1 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
3 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
4 * ipa-prop.c (ipa_set_jf_known_type): Return early when
6 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
8 (detect_type_change): Return true when giving up early.
9 (compute_complex_assign_jump_func): Fix type parameter of
11 (compute_complex_ancestor_jump_func): Likewise.
12 (update_jump_functions_after_inlining): Fix updating of
14 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type
17 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
19 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
20 inline clones when edge disappears.
22 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
25 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
26 Split 64-bit moves into 2 patterns. Do not allow the use of
27 direct move for TDmode in little endian, since the decimal value
28 has little endian bytes within a word, but the 64-bit pieces are
29 ordered in a big endian fashion, and normal subreg's of TDmode are
31 (mov<mode>_64bit_dm): Likewise.
32 (movtd_64bit_nodm): Likewise.
34 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
36 PR tree-optimization/60174
37 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
38 statement of an SSA_NAME that occurs in an abnormal PHI node.
40 2014-02-18 Jakub Jelinek <jakub@redhat.com>
43 * final.c (SEEN_BB): Remove.
44 (SEEN_NOTE, SEEN_EMITTED): Renumber.
45 (final_scan_insn): Don't force_source_line on second
46 NOTE_INSN_BASIC_BLOCK.
48 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
51 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
52 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
53 (type_natural_mode): Warn ABI change when %zmm register is not
54 available for AVX512F vector value passing.
56 2014-02-18 Kai Tietz <ktietz@redhat.com>
59 * config/i386/i386.c (ix86_expand_prologue): Use value in
60 rax register as displacement when restoring %r10 or %rax.
61 Fix wrong offset when restoring both registers.
63 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
65 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
66 assertion with conditional return.
68 2014-02-18 Jakub Jelinek <jakub@redhat.com>
69 Uros Bizjak <ubizjak@gmail.com>
72 * config/i386/driver-i386.c (host_detect_local_cpu): If
73 YMM state is not saved by the OS, also clear has_f16c. Move
74 CPUID 0x80000001 handling before YMM state saving checking.
76 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
78 PR rtl-optimization/58960
79 * haifa-sched.c (alloc_global_sched_pressure_data): New,
81 (sched_init): ... here.
82 (free_global_sched_pressure_data): New, factored out from ...
83 (sched_finish): ... here.
84 * sched-int.h (free_global_sched_pressure_data): Declare.
85 * sched-rgn.c (nr_regions_initial): New static global.
86 (haifa_find_rgns): Initialize it.
87 (schedule_region): Disable sched-pressure for the newly
90 2014-02-17 Richard Biener <rguenther@suse.de>
92 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
93 release SSA defs of pattern stmts.
95 2014-02-17 Richard Biener <rguenther@suse.de>
97 * tree-inline.c (expand_call_inline): Release the virtual
98 operand defined by the call we are about to inline.
100 2014-02-17 Richard Biener <rguenther@suse.de>
102 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
104 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
105 Ilya Tocar <ilya.tocar@intel.com>
107 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
108 arguments order in builtin.
109 (_mm512_permutexvar_epi64): Ditto.
110 (_mm512_mask_permutexvar_epi64): Ditto
111 (_mm512_maskz_permutexvar_epi32): Ditto
112 (_mm512_permutexvar_epi32): Ditto
113 (_mm512_mask_permutexvar_epi32): Ditto
115 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
117 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
118 (p8_vmrgow): Likewise.
120 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
122 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
125 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
128 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
129 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
130 into 64-bit and 32-bit moves. On 64-bit moves, add support for
131 using direct move instructions on ISA 2.07. Also adjust
132 instruction length for 64-bit.
133 (mov<mode>_64bit, TFmode/TDmode): Likewise.
134 (mov<mode>_32bit, TFmode/TDmode): Likewise.
136 2014-02-15 Alan Modra <amodra@gmail.com>
140 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
141 find_replacement on parts of insn rtl that might be reloaded.
143 2014-02-15 Richard Biener <rguenther@suse.de>
145 PR tree-optimization/60183
146 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
147 (tree_ssa_phiprop): Calculate and free post-dominators.
149 2014-02-14 Jeff Law <law@redhat.com>
151 PR rtl-optimization/60131
152 * ree.c (get_extended_src_reg): New function.
153 (combine_reaching_defs): Use it rather than assuming location of REG.
154 (find_and_remove_re): Verify first operand of extension is
155 a REG before adding the insns to the copy list.
157 2014-02-14 Roland McGrath <mcgrathr@google.com>
159 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
160 * configure: Regenerated.
161 * config.in: Regenerated.
162 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
163 instead of ASM_SHORT.
165 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
166 Richard Earnshaw <rearnsha@arm.com>
168 PR rtl-optimization/59535
169 * lra-constraints.c (process_alt_operands): Encourage alternative
170 when unassigned pseudo class is superset of the alternative class.
171 (inherit_reload_reg): Don't inherit when optimizing for code size.
172 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
173 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
174 modes not less than 4 for Thumb1.
176 2014-02-14 Kyle McMartin <kyle@redhat.com>
179 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
181 2014-02-14 Richard Biener <rguenther@suse.de>
183 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
184 (get_frame_arg): Drop the assert with langhook types_compatible_p.
185 Do not strip INDIRECT_REFs.
187 2014-02-14 Richard Biener <rguenther@suse.de>
190 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
191 DECL_FUNCTION_SPECIFIC_TARGET.
192 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
193 * tree-streamer-out.c (pack_ts_target_option): Remove.
194 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
195 (write_ts_function_decl_tree_pointers): Do not stream
196 DECL_FUNCTION_SPECIFIC_TARGET.
197 * tree-streamer-in.c (unpack_ts_target_option): Remove.
198 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
199 (lto_input_ts_function_decl_tree_pointers): Do not stream
200 DECL_FUNCTION_SPECIFIC_TARGET.
202 2014-02-14 Jakub Jelinek <jakub@redhat.com>
204 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
205 (get_initial_def_for_induction, vectorizable_induction): Ignore
206 debug stmts when looking for exit_phi.
207 (vectorizable_live_operation): Fix up condition.
209 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
211 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
212 nreverse() because it changes the content of original tree list.
214 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
216 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
217 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
219 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
221 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
222 GNU coding standards.
224 2014-02-13 Jakub Jelinek <jakub@redhat.com>
227 * dwarf2out.c (gen_subprogram_die): Don't call
228 add_calling_convention_attribute if subr_die is old_die.
230 2014-02-13 Sharad Singhai <singhai@google.com>
232 * doc/optinfo.texi: Fix order of nodes.
234 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
236 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
237 operands[2], not operands[3].
239 2014-02-13 Richard Biener <rguenther@suse.de>
242 * doc/install.texi (ISL): Update recommended version to 0.12.2,
243 mention the possibility of an in-tree build.
244 (CLooG): Update recommended version to 0.18.1, mention the
245 possibility of an in-tree build and clarify that the ISL
246 bundled with CLooG does not work.
248 2014-02-13 Jakub Jelinek <jakub@redhat.com>
251 * expr.c (compress_float_constant): If x is a hard register,
252 extend into a pseudo and then move to x.
254 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
256 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
257 caused by bad second argument to warning_at() with -mhotpatch and
258 nested functions (e.g. with gfortran).
260 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
262 * opts.c (option_name): Remove "enabled by default" rider.
264 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
266 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
268 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
269 Uros Bizjak <ubizjak@gmail.com>
272 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
273 * configure: Regenerated.
275 2014-02-12 Richard Biener <rguenther@suse.de>
277 * vec.c (vec_prefix::calculate_allocation): Move as
278 inline variant to vec.h.
279 (vec_prefix::calculate_allocation_1): New out-of-line version.
280 * vec.h (vec_prefix::calculate_allocation_1): Declare.
281 (vec_prefix::m_has_auto_buf): Rename to ...
282 (vec_prefix::m_using_auto_storage): ... this.
283 (vec_prefix::calculate_allocation): Inline the easy cases
284 and dispatch to calculate_allocation_1 which doesn't need the
286 (va_heap::reserve): Use gcc_checking_assert.
287 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
288 m_using_auto_storage.
289 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
291 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
292 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
293 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
295 2014-02-12 Richard Biener <rguenther@suse.de>
297 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
298 when we found a dependence.
300 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
302 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
304 (maybe_fold_stmt): ... into this new function.
305 * omp-low.c (lower_omp): Update comment.
307 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
310 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
313 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
315 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
316 identifiers in comments.
317 (cortexa53_extra_costs): Likewise.
318 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
319 (cortexa7_extra_costs): Likewise.
320 (cortexa12_extra_costs): Likewise.
321 (cortexa15_extra_costs): Likewise.
322 (v7m_extra_costs): Likewise.
324 2014-02-12 Richard Biener <rguenther@suse.de>
327 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
328 of posix_memalign being successful.
329 (lower_stmt): Restrict lowering of posix_memalign to when
330 -ftree-bit-ccp is enabled.
332 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
334 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
336 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
338 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
340 PR rtl-optimization/60116
341 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
342 other_insn once the combination has been validated.
344 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
347 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
349 * ipa-devirt.c: Include demangle.h
350 (odr_violation_reported): New static variable.
351 (add_type_duplicate): Update odr_violations.
352 (maybe_record_node): Add completep parameter; update it.
353 (record_target_from_binfo): Add COMPLETEP parameter;
355 (possible_polymorphic_call_targets_1): Likewise.
356 (struct polymorphic_call_target_d): Add nonconstruction_targets;
357 rename FINAL to COMPLETE.
358 (record_targets_from_bases): Sanity check we found the binfo;
359 fix COMPLETEP updating.
360 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
361 parameter, fix computing of COMPLETEP.
362 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
363 at LTO time do demangling.
364 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
365 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
367 (gimple_get_virt_method_for_binfo): Likewise.
368 * gimple-fold.h (gimple_get_virt_method_for_binfo,
369 gimple_get_virt_method_for_vtable): Update prototypes.
371 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
374 * genautomata.c (add_presence_absence): Fix typo with
375 {final_}presence_list.
377 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
380 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
381 for VSX/Altivec vectors that land in GPR registers.
383 2014-02-11 Richard Henderson <rth@redhat.com>
384 Jakub Jelinek <jakub@redhat.com>
387 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
388 around drhs if type conversion to lacc->type is not useless.
390 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
392 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
394 (cortex-a57.cortex-a53): Likewise.
395 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
397 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
399 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
402 2014-02-11 Renlin Li <Renlin.Li@arm.com>
404 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
405 add_options_for_arm_vfp3.
407 2014-02-11 Jeff Law <law@redhat.com>
410 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
411 object with an undesirable mode.
413 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
416 * config/i386/sol2-9.h: New file.
417 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
418 *-*-solaris2.9*): Use it.
420 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
422 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
423 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
425 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
427 * config/microblaze/microblaze.c: Extend mcpu version format
429 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
431 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
433 2014-02-10 Richard Henderson <rth@redhat.com>
436 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
437 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
438 ms-abi vs -mno-accumulate-outgoing-args.
439 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
440 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
443 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
446 * cfgexpand.c (expand_asm_operands): Attach source location to
447 ASM_INPUT rtx objects.
448 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
450 2014-02-10 Nick Clifton <nickc@redhat.com>
452 * config/mn10300/mn10300.c (popcount): New function.
453 (mn10300_expand_prologue): Include saved registers in stack usage
456 2014-02-10 Jeff Law <law@redhat.com>
459 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
460 when changing the SET_DEST of a prior insn to avoid an input reload.
462 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
464 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
465 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
466 -mcall-openbsd, or -mcall-linux.
467 (CC1_ENDIAN_BIG_SPEC): Remove.
468 (CC1_ENDIAN_LITTLE_SPEC): Remove.
469 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
470 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
471 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
472 and %cc1_endian_default.
473 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
475 2014-02-10 Richard Biener <rguenther@suse.de>
477 PR tree-optimization/60115
478 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
479 MEM_REF handling. Properly verify that the accesses are not
480 out of the objects bound.
482 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
484 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
487 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
489 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
490 proper constants and fix formatting.
