1 2014-02-20 Martin Jambor <mjambor@suse.cz>
4 * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
5 info when checking whether lattices are bottom.
7 2014-02-20 Richard Biener <rguenther@suse.de>
10 * tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
13 2014-02-20 Jan Hubicka <hubicka@ucw.cz>
16 * ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale parameter
17 specifying the scaling.
18 (inline_call): Update.
19 (want_inline_recursively): Guard division by zero.
20 (recursive_inlining): Update.
21 * ipa-inline.h (clone_inlined_nodes): Update.
23 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
26 * config/i386/i386.c (classify_argument): Pass structures of size
27 64 bytes or less in register.
29 2014-02-20 Ilya Tocar <ilya.tocar@intel.com>
30 Kirill Yukhin <kirill.yukhin@intel.com>
32 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
33 (_mm_rcp28_round_ss): Ditto.
34 (_mm_rsqrt28_round_sd): Ditto.
35 (_mm_rsqrt28_round_ss): Ditto.
36 * config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
37 (_mm_rcp14_round_ss): Ditto.
38 (_mm_rsqrt14_round_sd): Ditto.
39 (_mm_rsqrt14_round_ss): Ditto.
40 * config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
41 the first input operand, get rid of match_dup.
42 (avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
44 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
46 (avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
47 operand as the first input operand, set type attribute.
48 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
50 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
51 operand as the first input operand, set type attribute.
53 2014-02-19 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
55 * config/rs6000/rs6000.c (vspltis_constant): Fix most significant
58 2014-02-19 H.J. Lu <hongjiu.lu@intel.com>
61 * config/i386/i386.c (construct_container): Remove TFmode check
62 for X86_64_INTEGER_CLASS.
64 2014-02-19 Uros Bizjak <ubizjak@gmail.com>
67 * config/i386/i386.c (type_natural_mode): Warn for ABI changes
68 only when -Wpsabi is enabled.
70 2014-02-19 Michael Hudson-Doyle <michael.hudson@linaro.org>
73 * config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
74 passing arrays in registers are the same as for structs, so remove the
75 special case for them.
77 2014-02-19 Eric Botcazou <ebotcazou@adacore.com>
79 * expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
80 destination type, extract only the valid bits if the source type is not
81 integral and has a different mode.
83 2014-02-19 Richard Biener <rguenther@suse.de>
86 * tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
89 2014-02-19 Richard Biener <rguenther@suse.de>
92 * ipa-prop.c: Include stringpool.h and tree-ssanames.h.
93 (ipa_modify_call_arguments): Emit an argument load explicitely and
94 preserve virtual SSA form there and for the replacement call.
95 Do not update SSA form nor free dominance info.
97 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
99 * ipa.c (function_and_variable_visibility): Also clear WEAK
100 flag when disolving COMDAT_GROUP.
102 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
104 * ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
105 * ipa-prop.c (ipa_set_jf_known_type): Return early when
107 (ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
108 do more sanity checks.
109 (detect_type_change): Return true when giving up early.
110 (compute_complex_assign_jump_func): Fix type parameter of
112 (compute_complex_ancestor_jump_func): Likewise.
113 (update_jump_functions_after_inlining): Fix updating of
115 * ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type to be NULL.
117 2014-02-18 Jan Hubicka <hubicka@ucw.cz>
119 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
120 inline clones when edge disappears.
122 2014-02-18 Michael Meissner <meissner@linux.vnet.ibm.com>
125 * config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
126 Split 64-bit moves into 2 patterns. Do not allow the use of
127 direct move for TDmode in little endian, since the decimal value
128 has little endian bytes within a word, but the 64-bit pieces are
129 ordered in a big endian fashion, and normal subreg's of TDmode are
131 (mov<mode>_64bit_dm): Likewise.
132 (movtd_64bit_nodm): Likewise.
134 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
136 PR tree-optimization/60174
137 * tree-ssa-reassoc.c (init_range_entry): Do not look into the defining
138 statement of an SSA_NAME that occurs in an abnormal PHI node.
140 2014-02-18 Jakub Jelinek <jakub@redhat.com>
143 * final.c (SEEN_BB): Remove.
144 (SEEN_NOTE, SEEN_EMITTED): Renumber.
145 (final_scan_insn): Don't force_source_line on second
146 NOTE_INSN_BASIC_BLOCK.
148 2014-02-18 Uros Bizjak <ubizjak@gmail.com>
151 * config/i386/i386.h (struct ix86_args): Add warn_avx512f.
152 * config/i386/i386.c (init_cumulative_args): Initialize warn_avx512f.
153 (type_natural_mode): Warn ABI change when %zmm register is not
154 available for AVX512F vector value passing.
156 2014-02-18 Kai Tietz <ktietz@redhat.com>
159 * config/i386/i386.c (ix86_expand_prologue): Use value in
160 rax register as displacement when restoring %r10 or %rax.
161 Fix wrong offset when restoring both registers.
163 2014-02-18 Eric Botcazou <ebotcazou@adacore.com>
165 * ipa-prop.c (compute_complex_ancestor_jump_func): Replace overzealous
166 assertion with conditional return.
168 2014-02-18 Jakub Jelinek <jakub@redhat.com>
169 Uros Bizjak <ubizjak@gmail.com>
172 * config/i386/driver-i386.c (host_detect_local_cpu): If
173 YMM state is not saved by the OS, also clear has_f16c. Move
174 CPUID 0x80000001 handling before YMM state saving checking.
176 2014-02-18 Andrey Belevantsev <abel@ispras.ru>
178 PR rtl-optimization/58960
179 * haifa-sched.c (alloc_global_sched_pressure_data): New,
180 factored out from ...
181 (sched_init): ... here.
182 (free_global_sched_pressure_data): New, factored out from ...
183 (sched_finish): ... here.
184 * sched-int.h (free_global_sched_pressure_data): Declare.
185 * sched-rgn.c (nr_regions_initial): New static global.
186 (haifa_find_rgns): Initialize it.
187 (schedule_region): Disable sched-pressure for the newly
190 2014-02-17 Richard Biener <rguenther@suse.de>
192 * tree-vect-stmts.c (free_stmt_vec_info): Clear BB and
193 release SSA defs of pattern stmts.
195 2014-02-17 Richard Biener <rguenther@suse.de>
197 * tree-inline.c (expand_call_inline): Release the virtual
198 operand defined by the call we are about to inline.
200 2014-02-17 Richard Biener <rguenther@suse.de>
202 * tree-ssa.c (verify_ssa): If verify_def found an error, ICE.
204 2014-02-17 Kirill Yukhin <kirill.yukhin@intel.com>
205 Ilya Tocar <ilya.tocar@intel.com>
207 * config/i386/avx512fintrin.h (_mm512_maskz_permutexvar_epi64): Swap
208 arguments order in builtin.
209 (_mm512_permutexvar_epi64): Ditto.
210 (_mm512_mask_permutexvar_epi64): Ditto
211 (_mm512_maskz_permutexvar_epi32): Ditto
212 (_mm512_permutexvar_epi32): Ditto
213 (_mm512_mask_permutexvar_epi32): Ditto
215 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
217 * config/rs6000/altivec.md (p8_vmrgew): Handle little endian targets.
218 (p8_vmrgow): Likewise.
220 2014-02-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
222 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Handle little
225 2014-02-15 Michael Meissner <meissner@linux.vnet.ibm.com>
228 * config/rs6000/rs6000.md (rreg): Add TFmode, TDmode constraints.
229 (mov<mode>_internal, TFmode/TDmode): Split TFmode/TDmode moves
230 into 64-bit and 32-bit moves. On 64-bit moves, add support for
231 using direct move instructions on ISA 2.07. Also adjust
232 instruction length for 64-bit.
233 (mov<mode>_64bit, TFmode/TDmode): Likewise.
234 (mov<mode>_32bit, TFmode/TDmode): Likewise.
236 2014-02-15 Alan Modra <amodra@gmail.com>
240 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
241 find_replacement on parts of insn rtl that might be reloaded.
243 2014-02-15 Richard Biener <rguenther@suse.de>
245 PR tree-optimization/60183
246 * tree-ssa-phiprop.c (propagate_with_phi): Avoid speculating loads.
247 (tree_ssa_phiprop): Calculate and free post-dominators.
249 2014-02-14 Jeff Law <law@redhat.com>
251 PR rtl-optimization/60131
252 * ree.c (get_extended_src_reg): New function.
253 (combine_reaching_defs): Use it rather than assuming location of REG.
254 (find_and_remove_re): Verify first operand of extension is
255 a REG before adding the insns to the copy list.
257 2014-02-14 Roland McGrath <mcgrathr@google.com>
259 * configure.ac (HAVE_AS_IX86_UD2): New test for 'ud2' mnemonic.
260 * configure: Regenerated.
261 * config.in: Regenerated.
262 * config/i386/i386.md (trap) [HAVE_AS_IX86_UD2]: Use the mnemonic
263 instead of ASM_SHORT.
265 2014-02-14 Vladimir Makarov <vmakarov@redhat.com>
266 Richard Earnshaw <rearnsha@arm.com>
268 PR rtl-optimization/59535
269 * lra-constraints.c (process_alt_operands): Encourage alternative
270 when unassigned pseudo class is superset of the alternative class.
271 (inherit_reload_reg): Don't inherit when optimizing for code size.
272 * config/arm/arm.h (MODE_BASE_REG_CLASS): Add version for LRA
273 returning CORE_REGS for anything but Thumb1 and BASE_REGS for
274 modes not less than 4 for Thumb1.
276 2014-02-14 Kyle McMartin <kyle@redhat.com>
279 * config/host-linux.c (TRY_EMPTY_VM_SPACE): Define for AArch64.
281 2014-02-14 Richard Biener <rguenther@suse.de>
283 * cilk-common.c (cilk_arrow): Build a MEM_REF, not an INDIRECT_REF.
284 (get_frame_arg): Drop the assert with langhook types_compatible_p.
285 Do not strip INDIRECT_REFs.
287 2014-02-14 Richard Biener <rguenther@suse.de>
290 * lto-streamer-out.c (DFS_write_tree_body): Do not follow
291 DECL_FUNCTION_SPECIFIC_TARGET.
292 (hash_tree): Do not hash DECL_FUNCTION_SPECIFIC_TARGET.
293 * tree-streamer-out.c (pack_ts_target_option): Remove.
294 (streamer_pack_tree_bitfields): Do not stream TS_TARGET_OPTION.
295 (write_ts_function_decl_tree_pointers): Do not stream
296 DECL_FUNCTION_SPECIFIC_TARGET.
297 * tree-streamer-in.c (unpack_ts_target_option): Remove.
298 (unpack_value_fields): Do not stream TS_TARGET_OPTION.
299 (lto_input_ts_function_decl_tree_pointers): Do not stream
300 DECL_FUNCTION_SPECIFIC_TARGET.
302 2014-02-14 Jakub Jelinek <jakub@redhat.com>
304 * tree-vect-loop.c (vect_is_slp_reduction): Don't set use_stmt twice.
305 (get_initial_def_for_induction, vectorizable_induction): Ignore
306 debug stmts when looking for exit_phi.
307 (vectorizable_live_operation): Fix up condition.
309 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
311 * config/nds32/nds32.c (nds32_asm_function_prologue): Do not use
312 nreverse() because it changes the content of original tree list.
314 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
316 * config/nds32/t-mlibs (MULTILIB_OPTIONS): Fix typo in comment.
317 * config/nds32/nds32.c (nds32_merge_decl_attributes): Likewise.
319 2014-02-14 Chung-Ju Wu <jasonwucj@gmail.com>
321 * config/nds32/nds32.c (nds32_naked_function_p): Follow the
322 GNU coding standards.
324 2014-02-13 Jakub Jelinek <jakub@redhat.com>
327 * dwarf2out.c (gen_subprogram_die): Don't call
328 add_calling_convention_attribute if subr_die is old_die.
330 2014-02-13 Sharad Singhai <singhai@google.com>
332 * doc/optinfo.texi: Fix order of nodes.
334 2014-02-13 Uros Bizjak <ubizjak@gmail.com>
336 * config/i386/sse.md (xop_vmfrcz<mode>2): Generate const0 in
337 operands[2], not operands[3].
339 2014-02-13 Richard Biener <rguenther@suse.de>
342 * doc/install.texi (ISL): Update recommended version to 0.12.2,
343 mention the possibility of an in-tree build.
344 (CLooG): Update recommended version to 0.18.1, mention the
345 possibility of an in-tree build and clarify that the ISL
346 bundled with CLooG does not work.
348 2014-02-13 Jakub Jelinek <jakub@redhat.com>
351 * expr.c (compress_float_constant): If x is a hard register,
352 extend into a pseudo and then move to x.
354 2014-02-13 Dominik Vogt <vogt@linux.vnet.ibm.com>
356 * config/s390/s390.c (s390_asm_output_function_label): Fix crash
357 caused by bad second argument to warning_at() with -mhotpatch and
358 nested functions (e.g. with gfortran).
360 2014-02-13 Richard Sandiford <rdsandiford@googlemail.com>
362 * opts.c (option_name): Remove "enabled by default" rider.
364 2014-02-12 John David Anglin <danglin@gcc.gnu.org>
366 * config/pa/pa.c (pa_option_override): Remove auto increment FIXME.
368 2014-02-12 H.J. Lu <hongjiu.lu@intel.com>
369 Uros Bizjak <ubizjak@gmail.com>
372 * configure.ac (HAVE_AS_GOTOFF_IN_DATA): Pass --32 to GNU assembler.
373 * configure: Regenerated.
