<mode>_internal,
3624 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3625 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3626 *<code><mode>3, *andnot<mode>3<mask_name>,
3627 <mask_codefor><code><mode>3<mask_name>): Likewise.
3628 * config/i386/subst.md (mask_mode512bit_condition,
3629 sd_mask_mode512bit_condition): Likewise.
3631 2014-01-02 Xinliang David Li <davidxl@google.com>
3633 PR tree-optimization/59303
3634 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3635 (dump_predicates): Better output format.
3636 (pred_equal_p): New function.
3637 (is_neq_relop_p): Ditto.
3638 (is_neq_zero_form_p): Ditto.
3639 (pred_expr_equal_p): Ditto.
3640 (pred_neg_p): Ditto.
3641 (simplify_pred): Ditto.
3642 (simplify_preds_2): Ditto.
3643 (simplify_preds_3): Ditto.
3644 (simplify_preds_4): Ditto.
3645 (simplify_preds): Ditto.
3647 (push_to_worklist): Ditto.
3648 (get_pred_info_from_cmp): Ditto.
3649 (is_degenerated_phi): Ditto.
3650 (normalize_one_pred_1): Ditto.
3651 (normalize_one_pred): Ditto.
3652 (normalize_one_pred_chain): Ditto.
3653 (normalize_preds): Ditto.
3654 (normalize_cond_1): Remove function.
3655 (normalize_cond): Ditto.
3656 (is_gcond_subset_of): Ditto.
3657 (is_subset_of_any): Ditto.
3658 (is_or_set_subset_of): Ditto.
3659 (is_and_set_subset_of): Ditto.
3660 (is_norm_cond_subset_of): Ditto.
3661 (pred_chain_length_cmp): Ditto.
3662 (convert_control_dep_chain_into_preds): Type change.
3663 (find_predicates): Ditto.
3664 (find_def_preds): Ditto.
3665 (destroy_predicates_vecs): Ditto.
3666 (find_matching_predicates_in_rest_chains): Ditto.
3667 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3668 (is_pred_expr_subset): Ditto.
3669 (is_pred_chain_subset_of): Ditto.
3670 (is_included_in): Ditto.
3671 (is_superset_of): Ditto.
3673 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3675 Update copyright years.
3677 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3679 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3680 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3681 config/arc/arc.md, config/arc/arc.opt,
3682 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3683 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3684 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3685 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3686 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3687 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3688 vtable-verify.c, vtable-verify.h: Use the standard form for the
3691 2014-01-02 Tobias Burnus <burnus@net-b.de>
3693 * gcc.c (process_command): Update copyright notice dates.
3694 * gcov-dump.c: Ditto.
3696 * doc/cpp.texi: Bump @copying's copyright year.
3697 * doc/cppinternals.texi: Ditto.
3698 * doc/gcc.texi: Ditto.
3699 * doc/gccint.texi: Ditto.
3700 * doc/gcov.texi: Ditto.
3701 * doc/install.texi: Ditto.
3702 * doc/invoke.texi: Ditto.
3704 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3706 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3708 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3710 * config/i386/sse.md (*mov<mode>_internal): Guard
3711 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3713 PR rtl-optimization/59647
3714 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3715 new_rtx into UNSIGNED_FLOAT rtxes.
3717 Copyright (C) 2014 Free Software Foundation, Inc.
3719 Copying and distribution of this file, with or without modification,
3720 are permitted in any medium without royalty provided the copyright
3721 notice and this notice are preserved.