491 (possible_polymorphic_call_targets): Fix formatting.
493 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
494 Ilya Tocar <ilya.tocar@intel.com>
496 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
497 (_mm512_loadu_epi32): Renamed into...
498 (_mm512_loadu_si512): This.
499 (_mm512_storeu_epi32): Renamed into...
500 (_mm512_storeu_si512): This.
501 (_mm512_maskz_ceil_ps): Removed.
502 (_mm512_maskz_ceil_pd): Ditto.
503 (_mm512_maskz_floor_ps): Ditto.
504 (_mm512_maskz_floor_pd): Ditto.
505 (_mm512_floor_round_ps): Ditto.
506 (_mm512_floor_round_pd): Ditto.
507 (_mm512_ceil_round_ps): Ditto.
508 (_mm512_ceil_round_pd): Ditto.
509 (_mm512_mask_floor_round_ps): Ditto.
510 (_mm512_mask_floor_round_pd): Ditto.
511 (_mm512_mask_ceil_round_ps): Ditto.
512 (_mm512_mask_ceil_round_pd): Ditto.
513 (_mm512_maskz_floor_round_ps): Ditto.
514 (_mm512_maskz_floor_round_pd): Ditto.
515 (_mm512_maskz_ceil_round_ps): Ditto.
516 (_mm512_maskz_ceil_round_pd): Ditto.
517 (_mm512_expand_pd): Ditto.
518 (_mm512_expand_ps): Ditto.
519 * config/i386/i386.c (ix86_builtins): Remove
520 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
522 * config/i386/predicates.md (const1256_operand): New.
523 (const_1_to_2_operand): Ditto.
524 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
525 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
526 (*avx512pf_gatherpf<mode>sf): Ditto.
527 (avx512pf_gatherpf<mode>df): Ditto.
528 (*avx512pf_gatherpf<mode>df_mask): Ditto.
529 (*avx512pf_gatherpf<mode>df): Ditto.
530 (avx512pf_scatterpf<mode>sf): Ditto.
531 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
532 (*avx512pf_scatterpf<mode>sf): Ditto.
533 (avx512pf_scatterpf<mode>df): Ditto.
534 (*avx512pf_scatterpf<mode>df_mask): Ditto.
535 (*avx512pf_scatterpf<mode>df): Ditto.
536 (avx512f_expand<mode>): Removed.
537 (<shift_insn><mode>3<mask_name>): Change predicate type.
539 2014-02-08 Jakub Jelinek <jakub@redhat.com>
541 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
542 not at the end of datarefs vector use ordered_remove to avoid
543 reordering datarefs vector.
546 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
547 mark local addressable non-static vars as GOVD_PRIVATE
548 instead of GOVD_LOCAL.
549 * omp-low.c (lower_omp_for): Move gimple_bind_vars
550 and BLOCK_VARS of gimple_bind_block to new_stmt rather
554 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
555 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
556 assume_aligned or alloc_align attributes.
557 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
558 arguments. Handle also assume_aligned and alloc_align attributes.
559 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
560 calls to functions with assume_aligned or alloc_align attributes.
561 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
563 2014-02-08 Terry Guo <terry.guo@arm.com>
565 * doc/invoke.texi: Document ARM -march=armv7e-m.
567 2014-02-08 Jakub Jelinek <jakub@redhat.com>
569 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
570 flag on __cilkrts_rethrow builtin.
573 * ipa-cp.c (determine_versionability): Fail at -O0
574 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
575 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
578 2014-02-04 Jakub Jelinek <jakub@redhat.com>
581 * tree-inline.c (copy_forbidden): Fail for
582 __attribute__((optimize (0))) functions.
584 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
586 * varpool.c: Include pointer-set.h.
587 (varpool_remove_unreferenced_decls): Variables in other partitions
588 will not be output; be however careful to not lose information
591 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
593 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
594 lookup in the vtable constructor.
596 2014-02-07 Jeff Law <law@redhat.com>
599 * config/m68k/m68k.md (ashldi_extsi): Turn into a
600 define_insn_and_split.
602 * ipa-inline.c (inline_small_functions): Fix typos.
604 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
606 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
607 (s390_can_use_return_insn): Declare.
608 * config/s390/s390.h (EPILOGUE_USES): Define.
609 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
611 (s390_chunkify_start): Handle return JUMP_LABELs.
612 (s390_early_mach): Emit a main_pool instruction on the entry edge.
613 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
614 (s390_can_use_return_insn): New functions.
615 (s390_fix_long_loop_prediction): Handle conditional returns.
616 (TARGET_SET_UP_BY_PROLOGUE): Define.
617 * config/s390/s390.md (ANY_RETURN): New code iterator.
618 (*creturn, *csimple_return, return, simple_return): New patterns.
620 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
622 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
623 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
624 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
625 REG_CFA_RESTORE list when deciding not to restore a register.
627 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
629 * config/s390/s390.c: Include tree-pass.h and context.h.
630 (s390_early_mach): New function, split out from...
631 (s390_emit_prologue): ...here.
632 (pass_data_s390_early_mach): New pass structure.
633 (pass_s390_early_mach): New class.
634 (s390_option_override): Create and register early_mach pass.
637 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
639 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
640 to match for the exit block.
642 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
644 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
645 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
646 Reject misaligned operands.
648 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
650 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
652 2014-02-07 Richard Biener <rguenther@suse.de>
655 * gimple-low.c (lower_builtin_posix_memalign): New function.
656 (lower_stmt): Call it to lower posix_memalign in a way
657 to make alignment info accessible.
659 2014-02-07 Jakub Jelinek <jakub@redhat.com>
662 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
663 __builtin_setjmp_receiver.
665 2014-02-07 Richard Biener <rguenther@suse.de>
668 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
669 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
670 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
671 Handle BUILT_IN_POSIX_MEMALIGN.
672 (find_func_clobbers): Likewise.
673 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
674 (call_may_clobber_ref_p_1): Likewise.
676 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
679 * ipa-devirt.c (record_target_from_binfo): Remove overactive
682 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
685 * lto-cgraph.c (lto_output_node): Use
686 symtab_get_symbol_partitioning_class.
687 (lto_output_varpool_node): likewise.
688 (symtab_get_symbol_partitioning_class): Move here from
690 * cgraph.h (symbol_partitioning_class): Likewise.
691 (symtab_get_symbol_partitioning_class): Declare.
693 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
695 * ggc.h (ggc_internal_cleared_alloc): New macro.
696 * vec.h (vec_safe_copy): Handle memory stats.
697 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
698 * target-globals.c (save_target_globals): Likewise.
700 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
703 * expr.c (emit_move_resolve_push): Export; be bit more selective
704 on when to clear alias set.
705 * expr.h (emit_move_resolve_push): Declare.
706 * function.h (struct function): Add tail_call_marked.
707 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
708 * config/i386/i386-protos.h (ix86_expand_push): Remove.
709 * config/i386/i386.md (TImode move expander): De not call
711 (FP push expanders): Preserve memory attributes.
712 * config/i386/sse.md (push<mode>1): Remove.
713 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
714 (ix86_expand_push): Remove.
715 * config/i386/mmx.md (push<mode>1): Remove.
717 2014-02-06 Jakub Jelinek <jakub@redhat.com>
719 PR rtl-optimization/60030
720 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
721 lopart with paradoxical subreg before shifting it up by hprec.
723 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
725 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
726 Remove extra newline at end of file.
727 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
728 (arm_issue_rate): Handle cortexa57.
729 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
730 (cortex-a57.cortex-a53): Likewise.
732 2014-02-06 Jakub Jelinek <jakub@redhat.com>
735 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
736 don't record in REG_FRAME_RELATED_EXPR registers not set in that
738 (arm_expand_prologue): Adjust all callers.
739 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
740 info, registers also at the lowest numbered registers side. Use
741 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
745 * var-tracking.c (adjust_mems): Before adding a SET to
746 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
748 2014-02-06 Alan Modra <amodra@gmail.com>
751 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
752 change SDmode to DDmode when lra_in_progress.
754 2014-02-06 Jakub Jelinek <jakub@redhat.com>
757 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
758 free_data_ref on the dr first, and before goto again also set dr
759 to the next dr. For simd_lane_access, free old datarefs[i] before
760 overwriting it. For get_vectype_for_scalar_type failure, don't
761 free_data_ref if simd_lane_access.
763 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
766 * tree.h (opts_for_fn): New inline function.
767 (opt_for_fn): Define.
768 * config/i386/i386.c (ix86_function_regparm): Use
769 opt_for_fn (decl, optimize) instead of optimize.
771 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
773 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
774 for SYMBOL_REF in large memory model.
776 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
778 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
780 (cortex-a57): Likewise.
781 (cortex-a57.cortex-a53): Likewise.
783 2014-02-06 Yury Gribov <y.gribov@samsung.com>
784 Kugan Vivekanandarajah <kuganv@linaro.org>
786 * config/arm/arm.c (arm_vector_alignment_reachable): Check
788 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
790 2014-02-06 Richard Biener <rguenther@suse.de>
792 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
793 set_loop_copy and initialize_original_copy_tables.
795 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
797 * config/aarch64/aarch64-simd.md
798 (aarch64_ashr_simddi): Change QI to SI.
800 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
801 Jakub Jelinek <jakub@redhat.com>
804 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
807 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
809 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
810 CODE_FOR_altivec_vpku[hw]um to
811 CODE_FOR_altivec_vpku[hw]um_direct.
812 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
813 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
814 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
815 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
817 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
819 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
820 generation for -maltivec=be.
821 (altivec_vsumsws): Simplify redundant test.
823 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
825 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
826 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
827 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
828 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
830 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
832 (altivec_vpks<VI_char>ss): Likewise.
833 (altivec_vpks<VI_char>us): Likewise.
834 (altivec_vpku<VI_char>us): Likewise.
835 (altivec_vpku<VI_char>um): Likewise.
836 (altivec_vpku<VI_char>um_direct): New (copy of
837 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
839 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
840 target is little endian and -maltivec=be is not specified.
841 (*altivec_vupkhs<VU_char>_direct): New (copy of
842 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
843 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
844 target is little endian and -maltivec=be is not specified.
845 (*altivec_vupkls<VU_char>_direct): New (copy of
846 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
847 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
848 little endian and -maltivec=be is not specified.
849 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
850 little endian and -maltivec=be is not specified.
852 2014-02-05 Richard Henderson <rth@redhat.com>
855 * combine-stack-adj.c: Revert r206943.
856 * sched-int.h (struct deps_desc): Add last_args_size.
857 * sched-deps.c (init_deps): Initialize it.
858 (sched_analyze_insn): Add OUTPUT dependencies between insns that
859 contain REG_ARGS_SIZE notes.
861 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
863 * lto-cgraph.c (asm_nodes_output): Make global.
864 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
865 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
866 (driver_handle_option): Handle OPT_fwpa.
868 2014-02-05 Jakub Jelinek <jakub@redhat.com>
871 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
872 a comment typo and formatting issue. If odr_hash hasn't been
873 created, return vNULL and set *completep to false.
876 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
877 bb with no successors.
879 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
882 * doc/invoke.texi (-march): Clarify documentation for ARM.
886 2014-02-05 Richard Biener <rguenther@suse.de>
888 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
889 when not vectorizing because of too many alias checks.
890 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
891 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
893 2014-02-05 Nick Clifton <nickc@redhat.com>
895 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
896 accept extended registers in any mode when compiling for the MN10300.
898 2014-02-05 Yury Gribov <y.gribov@samsung.com>
900 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
901 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
902 sanitization attributes.
903 (can_inline_edge_p): Likewise.
904 (sanitize_attrs_match_for_inline_p): New function.
906 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
908 * ipa-prop.c (detect_type_change): Shor circuit testing of
909 type changes on THIS pointer.
911 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
914 * config/pa/pa.c (legitimize_tls_address): Return original address
915 if not passed a SYMBOL_REF rtx.
916 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
918 (pa_emit_move_sequence): Simplify TLS source operands.
919 (pa_legitimate_constant_p): Reject all TLS constants.