375 2014-02-12 Richard Biener <rguenther@suse.de>
377 * vec.c (vec_prefix::calculate_allocation): Move as
378 inline variant to vec.h.
379 (vec_prefix::calculate_allocation_1): New out-of-line version.
380 * vec.h (vec_prefix::calculate_allocation_1): Declare.
381 (vec_prefix::m_has_auto_buf): Rename to ...
382 (vec_prefix::m_using_auto_storage): ... this.
383 (vec_prefix::calculate_allocation): Inline the easy cases
384 and dispatch to calculate_allocation_1 which doesn't need the
386 (va_heap::reserve): Use gcc_checking_assert.
387 (vec<T, A, vl_embed>::embedded_init): Add argument to initialize
388 m_using_auto_storage.
389 (auto_vec): Change m_vecpfx member to a vec<T, va_heap, vl_embed>
391 (vec<T, va_heap, vl_ptr>::reserve): Remove redundant check.
392 (vec<T, va_heap, vl_ptr>::release): Avoid casting.
393 (vec<T, va_heap, vl_ptr>::using_auto_storage): Simplify.
395 2014-02-12 Richard Biener <rguenther@suse.de>
397 * gcse.c (compute_transp): break from loop over canon_modify_mem_list
398 when we found a dependence.
400 2014-02-12 Thomas Schwinge <thomas@codesourcery.com>
402 * gimplify.c (gimplify_call_expr, gimplify_modify_expr): Move
404 (maybe_fold_stmt): ... into this new function.
405 * omp-low.c (lower_omp): Update comment.
407 * omp-low.c (lower_omp_target): Add clobber for sizes array, after
410 * omp-low.c (diagnose_sb_0): Make sure label_ctx is valid to
413 2014-02-12 James Greenhalgh <james.greenhalgh@arm.com>
415 * config/arm/aarch-cost-tables.h (generic_extra_costs): Fix
416 identifiers in comments.
417 (cortexa53_extra_costs): Likewise.
418 * config/arm/arm.c (cortexa9_extra_costs): Fix identifiers in comments.
419 (cortexa7_extra_costs): Likewise.
420 (cortexa12_extra_costs): Likewise.
421 (cortexa15_extra_costs): Likewise.
422 (v7m_extra_costs): Likewise.
424 2014-02-12 Richard Biener <rguenther@suse.de>
427 * gimple-low.c (lower_builtin_posix_memalign): Lower conditional
428 of posix_memalign being successful.
429 (lower_stmt): Restrict lowering of posix_memalign to when
430 -ftree-bit-ccp is enabled.
432 2014-02-12 Senthil Kumar Selvaraj <senthil_kumar.selvaraj@atmel.com>
434 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Pass vNULL for
436 * config/spu/spu-c.c (spu_resolve_overloaded_builtin): Likewise.
438 2014-02-12 Eric Botcazou <ebotcazou@adacore.com>
440 PR rtl-optimization/60116
441 * combine.c (try_combine): Also remove dangling REG_DEAD notes on the
442 other_insn once the combination has been validated.
444 2014-02-11 Jan Hubicka <hubicka@ucw.cz>
447 * ipa-utils.h (possible_polymorphic_call_targets): Update prototype
449 * ipa-devirt.c: Include demangle.h
450 (odr_violation_reported): New static variable.
451 (add_type_duplicate): Update odr_violations.
452 (maybe_record_node): Add completep parameter; update it.
453 (record_target_from_binfo): Add COMPLETEP parameter;
455 (possible_polymorphic_call_targets_1): Likewise.
456 (struct polymorphic_call_target_d): Add nonconstruction_targets;
457 rename FINAL to COMPLETE.
458 (record_targets_from_bases): Sanity check we found the binfo;
459 fix COMPLETEP updating.
460 (possible_polymorphic_call_targets): Add NONCONSTRUTION_TARGETSP
461 parameter, fix computing of COMPLETEP.
462 (dump_possible_polymorphic_call_targets): Imrove readability of dump;
463 at LTO time do demangling.
464 (ipa_devirt): Use nonconstruction_targets; Improve dumps.
465 * gimple-fold.c (gimple_get_virt_method_for_vtable): Add can_refer
467 (gimple_get_virt_method_for_binfo): Likewise.
468 * gimple-fold.h (gimple_get_virt_method_for_binfo,
469 gimple_get_virt_method_for_vtable): Update prototypes.
471 2014-02-11 Vladimir Makarov <vmakarov@redhat.com>
474 * genautomata.c (add_presence_absence): Fix typo with
475 {final_}presence_list.
477 2014-02-11 Michael Meissner <meissner@linux.vnet.ibm.com>
480 * config/rs6000/rs6000.md (128-bit GPR splitter): Add a splitter
481 for VSX/Altivec vectors that land in GPR registers.
483 2014-02-11 Richard Henderson <rth@redhat.com>
484 Jakub Jelinek <jakub@redhat.com>
487 * tree-sra.c (load_assign_lhs_subreplacements): Add VIEW_CONVERT_EXPR
488 around drhs if type conversion to lacc->type is not useless.
490 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
492 * config/aarch64/aarch64-cores.def (cortex-a57): Use cortexa57
494 (cortex-a57.cortex-a53): Likewise.
495 * config/aarch64/aarch64.c (cortexa57_tunings): New tuning struct.
497 2014-02-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
499 * config/arm/thumb2.md (*thumb2_movhi_insn): Add alternatives for
502 2014-02-11 Renlin Li <Renlin.Li@arm.com>
504 * doc/sourcebuild.texi: Document check_effective_target_arm_vfp3_ok and
505 add_options_for_arm_vfp3.
507 2014-02-11 Jeff Law <law@redhat.com>
510 * expr.c (expand_expr_addr_expr_1): Handle expand_expr returning an
511 object with an undesirable mode.
513 2014-02-11 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
516 * config/i386/sol2-9.h: New file.
517 * config.gcc (i[34567]86-*-solaris2* | x86_64-*-solaris2.1[0-9]*,
518 *-*-solaris2.9*): Use it.
520 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
522 * config/microblaze/microblaze.md: Add movsi4_rev insn pattern.
523 * config/microblaze/predicates.md: Add reg_or_mem_operand predicate.
525 2014-02-10 Nagaraju Mekala <nagaraju.mekala@xilinx.com>
527 * config/microblaze/microblaze.c: Extend mcpu version format
529 2014-02-10 David Holsgrove <david.holsgrove@xilinx.com>
531 * config/microblaze/microblaze.h: Define SIZE_TYPE and PTRDIFF_TYPE.
533 2014-02-10 Richard Henderson <rth@redhat.com>
536 * calls.c (expand_call): Don't double-push for reg_parm_stack_space.
537 * config/i386/i386.c (init_cumulative_args): Remove sorry for 64-bit
538 ms-abi vs -mno-accumulate-outgoing-args.
539 (ix86_expand_prologue): Unconditionally call ix86_eax_live_at_start_p.
540 * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Fix comment with
543 2014-02-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
546 * cfgexpand.c (expand_asm_operands): Attach source location to
547 ASM_INPUT rtx objects.
548 * print-rtl.c (print_rtx): Check for UNKNOWN_LOCATION.
550 2014-02-10 Nick Clifton <nickc@redhat.com>
552 * config/mn10300/mn10300.c (popcount): New function.
553 (mn10300_expand_prologue): Include saved registers in stack usage
556 2014-02-10 Jeff Law <law@redhat.com>
559 * reload1.c (emit_input_reload_insns): Do not create invalid RTL
560 when changing the SET_DEST of a prior insn to avoid an input reload.
562 2014-02-10 Ulrich Weigand <Ulrich.Weigand@de.ibm.com>
564 * config/rs6000/sysv4.h (ENDIAN_SELECT): Do not attempt to enforce
565 big-endian mode for -mcall-aixdesc, -mcall-freebsd, -mcall-netbsd,
566 -mcall-openbsd, or -mcall-linux.
567 (CC1_ENDIAN_BIG_SPEC): Remove.
568 (CC1_ENDIAN_LITTLE_SPEC): Remove.
569 (CC1_ENDIAN_DEFAULT_SPEC): Remove.
570 (CC1_SPEC): Remove (always empty) %cc1_endian_... spec.
571 (SUBTARGET_EXTRA_SPECS): Remove %cc1_endian_big, %cc1_endian_little,
572 and %cc1_endian_default.
573 * config/rs6000/sysv4le.h (CC1_ENDIAN_DEFAULT_SPEC): Remove.
575 2014-02-10 Richard Biener <rguenther@suse.de>
577 PR tree-optimization/60115
578 * tree-eh.c (tree_could_trap_p): Unify TARGET_MEM_REF and
579 MEM_REF handling. Properly verify that the accesses are not
580 out of the objects bound.
582 2014-02-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
584 * config/aarch64/aarch64.c (aarch64_override_options): Fix typo from
587 2014-02-10 Eric Botcazou <ebotcazou@adacore.com>
589 * ipa-devirt.c (get_polymorphic_call_info_from_invariant): Return
590 proper constants and fix formatting.
591 (possible_polymorphic_call_targets): Fix formatting.
593 2014-02-10 Kirill Yukhin <kirill.yukhin@intel.com>
594 Ilya Tocar <ilya.tocar@intel.com>
596 * config/i386/avx512fintrin.h (_mm512_storeu_epi64): Removed.
597 (_mm512_loadu_epi32): Renamed into...
598 (_mm512_loadu_si512): This.
599 (_mm512_storeu_epi32): Renamed into...
600 (_mm512_storeu_si512): This.
601 (_mm512_maskz_ceil_ps): Removed.
602 (_mm512_maskz_ceil_pd): Ditto.
603 (_mm512_maskz_floor_ps): Ditto.
604 (_mm512_maskz_floor_pd): Ditto.
605 (_mm512_floor_round_ps): Ditto.
606 (_mm512_floor_round_pd): Ditto.
607 (_mm512_ceil_round_ps): Ditto.
608 (_mm512_ceil_round_pd): Ditto.
609 (_mm512_mask_floor_round_ps): Ditto.
610 (_mm512_mask_floor_round_pd): Ditto.
611 (_mm512_mask_ceil_round_ps): Ditto.
612 (_mm512_mask_ceil_round_pd): Ditto.
613 (_mm512_maskz_floor_round_ps): Ditto.
614 (_mm512_maskz_floor_round_pd): Ditto.
615 (_mm512_maskz_ceil_round_ps): Ditto.
616 (_mm512_maskz_ceil_round_pd): Ditto.
617 (_mm512_expand_pd): Ditto.
618 (_mm512_expand_ps): Ditto.
619 * config/i386/i386.c (ix86_builtins): Remove
620 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK.
622 * config/i386/predicates.md (const1256_operand): New.
623 (const_1_to_2_operand): Ditto.
624 * config/i386/sse.md (avx512pf_gatherpf<mode>sf): Change hint value.
625 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
626 (*avx512pf_gatherpf<mode>sf): Ditto.
627 (avx512pf_gatherpf<mode>df): Ditto.
628 (*avx512pf_gatherpf<mode>df_mask): Ditto.
629 (*avx512pf_gatherpf<mode>df): Ditto.
630 (avx512pf_scatterpf<mode>sf): Ditto.
631 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
632 (*avx512pf_scatterpf<mode>sf): Ditto.
633 (avx512pf_scatterpf<mode>df): Ditto.
634 (*avx512pf_scatterpf<mode>df_mask): Ditto.
635 (*avx512pf_scatterpf<mode>df): Ditto.
636 (avx512f_expand<mode>): Removed.
637 (<shift_insn><mode>3<mask_name>): Change predicate type.
639 2014-02-08 Jakub Jelinek <jakub@redhat.com>
641 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers
642 not at the end of datarefs vector use ordered_remove to avoid
643 reordering datarefs vector.
646 * gimplify.c (gimplify_bind_expr): In ORT_SIMD region
647 mark local addressable non-static vars as GOVD_PRIVATE
648 instead of GOVD_LOCAL.
649 * omp-low.c (lower_omp_for): Move gimple_bind_vars
650 and BLOCK_VARS of gimple_bind_block to new_stmt rather
654 * tree-ssa-ccp.c (surely_varying_stmt_p): Don't return true
655 if TYPE_ATTRIBUTES (gimple_call_fntype ()) contain
656 assume_aligned or alloc_align attributes.
657 (bit_value_assume_aligned): Add ATTR, PTRVAL and ALLOC_ALIGN
658 arguments. Handle also assume_aligned and alloc_align attributes.
659 (evaluate_stmt): Adjust bit_value_assume_aligned caller. Handle
660 calls to functions with assume_aligned or alloc_align attributes.
661 * doc/extend.texi: Document assume_aligned and alloc_align attributes.
663 2014-02-08 Terry Guo <terry.guo@arm.com>
665 * doc/invoke.texi: Document ARM -march=armv7e-m.
667 2014-02-08 Jakub Jelinek <jakub@redhat.com>
669 * cilk-common.c (cilk_init_builtins): Clear TREE_NOTHROW
670 flag on __cilkrts_rethrow builtin.
673 * ipa-cp.c (determine_versionability): Fail at -O0
674 or __attribute__((optimize (0))) or -fno-ipa-cp functions.
675 * tree-sra.c (ipa_sra_preliminary_function_checks): Similarly.
678 2014-02-04 Jakub Jelinek <jakub@redhat.com>
681 * tree-inline.c (copy_forbidden): Fail for
682 __attribute__((optimize (0))) functions.
684 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
686 * varpool.c: Include pointer-set.h.
687 (varpool_remove_unreferenced_decls): Variables in other partitions
688 will not be output; be however careful to not lose information
691 2014-02-07 Jan Hubicka <hubicka@ucw.cz>
693 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do O(1)
694 lookup in the vtable constructor.
696 2014-02-07 Jeff Law <law@redhat.com>
699 * config/m68k/m68k.md (ashldi_extsi): Turn into a
700 define_insn_and_split.
702 * ipa-inline.c (inline_small_functions): Fix typos.
704 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
706 * config/s390/s390-protos.h (s390_can_use_simple_return_insn)
707 (s390_can_use_return_insn): Declare.
708 * config/s390/s390.h (EPILOGUE_USES): Define.
709 * config/s390/s390.c (s390_mainpool_start): Allow two main_pool
711 (s390_chunkify_start): Handle return JUMP_LABELs.
712 (s390_early_mach): Emit a main_pool instruction on the entry edge.
713 (s300_set_up_by_prologue, s390_can_use_simple_return_insn)
714 (s390_can_use_return_insn): New functions.
715 (s390_fix_long_loop_prediction): Handle conditional returns.
716 (TARGET_SET_UP_BY_PROLOGUE): Define.
717 * config/s390/s390.md (ANY_RETURN): New code iterator.
718 (*creturn, *csimple_return, return, simple_return): New patterns.
720 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
722 * config/s390/s390.c (s390_restore_gprs_from_fprs): Add REG_CFA_RESTORE
723 notes to each restore. Also add REG_CFA_DEF_CFA when restoring %r15.
724 (s390_optimize_prologue): Don't clear RTX_FRAME_RELATED_P. Update the
725 REG_CFA_RESTORE list when deciding not to restore a register.
727 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
729 * config/s390/s390.c: Include tree-pass.h and context.h.
730 (s390_early_mach): New function, split out from...
731 (s390_emit_prologue): ...here.
732 (pass_data_s390_early_mach): New pass structure.
733 (pass_s390_early_mach): New class.
734 (s390_option_override): Create and register early_mach pass.
737 2014-02-07 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
739 * var-tracking.c (vt_stack_adjustments): Don't require stack_adjusts
740 to match for the exit block.
742 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
744 * config/s390/s390.md ("atomic_load<mode>", "atomic_store<mode>")
745 ("atomic_compare_and_swap<mode>", "atomic_fetch_<atomic><mode>"):
746 Reject misaligned operands.
748 2014-02-07 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
750 * optabs.c (expand_atomic_compare_and_swap): Allow expander to fail.
752 2014-02-07 Richard Biener <rguenther@suse.de>
755 * gimple-low.c (lower_builtin_posix_memalign): New function.
756 (lower_stmt): Call it to lower posix_memalign in a way
757 to make alignment info accessible.
759 2014-02-07 Jakub Jelinek <jakub@redhat.com>
762 * tree.c (build_common_builtin_nodes): Set ECF_LEAF for
763 __builtin_setjmp_receiver.
765 2014-02-07 Richard Biener <rguenther@suse.de>
768 * builtin-types.def (BT_FN_INT_PTRPTR_SIZE_SIZE): Add.
769 * builtins.def (BUILT_IN_POSIX_MEMALIGN): Likewise.
770 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
771 Handle BUILT_IN_POSIX_MEMALIGN.
772 (find_func_clobbers): Likewise.
773 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
774 (call_may_clobber_ref_p_1): Likewise.
776 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
779 * ipa-devirt.c (record_target_from_binfo): Remove overactive
782 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
785 * lto-cgraph.c (lto_output_node): Use
786 symtab_get_symbol_partitioning_class.
787 (lto_output_varpool_node): likewise.
788 (symtab_get_symbol_partitioning_class): Move here from
790 * cgraph.h (symbol_partitioning_class): Likewise.
791 (symtab_get_symbol_partitioning_class): Declare.
793 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
795 * ggc.h (ggc_internal_cleared_alloc): New macro.
796 * vec.h (vec_safe_copy): Handle memory stats.
797 * omp-low.c (simd_clone_struct_alloc): Use ggc_internal_cleared_alloc.
798 * target-globals.c (save_target_globals): Likewise.
800 2014-02-06 Jan Hubicka <hubicka@ucw.cz>
803 * expr.c (emit_move_resolve_push): Export; be bit more selective
804 on when to clear alias set.
805 * expr.h (emit_move_resolve_push): Declare.
806 * function.h (struct function): Add tail_call_marked.
807 * tree-tailcall.c (optimize_tail_call): Set tail_call_marked.
808 * config/i386/i386-protos.h (ix86_expand_push): Remove.
809 * config/i386/i386.md (TImode move expander): De not call
811 (FP push expanders): Preserve memory attributes.
812 * config/i386/sse.md (push<mode>1): Remove.
813 * config/i386/i386.c (ix86_expand_vector_move): Handle push operation.
814 (ix86_expand_push): Remove.
815 * config/i386/mmx.md (push<mode>1): Remove.
817 2014-02-06 Jakub Jelinek <jakub@redhat.com>
819 PR rtl-optimization/60030
820 * internal-fn.c (ubsan_expand_si_overflow_mul_check): Surround
821 lopart with paradoxical subreg before shifting it up by hprec.
823 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
825 * config/arm/aarch-cost-tables.h (cortexa57_extra_costs): New table.
826 Remove extra newline at end of file.
827 * config/arm/arm.c (arm_cortex_a57_tune): New tuning struct.
828 (arm_issue_rate): Handle cortexa57.
829 * config/arm/arm-cores.def (cortex-a57): Use cortex_a57 tuning.
830 (cortex-a57.cortex-a53): Likewise.
832 2014-02-06 Jakub Jelinek <jakub@redhat.com>
835 * config/arm/arm.c (emit_multi_reg_push): Add dwarf_regs_mask argument,
836 don't record in REG_FRAME_RELATED_EXPR registers not set in that
838 (arm_expand_prologue): Adjust all callers.
839 (arm_unwind_emit_sequence): Allow saved, but not important for unwind
840 info, registers also at the lowest numbered registers side. Use
841 gcc_assert instead of abort, and SET_SRC/SET_DEST macros instead of
845 * var-tracking.c (adjust_mems): Before adding a SET to
846 amd->side_effects, adjust it's SET_SRC using simplify_replace_fn_rtx.
848 2014-02-06 Alan Modra <amodra@gmail.com>
851 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_mode): Only
852 change SDmode to DDmode when lra_in_progress.
854 2014-02-06 Jakub Jelinek <jakub@redhat.com>
857 * tree-vect-data-refs.c (vect_analyze_data_refs): For clobbers, call
858 free_data_ref on the dr first, and before goto again also set dr
859 to the next dr. For simd_lane_access, free old datarefs[i] before
860 overwriting it. For get_vectype_for_scalar_type failure, don't
861 free_data_ref if simd_lane_access.
863 * Makefile.in (prefix.o, cppbuiltin.o): Depend on $(BASEVER).
866 * tree.h (opts_for_fn): New inline function.
867 (opt_for_fn): Define.
868 * config/i386/i386.c (ix86_function_regparm): Use
869 opt_for_fn (decl, optimize) instead of optimize.
871 2014-02-06 Marcus Shawcroft <marcus.shawcroft@arm.com>
873 * config/aarch64/aarch64.c (aarch64_classify_symbol): Fix logic
874 for SYMBOL_REF in large memory model.
876 2014-02-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
878 * config/aarch64/aarch64-cores.def (cortex-a53): Specify CRC32
880 (cortex-a57): Likewise.
881 (cortex-a57.cortex-a53): Likewise.
883 2014-02-06 Yury Gribov <y.gribov@samsung.com>
884 Kugan Vivekanandarajah <kuganv@linaro.org>
886 * config/arm/arm.c (arm_vector_alignment_reachable): Check
888 * config/arm/arm.c (arm_builtin_support_vector_misalignment): Likewise.
890 2014-02-06 Richard Biener <rguenther@suse.de>
892 * tree-cfg.c (gimple_duplicate_sese_region): Fix ordering of
893 set_loop_copy and initialize_original_copy_tables.
895 2014-02-06 Alex Velenko <Alex.Velenko@arm.com>
897 * config/aarch64/aarch64-simd.md
898 (aarch64_ashr_simddi): Change QI to SI.
900 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
901 Jakub Jelinek <jakub@redhat.com>
904 * ipa-inline-analysis.c (compute_bb_predicates): Ensure monotonicity
907 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
909 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Change
910 CODE_FOR_altivec_vpku[hw]um to
911 CODE_FOR_altivec_vpku[hw]um_direct.
912 * config/rs6000/altivec.md (vec_unpacks_hi_<VP_small_lc>): Change
913 UNSPEC_VUNPACK_HI_SIGN to UNSPEC_VUNPACK_HI_SIGN_DIRECT.
914 (vec_unpacks_lo_<VP_small_lc>): Change UNSPEC_VUNPACK_LO_SIGN to
915 UNSPEC_VUNPACK_LO_SIGN_DIRECT.
917 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
919 * config/rs6000/altivec.md (altivec_vsum2sws): Adjust code
920 generation for -maltivec=be.
921 (altivec_vsumsws): Simplify redundant test.
923 2014-02-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
925 * altivec.md (UNSPEC_VPACK_UNS_UNS_MOD_DIRECT): New unspec.
926 (UNSPEC_VUNPACK_HI_SIGN_DIRECT): Likewise.
927 (UNSPEC_VUNPACK_LO_SIGN_DIRECT): Likewise.
928 (mulv8hi3): Use gen_altivec_vpkuwum_direct instead of
930 (altivec_vpkpx): Test for VECTOR_ELT_ORDER_BIG instead of for
932 (altivec_vpks<VI_char>ss): Likewise.
933 (altivec_vpks<VI_char>us): Likewise.
934 (altivec_vpku<VI_char>us): Likewise.
935 (altivec_vpku<VI_char>um): Likewise.
936 (altivec_vpku<VI_char>um_direct): New (copy of
937 altivec_vpku<VI_char>um that still relies on BYTES_BIG_ENDIAN, for
939 (altivec_vupkhs<VU_char>): Emit vupkls* instead of vupkhs* when
940 target is little endian and -maltivec=be is not specified.
941 (*altivec_vupkhs<VU_char>_direct): New (copy of
942 altivec_vupkhs<VU_char> that always emits vupkhs*, for internal use).
943 (altivec_vupkls<VU_char>): Emit vupkhs* instead of vupkls* when
944 target is little endian and -maltivec=be is not specified.
945 (*altivec_vupkls<VU_char>_direct): New (copy of
946 altivec_vupkls<VU_char> that always emits vupkls*, for internal use).
947 (altivec_vupkhpx): Emit vupklpx instead of vupkhpx when target is
948 little endian and -maltivec=be is not specified.
949 (altivec_vupklpx): Emit vupkhpx instead of vupklpx when target is
950 little endian and -maltivec=be is not specified.
952 2014-02-05 Richard Henderson <rth@redhat.com>
955 * combine-stack-adj.c: Revert r206943.
956 * sched-int.h (struct deps_desc): Add last_args_size.
957 * sched-deps.c (init_deps): Initialize it.
958 (sched_analyze_insn): Add OUTPUT dependencies between insns that
959 contain REG_ARGS_SIZE notes.
961 2014-02-05 Jan Hubicka <hubicka@ucw.cz>
963 * lto-cgraph.c (asm_nodes_output): Make global.
964 * lto-wrapper.c (run_gcc): Pass down paralelizm to WPA.
965 * gcc.c (AS_NEEDS_DASH_FOR_PIPED_INPUT): Allow WPA parameter
966 (driver_handle_option): Handle OPT_fwpa.
968 2014-02-05 Jakub Jelinek <jakub@redhat.com>
971 * ipa-devirt.c (possible_polymorphic_call_targets): Fix
972 a comment typo and formatting issue. If odr_hash hasn't been
973 created, return vNULL and set *completep to false.
976 * tree-eh.c (cleanup_empty_eh): Bail out on totally empty
977 bb with no successors.
979 2014-02-05 James Greenhalgh <james.greenhalgh@arm.com>
982 * doc/invoke.texi (-march): Clarify documentation for ARM.
986 2014-02-05 Richard Biener <rguenther@suse.de>
988 * tree-vect-loop.c (vect_analyze_loop_2): Be more informative
989 when not vectorizing because of too many alias checks.
990 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
991 Add more verboseness, avoid duplicate MSG_MISSED_OPTIMIZATION.
993 2014-02-05 Nick Clifton <nickc@redhat.com>
995 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Do not
996 accept extended registers in any mode when compiling for the MN10300.
998 2014-02-05 Yury Gribov <y.gribov@samsung.com>
1000 * cif-code.def (ATTRIBUTE_MISMATCH): New CIF code.
1001 * ipa-inline.c (report_inline_failed_reason): Handle mismatched
1002 sanitization attributes.
1003 (can_inline_edge_p): Likewise.
1004 (sanitize_attrs_match_for_inline_p): New function.
1006 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1008 * ipa-prop.c (detect_type_change): Shor circuit testing of
1009 type changes on THIS pointer.
1011 2014-02-04 John David Anglin <danglin@gcc.gnu.org>
1014 * config/pa/pa.c (legitimize_tls_address): Return original address
1015 if not passed a SYMBOL_REF rtx.
1016 (hppa_legitimize_address): Call legitimize_tls_address for all TLS
1018 (pa_emit_move_sequence): Simplify TLS source operands.
1019 (pa_legitimate_constant_p): Reject all TLS constants.
1020 * config/pa/pa.h (PA_SYMBOL_REF_TLS_P): Correct comment.