920 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
921 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
923 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
925 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
926 groups when we know they are controlled by LTO.
927 * varasm.c (default_binds_local_p_1): If object is in other partition,
928 it will be resolved locally.
930 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
932 * config/host-linux.c (linux_gt_pch_use_address): Don't
933 use SSIZE_MAX because it is not always defined.
935 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
938 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
939 threshold for pseudo splitting.
940 (update_ebb_live_info): Process call argument hard registers and
941 hard registers from insn definition too.
942 (max_small_class_regs_num): New constant.
943 (inherit_in_ebb): Update live hard regs through EBBs. Update
944 reloads_num only for small register classes. Don't split for
947 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
950 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
953 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
955 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
958 2014-02-04 Marek Polacek <polacek@redhat.com>
960 * gdbinit.in (pel): Define.
962 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
964 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
967 2014-02-04 Richard Biener <rguenther@suse.de>
970 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
971 in function context local.
972 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
973 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
974 similar to LTO_imported_decl_ref.
976 2014-02-04 Jakub Jelinek <jakub@redhat.com>
978 PR tree-optimization/60002
979 * cgraphclones.c (build_function_decl_skip_args): Clear
982 PR tree-optimization/60023
983 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
984 false to gsi_replace.
985 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
986 has been in some EH region and vec_stmt could throw, add
987 vec_stmt into the same EH region.
988 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
989 has no lhs, ignore it.
990 * internal-fn.c (expand_MASK_LOAD): Likewise.
993 * tree-inline.c (copy_forbidden): Fail for
994 __attribute__((optimize (0))) functions.
997 * omp-low.c (simd_clone_struct_copy): If from->inbranch
998 is set, copy one less argument.
999 (expand_simd_clones): Don't subtract clone_info->inbranch
1000 from simd_clone_struct_alloc argument.
1002 PR rtl-optimization/57915
1003 * recog.c (simplify_while_replacing): If all unary/binary/relational
1004 operation arguments are constant, attempt to simplify those.
1007 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1008 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1010 2014-02-04 Richard Biener <rguenther@suse.de>
1012 PR tree-optimization/60012
1013 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1014 TBAA disambiguation to all DDRs.
1016 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1019 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1020 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1022 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1025 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1027 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1029 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1030 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1032 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1035 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1036 to figure out targets of polymorphic calls with known decl.
1037 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1038 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1039 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1040 (get_polymorphic_call_info): ... here.
1041 (get_polymorphic_call_info_from_invariant): New function.
1043 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1045 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1046 lookup via vtable pointer; check for type consistency
1047 and turn inconsitent facts into UNREACHABLE.
1048 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1049 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1050 type inconsistent querries; return UNREACHABLE instead.
1052 2014-02-03 Richard Henderson <rth@twiddle.net>
1055 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1056 already processed this node.
1057 (normalize_one_pred_1): Pass along mark_set.
1058 (normalize_one_pred): Create and destroy a pointer_set_t.
1059 (normalize_one_pred_chain): Likewise.
1061 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1063 PR gcov-profile/58602
1064 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1066 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1069 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1070 -fno-devirtualize; try to devirtualize by the knowledge of
1071 virtual table pointer given by aggregate propagation.
1072 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1073 (ipa_print_node_jump_functions): Dump also offset that
1074 is relevant for polymorphic calls.
1075 (determine_known_aggregate_parts): Add arg_type parameter; use it
1076 instead of determining the type from pointer type.
1077 (ipa_compute_jump_functions_for_edge): Update call of
1078 determine_known_aggregate_parts.
1079 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1080 (gimple_get_virt_method_for_binfo): ... here; simplify using
1081 vtable_pointer_value_to_vtable.
1082 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1083 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1084 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1085 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1087 (vtable_pointer_value_to_binfo): ... here.
1088 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1090 2014-02-03 Teresa Johnson <tejohnson@google.com>
1092 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1093 redef of outer loop index variable.
1095 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1099 * doc/extend.texi (Function Attributes): Typo.
1101 2014-02-03 Cong Hou <congh@google.com>
1103 PR tree-optimization/60000
1104 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1105 if the vectorized statement is a store. A store statement can only
1106 appear at the end of pattern statements.
1108 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1110 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1111 (ix86_option_override_internal): Default long double to 64-bit for
1112 32-bit Bionic and to 128-bit for 64-bit Bionic.
1114 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1115 TARGET_LONG_DOUBLE_128 is true.
1116 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1118 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1119 (mlong-double-64): Negate -mlong-double-128.
1120 (mlong-double-128): New option.
1122 * config/i386/i386-c.c (ix86_target_macros): Define
1123 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1125 * doc/invoke.texi: Document -mlong-double-128.
1127 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1129 PR rtl-optimization/60024
1130 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1132 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1134 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1136 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1138 PR rtl-optimization/57662
1139 * sel-sched.c (code_motion_path_driver): Do not mark already not
1140 existing blocks in the visiting bitmap.
1142 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1144 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1145 on the insn being emitted.
1147 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1148 Will Deacon <will.deacon@arm.com>
1150 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1152 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1154 * config/arm/arm-tables.opt: Regenerate.
1156 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1158 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1159 for vector types other than V16QImode.
1160 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1161 define_expand, and call altivec_expand_vec_perm_le when producing
1162 code with little endian element order.
1163 (*altivec_vperm_<mode>_internal): New insn having previous
1164 behavior of altivec_vperm_<mode>.
1165 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1166 altivec_expand_vec_perm_le when producing code with little endian
1168 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1169 behavior of altivec_vperm_<mode>_uns.
1171 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1173 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1174 (altivec_vsumsws): Add handling for -maltivec=be with a little
1176 (altivec_vsumsws_direct): New.
1177 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1178 gen_altivec_vsumsws.
1180 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1182 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1183 vtable_pointer_value_to_binfo): New functions.
1184 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1185 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1187 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1189 * config/nios2/nios2.md (load_got_register): Initialize GOT
1190 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1191 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1193 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1195 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1196 preserverd by passthrough, do not propagate the type.
1198 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1200 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1201 (mips_atomic_assign_expand_fenv): New function.
1202 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1204 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1206 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1207 (__builtin_mips_set_fcsr): Likewise.
1208 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1209 MIPS_USI_FTYPE_VOID.
1210 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1211 (mips16_expand_set_fcsr): Likewise.
1212 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1213 (mips16_set_fcsr_stub): Likewise.
1214 (mips16_get_fcsr_one_only_stub): New class.
1215 (mips16_set_fcsr_one_only_stub): Likewise.
1216 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1217 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1218 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1219 (hard_float): New availability predicate.
1220 (mips_builtins): Add get_fcsr and set_fcsr.
1221 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1222 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1223 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1224 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1225 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1228 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1230 * config/mips/mips.c (mips_one_only_stub): New class.
1231 (mips_need_mips16_rdhwr_p): Replace with...
1232 (mips16_rdhwr_stub): ...this new variable.
1233 (mips16_stub_call_address): New function.
1234 (mips16_rdhwr_one_only_stub): New class.
1235 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1236 (mips_output_mips16_rdhwr): Delete.
1237 (mips_finish_stub): New function.
1238 (mips_code_end): Use it to handle rdhwr stubs.
1240 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1243 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1244 when calculating size of integer atomic types.
1246 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1248 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1250 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1252 PR tree-optimization/60003
1253 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1254 * profile.c (branch_prob): Use gimple_call_builtin_p
1255 to check for BUILT_IN_SETJMP_RECEIVER.
1256 * tree-inline.c (copy_bb): Call notice_special_calls.
1258 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1261 * lra-constraints.c (process_alt_operands): Update reload_sum only
1264 2014-01-31 Richard Henderson <rth@redhat.com>
1267 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1268 until after else_eh is processed.
1270 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1272 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1273 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1274 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1275 in smmintrin.h, remove them.
1276 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1277 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1278 * config/i386/i386.md (ROUND_SAE): Fix value.
1279 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1280 (const48_operand): New.
1281 * config/i386/subst.md (round), (round_expand): Use
1282 const_4_or_8_to_11_operand.
1283 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1285 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1287 * config/i386/constraints.md (Yk): Swap meaning with k.
1288 * config/i386/i386.md (movhi_internal): Change Yk to k.
1289 (movqi_internal): Ditto.
1290 (*k<logic><mode>): Ditto.
1293 (kandn<mode>): Ditto.
1294 (*<code>hi_1): Ditto.
1295 (*<code>qi_1): Ditto.
1296 (kxnor<mode>): Ditto.
1297 (kortestzhi): Ditto.
1298 (kortestchi): Ditto.
1300 (*one_cmplhi2_1): Ditto.
1301 (*one_cmplqi2_1): Ditto.
1302 * config/i386/sse.md (): Change k to Yk.
1303 (avx512f_load<mode>_mask): Ditto.
1304 (avx512f_blendm<mode>): Ditto.
1305 (avx512f_store<mode>_mask): Ditto.
1306 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1307 (avx512f_storedqu<mode>_mask): Ditto.
1308 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1310 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1311 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1312 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1313 (avx512f_maskcmp<mode>3): Ditto.
1314 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1315 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1316 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1317 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1318 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1319 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1320 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1321 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1322 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1323 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1324 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1325 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1326 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1327 (vec_extract_lo_<mode>_maskm): Ditto.
1328 (vec_extract_hi_<mode>_maskm): Ditto.
1329 (avx512f_vternlog<mode>_mask): Ditto.
1330 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1331 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1332 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1333 (avx512f_<code>v8div16qi2_mask): Ditto.
1334 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1335 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1336 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1337 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1338 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1339 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1340 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1341 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1342 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1343 (avx512cd_maskb_vec_dupv8di): Ditto.
1344 (avx512cd_maskw_vec_dupv16si): Ditto.
1345 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1346 (avx512f_vpermi2var<mode>3_mask): Ditto.
1347 (avx512f_vpermi2var<mode>3_mask): Ditto.
1348 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1349 (*avx512f_gathersi<mode>): Ditto.
1350 (*avx512f_gathersi<mode>_2): Ditto.
1351 (*avx512f_gatherdi<mode>): Ditto.
1352 (*avx512f_gatherdi<mode>_2): Ditto.
1353 (*avx512f_scattersi<mode>): Ditto.
1354 (*avx512f_scatterdi<mode>): Ditto.
1355 (avx512f_compress<mode>_mask): Ditto.
1356 (avx512f_compressstore<mode>_mask): Ditto.
1357 (avx512f_expand<mode>_mask): Ditto.
1358 * config/i386/subst.md (mask): Change k to Yk.
1359 (mask_scalar_merge): Ditto.
1362 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1364 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1366 2014-01-31 Richard Biener <rguenther@suse.de>
1369 * builtins.c (fold_builtin_memory_op): Make sure to not
1370 use a floating-point mode or a boolean or enumeral type for
1373 2014-01-30 DJ Delorie <dj@redhat.com>
1375 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1376 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1377 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1378 whenever main() has an epilogue.
1380 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1382 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1383 unused variable "field".
1384 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1385 (vsx_mergeh_<mode>): Likewise.
1386 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1387 (altivec_vmrghh): Likewise.
1388 (altivec_vmrghw): Likewise.
1389 (altivec_vmrglb): Likewise.
1390 (altivec_vmrglh): Likewise.
1391 (altivec_vmrglw): Likewise.
1392 (altivec_vspltb): Add missing uses.
1393 (altivec_vsplth): Likewise.
1394 (altivec_vspltw): Likewise.
1395 (altivec_vspltsf): Likewise.
1397 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1400 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1401 frame related instructions.
1403 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1405 PR rtl-optimization/59959
1406 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1407 any reload of register whose subreg is invalid.
1409 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1411 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1412 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1413 Add missing return type - void.
1415 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1417 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1418 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1419 remove element index adjustment for endian (now handled in vsx.md
1421 (altivec_expand_vec_perm_const): Use
1422 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1423 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1424 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1425 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1426 define_expand and a new define_insn *altivec_vspltb_internal;
1427 adjust for -maltivec=be on a little endian target.
1428 (altivec_vspltb_direct): New.