1021 (CONSTANT_ADDRESS_P): Reject TLS CONST addresses.
1023 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1025 * ipa.c (function_and_variable_visibility): Decompose DECL_ONE_ONLY
1026 groups when we know they are controlled by LTO.
1027 * varasm.c (default_binds_local_p_1): If object is in other partition,
1028 it will be resolved locally.
1030 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1032 * config/host-linux.c (linux_gt_pch_use_address): Don't
1033 use SSIZE_MAX because it is not always defined.
1035 2014-02-04 Vladimir Makarov <vmakarov@redhat.com>
1038 * lra-constraints.c (need_for_split_p): Use more 3 reloads as
1039 threshold for pseudo splitting.
1040 (update_ebb_live_info): Process call argument hard registers and
1041 hard registers from insn definition too.
1042 (max_small_class_regs_num): New constant.
1043 (inherit_in_ebb): Update live hard regs through EBBs. Update
1044 reloads_num only for small register classes. Don't split for
1047 2014-02-04 Markus Trippelsdorf <markus@trippelsdorf.de>
1050 * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that target
1053 2014-02-04 Jan Hubicka <hubicka@ucw.cz>
1055 * gimple-fold.c (can_refer_decl_in_current_unit_p): Default
1058 2014-02-04 Marek Polacek <polacek@redhat.com>
1060 * gdbinit.in (pel): Define.
1062 2014-02-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1064 * doc/invoke.texi (fstrict-volatile-bitfields): Clarify current
1067 2014-02-04 Richard Biener <rguenther@suse.de>
1070 * lto-streamer-out.c (tree_is_indexable): Force NAMELIST_DECLs
1071 in function context local.
1072 (lto_output_tree_ref): Do not write trees from lto_output_tree_ref.
1073 * lto-streamer-in.c (lto_input_tree_ref): Handle LTO_namelist_decl_ref
1074 similar to LTO_imported_decl_ref.
1076 2014-02-04 Jakub Jelinek <jakub@redhat.com>
1078 PR tree-optimization/60002
1079 * cgraphclones.c (build_function_decl_skip_args): Clear
1082 PR tree-optimization/60023
1083 * tree-if-conv.c (predicate_mem_writes): Pass true instead of
1084 false to gsi_replace.
1085 * tree-vect-stmts.c (vect_finish_stmt_generation): If stmt
1086 has been in some EH region and vec_stmt could throw, add
1087 vec_stmt into the same EH region.
1088 * tree-data-ref.c (get_references_in_stmt): If IFN_MASK_LOAD
1089 has no lhs, ignore it.
1090 * internal-fn.c (expand_MASK_LOAD): Likewise.
1093 * tree-inline.c (copy_forbidden): Fail for
1094 __attribute__((optimize (0))) functions.
1097 * omp-low.c (simd_clone_struct_copy): If from->inbranch
1098 is set, copy one less argument.
1099 (expand_simd_clones): Don't subtract clone_info->inbranch
1100 from simd_clone_struct_alloc argument.
1102 PR rtl-optimization/57915
1103 * recog.c (simplify_while_replacing): If all unary/binary/relational
1104 operation arguments are constant, attempt to simplify those.
1107 * expmed.c (expand_mult): For MODE_VECTOR_INT multiplication
1108 if there is no vashl<mode>3 or ashl<mode>3 insn, skip_synth.
1110 2014-02-04 Richard Biener <rguenther@suse.de>
1112 PR tree-optimization/60012
1113 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Apply
1114 TBAA disambiguation to all DDRs.
1116 2014-02-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1119 * config/sol2.h (LINK_LIBGCC_MAPFILE_SPEC): Define.
1120 (LINK_SPEC): Use it for -shared, -shared-libgcc.
1122 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1125 * tree.c (get_binfo_at_offset): Do not get confused by empty classes;
1127 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1129 * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Remove.
1130 * gimple-fold.h (gimple_extract_devirt_binfo_from_cst): Remove.
1132 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1135 * ipa-cp.c (ipa_get_indirect_edge_target_1): Use ipa-devirt
1136 to figure out targets of polymorphic calls with known decl.
1137 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1138 * ipa-utils.h (get_polymorphic_call_info_from_invariant): Declare.
1139 * ipa-devirt.c (get_polymorphic_call_info_for_decl): Break out from ...
1140 (get_polymorphic_call_info): ... here.
1141 (get_polymorphic_call_info_from_invariant): New function.
1143 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1145 * ipa-cp.c (ipa_get_indirect_edge_target_1): Do direct
1146 lookup via vtable pointer; check for type consistency
1147 and turn inconsitent facts into UNREACHABLE.
1148 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1149 * gimple-fold.c (gimple_get_virt_method_for_vtable): Do not ICE on
1150 type inconsistent querries; return UNREACHABLE instead.
1152 2014-02-03 Richard Henderson <rth@twiddle.net>
1155 * tree-ssa-uninit.c (push_to_worklist): Don't re-push if we've
1156 already processed this node.
1157 (normalize_one_pred_1): Pass along mark_set.
1158 (normalize_one_pred): Create and destroy a pointer_set_t.
1159 (normalize_one_pred_chain): Likewise.
1161 2014-02-03 Laurent Aflonsi <laurent.alfonsi@st.com>
1163 PR gcov-profile/58602
1164 * gcc/gcov-io.c (gcov_open): Open with truncation when mode < 0.
1166 2014-02-03 Jan Hubicka <hubicka@ucw.cz>
1169 * ipa-cp.c (ipa_get_indirect_edge_target_1): Give up on
1170 -fno-devirtualize; try to devirtualize by the knowledge of
1171 virtual table pointer given by aggregate propagation.
1172 * ipa-prop.c (try_make_edge_direct_virtual_call): Likewise.
1173 (ipa_print_node_jump_functions): Dump also offset that
1174 is relevant for polymorphic calls.
1175 (determine_known_aggregate_parts): Add arg_type parameter; use it
1176 instead of determining the type from pointer type.
1177 (ipa_compute_jump_functions_for_edge): Update call of
1178 determine_known_aggregate_parts.
1179 * gimple-fold.c (gimple_get_virt_method_for_vtable): Break out from ...
1180 (gimple_get_virt_method_for_binfo): ... here; simplify using
1181 vtable_pointer_value_to_vtable.
1182 * gimple-fold.h (gimple_get_virt_method_for_vtable): Declare.
1183 * ipa-devirt.c (subbinfo_with_vtable_at_offset): Turn OFFSET parameter
1184 to unsigned HOST_WIDE_INT; use vtable_pointer_value_to_vtable.
1185 (vtable_pointer_value_to_vtable): Break out from ...; handle also
1187 (vtable_pointer_value_to_binfo): ... here.
1188 * ipa-utils.h (vtable_pointer_value_to_vtable): Declare.
1190 2014-02-03 Teresa Johnson <tejohnson@google.com>
1192 * tree-vect-slp.c (vect_supported_load_permutation_p): Avoid
1193 redef of outer loop index variable.
1195 2014-02-03 Marc Glisse <marc.glisse@inria.fr>
1199 * doc/extend.texi (Function Attributes): Typo.
1201 2014-02-03 Cong Hou <congh@google.com>
1203 PR tree-optimization/60000
1204 * tree-vect-loop.c (vect_transform_loop): Set pattern_def_seq to NULL
1205 if the vectorized statement is a store. A store statement can only
1206 appear at the end of pattern statements.
1208 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1210 * config/i386/i386.c (flag_opts): Add -mlong-double-128.
1211 (ix86_option_override_internal): Default long double to 64-bit for
1212 32-bit Bionic and to 128-bit for 64-bit Bionic.
1214 * config/i386/i386.h (LONG_DOUBLE_TYPE_SIZE): Use 128 if
1215 TARGET_LONG_DOUBLE_128 is true.
1216 (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Likewise.
1218 * config/i386/i386.opt (mlong-double-80): Negate -mlong-double-64.
1219 (mlong-double-64): Negate -mlong-double-128.
1220 (mlong-double-128): New option.
1222 * config/i386/i386-c.c (ix86_target_macros): Define
1223 __LONG_DOUBLE_128__ for TARGET_LONG_DOUBLE_128.
1225 * doc/invoke.texi: Document -mlong-double-128.
1227 2014-02-03 H.J. Lu <hongjiu.lu@intel.com>
1229 PR rtl-optimization/60024
1230 * sel-sched.c (init_regs_for_mode): Check if mode is OK first.
1232 2014-02-03 Markus Trippelsdorf <markus@trippelsdorf.de>
1234 * doc/invoke.texi (fprofile-reorder-functions): Fix typo.
1236 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1238 PR rtl-optimization/57662
1239 * sel-sched.c (code_motion_path_driver): Do not mark already not
1240 existing blocks in the visiting bitmap.
1242 2014-02-03 Andrey Belevantsev <abel@ispras.ru>
1244 * sel-sched-ir.c (sel_gen_insn_from_expr_after): Reset INSN_DELETED_P
1245 on the insn being emitted.
1247 2014-02-03 James Greenhalgh <james.greenhalgh@arm.com>
1248 Will Deacon <will.deacon@arm.com>
1250 * doc/gimple.texi (gimple_asm_clear_volatile): Remove.
1252 2014-02-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1254 * config/arm/arm-tables.opt: Regenerate.
1256 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1258 * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): Generalize
1259 for vector types other than V16QImode.
1260 * config/rs6000/altivec.md (altivec_vperm_<mode>): Change to a
1261 define_expand, and call altivec_expand_vec_perm_le when producing
1262 code with little endian element order.
1263 (*altivec_vperm_<mode>_internal): New insn having previous
1264 behavior of altivec_vperm_<mode>.
1265 (altivec_vperm_<mode>_uns): Change to a define_expand, and call
1266 altivec_expand_vec_perm_le when producing code with little endian
1268 (*altivec_vperm_<mode>_uns_internal): New insn having previous
1269 behavior of altivec_vperm_<mode>_uns.
1271 2014-02-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1273 * config/rs6000/altivec.md (UNSPEC_VSUMSWS_DIRECT): New unspec.
1274 (altivec_vsumsws): Add handling for -maltivec=be with a little
1276 (altivec_vsumsws_direct): New.
1277 (reduc_splus_<mode>): Call gen_altivec_vsumsws_direct instead of
1278 gen_altivec_vsumsws.
1280 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1282 * ipa-devirt.c (subbinfo_with_vtable_at_offset,
1283 vtable_pointer_value_to_binfo): New functions.
1284 * ipa-utils.h (vtable_pointer_value_to_binfo): Declare.
1285 * ipa-prop.c (extr_type_from_vtbl_ptr_store): Use it.
1287 2014-02-02 Sandra Loosemore <sandra@codesourcery.com>
1289 * config/nios2/nios2.md (load_got_register): Initialize GOT
1290 pointer from _gp_got instead of _GLOBAL_OFFSET_TABLE_.
1291 * config/nios2/nios2.c (nios2_function_profiler): Likewise.
1293 2014-02-02 Jan Hubicka <hubicka@ucw.cz>
1295 * ipa-prop.c (update_jump_functions_after_inlining): When type is not
1296 preserverd by passthrough, do not propagate the type.
1298 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1300 * config/mips/mips.c (MIPS_GET_FCSR, MIPS_SET_FCSR): New macros.
1301 (mips_atomic_assign_expand_fenv): New function.
1302 (TARGET_ATOMIC_ASSIGN_EXPAND_FENV): Define.
1304 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1306 * doc/extend.texi (__builtin_mips_get_fcsr): Document.
1307 (__builtin_mips_set_fcsr): Likewise.
1308 * config/mips/mips-ftypes.def: Add MIPS_VOID_FTYPE_USI and
1309 MIPS_USI_FTYPE_VOID.
1310 * config/mips/mips-protos.h (mips16_expand_get_fcsr): Declare
1311 (mips16_expand_set_fcsr): Likewise.
1312 * config/mips/mips.c (mips16_get_fcsr_stub): New variable.
1313 (mips16_set_fcsr_stub): Likewise.
1314 (mips16_get_fcsr_one_only_stub): New class.
1315 (mips16_set_fcsr_one_only_stub): Likewise.
1316 (mips16_expand_get_fcsr, mips16_expand_set_fcsr): New functions.
1317 (mips_code_end): Output the get_fcsr and set_fcsr stubs, if needed.
1318 (BUILTIN_AVAIL_MIPS16, AVAIL_ALL): New macros.
1319 (hard_float): New availability predicate.
1320 (mips_builtins): Add get_fcsr and set_fcsr.
1321 (mips_expand_builtin): Check BUILTIN_AVAIL_MIPS16.
1322 * config/mips/mips.md (UNSPEC_GET_FCSR, UNSPEC_SET_FCSR): New unspecs.
1323 (GET_FCSR_REGNUM, SET_FCSR_REGNUM): New constants.
1324 (mips_get_fcsr, *mips_get_fcsr, mips_get_fcsr_mips16_<mode>)
1325 (mips_set_fcsr, *mips_set_fcsr, mips_set_fcsr_mips16_<mode>): New
1328 2014-02-02 Richard Sandiford <rdsandiford@googlemail.com>
1330 * config/mips/mips.c (mips_one_only_stub): New class.
1331 (mips_need_mips16_rdhwr_p): Replace with...
1332 (mips16_rdhwr_stub): ...this new variable.
1333 (mips16_stub_call_address): New function.
1334 (mips16_rdhwr_one_only_stub): New class.
1335 (mips_expand_thread_pointer): Use mips16_stub_call_address.
1336 (mips_output_mips16_rdhwr): Delete.