1429 (altivec_vsplth): Divide into a define_expand and a new
1430 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1431 little endian target.
1432 (altivec_vsplth_direct): New.
1433 (altivec_vspltw): Divide into a define_expand and a new
1434 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1435 little endian target.
1436 (altivec_vspltw_direct): New.
1437 (altivec_vspltsf): Divide into a define_expand and a new
1438 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1439 a little endian target.
1441 2014-01-30 Richard Biener <rguenther@suse.de>
1443 PR tree-optimization/59993
1444 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1445 can propagate form the earlier stmt and avoid the transform
1446 when the intermediate result is needed.
1448 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1450 * README.Portability: Fix typo.
1452 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1454 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1455 comparison_operator with ordered_comparison_operator.
1457 2014-01-30 Nick Clifton <nickc@redhat.com>
1459 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1460 Rename to mn10300_store_multiple_regs.
1461 * config/mn10300/mn10300.c: Likewise.
1462 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1463 store_multiple_regs.
1464 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1465 Call mn10300_store_multiple_regs.
1467 2014-01-30 Nick Clifton <nickc@redhat.com>
1468 DJ Delorie <dj@redhat.com>
1470 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1471 %fp 2 to keep registers after it properly word-aligned.
1472 (rl78_alloc_physical_registers_umul): Handle the case where both
1473 input operands are the same.
1475 2014-01-30 Richard Biener <rguenther@suse.de>
1477 PR tree-optimization/59903
1478 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1481 2014-01-30 Jason Merrill <jason@redhat.com>
1484 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1487 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1489 2014-01-30 Richard Biener <rguenther@suse.de>
1491 PR tree-optimization/59951
1492 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1494 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1497 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1498 SFmode to DFmode case.
1500 2014-01-29 DJ Delorie <dj@redhat.com>
1502 * config/msp430/msp430.opt (-minrt): New.
1503 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1505 (ENDFILE_SPEC): Likewise.
1507 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1509 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1510 (estimate_function_body_sizes): Use it.
1512 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1515 * dwarf2out.c (is_cxx_auto): New.
1516 (is_base_type): Use it.
1517 (gen_type_die_with_usage): Likewise.
1519 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1521 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1522 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1523 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1524 -maltivec=be with LE targets.
1525 (vsx_mergeh_<mode>): Likewise.
1526 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1527 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1528 (altivec_vmrghb): Replace with define_expand and new
1529 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1530 (altivec_vmrghb_direct): New define_insn.
1531 (altivec_vmrghh): Replace with define_expand and new
1532 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1533 (altivec_vmrghh_direct): New define_insn.
1534 (altivec_vmrghw): Replace with define_expand and new
1535 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1536 (altivec_vmrghw_direct): New define_insn.
1537 (*altivec_vmrghsf): Adjust for endianness.
1538 (altivec_vmrglb): Replace with define_expand and new
1539 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1540 (altivec_vmrglb_direct): New define_insn.
1541 (altivec_vmrglh): Replace with define_expand and new
1542 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1543 (altivec_vmrglh_direct): New define_insn.
1544 (altivec_vmrglw): Replace with define_expand and new
1545 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1546 (altivec_vmrglw_direct): New define_insn.
1547 (*altivec_vmrglsf): Adjust for endianness.
1548 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1549 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1550 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1551 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1552 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1553 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1554 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1555 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1557 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1559 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1560 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1563 2014-01-29 Richard Biener <rguenther@suse.de>
1565 PR tree-optimization/58742
1566 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1567 associate_pointerplus_align.
1568 (associate_pointerplus_diff): New function.
1569 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1570 and associate_pointerplus_diff.
1572 2014-01-29 Richard Biener <rguenther@suse.de>
1574 * lto-streamer.h (LTO_major_version): Bump to 3.
1575 (LTO_minor_version): Reset to 0.
1577 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1579 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1580 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1581 (arm_file_start): Generate correct asm header for armv7ve.
1582 * config/arm/bpabi.h: Add multilib support for armv7ve.
1583 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1584 and cortex-a15 to armv7ve.
1585 * config/arm/t-aprofile: Add multilib support for armv7ve.
1586 * doc/invoke.texi: Document -march=armv7ve.
1588 2014-01-29 Richard Biener <rguenther@suse.de>
1590 PR tree-optimization/58742
1591 * tree-ssa-forwprop.c (associate_plusminus): Return true
1592 if we changed sth, defer EH cleanup to ...
1593 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1594 (simplify_mult): New function.
1596 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1599 PR tree-optimization/59920
1600 * tree.c (build_common_builtin_nodes): Remove
1601 __builtin_setjmp_dispatcher initialization.
1602 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1603 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1604 instead of gsi_after_labels + manually skipping debug stmts.
1605 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1606 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1607 * tree-inline.c (copy_edges_for_bb): Remove
1608 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1609 argument. Ignore computed_goto_p stmts. Don't call
1610 make_abnormal_goto_edges. If a call might need abnormal edges
1611 for non-local gotos, see if it already has an edge to
1612 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1613 with true argument, don't do anything then, otherwise add
1614 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1615 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1617 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1618 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1619 (lower_stmt): Don't set data->calls_builtin_setjmp.
1620 (lower_builtin_setjmp): Adjust comment.
1621 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1622 * tree-cfg.c (found_computed_goto): Remove.
1623 (factor_computed_gotos): Remove.
1624 (make_goto_expr_edges): Return bool, true for computed gotos.
1625 Don't call make_abnormal_goto_edges.
1626 (build_gimple_cfg): Don't set found_computed_goto, don't call
1627 factor_computed_gotos.
1628 (computed_goto_p): No longer static.
1629 (make_blocks): Don't set found_computed_goto.
1630 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1631 (make_edges): If make_goto_expr_edges returns true, push bb
1632 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1633 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1634 vector. Record mapping between bbs and OpenMP regions if there
1635 are any, adjust make_gimple_omp_edges caller. Call
1636 handle_abnormal_edges.
1637 (make_abnormal_goto_edges): Remove.
1638 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1639 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1640 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1641 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1642 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1643 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1644 filling *region also set *region_idx to (*region)->entry->index.
1647 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1648 For REGs set ORIGINAL_REGNO.
1650 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1652 * doc/md.texi: Mention that a target shouldn't implement
1653 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1656 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1658 PR tree-optimization/59594
1659 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1660 a copy of the datarefs vector rather than the vector itself.
1662 2014-01-28 Jason Merrill <jason@redhat.com>
1665 * dwarf2out.c (auto_die): New static.
1666 (gen_type_die_with_usage): Handle C++1y 'auto'.
1667 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1670 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1673 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1674 (SPEC_X32): Likewise.
1675 (SPEC_64): Likewise.
1676 * config/i386/i386.c (ix86_option_override_internal): Turn off
1677 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1679 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1680 * config/i386/i386.h (TARGET_16BIT): New macro.
1681 (TARGET_16BIT_P): Likewise.
1682 * config/i386/i386.opt: Add m16.
1683 * doc/invoke.texi: Document -m16.
1685 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1687 PR preprocessor/59935
1688 * input.c (location_get_source_line): Bail out on when line number
1689 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1691 2014-01-28 Richard Biener <rguenther@suse.de>
1693 PR tree-optimization/58742
1694 * tree-ssa-forwprop.c (associate_plusminus): Handle
1695 pointer subtraction of the form (T)(P + A) - (T)P.
1697 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1699 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1702 2014-01-28 Richard Biener <rguenther@suse.de>
1705 2014-01-28 Richard Biener <rguenther@suse.de>
1707 PR rtl-optimization/45364
1708 PR rtl-optimization/59890
1709 * var-tracking.c (local_get_addr_clear_given_value): Handle
1710 already cleared slot.
1711 (val_reset): Handle not allocated local_get_addr_cache.
1712 (vt_find_locations): Use post-order on the inverted CFG.
1714 2014-01-28 Richard Biener <rguenther@suse.de>
1716 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1718 2014-01-28 Richard Biener <rguenther@suse.de>
1720 PR rtl-optimization/45364
1721 PR rtl-optimization/59890
1722 * var-tracking.c (local_get_addr_clear_given_value): Handle
1723 already cleared slot.
1724 (val_reset): Handle not allocated local_get_addr_cache.
1725 (vt_find_locations): Use post-order on the inverted CFG.
1727 2014-01-28 Alan Modra <amodra@gmail.com>
1729 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1730 * configure.ac <recursive call for build != host>: Define
1731 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1732 and LD_FOR_BUILD too.
1733 * configure: Regenerate.
1735 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1737 * config/i386/i386.c (get_builtin_code_for_version): Separate
1738 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1739 Broadwell from Haswell.
1741 2014-01-27 Steve Ellcey <sellcey@mips.com>
1743 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1744 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1745 * config/mips/mips.c (mips_option_override): Change setting
1747 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1748 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1749 Change from Mask to Var.
1751 2014-01-27 Jeff Law <law@redhat.com>
1753 * ipa-inline.c (inline_small_functions): Fix typo.
1755 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1757 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1758 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1759 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1760 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1761 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1762 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1763 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1764 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1765 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1766 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1767 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1768 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1769 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1770 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1771 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1772 (_mm512_storeu_epi64): Ditto.
1773 (_mm512_cmpge_epi32_mask): Ditto.
1774 (_mm512_cmpge_epu32_mask): Ditto.
1775 (_mm512_cmpge_epi64_mask): Ditto.
1776 (_mm512_cmpge_epu64_mask): Ditto.
1777 (_mm512_cmple_epi32_mask): Ditto.
1778 (_mm512_cmple_epu32_mask): Ditto.
1779 (_mm512_cmple_epi64_mask): Ditto.
1780 (_mm512_cmple_epu64_mask): Ditto.
1781 (_mm512_cmplt_epi32_mask): Ditto.
1782 (_mm512_cmplt_epu32_mask): Ditto.
1783 (_mm512_cmplt_epi64_mask): Ditto.
1784 (_mm512_cmplt_epu64_mask): Ditto.
1785 (_mm512_cmpneq_epi32_mask): Ditto.
1786 (_mm512_cmpneq_epu32_mask): Ditto.
1787 (_mm512_cmpneq_epi64_mask): Ditto.
1788 (_mm512_cmpneq_epu64_mask): Ditto.
1789 (_mm512_expand_pd): Ditto.
1790 (_mm512_expand_ps): Ditto.
1791 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1792 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1793 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1794 * config/i386/i386.c (ix86_builtins): Add
1795 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1796 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1797 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1798 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1799 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1800 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1801 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1802 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1803 IX86_BUILTIN_PMOVUSQW512_MEM.
1804 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1805 __builtin_ia32_pmovsqd512mem_mask,
1806 __builtin_ia32_pmovqd512mem_mask,
1807 __builtin_ia32_pmovusqw512mem_mask,
1808 __builtin_ia32_pmovsqw512mem_mask,
1809 __builtin_ia32_pmovqw512mem_mask,
1810 __builtin_ia32_pmovusdw512mem_mask,
1811 __builtin_ia32_pmovsdw512mem_mask,
1812 __builtin_ia32_pmovdw512mem_mask,
1813 __builtin_ia32_pmovqb512mem_mask,
1814 __builtin_ia32_pmovusqb512mem_mask,
1815 __builtin_ia32_pmovsqb512mem_mask,
1816 __builtin_ia32_pmovusdb512mem_mask,
1817 __builtin_ia32_pmovsdb512mem_mask,
1818 __builtin_ia32_pmovdb512mem_mask.
1819 (bdesc_args): Add __builtin_ia32_expanddf512,
1820 __builtin_ia32_expandsf512.
1821 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1822 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1823 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1824 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1825 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1826 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1827 (avx512f_<code>v8div16qi2_mask_store): This.
1828 (avx512f_expand<mode>): New.
1830 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1832 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1834 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1835 (_mm512_prefetch_i32scatter_pd): Ditto.
1836 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1837 (_mm512_prefetch_i64scatter_pd): Ditto.
1838 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1839 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1840 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1841 (_mm512_prefetch_i32scatter_ps): Ditto.
1842 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1843 (_mm512_prefetch_i64scatter_ps): Ditto.
1844 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1845 * config/i386/i386-builtin-types.def: Define
1846 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1847 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1848 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1849 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1850 IX86_BUILTIN_SCATTERPFQPD.
1851 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1852 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1853 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1854 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1855 __builtin_ia32_scatterpfqps.
1856 (ix86_expand_builtin): Expand new built-ins.
1857 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1858 fix memory access data type.
1859 (*avx512pf_gatherpf<mode>_mask): Ditto.
1860 (*avx512pf_gatherpf<mode>): Ditto.
1861 (avx512pf_scatterpf<mode>): Ditto.
1862 (*avx512pf_scatterpf<mode>_mask): Ditto.
1863 (*avx512pf_scatterpf<mode>): Ditto.
1864 (GATHER_SCATTER_SF_MEM_MODE): New.
1865 (avx512pf_gatherpf<mode>df): Ditto.
1866 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1867 (*avx512pf_scatterpf<mode>df): Ditto.
1869 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1872 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1873 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1876 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1878 * common/config/arm/arm-common.c
1879 (arm_rewrite_mcpu): Handle multiple names.
1881 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1883 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1885 * gimple-builder.h (create_gimple_tmp): Delete.
1887 2014-01-27 Christian Bruel <christian.bruel@st.com>
1889 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1892 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1894 * config/pa/pa.md (call): Generate indirect long calls to non-local
1895 functions when outputing 32-bit code.
1896 (call_value): Likewise except for special call to buggy powf function.
1898 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1899 portable runtime and PIC indirect calls.
1900 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
1901 and PIC call sequences. Use ldo instead of blr to set return register
1902 in PIC call sequence.
1904 2014-01-25 Walter Lee <walt@tilera.com>
1906 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
1907 avoid clobbering a live register.
1909 2014-01-25 Walter Lee <walt@tilera.com>
1911 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
1912 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
1913 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
1914 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
1916 2014-01-25 Walter Lee <walt@tilera.com>
1918 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
1919 arguments on even registers.
1920 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
1922 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
1923 (BIGGEST_ALIGNMENT): Ditto.
1924 (BIGGEST_FIELD_ALIGNMENT): Ditto.
1926 2014-01-25 Walter Lee <walt@tilera.com>
1928 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
1929 insns before bundling.
1930 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
1932 2014-01-25 Walter Lee <walt@tilera.com>
1934 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
1935 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
1936 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
1938 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1940 * config/mips/constraints.md (kl): Delete.
1941 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
1942 define expands, using...
1943 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
1944 instructions for MIPS16.
1945 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
1946 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
1948 2014-01-25 Walter Lee <walt@tilera.com>
1950 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
1954 2014-01-25 Walter Lee <walt@tilera.com>
1956 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
1957 (TARGET_EXPAND_TO_RTL_HOOK): Define.
1959 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
1961 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
1964 2014-01-25 Jakub Jelinek <jakub@redhat.com>
1966 * print-rtl.c (in_call_function_usage): New var.
1967 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
1968 EXPR_LIST mode as mode and not as reg note name.
1971 * cfgloopmanip.c (copy_loop_info): If
1972 loop->warned_aggressive_loop_optimizations, make sure
1973 the flag is set in target loop too.
1975 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
1977 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
1979 * builtins.def: Likewise.
1980 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
1981 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
1982 * ira.c (ira_setup_eliminable_regset): Likewise.
1983 * omp-low.c (gate_expand_omp): Likewise.
1984 (execute_lower_omp): Likewise.
1985 (diagnose_sb_0): Likewise.
1986 (gate_diagnose_omp_blocks): Likewise.
1987 (simd_clone_clauses_extract): Likewise.
1990 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1992 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
1993 correction for little endian...
1994 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
1997 2014-01-24 Jeff Law <law@redhat.com>
1999 PR tree-optimization/59919
2000 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2001 for non-returning calls.
2003 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2005 * common/config/aarch64/aarch64-common.c
2006 (aarch64_rewrite_mcpu): Handle multiple names.
2007 * config/aarch64/aarch64.h
2008 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2010 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2012 * input.c (add_file_to_cache_tab): Handle the case where fopen
2015 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2018 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2019 from push operand if code of push isn't PRE_DEC.
2021 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2024 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2025 -mquad-memory-atomic. Update -mquad-memory documentation to say
2026 it is only used for non-atomic loads/stores.
2028 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2029 -mquad-memory or -mquad-memory-atomic switches.
2031 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2032 -mquad-memory-atomic to ISA 2.07 support.
2034 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2035 to separate support of normal quad word memory operations (ldq, stq)
2036 from the atomic quad word memory operations.
2038 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2039 support to separate non-atomic quad word operations from atomic
2040 quad word operations. Disable non-atomic quad word operations in
2041 little endian mode so that we don't have to swap words after the
2042 load and before the store.
2043 (quad_load_store_p): Add comment about atomic quad word support.
2044 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2045 options printed with -mdebug=reg.
2047 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2048 -mquad-memory-atomic as the test for whether we have quad word
2049 atomic instructions.
2050 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2051 or -mp8-vector are used, allow byte/half-word atomic operations.
2053 * config/rs6000/sync.md (load_lockedti): Insure that the address
2054 is a proper indexed or indirect address for the lqarx instruction.
2055 On little endian systems, swap the hi/lo registers after the lqarx
2057 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2058 insure the address is valid for the lqarx instruction.
2059 (store_conditionalti): Insure that the address is a proper indexed
2060 or indirect address for the stqcrx. instruction. On little endian
2061 systems, swap the hi/lo registers before doing the stqcrx.
2063 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2064 insure the address is valid for the stqcrx. instruction.
2066 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2067 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2068 type of quad memory support is available.
2070 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2073 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2074 there is a danger of looping.
2076 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2078 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2079 force flag_ira_loop_pressure if set via command line.
2081 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2083 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2084 (ashr_simd): New builtin handling DI mode.
2085 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2086 (aarch64_sshr_simddi): New match pattern.
2087 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2088 (vshrd_n_s64): Likewise.
2089 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2091 2014-01-23 Nick Clifton <nickc@redhat.com>
2093 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2094 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2095 favour of mcu specific scripts.
2096 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2099 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2100 Alex Velenko <Alex.Velenko@arm.com>
2102 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2103 (vaddv_s16): Likewise.
2104 (vaddv_s32): Likewise.
2105 (vaddv_u8): Likewise.
2106 (vaddv_u16): Likewise.
2107 (vaddv_u32): Likewise.
2108 (vaddvq_s8): Likewise.
2109 (vaddvq_s16): Likewise.
2110 (vaddvq_s32): Likewise.
2111 (vaddvq_s64): Likewise.
2112 (vaddvq_u8): Likewise.
2113 (vaddvq_u16): Likewise.
2114 (vaddvq_u32): Likewise.
2115 (vaddvq_u64): Likewise.
2116 (vaddv_f32): Likewise.
2117 (vaddvq_f32): Likewise.
2118 (vaddvq_f64): Likewise.
2119 (vmaxv_f32): Likewise.
2120 (vmaxv_s8): Likewise.
2121 (vmaxv_s16): Likewise.
2122 (vmaxv_s32): Likewise.
2123 (vmaxv_u8): Likewise.
2124 (vmaxv_u16): Likewise.
2125 (vmaxv_u32): Likewise.
2126 (vmaxvq_f32): Likewise.
2127 (vmaxvq_f64): Likewise.
2128 (vmaxvq_s8): Likewise.
2129 (vmaxvq_s16): Likewise.
2130 (vmaxvq_s32): Likewise.
2131 (vmaxvq_u8): Likewise.
2132 (vmaxvq_u16): Likewise.
2133 (vmaxvq_u32): Likewise.
2134 (vmaxnmv_f32): Likewise.
2135 (vmaxnmvq_f32): Likewise.
2136 (vmaxnmvq_f64): Likewise.
2137 (vminv_f32): Likewise.
2138 (vminv_s8): Likewise.
2139 (vminv_s16): Likewise.
2140 (vminv_s32): Likewise.
2141 (vminv_u8): Likewise.
2142 (vminv_u16): Likewise.
2143 (vminv_u32): Likewise.
2144 (vminvq_f32): Likewise.
2145 (vminvq_f64): Likewise.
2146 (vminvq_s8): Likewise.
2147 (vminvq_s16): Likewise.
2148 (vminvq_s32): Likewise.
2149 (vminvq_u8): Likewise.
2150 (vminvq_u16): Likewise.
2151 (vminvq_u32): Likewise.
2152 (vminnmv_f32): Likewise.
2153 (vminnmvq_f32): Likewise.
2154 (vminnmvq_f64): Likewise.
2156 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2158 * config/aarch64/aarch64-simd.md
2159 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2160 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2161 (*aarch64_mul3_elt<mode>): Likewise.
2162 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2163 (*aarch64_mul3_elt_to_64v2df): Likewise.
2164 (*aarch64_mla_elt<mode>): Likewise.
2165 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2166 (*aarch64_mls_elt<mode>): Likewise.
2167 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2168 (*aarch64_fma4_elt<mode>): Likewise.
2169 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2170 (*aarch64_fma4_elt_to_64v2df): Likewise.
2171 (*aarch64_fnma4_elt<mode>): Likewise.
2172 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2173 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2174 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2175 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2176 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2177 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2178 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2179 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2180 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2182 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2184 * config/aarch64/aarch64-simd.md
2185 (aarch64_be_checked_get_lane<mode>): New define_expand.
2186 * config/aarch64/aarch64-simd-builtins.def
2187 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2188 New builtin definition.
2189 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2190 Use new safe be builtin.
2192 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2194 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2196 (aarch64_be_st1<mode>): Likewise.
2197 (aarch_ld1<VALL:mode>): Define_expand modified.
2198 (aarch_st1<VALL:mode>): Likewise.
2199 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2200 (UNSPEC_ST1): Likewise.
2202 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2204 * config/microblaze/microblaze.md: Add trap insn and attribute
2206 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2208 PR preprocessor/58580
2209 * input.h (location_get_source_line): Take an additional line_size
2211 (void diagnostics_file_cache_fini): Declare new function.
2212 * input.c (struct fcache): New type.
2213 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2214 New static constants.
2215 (diagnostic_file_cache_init, total_lines_num)
2216 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2217 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2218 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2219 (get_next_line, read_next_line, goto_next_line, read_line_num):
2220 New static function definitions.
2221 (diagnostic_file_cache_fini): New function.
2222 (location_get_source_line): Take an additional output line_len
2223 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2225 * diagnostic.c (diagnostic_finish): Call
2226 diagnostic_file_cache_fini.
2227 (adjust_line): Take an additional input parameter for the length
2228 of the line, rather than calculating it with strlen.
2229 (diagnostic_show_locus): Adjust the use of
2230 location_get_source_line and adjust_line with respect to their new
2231 signature. While displaying a line now, do not stop at the first
2232 null byte. Rather, display the zero byte as a space and keep
2233 going until we reach the size of the line.
2234 * Makefile.in: Add vec.o to OBJS-libcommon
2236 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2237 Ilya Tocar <ilya.tocar@intel.com>
2239 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2240 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2241 (__builtin_ia32_kmov16): Ditto.
2242 * config/i386/i386.md (UNSPEC_KMOV): New.
2245 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2247 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2248 (_mm512_storeu_si512): Ditto.
2250 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2253 * rtl.h (get_referenced_operands): Declare.
2254 * recog.c (get_referenced_operands): New function.
2255 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2256 operands have been referenced when recording LO_SUM references.
2258 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2260 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2262 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2264 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2265 Enable for generic and recent AMD targets.
2267 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2269 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2270 ARG_SIZE note when adjustment was eliminated.
2272 2014-01-22 Jeff Law <law@redhat.com>
2274 PR tree-optimization/59597
2275 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2276 in file. Accept new argument REGISTERING and use it to modify
2277 dump output appropriately.
2278 (register_jump_thread): Corresponding changes.
2279 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2280 thread paths involving joiner blocks. Add code to dump cancelled
2281 jump threading paths.