1337 (mips_finish_stub): New function.
1338 (mips_code_end): Use it to handle rdhwr stubs.
1340 2014-02-02 Uros Bizjak <ubizjak@gmail.com>
1343 * config/i386/i386.c (classify_argument): Fix handling of bit_offset
1344 when calculating size of integer atomic types.
1346 2014-02-02 H.J. Lu <hongjiu.lu@intel.com>
1348 * ipa-inline-analysis.c (true_predicate_p): Fix a typo in comments.
1350 2014-02-01 Jakub Jelinek <jakub@redhat.com>
1352 PR tree-optimization/60003
1353 * gimple-low.c (lower_builtin_setjmp): Set cfun->has_nonlocal_label.
1354 * profile.c (branch_prob): Use gimple_call_builtin_p
1355 to check for BUILT_IN_SETJMP_RECEIVER.
1356 * tree-inline.c (copy_bb): Call notice_special_calls.
1358 2014-01-31 Vladimir Makarov <vmakarov@redhat.com>
1361 * lra-constraints.c (process_alt_operands): Update reload_sum only
1364 2014-01-31 Richard Henderson <rth@redhat.com>
1367 * tree-eh.c (lower_try_finally_switch): Delay lowering finally block
1368 until after else_eh is processed.
1370 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1372 * config/i386/avx512fintrin.h (_MM_FROUND_TO_NEAREST_INT),
1373 (_MM_FROUND_TO_NEG_INF), (_MM_FROUND_TO_POS_INF),
1374 (_MM_FROUND_TO_ZERO), (_MM_FROUND_CUR_DIRECTION): Are already defined
1375 in smmintrin.h, remove them.
1376 (_MM_FROUND_NO_EXC): Same as above, bit also wrong value.
1377 * config/i386/i386.c (ix86_print_operand): Split sae and rounding.
1378 * config/i386/i386.md (ROUND_SAE): Fix value.
1379 * config/i386/predicates.md (const_4_or_8_to_11_operand): New.
1380 (const48_operand): New.
1381 * config/i386/subst.md (round), (round_expand): Use
1382 const_4_or_8_to_11_operand.
1383 (round_saeonly), (round_saeonly_expand): Use const48_operand.
1385 2014-01-31 Ilya Tocar <ilya.tocar@intel.com>
1387 * config/i386/constraints.md (Yk): Swap meaning with k.
1388 * config/i386/i386.md (movhi_internal): Change Yk to k.
1389 (movqi_internal): Ditto.
1390 (*k<logic><mode>): Ditto.
1393 (kandn<mode>): Ditto.
1394 (*<code>hi_1): Ditto.
1395 (*<code>qi_1): Ditto.
1396 (kxnor<mode>): Ditto.
1397 (kortestzhi): Ditto.
1398 (kortestchi): Ditto.
1400 (*one_cmplhi2_1): Ditto.
1401 (*one_cmplqi2_1): Ditto.
1402 * config/i386/sse.md (): Change k to Yk.
1403 (avx512f_load<mode>_mask): Ditto.
1404 (avx512f_blendm<mode>): Ditto.
1405 (avx512f_store<mode>_mask): Ditto.
1406 (avx512f_storeu<ssemodesuffix>512_mask): Ditto.
1407 (avx512f_storedqu<mode>_mask): Ditto.
1408 (avx512f_cmp<mode>3<mask_scalar_merge_name><round_saeonly_name>):
1410 (avx512f_ucmp<mode>3<mask_scalar_merge_name>): Ditto.
1411 (avx512f_vmcmp<mode>3<round_saeonly_name>): Ditto.
1412 (avx512f_vmcmp<mode>3_mask<round_saeonly_name>): Ditto.
1413 (avx512f_maskcmp<mode>3): Ditto.
1414 (avx512f_fmadd_<mode>_mask<round_name>): Ditto.
1415 (avx512f_fmadd_<mode>_mask3<round_name>): Ditto.
1416 (avx512f_fmsub_<mode>_mask<round_name>): Ditto.
1417 (avx512f_fmsub_<mode>_mask3<round_name>): Ditto.
1418 (avx512f_fnmadd_<mode>_mask<round_name>): Ditto.
1419 (avx512f_fnmadd_<mode>_mask3<round_name>): Ditto.
1420 (avx512f_fnmsub_<mode>_mask<round_name>): Ditto.
1421 (avx512f_fnmsub_<mode>_mask3<round_name>): Ditto.
1422 (avx512f_fmaddsub_<mode>_mask<round_name>): Ditto.
1423 (avx512f_fmaddsub_<mode>_mask3<round_name>): Ditto.
1424 (avx512f_fmsubadd_<mode>_mask<round_name>): Ditto.
1425 (avx512f_fmsubadd_<mode>_mask3<round_name>): Ditto.
1426 (avx512f_vextract<shuffletype>32x4_1_maskm): Ditto.
1427 (vec_extract_lo_<mode>_maskm): Ditto.
1428 (vec_extract_hi_<mode>_maskm): Ditto.
1429 (avx512f_vternlog<mode>_mask): Ditto.
1430 (avx512f_fixupimm<mode>_mask<round_saeonly_name>): Ditto.
1431 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Ditto.
1432 (avx512f_<code><pmov_src_lower><mode>2_mask): Ditto.
1433 (avx512f_<code>v8div16qi2_mask): Ditto.
1434 (avx512f_<code>v8div16qi2_mask_store): Ditto.
1435 (avx512f_eq<mode>3<mask_scalar_merge_name>_1): Ditto.
1436 (avx512f_gt<mode>3<mask_scalar_merge_name>): Ditto.
1437 (avx512f_testm<mode>3<mask_scalar_merge_name>): Ditto.
1438 (avx512f_testnm<mode>3<mask_scalar_merge_name>): Ditto.
1439 (*avx512pf_gatherpf<mode>sf_mask): Ditto.
1440 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1441 (*avx512pf_scatterpf<mode>sf_mask): Ditto.
1442 (*avx512pf_scatterpf<mode>df_mask): Ditto.
1443 (avx512cd_maskb_vec_dupv8di): Ditto.
1444 (avx512cd_maskw_vec_dupv16si): Ditto.
1445 (avx512f_vpermi2var<mode>3_maskz): Ditto.
1446 (avx512f_vpermi2var<mode>3_mask): Ditto.
1447 (avx512f_vpermi2var<mode>3_mask): Ditto.
1448 (avx512f_vpermt2var<mode>3_maskz): Ditto.
1449 (*avx512f_gathersi<mode>): Ditto.
1450 (*avx512f_gathersi<mode>_2): Ditto.
1451 (*avx512f_gatherdi<mode>): Ditto.
1452 (*avx512f_gatherdi<mode>_2): Ditto.
1453 (*avx512f_scattersi<mode>): Ditto.
1454 (*avx512f_scatterdi<mode>): Ditto.
1455 (avx512f_compress<mode>_mask): Ditto.
1456 (avx512f_compressstore<mode>_mask): Ditto.
1457 (avx512f_expand<mode>_mask): Ditto.
1458 * config/i386/subst.md (mask): Change k to Yk.
1459 (mask_scalar_merge): Ditto.
1462 2014-01-31 Marc Glisse <marc.glisse@inria.fr>
1464 * doc/extend.texi (Vector Extensions): Document ?: in C++.
1466 2014-01-31 Richard Biener <rguenther@suse.de>
1469 * builtins.c (fold_builtin_memory_op): Make sure to not
1470 use a floating-point mode or a boolean or enumeral type for
1473 2014-01-30 DJ Delorie <dj@redhat.com>
1475 * config/msp430/msp430.h (LIB_SPEC): Add -lcrt
1476 * config/msp430/msp430.md (msp430_refsym_need_exit): New.
1477 * config/msp430/msp430.c (msp430_expand_epilogue): Call it
1478 whenever main() has an epilogue.
1480 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1482 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Remove
1483 unused variable "field".
1484 * config/rs6000/vsx.md (vsx_mergel_<mode>): Add missing DONE.
1485 (vsx_mergeh_<mode>): Likewise.
1486 * config/rs6000/altivec.md (altivec_vmrghb): Likewise.
1487 (altivec_vmrghh): Likewise.
1488 (altivec_vmrghw): Likewise.
1489 (altivec_vmrglb): Likewise.
1490 (altivec_vmrglh): Likewise.
1491 (altivec_vmrglw): Likewise.
1492 (altivec_vspltb): Add missing uses.
1493 (altivec_vsplth): Likewise.
1494 (altivec_vspltw): Likewise.
1495 (altivec_vspltsf): Likewise.
1497 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1500 * ifcvt.c (cond_exec_process_insns): Don't conditionalize
1501 frame related instructions.
1503 2014-01-30 Vladimir Makarov <vmakarov@redhat.com>
1505 PR rtl-optimization/59959
1506 * lra-constrains.c (simplify_operand_subreg): Assign NO_REGS to
1507 any reload of register whose subreg is invalid.
1509 2014-01-30 Jakub Jelinek <jakub@redhat.com>
1511 * config/i386/f16cintrin.h (_cvtsh_ss): Avoid -Wnarrowing warning.
1512 * config/i386/avx512fintrin.h (_mm512_mask_cvtusepi64_storeu_epi32):
1513 Add missing return type - void.
1515 2014-01-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1517 * gcc/config/rs6000/rs6000.c (rs6000_expand_vector_init): Use
1518 gen_vsx_xxspltw_v4sf_direct instead of gen_vsx_xxspltw_v4sf;
1519 remove element index adjustment for endian (now handled in vsx.md
1521 (altivec_expand_vec_perm_const): Use
1522 gen_altivec_vsplt[bhw]_direct instead of gen_altivec_vsplt[bhw].
1523 * gcc/config/rs6000/vsx.md (UNSPEC_VSX_XXSPLTW): New unspec.
1524 (vsx_xxspltw_<mode>): Adjust element index for little endian.
1525 * gcc/config/rs6000/altivec.md (altivec_vspltb): Divide into a
1526 define_expand and a new define_insn *altivec_vspltb_internal;
1527 adjust for -maltivec=be on a little endian target.
1528 (altivec_vspltb_direct): New.
1529 (altivec_vsplth): Divide into a define_expand and a new
1530 define_insn *altivec_vsplth_internal; adjust for -maltivec=be on a
1531 little endian target.
1532 (altivec_vsplth_direct): New.
1533 (altivec_vspltw): Divide into a define_expand and a new
1534 define_insn *altivec_vspltw_internal; adjust for -maltivec=be on a
1535 little endian target.
1536 (altivec_vspltw_direct): New.
1537 (altivec_vspltsf): Divide into a define_expand and a new
1538 define_insn *altivec_vspltsf_internal; adjust for -maltivec=be on
1539 a little endian target.
1541 2014-01-30 Richard Biener <rguenther@suse.de>
1543 PR tree-optimization/59993
1544 * tree-ssa-forwprop.c (associate_pointerplus): Check we
1545 can propagate form the earlier stmt and avoid the transform
1546 when the intermediate result is needed.
1548 2014-01-30 Alangi Derick <alangiderick@gmail.com>
1550 * README.Portability: Fix typo.
1552 2014-01-30 David Holsgrove <david.holsgrove@xilinx.com>
1554 * config/microblaze/microblaze.md(cstoresf4, cbranchsf4): Replace
1555 comparison_operator with ordered_comparison_operator.
1557 2014-01-30 Nick Clifton <nickc@redhat.com>
1559 * config/mn10300/mn10300-protos.h (mn10300_store_multiple_operation_p):
1560 Rename to mn10300_store_multiple_regs.
1561 * config/mn10300/mn10300.c: Likewise.
1562 * config/mn10300/mn10300.md (store_movm): Fix typo: call
1563 store_multiple_regs.
1564 * config/mn10300/predicates.md (mn10300_store_multiple_operation):
1565 Call mn10300_store_multiple_regs.
1567 2014-01-30 Nick Clifton <nickc@redhat.com>
1568 DJ Delorie <dj@redhat.com>
1570 * config/rl78/rl78.c (register_sizes): Make the "upper half" of
1571 %fp 2 to keep registers after it properly word-aligned.
1572 (rl78_alloc_physical_registers_umul): Handle the case where both
1573 input operands are the same.
1575 2014-01-30 Richard Biener <rguenther@suse.de>
1577 PR tree-optimization/59903
1578 * tree-vect-loop.c (vect_transform_loop): Guard multiple-types
1581 2014-01-30 Jason Merrill <jason@redhat.com>
1584 * tree.c (walk_type_fields): Handle VECTOR_TYPE.
1587 * cgraphunit.c (expand_thunk): Copy volatile arg to a temporary.
1589 2014-01-30 Richard Biener <rguenther@suse.de>
1591 PR tree-optimization/59951
1592 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Skip uses in debug insns.
1594 2014-01-30 Savin Zlobec <savin.zlobec@gmail.com>
1597 * config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
1598 SFmode to DFmode case.
1600 2014-01-29 DJ Delorie <dj@redhat.com>
1602 * config/msp430/msp430.opt (-minrt): New.
1603 * config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
1605 (ENDFILE_SPEC): Likewise.
1607 2014-01-29 Jan Hubicka <hubicka@ucw.cz>
1609 * ipa-inline-analysis.c (clobber_only_eh_bb_p): New function.
1610 (estimate_function_body_sizes): Use it.
1612 2014-01-29 Paolo Carlini <paolo.carlini@oracle.com>
1615 * dwarf2out.c (is_cxx_auto): New.
1616 (is_base_type): Use it.
1617 (gen_type_die_with_usage): Likewise.