2283 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2285 PR rtl-optimization/59477
2286 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2287 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2289 2014-01-22 Tom Tromey <tromey@redhat.com>
2291 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2293 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2295 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2297 PR rtl-optimization/59896
2298 * lra-constraints.c (process_alt_operands): Check unused note for
2299 matched operands of insn with no output reloads.
2301 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2303 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2304 (mips_move_from_gpr_cost): Likewise.
2306 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2308 PR rtl-optimization/59858
2309 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2310 ira_class_hard_regs_num.
2311 (process_alt_operands): Increase reject for dying matched operand.
2313 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2316 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2317 smaller than size, perform several stores or loads and stores
2318 at dst + count - size to store or copy all of size bytes, rather
2319 than just last modesize bytes.
2321 2014-01-20 DJ Delorie <dj@redhat.com>
2323 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2324 that CLOBBERs are REGs before propogating their values.
2326 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2329 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2330 (cgraph_inline_failed_type): New function.
2331 * cgraph.h (DEFCIFCODE): Add type.
2332 (cgraph_inline_failed_type_t): New enum.
2333 (cgraph_inline_failed_type): New prototype.
2334 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2335 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2336 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2337 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2338 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2339 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2340 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2341 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2342 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2343 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2344 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2345 OPTIMIZATION_MISMATCH.
2346 * tree-inline.c (expand_call_inline): Emit errors during
2347 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2349 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2352 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2353 mode attribute in insn output.
2355 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2357 * output.h (output_constant): Delete.
2358 * varasm.c (output_constant): Make private.
2360 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2362 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2364 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2367 * tree.h (fold_builtin_strcat): New prototype.
2368 * builtins.c (fold_builtin_strcat): No longer static. Add len
2369 argument, if non-NULL, don't call c_strlen. Optimize
2370 directly into __builtin_memcpy instead of __builtin_strcpy.
2371 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2372 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2374 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2376 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2377 for SImode_address_operand operands, having only a REG argument.
2379 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2381 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2382 loader name using mbig-endian.
2383 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2385 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2387 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2391 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2393 * config/aarch64/aarch64-protos.h
2394 (aarch64_cannot_change_mode_class_ptr): Declare.
2395 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2396 aarch64_cannot_change_mode_class_ptr): New.
2397 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2398 backend hook aarch64_cannot_change_mode_class.
2400 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2402 * common/config/aarch64/aarch64-common.c
2403 (aarch64_handle_option): Don't handle any option order logic here.
2404 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2405 selected_cpu, warn on architecture version mismatch.
2406 (aarch64_override_options): Fix parsing order for option strings.
2408 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2409 Iain Sandoe <iain@codesourcery.com>
2412 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2413 warning. Amend comment to reflect current functionality.
2415 2014-01-20 Richard Biener <rguenther@suse.de>
2418 * builtins.c (fold_builtin_strcat): Remove case better handled
2419 by tree-ssa-strlen.c.
2421 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2423 * config/aarch64/aarch64.opt
2424 (mcpu, march, mtune): Make case-insensitive.
2426 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2429 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2430 if operands[1] is a REG or ZERO_EXTEND of a REG.
2432 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2434 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2436 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2438 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2439 long non-pic millicode calls.
2441 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2443 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2445 2014-01-19 Kito Cheng <kito@0xlab.org>
2447 * builtins.c (expand_movstr): Check movstr expand done or fail.
2449 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2450 H.J. Lu <hongjiu.lu@intel.com>
2453 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2454 to DImode for zero-extended addresses.
2456 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2458 PR rtl-optimization/57763
2459 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2460 on the new indirect jump_insn and increment LABEL_NUSES (label).
2462 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2466 * config.gcc (x86_archs): New variable.
2467 (x86_64_archs): Likewise.
2468 (x86_cpus): Likewise.
2469 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2470 --with-arch/--with-cpu= options.
2471 Support --with-arch=/--with-cpu={nehalem,westmere,
2472 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2474 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2476 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2477 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2479 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2481 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2483 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2486 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2487 clear cpp_get_options (parse_in)->warn_unused_macros for
2488 ix86_target_macros_internal with cpp_define.
2490 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2492 * jump.c (delete_related_insns): Keep (use (insn))s.
2493 * reorg.c (redundant_insn): Check for barriers too.
2495 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2497 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2499 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2501 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2502 call to $$dyncall when TARGET_LONG_CALLS is true.
2504 2014-01-17 Jeff Law <law@redhat.com>
2506 * ree.c (combine_set_extension): Temporarily disable test for
2507 changing number of hard registers.
2509 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2512 * ipa-inline-analysis.c (inline_free_summary):
2513 Do not free summary of aliases.
2515 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2518 * gimplify.c (gimplify_expr): Use create_tmp_var
2519 instead of create_tmp_var_raw. If cond doesn't have
2520 integral type, don't add the IFN_ANNOTATE builtin at all.
2522 2014-01-17 Martin Jambor <mjambor@suse.cz>
2525 * ipa-cp.c (prev_edge_clone): New variable.
2526 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2527 Also resize prev_edge_clone vector.
2528 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2529 (ipcp_edge_removal_hook): New function.
2530 (ipcp_driver): Register ipcp_edge_removal_hook.
2532 2014-01-17 Andrew Pinski <apinski@cavium.com>
2533 Steve Ellcey <sellcey@mips.com>
2536 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2539 2014-01-17 Jeff Law <law@redhat.com>
2542 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2543 so that pass_ccp runs first.
2545 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2547 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2548 (ix86_adjust_cost): Use !TARGET_XXX.
2549 (do_reorder_for_imul): Likewise.
2550 (swap_top_of_ready_list): Likewise.
2551 (ix86_sched_reorder): Likewise.
2553 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2555 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2556 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2557 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2558 (intel_memset): New. Duplicate slm_memset.
2559 (intel_cost): New. Duplicate slm_cost.
2560 (m_INTEL): New macro.
2561 (processor_target_table): Add "intel".
2562 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2563 with PROCESSOR_INTEL for "intel".
2564 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2565 PROCESSOR_SILVERMONT.
2566 (ix86_issue_rate): Likewise.
2567 (ix86_adjust_cost): Likewise.
2568 (ia32_multipass_dfa_lookahead): Likewise.
2569 (swap_top_of_ready_list): Likewise.
2570 (ix86_sched_reorder): Likewise.
2571 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2572 instead of TARGET_OPT_AGU.
2573 * config/i386/i386.h (TARGET_INTEL): New.
2574 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2575 (processor_type): Add PROCESSOR_INTEL.
2576 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2577 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2579 2014-01-17 Marek Polacek <polacek@redhat.com>
2582 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2585 2014-01-17 Richard Biener <rguenther@suse.de>
2587 PR tree-optimization/46590
2588 * opts.c (default_options_table): Add entries for
2589 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2590 all enabled at -O1 but not for -Og.
2591 * common.opt (fbranch-count-reg): Remove Init(1).
2592 (fmove-loop-invariants): Likewise.
2593 (ftree-pta): Likewise.
2595 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2597 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2598 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2599 decls to at least the GCC 4.8 used alignments.
2602 * tree-nested.c (convert_nonlocal_reference_stmt,
2603 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2604 of GIMPLE_BIND stmts, adjust associated decls.
2606 2014-01-17 Richard Biener <rguenther@suse.de>
2608 PR tree-optimization/46590
2609 * vec.h (vec<>::bseach): New member function implementing
2610 binary search according to C89 bsearch.
2611 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2612 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2613 bitmap pointer again. Make accesses_in_loop a flat array.
2614 (mem_ref_obstack): New global.
2615 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2616 (mark_ref_stored): Likewise.
2617 (ref_indep_loop_p_2): Likewise.
2618 (set_ref_stored_in_loop): New helper function.
2619 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2620 (memref_free): Adjust.
2621 (record_mem_ref_loc): Simplify.
2622 (gather_mem_refs_stmt): Adjust.
2623 (sort_locs_in_loop_postorder_cmp): New function.
2624 (analyze_memory_references): Sort accesses_in_loop after
2625 loop postorder number.
2626 (find_ref_loc_in_loop_cmp): New function.
2627 (for_all_locs_in_loop): Find relevant cluster of locs in
2628 accesses_in_loop and iterate without recursion.
2629 (execute_sm): Avoid uninit warning.
2630 (struct ref_always_accessed): Simplify.
2631 (ref_always_accessed::operator ()): Likewise.
2632 (ref_always_accessed_p): Likewise.
2633 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2634 loop postorder numbers here.
2635 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2638 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2641 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2642 on decls for which assemble_alias has been called.
2644 2014-01-17 Nick Clifton <nickc@redhat.com>
2646 * config/msp430/msp430.opt: (mcpu): New option.
2647 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2648 (msp430_option_override): Parse target_cpu. If the MCU name
2649 matches a generic string, clear target_mcu.
2650 (msp430_attr): Allow numeric interrupt values up to 63.
2651 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2652 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2654 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2656 * config/msp430/msp430.md (popm): Use %J rather than %I.
2657 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2658 (addhi_cy_i): Use immediate_operand for operand 2.
2659 * doc/invoke.texi: Document -mcpu option.
2661 2014-01-17 Richard Biener <rguenther@suse.de>
2663 PR rtl-optimization/38518
2664 * df.h (df_analyze_loop): Declare.
2665 * df-core.c: Include cfgloop.h.
2666 (df_analyze_1): Split out main part of df_analyze.
2667 (df_analyze): Adjust.
2668 (loop_inverted_post_order_compute): New function.
2669 (loop_post_order_compute): Likewise.
2670 (df_analyze_loop): New function avoiding whole-function
2672 * loop-invariant.c (find_defs): Use df_analyze_loop.
2673 (find_invariants): Adjust.
2674 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2676 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2678 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2679 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2681 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2683 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2684 traversal when removing references.
2686 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2689 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2691 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2694 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2695 pushing a reload for an autoinc when we had previously reloaded an
2696 inner part of the address.
2698 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2700 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2702 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2703 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2704 when not giving up or versioning for alias only because of
2706 (vect_analyze_data_ref_dependences): Set to true.
2707 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2709 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2710 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2714 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2717 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2718 operand 0 predicate for gathers, use a new pseudo as subtarget.
2720 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2723 * lra-constraints.c (process_alt_operands): Add printing debug info.
2724 Check absence of input/output reloads for matched operands too.
2726 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2728 PR rtl-optimization/59835
2729 * ira.c (ira_init_register_move_cost): Increase cost for
2732 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2734 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2736 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2739 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2740 non-register objects. Use gen_(high/low)part more consistently.
2743 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2746 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2747 endian support, remove tests for WORDS_BIG_ENDIAN.
2748 (p8_mfvsrd_3_<mode>): Likewise.
2749 (reload_gpr_from_vsx<mode>): Likewise.
2750 (reload_gpr_from_vsxsf): Likewise.
2751 (p8_mfvsrd_4_disf): Likewise.
2753 2014-01-16 Richard Biener <rguenther@suse.de>
2755 PR rtl-optimization/46590
2756 * lcm.c (compute_antinout_edge): Use postorder iteration.
2757 (compute_laterin): Use inverted postorder iteration.
2759 2014-01-16 Nick Clifton <nickc@redhat.com>
2762 * varasm.c (output_constant): Return the number of bytes actually
2764 (output_constructor_array_range): Update the field size with the
2765 number of bytes emitted by output_constant.
2766 (output_constructor_regular_field): Likewise. Also do not
2767 complain if the total number of bytes emitted is now greater
2768 than the expected fieldpos.
2769 * output.h (output_constant): Update prototype and descriptive comment.
2771 2014-01-16 Marek Polacek <polacek@redhat.com>
2774 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2775 it is error_mark_node.
2777 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2779 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2780 VALID_AVX256_REG_OR_OI_MODE.
2782 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2784 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2785 current procedure should be profiled.
2787 2014-01-15 Andrew Pinski <apinski@cavium.com>
2789 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2790 of moving from/to the STACK_REG register class.
2792 2014-01-15 Richard Henderson <rth@redhat.com>
2795 * reginfo.c (global_regs_decl): Globalize.
2796 * rtl.h (global_regs_decl): Declare.
2797 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2798 reserved via global_regs.
2800 2014-01-15 Teresa Johnson <tejohnson@google.com>
2802 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2804 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2806 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2807 and vmulosh rather than call gen_vec_widen_smult_*.
2808 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2809 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2810 (vec_widen_smult_even_v16qi): Likewise.
2811 (vec_widen_umult_even_v8hi): Likewise.
2812 (vec_widen_smult_even_v8hi): Likewise.
2813 (vec_widen_umult_odd_v16qi): Likewise.
2814 (vec_widen_smult_odd_v16qi): Likewise.
2815 (vec_widen_umult_odd_v8hi): Likewise.
2816 (vec_widen_smult_odd_v8hi): Likewise.
2817 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2818 vmuloub rather than call gen_vec_widen_umult_*.
2819 (vec_widen_umult_lo_v16qi): Likewise.
2820 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2821 vmulosb rather than call gen_vec_widen_smult_*.
2822 (vec_widen_smult_lo_v16qi): Likewise.
2823 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2824 rather than call gen_vec_widen_umult_*.
2825 (vec_widen_umult_lo_v8hi): Likewise.
2826 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2827 rather than call gen_vec_widen_smult_*.
2828 (vec_widen_smult_lo_v8hi): Likewise.
2830 2014-01-15 Jeff Law <law@redhat.com>
2832 PR tree-optimization/59747
2833 * ree.c (find_and_remove_re): Properly handle case where a second
2834 eliminated extension requires widening a copy created for elimination
2835 of a prior extension.
2836 (combine_set_extension): Ensure that the number of hard regs needed
2837 for a destination register does not change when we widen it.
2839 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2841 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2842 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2843 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2844 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2845 (avr-*-rtems*): Likewise.
2846 (bfin*-rtems*): Likewise.
2847 (moxie-*-rtems*): Likewise.
2848 (h8300-*-rtems*): Likewise.
2849 (i[34567]86-*-rtems*): Likewise.
2850 (lm32-*-rtems*): Likewise.
2851 (m32r-*-rtems*): Likewise.
2852 (m68k-*-rtems*): Likewise.
2853 (microblaze*-*-rtems*): Likewise.
2854 (mips*-*-rtems*): Likewise.
2855 (powerpc-*-rtems*): Likewise.
2856 (sh-*-rtems*): Likewise.
2857 (sparc-*-rtems*): Likewise.
2858 (sparc64-*-rtems*): Likewise.
2859 (v850-*-rtems*): Likewise.
2860 (m32c-*-rtems*): Likewise.
2862 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2864 PR rtl-optimization/59511
2865 * ira.c (ira_init_register_move_cost): Use memory costs for some
2866 cases of register move cost calculations.
2867 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2868 instead of BB frequency.
2869 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2870 * lra-assigns.c (find_hard_regno_for): Ditto.
2872 2014-01-15 Richard Biener <rguenther@suse.de>
2874 PR tree-optimization/59822
2875 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2876 (vectorizable_load): Use it to hoist defs of uses of invariant
2877 loads out of the loop.
2879 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2880 Kugan Vivekanandarajah <kuganv@linaro.org>
2883 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2886 2014-01-15 Richard Biener <rguenther@suse.de>
2888 PR rtl-optimization/59802
2889 * lcm.c (compute_available): Use inverted postorder to seed
2890 the initial worklist.
2892 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2895 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2896 ADDR_REGS for invalid symrefs in non-PIC code.
2898 2014-01-15 Jakub Jelinek <jakub@redhat.com>
2901 * builtins.c (determine_block_size): Initialize *probable_max_size
2902 even if len_rtx is CONST_INT.
2904 2014-01-14 Andrew Pinski <apinski@cavium.com>
2906 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
2907 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
2908 (cortexa53_tunings): Likewise.
2909 (aarch64_sched_issue_rate): New function.
2910 (TARGET_SCHED_ISSUE_RATE): Define.
2912 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2914 * ira-costs.c (find_costs_and_classes): Add missed
2915 ira_init_register_move_cost_if_necessary.
2917 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
2920 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
2922 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
2925 * config/i386/i386.c (type_natural_mode): Add a bool parameter
2926 to indicate if type is used for function return value. Warn ABI
2927 change if the vector mode isn't available for function return value.
2928 (ix86_function_arg_advance): Pass false to type_natural_mode.
2929 (ix86_function_arg): Likewise.
2930 (ix86_gimplify_va_arg): Likewise.
2931 (function_arg_32): Don't warn ABI change.
2932 (ix86_function_value): Pass true to type_natural_mode.
2933 (ix86_return_in_memory): Likewise.
2934 (ix86_struct_value_rtx): Removed.
2935 (TARGET_STRUCT_VALUE_RTX): Likewise.
2937 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
2939 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
2940 converting a conditional jump into a conditional return.
2942 2014-01-14 Richard Biener <rguenther@suse.de>
2944 PR tree-optimization/58921
2945 PR tree-optimization/59006
2946 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
2947 hoisting invariant stmts.
2948 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
2949 invariant loads on the preheader edge if possible.
2951 2014-01-14 Joey Ye <joey.ye@arm.com>
2953 * doc/plugin.texi (Building GCC plugins): Update to C++.
2955 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
2957 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
2958 (_mm_rcp28_round_ss): Ditto.
2959 (_mm_rsqrt28_round_sd): Ditto.
2960 (_mm_rsqrt28_round_ss): Ditto.
2961 (_mm_rcp28_sd): Ditto.
2962 (_mm_rcp28_ss): Ditto.
2963 (_mm_rsqrt28_sd): Ditto.
2964 (_mm_rsqrt28_ss): Ditto.
2965 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
2966 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
2967 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
2968 (IX86_BUILTIN_RCP28SD): Ditto.
2969 (IX86_BUILTIN_RCP28SS): Ditto.
2970 (IX86_BUILTIN_RSQRT28SD): Ditto.
2971 (IX86_BUILTIN_RSQRT28SS): Ditto.
2972 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
2973 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
2974 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
2975 (ix86_expand_special_args_builtin): Expand new FTYPE.
2976 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
2977 (srcp14<mode>): Make insn unary.
2978 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
2979 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
2980 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
2981 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
2982 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
2983 Fix rounding: make it SAE only.
2984 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
2986 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
2988 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
2989 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
2990 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
2991 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
2992 (round_saeonly_mask_scalar_operand4): Ditto.
2993 (round_saeonly_mask_scalar_op3): Ditto.
2994 (round_saeonly_mask_scalar_op4): Ditto.
2996 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2998 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2999 Implement -maltivec=be for vec_insert and vec_extract.
3001 2014-01-10 DJ Delorie <dj@redhat.com>
3003 * config/msp430/msp430.md (call_internal): Don't allow memory
3004 references with SP as the base register.
3005 (call_value_internal): Likewise.
3006 * config/msp430/constraints.md (Yc): New. For memory references
3007 that don't use SP as a base register.
3009 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3010 "an integer without a # prefix"
3011 * config/msp430/msp430.md (epilogue_helper): Use it.
3013 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3016 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3017 AVX512F gather builtins.
3018 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3019 on gather decls with INTEGER_TYPE masktype.
3020 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3021 directly into the builtin rather than hoisting it before loop.
3023 PR tree-optimization/59387
3024 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3025 (scev_const_prop): If folded_casts and type has undefined overflow,
3026 use force_gimple_operand instead of force_gimple_operand_gsi and
3027 for each added stmt if it is assign with
3028 arith_code_with_undefined_signed_overflow, call
3029 rewrite_to_defined_overflow.
3030 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3031 gimple-fold.h instead.
3032 (arith_code_with_undefined_signed_overflow,
3033 rewrite_to_defined_overflow): Moved to ...
3034 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3035 rewrite_to_defined_overflow): ... here. No longer static.
3036 Include gimplify-me.h.
3037 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3038 rewrite_to_defined_overflow): New prototypes.
3040 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3042 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3044 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3046 * builtins.c (get_object_alignment_2): Minor tweak.
3047 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3049 2014-01-13 Christian Bruel <christian.bruel@st.com>
3051 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3052 optimized non constant lengths.
3054 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3057 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3058 load as __atomic_load_N if possible.
3060 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3062 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3064 (rs6000_expand_builtin): Adjust call.
3066 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3069 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3070 * config/rs6000/rs6000.c: Include target-globals.h.
3071 (rs6000_set_current_function): Instead of doing target_reinit
3072 unconditionally, use save_target_globals_default_opts and
3073 restore_target_globals.
3075 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3077 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3078 (rs6000_expand_builtin): Handle mffs and mtfsf.
3079 (rs6000_init_builtins): Define mffs and mtfsf.
3080 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3081 (rs6000_mffs): New pattern.
3082 (rs6000_mtfsf): New pattern.
3084 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3086 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3087 Start narrowing with START. Apply candidate-use pair
3088 and check overall cost in narrowing.
3089 (iv_ca_prune): Pass new argument.
3091 2014-01-10 Jeff Law <law@redhat.com>
3094 * ree.c (combine_reaching_defs): Ensure the defining statement
3095 occurs before the extension when optimizing extensions with
3096 different source and destination hard registers.
3098 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3101 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3102 vtables into the type inheritance graph.
3104 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3106 PR rtl-optimization/59754
3107 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3108 modes in the REGNO != REGNO case.
3110 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3112 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3114 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3116 PR tree-optimization/59745
3117 * tree-predcom.c (tree_predictive_commoning_loop): Call
3118 free_affine_expand_cache if giving up because components is NULL.
3120 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3121 GC in payload of target_globals struct instead of allocating them on
3122 the heap and the larger structs separately using GC.
3123 * target-globals.h (struct target_globals): Make regs, hard_regs,
3124 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3125 of GTY((skip)) and change type to void *.
3126 (reset_target_globals): Cast loads from those fields to corresponding
3129 2014-01-10 Steve Ellcey <sellcey@mips.com>
3132 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3133 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3134 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3136 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3139 * aarch64-modes.def (CC_Zmode): New flags mode.
3140 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3141 represents an equality.
3142 (aarch64_get_condition_code): Handle CC_Zmode.
3143 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3145 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3147 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3148 extraction in good case.
3150 2014-01-10 Richard Biener <rguenther@suse.de>
3152 PR tree-optimization/59374
3153 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3154 checking after SLP discovery. Mark stmts not participating
3155 in any SLP instance properly.
3157 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3159 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3160 when handling a SET rtx.
3162 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3164 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3165 (cortex-a57): Likewise.
3166 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3168 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3170 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3171 non-iwmmxt builtins.
3173 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3177 * ipa-devirt.c record_target_from_binfo): Take as argument
3178 stack of binfos and lookup matching one for virtual inheritance.
3179 (possible_polymorphic_call_targets_1): Update.
3181 2014-01-10 Huacai Chen <chenhc@lemote.com>
3183 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3184 kernel strings for Loongson-2E/2F/3A.
3186 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3189 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3190 is_gimple_call before calling gimple_call_internal_p.
3192 2014-01-09 Steve Ellcey <sellcey@mips.com>
3194 * Makefile.in (TREE_FLOW_H): Remove.
3195 (TREE_SSA_H): Add file names from tree-flow.h.
3196 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3197 * tree.h: Remove tree-flow.h reference.
3198 * hash-table.h: Remove tree-flow.h reference.
3199 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3200 reference with tree-ssa-loop.h.
3202 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3204 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3205 default element-order behavior for -maltivec.
3206 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3207 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3208 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3209 when targeting big endian, at least for now.
3210 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3212 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3215 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3216 var satisfies use_register_for_decl, just take into account type
3217 alignment, rather than decl alignment.
3219 PR tree-optimization/59622
3220 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3221 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3222 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3223 Don't devirtualize for inplace at all. For targets.length () == 1,
3224 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3226 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3228 * config/i386/i386.md (cpu): Remove the unused btver1.
3230 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3232 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3234 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3237 * tree-core.h (struct target_globals): New forward declaration.