1619 2014-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1621 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Use
1622 CODE_FOR_altivec_vmrg*_direct rather than CODE_FOR_altivec_vmrg*.
1623 * config/rs6000/vsx.md (vsx_mergel_<mode>): Adjust for
1624 -maltivec=be with LE targets.
1625 (vsx_mergeh_<mode>): Likewise.
1626 * config/rs6000/altivec.md (UNSPEC_VMRG[HL]_DIRECT): New unspecs.
1627 (mulv8hi3): Use gen_altivec_vmrg[hl]w_direct.
1628 (altivec_vmrghb): Replace with define_expand and new
1629 *altivec_vmrghb_internal insn; adjust for -maltivec=be with LE targets.
1630 (altivec_vmrghb_direct): New define_insn.
1631 (altivec_vmrghh): Replace with define_expand and new
1632 *altivec_vmrghh_internal insn; adjust for -maltivec=be with LE targets.
1633 (altivec_vmrghh_direct): New define_insn.
1634 (altivec_vmrghw): Replace with define_expand and new
1635 *altivec_vmrghw_internal insn; adjust for -maltivec=be with LE targets.
1636 (altivec_vmrghw_direct): New define_insn.
1637 (*altivec_vmrghsf): Adjust for endianness.
1638 (altivec_vmrglb): Replace with define_expand and new
1639 *altivec_vmrglb_internal insn; adjust for -maltivec=be with LE targets.
1640 (altivec_vmrglb_direct): New define_insn.
1641 (altivec_vmrglh): Replace with define_expand and new
1642 *altivec_vmrglh_internal insn; adjust for -maltivec=be with LE targets.
1643 (altivec_vmrglh_direct): New define_insn.
1644 (altivec_vmrglw): Replace with define_expand and new
1645 *altivec_vmrglw_internal insn; adjust for -maltivec=be with LE targets.
1646 (altivec_vmrglw_direct): New define_insn.
1647 (*altivec_vmrglsf): Adjust for endianness.
1648 (vec_widen_umult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1649 (vec_widen_umult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1650 (vec_widen_smult_hi_v16qi): Use gen_altivec_vmrghh_direct.
1651 (vec_widen_smult_lo_v16qi): Use gen_altivec_vmrglh_direct.
1652 (vec_widen_umult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1653 (vec_widen_umult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1654 (vec_widen_smult_hi_v8hi): Use gen_altivec_vmrghw_direct.
1655 (vec_widen_smult_lo_v8hi): Use gen_altivec_vmrglw_direct.
1657 2014-01-29 Marcus Shawcroft <marcus.shawcroft@arm.com>
1659 * config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
1660 (aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
1663 2014-01-29 Richard Biener <rguenther@suse.de>
1665 PR tree-optimization/58742
1666 * tree-ssa-forwprop.c (associate_pointerplus): Rename to
1667 associate_pointerplus_align.
1668 (associate_pointerplus_diff): New function.
1669 (associate_pointerplus): Likewise. Call associate_pointerplus_align
1670 and associate_pointerplus_diff.
1672 2014-01-29 Richard Biener <rguenther@suse.de>
1674 * lto-streamer.h (LTO_major_version): Bump to 3.
1675 (LTO_minor_version): Reset to 0.
1677 2014-01-29 Renlin Li <Renlin.Li@arm.com>
1679 * config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
1680 * config/arm/arm.c (FL_FOR_ARCH7VE): New.
1681 (arm_file_start): Generate correct asm header for armv7ve.
1682 * config/arm/bpabi.h: Add multilib support for armv7ve.
1683 * config/arm/driver-arm.c: Change the architectures of cortex-a7
1684 and cortex-a15 to armv7ve.
1685 * config/arm/t-aprofile: Add multilib support for armv7ve.
1686 * doc/invoke.texi: Document -march=armv7ve.
1688 2014-01-29 Richard Biener <rguenther@suse.de>
1690 PR tree-optimization/58742
1691 * tree-ssa-forwprop.c (associate_plusminus): Return true
1692 if we changed sth, defer EH cleanup to ...
1693 (ssa_forward_propagate_and_combine): ... here. Call simplify_mult.
1694 (simplify_mult): New function.
1696 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1699 PR tree-optimization/59920
1700 * tree.c (build_common_builtin_nodes): Remove
1701 __builtin_setjmp_dispatcher initialization.
1702 * omp-low.h (make_gimple_omp_edges): Add a new int * argument.
1703 * profile.c (branch_prob): Use gsi_start_nondebug_after_labels_bb
1704 instead of gsi_after_labels + manually skipping debug stmts.
1705 Don't ignore bbs with BUILT_IN_SETJMP_DISPATCHER, instead
1706 ignore bbs with IFN_ABNORMAL_DISPATCHER.
1707 * tree-inline.c (copy_edges_for_bb): Remove
1708 can_make_abnormal_goto argument, instead add abnormal_goto_dest
1709 argument. Ignore computed_goto_p stmts. Don't call
1710 make_abnormal_goto_edges. If a call might need abnormal edges
1711 for non-local gotos, see if it already has an edge to
1712 IFN_ABNORMAL_DISPATCHER or if it is IFN_ABNORMAL_DISPATCHER
1713 with true argument, don't do anything then, otherwise add
1714 EDGE_ABNORMAL from the call's bb to abnormal_goto_dest.
1715 (copy_cfg_body): Compute abnormal_goto_dest, adjust copy_edges_for_bb
1717 * gimple-low.c (struct lower_data): Remove calls_builtin_setjmp.
1718 (lower_function_body): Don't emit __builtin_setjmp_dispatcher.
1719 (lower_stmt): Don't set data->calls_builtin_setjmp.
1720 (lower_builtin_setjmp): Adjust comment.
1721 * builtins.def (BUILT_IN_SETJMP_DISPATCHER): Remove.
1722 * tree-cfg.c (found_computed_goto): Remove.
1723 (factor_computed_gotos): Remove.
1724 (make_goto_expr_edges): Return bool, true for computed gotos.
1725 Don't call make_abnormal_goto_edges.
1726 (build_gimple_cfg): Don't set found_computed_goto, don't call
1727 factor_computed_gotos.
1728 (computed_goto_p): No longer static.
1729 (make_blocks): Don't set found_computed_goto.
1730 (get_abnormal_succ_dispatcher, handle_abnormal_edges): New functions.
1731 (make_edges): If make_goto_expr_edges returns true, push bb
1732 into ab_edge_goto vector, for stmt_can_make_abnormal_goto calls
1733 instead of calling make_abnormal_goto_edges push bb into ab_edge_call
1734 vector. Record mapping between bbs and OpenMP regions if there
1735 are any, adjust make_gimple_omp_edges caller. Call
1736 handle_abnormal_edges.
1737 (make_abnormal_goto_edges): Remove.
1738 * tree-cfg.h (make_abnormal_goto_edges): Remove.
1739 (computed_goto_p, get_abnormal_succ_dispatcher): New prototypes.
1740 * internal-fn.c (expand_ABNORMAL_DISPATCHER): New function.
1741 * builtins.c (expand_builtin): Don't handle BUILT_IN_SETJMP_DISPATCHER.
1742 * internal-fn.def (ABNORMAL_DISPATCHER): New.
1743 * omp-low.c (make_gimple_omp_edges): Add region_idx argument, when
1744 filling *region also set *region_idx to (*region)->entry->index.
1747 * read-rtl.c (read_rtx_code): Clear all of RTX_CODE_SIZE (code).
1748 For REGs set ORIGINAL_REGNO.
1750 2014-01-29 Bingfeng Mei <bmei@broadcom.com>
1752 * doc/md.texi: Mention that a target shouldn't implement
1753 vec_widen_(s|u)mul_even/odd pair if it is less efficient
1756 2014-01-29 Jakub Jelinek <jakub@redhat.com>
1758 PR tree-optimization/59594
1759 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Sort
1760 a copy of the datarefs vector rather than the vector itself.
1762 2014-01-28 Jason Merrill <jason@redhat.com>
1765 * dwarf2out.c (auto_die): New static.
1766 (gen_type_die_with_usage): Handle C++1y 'auto'.
1767 (gen_subprogram_die): If in-class DIE had 'auto', emit type again
1770 2014-01-28 H.J. Lu <hongjiu.lu@intel.com>
1773 * config/i386/gnu-user64.h (SPEC_32): Add "m16|" to "m32".
1774 (SPEC_X32): Likewise.
1775 (SPEC_64): Likewise.
1776 * config/i386/i386.c (ix86_option_override_internal): Turn off
1777 OPTION_MASK_ISA_64BIT, OPTION_MASK_ABI_X32 and OPTION_MASK_ABI_64
1779 (x86_file_start): Output .code16gcc for TARGET_16BIT.
1780 * config/i386/i386.h (TARGET_16BIT): New macro.
1781 (TARGET_16BIT_P): Likewise.
1782 * config/i386/i386.opt: Add m16.
1783 * doc/invoke.texi: Document -m16.
1785 2014-01-28 Jakub Jelinek <jakub@redhat.com>
1787 PR preprocessor/59935
1788 * input.c (location_get_source_line): Bail out on when line number
1789 is zero, and test the return value of lookup_or_add_file_to_cache_tab.
1791 2014-01-28 Richard Biener <rguenther@suse.de>
1793 PR tree-optimization/58742
1794 * tree-ssa-forwprop.c (associate_plusminus): Handle
1795 pointer subtraction of the form (T)(P + A) - (T)P.
1797 2014-01-28 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1799 * config/arm/arm.c (arm_new_rtx_costs): Remove useless statement
1802 2014-01-28 Richard Biener <rguenther@suse.de>
1805 2014-01-28 Richard Biener <rguenther@suse.de>
1807 PR rtl-optimization/45364
1808 PR rtl-optimization/59890
1809 * var-tracking.c (local_get_addr_clear_given_value): Handle
1810 already cleared slot.
1811 (val_reset): Handle not allocated local_get_addr_cache.
1812 (vt_find_locations): Use post-order on the inverted CFG.
1814 2014-01-28 Richard Biener <rguenther@suse.de>
1816 * tree-data-ref.h (ddr_is_anti_dependent, ddrs_have_anti_deps): Remove.
1818 2014-01-28 Richard Biener <rguenther@suse.de>
1820 PR rtl-optimization/45364
1821 PR rtl-optimization/59890
1822 * var-tracking.c (local_get_addr_clear_given_value): Handle
1823 already cleared slot.
1824 (val_reset): Handle not allocated local_get_addr_cache.
1825 (vt_find_locations): Use post-order on the inverted CFG.
1827 2014-01-28 Alan Modra <amodra@gmail.com>
1829 * Makefile.in (BUILD_CPPFLAGS): Do not use ALL_CPPFLAGS.
1830 * configure.ac <recursive call for build != host>: Define
1831 GENERATOR_FILE. Comment. Use CXX_FOR_BUILD, CXXFLAGS_FOR_BUILD
1832 and LD_FOR_BUILD too.
1833 * configure: Regenerate.
1835 2014-01-27 Allan Sandfeld Jensen <sandfeld@kde.org>
1837 * config/i386/i386.c (get_builtin_code_for_version): Separate
1838 Westmere from Nehalem, Ivy Bridge from Sandy Bridge and
1839 Broadwell from Haswell.
1841 2014-01-27 Steve Ellcey <sellcey@mips.com>
1843 * common/config/mips/mips-common.c (TARGET_DEFAULT_TARGET_FLAGS):
1844 Remove TARGET_FP_EXCEPTIONS_DEFAULT and MASK_FUSED_MADD.
1845 * config/mips/mips.c (mips_option_override): Change setting
1847 * config/mips/mips.h (TARGET_FP_EXCEPTIONS_DEFAULT): Remove.
1848 * config/mips/mips.opt (DSP, DSPR2, FP_EXCEPTIONS, FUSED_MADD, MIPS3D):
1849 Change from Mask to Var.
1851 2014-01-27 Jeff Law <law@redhat.com>
1853 * ipa-inline.c (inline_small_functions): Fix typo.
1855 2014-01-27 Ilya Tocar <ilya.tocar@intel.com>
1857 * config/i386/avx512fintrin.h (_mm512_mask_cvtepi32_storeu_epi8): New.
1858 (_mm512_mask_cvtsepi32_storeu_epi8): Ditto.
1859 (_mm512_mask_cvtusepi32_storeu_epi8): Ditto.
1860 (_mm512_mask_cvtepi32_storeu_epi16): Ditto.
1861 (_mm512_mask_cvtsepi32_storeu_epi16): Ditto.
1862 (_mm512_mask_cvtusepi32_storeu_epi16): Ditto.
1863 (_mm512_mask_cvtepi64_storeu_epi32): Ditto.
1864 (_mm512_mask_cvtsepi64_storeu_epi32): Ditto.
1865 (_mm512_mask_cvtusepi64_storeu_epi32): Ditto.
1866 (_mm512_mask_cvtepi64_storeu_epi16): Ditto.
1867 (_mm512_mask_cvtsepi64_storeu_epi16): Ditto.
1868 (_mm512_mask_cvtusepi64_storeu_epi16): Ditto.
1869 (_mm512_mask_cvtepi64_storeu_epi8): Ditto.
1870 (_mm512_mask_cvtsepi64_storeu_epi8): Ditto.
1871 (_mm512_mask_cvtusepi64_storeu_epi8): Ditto.
1872 (_mm512_storeu_epi64): Ditto.
1873 (_mm512_cmpge_epi32_mask): Ditto.
1874 (_mm512_cmpge_epu32_mask): Ditto.
1875 (_mm512_cmpge_epi64_mask): Ditto.
1876 (_mm512_cmpge_epu64_mask): Ditto.
1877 (_mm512_cmple_epi32_mask): Ditto.
1878 (_mm512_cmple_epu32_mask): Ditto.
1879 (_mm512_cmple_epi64_mask): Ditto.
1880 (_mm512_cmple_epu64_mask): Ditto.
1881 (_mm512_cmplt_epi32_mask): Ditto.
1882 (_mm512_cmplt_epu32_mask): Ditto.
1883 (_mm512_cmplt_epi64_mask): Ditto.
1884 (_mm512_cmplt_epu64_mask): Ditto.
1885 (_mm512_cmpneq_epi32_mask): Ditto.
1886 (_mm512_cmpneq_epu32_mask): Ditto.
1887 (_mm512_cmpneq_epi64_mask): Ditto.
1888 (_mm512_cmpneq_epu64_mask): Ditto.
1889 (_mm512_expand_pd): Ditto.
1890 (_mm512_expand_ps): Ditto.
1891 * config/i386/i386-builtin-types.def: Add PV16QI, PV16QI, PV16HI,
1892 VOID_PV8SI_V8DI_QI, VOID_PV8HI_V8DI_QI, VOID_PV16QI_V8DI_QI,
1893 VOID_PV16QI_V16SI_HI, VOID_PV16HI_V16SI_HI.
1894 * config/i386/i386.c (ix86_builtins): Add
1895 IX86_BUILTIN_EXPANDPD512_NOMASK, IX86_BUILTIN_EXPANDPS512_NOMASK,
1896 IX86_BUILTIN_PMOVDB512_MEM, IX86_BUILTIN_PMOVDW512_MEM,
1897 IX86_BUILTIN_PMOVQB512_MEM, IX86_BUILTIN_PMOVQD512_MEM,
1898 IX86_BUILTIN_PMOVQW512_MEM, IX86_BUILTIN_PMOVSDB512_MEM,
1899 IX86_BUILTIN_PMOVSDW512_MEM, IX86_BUILTIN_PMOVSQB512_MEM,
1900 IX86_BUILTIN_PMOVSQD512_MEM, IX86_BUILTIN_PMOVSQW512_MEM,
1901 IX86_BUILTIN_PMOVUSDB512_MEM, IX86_BUILTIN_PMOVUSDW512_MEM,
1902 IX86_BUILTIN_PMOVUSQB512_MEM, IX86_BUILTIN_PMOVUSQD512_MEM,
1903 IX86_BUILTIN_PMOVUSQW512_MEM.
1904 (bdesc_special_args): Add __builtin_ia32_pmovusqd512mem_mask,
1905 __builtin_ia32_pmovsqd512mem_mask,
1906 __builtin_ia32_pmovqd512mem_mask,
1907 __builtin_ia32_pmovusqw512mem_mask,
1908 __builtin_ia32_pmovsqw512mem_mask,
1909 __builtin_ia32_pmovqw512mem_mask,
1910 __builtin_ia32_pmovusdw512mem_mask,
1911 __builtin_ia32_pmovsdw512mem_mask,
1912 __builtin_ia32_pmovdw512mem_mask,
1913 __builtin_ia32_pmovqb512mem_mask,
1914 __builtin_ia32_pmovusqb512mem_mask,
1915 __builtin_ia32_pmovsqb512mem_mask,
1916 __builtin_ia32_pmovusdb512mem_mask,
1917 __builtin_ia32_pmovsdb512mem_mask,
1918 __builtin_ia32_pmovdb512mem_mask.
1919 (bdesc_args): Add __builtin_ia32_expanddf512,
1920 __builtin_ia32_expandsf512.
1921 (ix86_expand_special_args_builtin): Handle VOID_FTYPE_PV8SI_V8DI_QI,
1922 VOID_FTYPE_PV8HI_V8DI_QI, VOID_FTYPE_PV16HI_V16SI_HI,
1923 VOID_FTYPE_PV16QI_V8DI_QI, VOID_FTYPE_PV16QI_V16SI_HI.
1924 * config/i386/sse.md (unspec): Add UNSPEC_EXPAND_NOMASK.
1925 (avx512f_<code><pmov_src_lower><mode>2_mask_store): New.
1926 (*avx512f_<code>v8div16qi2_store_mask): Renamed to ...
1927 (avx512f_<code>v8div16qi2_mask_store): This.
1928 (avx512f_expand<mode>): New.
1930 2014-01-27 Kirill Yukhin <kirill.yukhin@intel.com>
1932 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i32gather_pd):
1934 (_mm512_mask_prefetch_i64gather_pd): Ditto.
1935 (_mm512_prefetch_i32scatter_pd): Ditto.
1936 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1937 (_mm512_prefetch_i64scatter_pd): Ditto.
1938 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1939 (_mm512_mask_prefetch_i32gather_ps): Fix operand type.
1940 (_mm512_mask_prefetch_i64gather_ps): Ditto.
1941 (_mm512_prefetch_i32scatter_ps): Ditto.
1942 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1943 (_mm512_prefetch_i64scatter_ps): Ditto.
1944 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1945 * config/i386/i386-builtin-types.def: Define
1946 VOID_FTYPE_QI_V8SI_PCINT64_INT_INT
1947 and VOID_FTYPE_QI_V8DI_PCINT64_INT_INT.
1948 * config/i386/i386.c (ix86_builtins): Define IX86_BUILTIN_GATHERPFQPD,
1949 IX86_BUILTIN_GATHERPFDPD, IX86_BUILTIN_SCATTERPFDPD,
1950 IX86_BUILTIN_SCATTERPFQPD.
1951 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_gatherpfdpd,
1952 __builtin_ia32_gatherpfdps, __builtin_ia32_gatherpfqpd,
1953 __builtin_ia32_gatherpfqps, __builtin_ia32_scatterpfdpd,
1954 __builtin_ia32_scatterpfdps, __builtin_ia32_scatterpfqpd,
1955 __builtin_ia32_scatterpfqps.
1956 (ix86_expand_builtin): Expand new built-ins.
1957 * config/i386/sse.md (avx512pf_gatherpf<mode>): Add SF suffix,
1958 fix memory access data type.
1959 (*avx512pf_gatherpf<mode>_mask): Ditto.
1960 (*avx512pf_gatherpf<mode>): Ditto.
1961 (avx512pf_scatterpf<mode>): Ditto.
1962 (*avx512pf_scatterpf<mode>_mask): Ditto.
1963 (*avx512pf_scatterpf<mode>): Ditto.
1964 (GATHER_SCATTER_SF_MEM_MODE): New.
1965 (avx512pf_gatherpf<mode>df): Ditto.
1966 (*avx512pf_gatherpf<mode>df_mask): Ditto.
1967 (*avx512pf_scatterpf<mode>df): Ditto.
1969 2014-01-27 Jakub Jelinek <jakub@redhat.com>
1972 * expmed.h (expmed_mode_index): Rework so that analysis and optimziers
1973 know when the MODE_PARTIAL_INT and MODE_VECTOR_INT cases can never be
1976 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1978 * common/config/arm/arm-common.c
1979 (arm_rewrite_mcpu): Handle multiple names.
1981 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
1983 2014-01-27 James Greenhalgh <james.greenhalgh@arm.com>
1985 * gimple-builder.h (create_gimple_tmp): Delete.
1987 2014-01-27 Christian Bruel <christian.bruel@st.com>
1989 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Fix remaining bytes after
1992 2014-01-26 John David Anglin <danglin@gcc.gnu.org>
1994 * config/pa/pa.md (call): Generate indirect long calls to non-local
1995 functions when outputing 32-bit code.
1996 (call_value): Likewise except for special call to buggy powf function.
1998 * config/pa/pa.c (pa_attr_length_indirect_call): Adjust length of
1999 portable runtime and PIC indirect calls.
2000 (pa_output_indirect_call): Remove unnecessary nop from portable runtime
2001 and PIC call sequences. Use ldo instead of blr to set return register
2002 in PIC call sequence.
2004 2014-01-25 Walter Lee <walt@tilera.com>
2006 * config/tilegx/sync.md (atomic_fetch_sub): Fix negation and
2007 avoid clobbering a live register.
2009 2014-01-25 Walter Lee <walt@tilera.com>
2011 * config/tilegx/tilegx-c.c (tilegx_cpu_cpp_builtins):
2012 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2}.
2013 * config/tilegx/tilepro-c.c (tilepro_cpu_cpp_builtins):
2014 Define __GCC_HAVE_SYNC_COMPARE_AND_SWAP_{1,2,4,8}.
2016 2014-01-25 Walter Lee <walt@tilera.com>
2018 * config/tilegx/tilegx.c (tilegx_function_arg): Start 16-byte
2019 arguments on even registers.
2020 (tilegx_gimplify_va_arg_expr): Align 16-byte var args to
2022 * config/tilegx/tilegx.h (STACK_BOUNDARY): Change to 16 bytes.
2023 (BIGGEST_ALIGNMENT): Ditto.
2024 (BIGGEST_FIELD_ALIGNMENT): Ditto.
2026 2014-01-25 Walter Lee <walt@tilera.com>
2028 * config/tilegx/tilegx.c (tilegx_gen_bundles): Delete barrier
2029 insns before bundling.
2030 * config/tilegx/tilegx.md (tile_network_barrier): Update comment.
2032 2014-01-25 Walter Lee <walt@tilera.com>
2034 * config/tilegx/tilegx.c (tilegx_expand_builtin): Set
2035 PREFETCH_SCHEDULE_BARRIER_P to true for prefetches.
2036 * config/tilepro/tilepro.c (tilepro_expand_builtin): Ditto.
2038 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2040 * config/mips/constraints.md (kl): Delete.
2041 * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Turn into
2042 define expands, using...
2043 (divmod<mode>4_mips16, udivmod<mode>4_mips16): ...these new
2044 instructions for MIPS16.
2045 (*divmod<mode>4, *udivmod<mode>4): New patterns, taken from the
2046 non-MIPS16 version of the old divmod<mode>4 and udivmod<mode>4.
2048 2014-01-25 Walter Lee <walt@tilera.com>
2050 * config/tilepro/tilepro.md (ctzdi2): Use register_operand predicate.
2054 2014-01-25 Walter Lee <walt@tilera.com>
2056 * config/tilegx/tilegx.c (tilegx_expand_to_rtl_hook): New.
2057 (TARGET_EXPAND_TO_RTL_HOOK): Define.
2059 2014-01-25 Richard Sandiford <rdsandiford@googlemail.com>
2061 * rtlanal.c (canonicalize_condition): Split out duplicated mode check.
2064 2014-01-25 Jakub Jelinek <jakub@redhat.com>
2066 * print-rtl.c (in_call_function_usage): New var.
2067 (print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
2068 EXPR_LIST mode as mode and not as reg note name.
2071 * cfgloopmanip.c (copy_loop_info): If
2072 loop->warned_aggressive_loop_optimizations, make sure
2073 the flag is set in target loop too.
2075 2014-01-24 Balaji V. Iyer <balaji.v.iyer@intel.com>
2077 * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
2079 * builtins.def: Likewise.
2080 * cilk.h (fn_contains_cilk_spawn_p): Likewise.
2081 * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
2082 * ira.c (ira_setup_eliminable_regset): Likewise.
2083 * omp-low.c (gate_expand_omp): Likewise.
2084 (execute_lower_omp): Likewise.
2085 (diagnose_sb_0): Likewise.
2086 (gate_diagnose_omp_blocks): Likewise.
2087 (simd_clone_clauses_extract): Likewise.
2090 2014-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2092 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
2093 correction for little endian...
2094 * config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
2097 2014-01-24 Jeff Law <law@redhat.com>
2099 PR tree-optimization/59919
2100 * tree-vrp.c (find_assert_locations_1): Do not register asserts
2101 for non-returning calls.
2103 2014-01-24 James Greenhalgh <james.greenhalgh@arm.com>
2105 * common/config/aarch64/aarch64-common.c
2106 (aarch64_rewrite_mcpu): Handle multiple names.
2107 * config/aarch64/aarch64.h
2108 (BIG_LITTLE_SPEC): Do not discard mcpu switches.
2110 2014-01-24 Dodji Seketeli <dodji@redhat.com>
2112 * input.c (add_file_to_cache_tab): Handle the case where fopen
2115 2014-01-23 H.J. Lu <hongjiu.lu@intel.com>
2118 * config/i386/i386.md (pushsf splitter): Get stack adjustment
2119 from push operand if code of push isn't PRE_DEC.
2121 2014-01-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2124 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
2125 -mquad-memory-atomic. Update -mquad-memory documentation to say
2126 it is only used for non-atomic loads/stores.
2128 * config/rs6000/predicates.md (quad_int_reg_operand): Allow either
2129 -mquad-memory or -mquad-memory-atomic switches.
2131 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
2132 -mquad-memory-atomic to ISA 2.07 support.
2134 * config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
2135 to separate support of normal quad word memory operations (ldq, stq)
2136 from the atomic quad word memory operations.
2138 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2139 support to separate non-atomic quad word operations from atomic
2140 quad word operations. Disable non-atomic quad word operations in
2141 little endian mode so that we don't have to swap words after the
2142 load and before the store.
2143 (quad_load_store_p): Add comment about atomic quad word support.
2144 (rs6000_opt_masks): Add -mquad-memory-atomic to the list of
2145 options printed with -mdebug=reg.
2147 * config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
2148 -mquad-memory-atomic as the test for whether we have quad word
2149 atomic instructions.
2150 (TARGET_SYNC_HI_QI): If either -mquad-memory-atomic, -mquad-memory,
2151 or -mp8-vector are used, allow byte/half-word atomic operations.