3238 (struct tree_target_option): Add globals field.
3239 * tree.h (TREE_TARGET_GLOBALS): Define.
3240 (prepare_target_option_nodes_for_pch): New prototype.
3241 * target-globals.h (struct target_globals): Define even if
3243 * tree.c (prepare_target_option_node_for_pch,
3244 prepare_target_option_nodes_for_pch): New functions.
3245 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3246 * config/i386/i386.c: Include target-globals.h.
3247 (ix86_set_current_function): Instead of doing target_reinit
3248 unconditionally, use save_target_globals_default_opts and
3249 restore_target_globals.
3251 2014-01-09 Richard Biener <rguenther@suse.de>
3253 PR tree-optimization/59715
3254 * tree-cfg.h (split_critical_edges): Declare.
3255 * tree-cfg.c (split_critical_edges): Export.
3256 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3258 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3260 * cfgexpand.c (expand_stack_vars): Optionally disable
3261 asan stack protection.
3262 (expand_used_vars): Likewise.
3263 (partition_stack_vars): Likewise.
3264 * asan.c (asan_emit_stack_protection): Optionally disable
3265 after return stack usage.
3266 (instrument_derefs): Optionally disable memory access instrumentation.
3267 (instrument_builtin_call): Likewise.
3268 (instrument_strlen_call): Likewise.
3269 (asan_protect_global): Optionally disable global variables protection.
3270 * doc/invoke.texi: Added doc for new options.
3271 * params.def: Added new options.
3272 * params.h: Likewise.
3274 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3276 PR rtl-optimization/59724
3277 * ifcvt.c (cond_exec_process_if_block): Don't call
3278 flow_find_head_matching_sequence with 0 longest_match.
3279 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3280 non-active insns if !stop_after.
3281 (try_head_merge_bb): Revert 2014-01-07 changes.
3283 2014-01-08 Jeff Law <law@redhat.com>
3285 * ree.c (get_sub_rtx): New function, extracted from...
3286 (merge_def_and_ext): Here.
3287 (combine_reaching_defs): Use get_sub_rtx.
3289 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3291 * cgraph.h (varpool_variable_node): Do not choke on null node.
3293 2014-01-08 Catherine Moore <clm@codesourcery.com>
3295 * config/mips/mips.md (simple_return): Attempt to use JRC
3297 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3299 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3301 PR rtl-optimization/59137
3302 * reorg.c (steal_delay_list_from_target): Call update_block for
3304 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3306 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3308 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3309 two duplicate entries.
3311 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3314 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3316 * config/mips/mips.c (mips_truncated_op_cost): New function.
3317 (mips_rtx_costs): Adjust test for BADDU.
3318 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3320 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3322 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3323 (*baddu_si): ...this new pattern.
3325 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3328 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3330 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3333 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3335 (expand_expr, expand_normal): Adjust.
3336 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3337 inner_reference_p. Use inner_reference_p to expand inner references.
3338 (store_expr): Adjust.
3339 * cfgexpand.c (expand_call_stmt): Adjust.
3341 2014-01-08 Rong Xu <xur@google.com>
3343 * gcov-io.c (gcov_var): Move from gcov-io.h.
3344 (gcov_position): Ditto.
3345 (gcov_is_error): Ditto.
3346 (gcov_rewrite): Ditto.
3347 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3348 only part to libgcc/libgcov.h.
3350 2014-01-08 Marek Polacek <polacek@redhat.com>
3353 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3355 2014-01-08 Marek Polacek <polacek@redhat.com>
3358 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3360 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3362 PR rtl-optimization/59649
3363 * stor-layout.c (get_mode_bounds): For BImode return
3364 0 and STORE_FLAG_VALUE.
3366 2014-01-08 Richard Biener <rguenther@suse.de>
3369 * gimple.h (is_gimple_builtin_call): Remove.
3370 (gimple_builtin_call_types_compatible_p): New.
3371 (gimple_call_builtin_p): New overload.
3372 * gimple.c (is_gimple_builtin_call): Remove.
3373 (validate_call): Rename to ...
3374 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3376 (validate_type): New static function.
3377 (gimple_call_builtin_p): New overload and adjust.
3378 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3379 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3380 (gimple_fold_stmt_to_constant_1): Likewise.
3381 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3383 2014-01-08 Richard Biener <rguenther@suse.de>
3386 * gimplify.c (gimplify_expr): Gimplify register-register type
3387 VIEW_CONVERT_EXPRs to separate stmts.
3389 2014-01-07 Jeff Law <law@redhat.com>
3392 * ree.c (combine_set_extension): Handle case where source
3393 and destination registers in an extension insn are different.
3394 (combine_reaching_defs): Allow source and destination registers
3395 in extension to be different under limited circumstances.
3396 (add_removable_extension): Remove restriction that the
3397 source and destination registers in the extension are the same.
3398 (find_and_remove_re): Emit a copy from the extension's
3399 destination to its source after the defining insn if
3400 the source and destination registers are different.
3403 * ifcvt.c (merge_if_block): If we are merging a block with more than
3404 one successor with a block with no successors, remove any BARRIER
3405 after the second block.
3407 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3409 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3411 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3414 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3415 for 14-bit register offsets when INT14_OK_STRICT is false.
3417 2014-01-07 Roland Stigge <stigge@antcom.de>
3418 Michael Meissner <meissner@linux.vnet.ibm.com>
3421 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3422 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3424 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3426 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3429 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3431 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3432 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3433 rtx is const0_rtx or not.
3435 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3438 * target-globals.c (save_target_globals): Remove this_fn_optab
3440 * toplev.c: Include optabs.h.
3441 (target_reinit): Temporarily restore the global options if another
3442 set of options are in force.
3444 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3446 PR rtl-optimization/58668
3447 * cfgcleanup.c (flow_find_cross_jump): Don't count
3448 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3449 to determine what is counted.
3450 (flow_find_head_matching_sequence): Use active_insn_p to determine
3452 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3454 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3455 determine what is counted.
3457 PR tree-optimization/59643
3458 * tree-predcom.c (split_data_refs_to_components): If one dr is
3459 read and one write, determine_offset fails and the write isn't
3460 in the bad component, just put the read into the bad component.
3462 2014-01-07 Mike Stump <mikestump@comcast.net>
3463 Jakub Jelinek <jakub@redhat.com>
3466 * tree-core.h (struct tree_optimization_option): Change optabs
3467 type from unsigned char * to void *.
3468 * optabs.c (init_tree_optimization_optabs): Adjust
3469 TREE_OPTIMIZATION_OPTABS initialization.
3471 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3474 * config/i386/i386.h (struct machine_function): Add
3475 no_drap_save_restore field.
3476 * config/i386/i386.c (ix86_save_reg): Use
3477 !cfun->machine->no_drap_save_restore instead of
3478 crtl->stack_realign_needed.
3479 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3480 this function clears frame_pointer_needed. Set
3481 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3482 and DRAP reg is needed.
3484 2014-01-06 Marek Polacek <polacek@redhat.com>
3487 * doc/implement-c.texi: Mention that other integer types are
3488 permitted as bit-field types in strictly conforming mode.
3490 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3492 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3495 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3497 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3499 2014-01-06 Martin Jambor <mjambor@suse.cz>
3502 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3504 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3506 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3510 * var-tracking.c (add_stores): Preserve the value of the source even if
3511 we don't record the store.
3513 2014-01-06 Terry Guo <terry.guo@arm.com>
3515 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3517 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3520 * config/darwin.c (darwin_function_section): Adjust return values to
3521 correspond to optimisation changes made in r206070.
3523 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3525 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3526 from prefetch_block tune setting.
3527 (nocona_cost): Correct size of prefetch block to 64.
3529 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3531 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3532 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3533 used to save the static chain register in the computation of the offset
3534 from which the FP registers need to be restored.
3536 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3538 PR tree-optimization/59519
3539 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3540 ICE if get_current_def (current_new_name) is already non-NULL, as long
3541 as it is a phi result of some other phi in *new_exit_bb that has
3544 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3545 or vmovdqu* for misaligned_operand.
3546 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3547 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3548 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3549 aligned_mem for AVX512F masked aligned load and store builtins and for
3552 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3554 PR tree-optimization/59651
3555 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3556 Address range for negative step should be added by TYPE_SIZE_UNIT.
3558 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3560 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3561 overlapping registers also for registers other than the stack pointer.
3563 2014-01-03 Marek Polacek <polacek@redhat.com>
3566 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3569 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3572 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3575 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3576 (push splitter): Use <P:MODE_SIZE> instead of
3577 GET_MODE_SIZE (<P:MODE>mode).
3578 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3579 (mov -1, reg peephole2): Likewise.
3580 * config/i386/sse.md (*mov<mode>_internal,
3581 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3582 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3583 *<code><mode>3, *andnot<mode>3<mask_name>,
3584 <mask_codefor><code><mode>3<mask_name>): Likewise.
3585 * config/i386/subst.md (mask_mode512bit_condition,
3586 sd_mask_mode512bit_condition): Likewise.
3588 2014-01-02 Xinliang David Li <davidxl@google.com>
3590 PR tree-optimization/59303
3591 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3592 (dump_predicates): Better output format.
3593 (pred_equal_p): New function.
3594 (is_neq_relop_p): Ditto.
3595 (is_neq_zero_form_p): Ditto.
3596 (pred_expr_equal_p): Ditto.
3597 (pred_neg_p): Ditto.
3598 (simplify_pred): Ditto.
3599 (simplify_preds_2): Ditto.
3600 (simplify_preds_3): Ditto.
3601 (simplify_preds_4): Ditto.
3602 (simplify_preds): Ditto.
3604 (push_to_worklist): Ditto.
3605 (get_pred_info_from_cmp): Ditto.
3606 (is_degenerated_phi): Ditto.
3607 (normalize_one_pred_1): Ditto.
3608 (normalize_one_pred): Ditto.
3609 (normalize_one_pred_chain): Ditto.
3610 (normalize_preds): Ditto.
3611 (normalize_cond_1): Remove function.
3612 (normalize_cond): Ditto.
3613 (is_gcond_subset_of): Ditto.
3614 (is_subset_of_any): Ditto.
3615 (is_or_set_subset_of): Ditto.
3616 (is_and_set_subset_of): Ditto.
3617 (is_norm_cond_subset_of): Ditto.
3618 (pred_chain_length_cmp): Ditto.
3619 (convert_control_dep_chain_into_preds): Type change.
3620 (find_predicates): Ditto.
3621 (find_def_preds): Ditto.
3622 (destroy_predicates_vecs): Ditto.
3623 (find_matching_predicates_in_rest_chains): Ditto.
3624 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3625 (is_pred_expr_subset): Ditto.
3626 (is_pred_chain_subset_of): Ditto.
3627 (is_included_in): Ditto.
3628 (is_superset_of): Ditto.
3630 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3632 Update copyright years.
3634 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3636 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3637 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3638 config/arc/arc.md, config/arc/arc.opt,
3639 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3640 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3641 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3642 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3643 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3644 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3645 vtable-verify.c, vtable-verify.h: Use the standard form for the
3648 2014-01-02 Tobias Burnus <burnus@net-b.de>
3650 * gcc.c (process_command): Update copyright notice dates.
3651 * gcov-dump.c: Ditto.
3653 * doc/cpp.texi: Bump @copying's copyright year.
3654 * doc/cppinternals.texi: Ditto.
3655 * doc/gcc.texi: Ditto.
3656 * doc/gccint.texi: Ditto.
3657 * doc/gcov.texi: Ditto.
3658 * doc/install.texi: Ditto.
3659 * doc/invoke.texi: Ditto.
3661 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3663 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3665 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3667 * config/i386/sse.md (*mov<mode>_internal): Guard
3668 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3670 PR rtl-optimization/59647
3671 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3672 new_rtx into UNSIGNED_FLOAT rtxes.
3674 Copyright (C) 2014 Free Software Foundation, Inc.
3676 Copying and distribution of this file, with or without modification,
3677 are permitted in any medium without royalty provided the copyright
3678 notice and this notice are preserved.