2153 * config/rs6000/sync.md (load_lockedti): Insure that the address
2154 is a proper indexed or indirect address for the lqarx instruction.
2155 On little endian systems, swap the hi/lo registers after the lqarx
2157 (load_lockedpti): Use indexed_or_indirect_operand predicate to
2158 insure the address is valid for the lqarx instruction.
2159 (store_conditionalti): Insure that the address is a proper indexed
2160 or indirect address for the stqcrx. instruction. On little endian
2161 systems, swap the hi/lo registers before doing the stqcrx.
2163 (store_conditionalpti): Use indexed_or_indirect_operand predicate to
2164 insure the address is valid for the stqcrx. instruction.
2166 * gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
2167 Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
2168 type of quad memory support is available.
2170 2014-01-23 Vladimir Makarov <vmakarov@redhat.com>
2173 * lra-constraints.c (simplify_operand_subreg): Spill pseudo if
2174 there is a danger of looping.
2176 2014-01-23 Pat Haugen <pthaugen@us.ibm.com>
2178 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2179 force flag_ira_loop_pressure if set via command line.
2181 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2183 * config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
2184 (ashr_simd): New builtin handling DI mode.
2185 * config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
2186 (aarch64_sshr_simddi): New match pattern.
2187 * config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
2188 (vshrd_n_s64): Likewise.
2189 * config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.
2191 2014-01-23 Nick Clifton <nickc@redhat.com>
2193 * config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
2194 (LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
2195 favour of mcu specific scripts.
2196 * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
2199 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2200 Alex Velenko <Alex.Velenko@arm.com>
2202 * config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
2203 (vaddv_s16): Likewise.
2204 (vaddv_s32): Likewise.
2205 (vaddv_u8): Likewise.
2206 (vaddv_u16): Likewise.
2207 (vaddv_u32): Likewise.
2208 (vaddvq_s8): Likewise.
2209 (vaddvq_s16): Likewise.
2210 (vaddvq_s32): Likewise.
2211 (vaddvq_s64): Likewise.
2212 (vaddvq_u8): Likewise.
2213 (vaddvq_u16): Likewise.
2214 (vaddvq_u32): Likewise.
2215 (vaddvq_u64): Likewise.
2216 (vaddv_f32): Likewise.
2217 (vaddvq_f32): Likewise.
2218 (vaddvq_f64): Likewise.
2219 (vmaxv_f32): Likewise.
2220 (vmaxv_s8): Likewise.
2221 (vmaxv_s16): Likewise.
2222 (vmaxv_s32): Likewise.
2223 (vmaxv_u8): Likewise.
2224 (vmaxv_u16): Likewise.
2225 (vmaxv_u32): Likewise.
2226 (vmaxvq_f32): Likewise.
2227 (vmaxvq_f64): Likewise.
2228 (vmaxvq_s8): Likewise.
2229 (vmaxvq_s16): Likewise.
2230 (vmaxvq_s32): Likewise.
2231 (vmaxvq_u8): Likewise.
2232 (vmaxvq_u16): Likewise.
2233 (vmaxvq_u32): Likewise.
2234 (vmaxnmv_f32): Likewise.
2235 (vmaxnmvq_f32): Likewise.
2236 (vmaxnmvq_f64): Likewise.
2237 (vminv_f32): Likewise.
2238 (vminv_s8): Likewise.
2239 (vminv_s16): Likewise.
2240 (vminv_s32): Likewise.
2241 (vminv_u8): Likewise.
2242 (vminv_u16): Likewise.
2243 (vminv_u32): Likewise.
2244 (vminvq_f32): Likewise.
2245 (vminvq_f64): Likewise.
2246 (vminvq_s8): Likewise.
2247 (vminvq_s16): Likewise.
2248 (vminvq_s32): Likewise.
2249 (vminvq_u8): Likewise.
2250 (vminvq_u16): Likewise.
2251 (vminvq_u32): Likewise.
2252 (vminnmv_f32): Likewise.
2253 (vminnmvq_f32): Likewise.
2254 (vminnmvq_f64): Likewise.
2256 2014-01-23 James Greenhalgh <james.greenhalgh@arm.com>
2258 * config/aarch64/aarch64-simd.md
2259 (aarch64_dup_lane<mode>): Correct lane number on big-endian.
2260 (aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
2261 (*aarch64_mul3_elt<mode>): Likewise.
2262 (*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
2263 (*aarch64_mul3_elt_to_64v2df): Likewise.
2264 (*aarch64_mla_elt<mode>): Likewise.
2265 (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
2266 (*aarch64_mls_elt<mode>): Likewise.
2267 (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
2268 (*aarch64_fma4_elt<mode>): Likewise.
2269 (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
2270 (*aarch64_fma4_elt_to_64v2df): Likewise.
2271 (*aarch64_fnma4_elt<mode>): Likewise.
2272 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2273 (*aarch64_fnma4_elt_to_64v2df): Likewise.
2274 (aarch64_sq<r>dmulh_lane<mode>): Likewise.
2275 (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
2276 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2277 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
2278 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2279 (aarch64_sqdmull_lane<mode>_internal): Likewise.
2280 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2282 2013-01-23 Alex Velenko <Alex.Velenko@arm.com>
2284 * config/aarch64/aarch64-simd.md
2285 (aarch64_be_checked_get_lane<mode>): New define_expand.
2286 * config/aarch64/aarch64-simd-builtins.def
2287 (BUILTIN_VALL (GETLANE, be_checked_get_lane, 0)):
2288 New builtin definition.
2289 * config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
2290 Use new safe be builtin.
2292 2014-01-23 Alex Velenko <Alex.Velenko@arm.com>
2294 * config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
2296 (aarch64_be_st1<mode>): Likewise.
2297 (aarch_ld1<VALL:mode>): Define_expand modified.
2298 (aarch_st1<VALL:mode>): Likewise.
2299 * config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
2300 (UNSPEC_ST1): Likewise.
2302 2014-01-23 David Holsgrove <david.holsgrove@xilinx.com>
2304 * config/microblaze/microblaze.md: Add trap insn and attribute
2306 2014-01-23 Dodji Seketeli <dodji@redhat.com>
2308 PR preprocessor/58580
2309 * input.h (location_get_source_line): Take an additional line_size
2311 (void diagnostics_file_cache_fini): Declare new function.
2312 * input.c (struct fcache): New type.
2313 (fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
2314 New static constants.
2315 (diagnostic_file_cache_init, total_lines_num)
2316 (lookup_file_in_cache_tab, evicted_cache_tab_entry)
2317 (add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
2318 (needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
2319 (get_next_line, read_next_line, goto_next_line, read_line_num):
2320 New static function definitions.
2321 (diagnostic_file_cache_fini): New function.
2322 (location_get_source_line): Take an additional output line_len
2323 parameter. Re-write using lookup_or_add_file_to_cache_tab and
2325 * diagnostic.c (diagnostic_finish): Call
2326 diagnostic_file_cache_fini.
2327 (adjust_line): Take an additional input parameter for the length
2328 of the line, rather than calculating it with strlen.
2329 (diagnostic_show_locus): Adjust the use of
2330 location_get_source_line and adjust_line with respect to their new
2331 signature. While displaying a line now, do not stop at the first
2332 null byte. Rather, display the zero byte as a space and keep
2333 going until we reach the size of the line.
2334 * Makefile.in: Add vec.o to OBJS-libcommon
2336 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2337 Ilya Tocar <ilya.tocar@intel.com>
2339 * config/i386/avx512fintrin.h (_mm512_kmov): New.
2340 * config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
2341 (__builtin_ia32_kmov16): Ditto.
2342 * config/i386/i386.md (UNSPEC_KMOV): New.
2345 2014-01-23 Kirill Yukhin <kirill.yukhin@intel.com>
2347 * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
2348 (_mm512_storeu_si512): Ditto.
2350 2014-01-23 Richard Sandiford <rdsandiford@googlemail.com>
2353 * rtl.h (get_referenced_operands): Declare.
2354 * recog.c (get_referenced_operands): New function.
2355 * config/mips/mips.c (mips_reorg_process_insns): Check which asm
2356 operands have been referenced when recording LO_SUM references.
2358 2014-01-22 David Holsgrove <david.holsgrove@xilinx.com>
2360 * config/microblaze/microblaze.md: Correct bswaphi2 insn.
2362 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2364 * config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
2365 Enable for generic and recent AMD targets.
2367 2014-01-22 Jan Hubicka <hubicka@ucw.cz>
2369 * combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
2370 ARG_SIZE note when adjustment was eliminated.
2372 2014-01-22 Jeff Law <law@redhat.com>
2374 PR tree-optimization/59597
2375 * tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
2376 in file. Accept new argument REGISTERING and use it to modify
2377 dump output appropriately.
2378 (register_jump_thread): Corresponding changes.
2379 (mark_threaded_blocks): Reinstate code to cancel unprofitable
2380 thread paths involving joiner blocks. Add code to dump cancelled
2381 jump threading paths.
2383 2014-01-22 Vladimir Makarov <vmakarov@redhat.com>
2385 PR rtl-optimization/59477
2386 * lra-constraints.c (inherit_in_ebb): Process call for living hard
2387 regs. Update reloads_num and potential_reload_hard_regs for all insns.
2389 2014-01-22 Tom Tromey <tromey@redhat.com>
2391 * config/i386/i386-interix.h (i386_pe_unique_section): Don't use
2393 * config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.
2395 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2397 PR rtl-optimization/59896
2398 * lra-constraints.c (process_alt_operands): Check unused note for
2399 matched operands of insn with no output reloads.
2401 2014-01-21 Richard Sandiford <rdsandiford@googlemail.com>
2403 * config/mips/mips.c (mips_move_to_gpr_cost): Add M16_REGS case.
2404 (mips_move_from_gpr_cost): Likewise.
2406 2014-01-21 Vladimir Makarov <vmakarov@redhat.com>
2408 PR rtl-optimization/59858
2409 * lra-constraints.c (SMALL_REGISTER_CLASS_P): Use
2410 ira_class_hard_regs_num.
2411 (process_alt_operands): Increase reject for dying matched operand.
2413 2014-01-21 Jakub Jelinek <jakub@redhat.com>
2416 * config/i386/i386.c (expand_small_movmem_or_setmem): If mode is
2417 smaller than size, perform several stores or loads and stores
2418 at dst + count - size to store or copy all of size bytes, rather
2419 than just last modesize bytes.
2421 2014-01-20 DJ Delorie <dj@redhat.com>
2423 * config/rl78/rl78.c (rl78_propogate_register_origins): Verify
2424 that CLOBBERs are REGs before propogating their values.
2426 2014-01-20 H.J. Lu <hongjiu.lu@intel.com>
2429 * cgraph.c (cgraph_inline_failed_string): Add type to DEFCIFCODE.
2430 (cgraph_inline_failed_type): New function.
2431 * cgraph.h (DEFCIFCODE): Add type.
2432 (cgraph_inline_failed_type_t): New enum.
2433 (cgraph_inline_failed_type): New prototype.
2434 * cif-code.def: Add CIF_FINAL_NORMAL to OK, FUNCTION_NOT_CONSIDERED,
2435 FUNCTION_NOT_OPTIMIZED, REDEFINED_EXTERN_INLINE,
2436 FUNCTION_NOT_INLINE_CANDIDATE, LARGE_FUNCTION_GROWTH_LIMIT,
2437 LARGE_STACK_FRAME_GROWTH_LIMIT, MAX_INLINE_INSNS_SINGLE_LIMIT,
2438 MAX_INLINE_INSNS_AUTO_LIMIT, INLINE_UNIT_GROWTH_LIMIT,
2439 RECURSIVE_INLINING, UNLIKELY_CALL, NOT_DECLARED_INLINED,
2440 OPTIMIZING_FOR_SIZE, ORIGINALLY_INDIRECT_CALL,
2441 INDIRECT_UNKNOWN_CALL, USES_COMDAT_LOCAL.
2442 Add CIF_FINAL_ERROR to UNSPECIFIED, BODY_NOT_AVAILABLE,
2443 FUNCTION_NOT_INLINABLE, OVERWRITABLE, MISMATCHED_ARGUMENTS,
2444 EH_PERSONALITY, NON_CALL_EXCEPTIONS, TARGET_OPTION_MISMATCH,
2445 OPTIMIZATION_MISMATCH.
2446 * tree-inline.c (expand_call_inline): Emit errors during
2447 early_inlining if cgraph_inline_failed_type returns CIF_FINAL_ERROR.
2449 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2452 * config/i386/sse.md (*andnot<mode>3<mask_name>): Handle MODE_V16SF
2453 mode attribute in insn output.
2455 2014-01-20 Eric Botcazou <ebotcazou@adacore.com>
2457 * output.h (output_constant): Delete.
2458 * varasm.c (output_constant): Make private.
2460 2014-01-20 Alex Velenko <Alex.Velenko@arm.com>
2462 * config/aarch64/aarch64-simd.md (vec_perm<mode>): Add BE check.
2464 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2467 * tree.h (fold_builtin_strcat): New prototype.
2468 * builtins.c (fold_builtin_strcat): No longer static. Add len
2469 argument, if non-NULL, don't call c_strlen. Optimize
2470 directly into __builtin_memcpy instead of __builtin_strcpy.
2471 (fold_builtin_2): Adjust fold_builtin_strcat caller.
2472 * gimple-fold.c (gimple_fold_builtin): Handle BUILT_IN_STRCAT.
2474 2014-01-20 Uros Bizjak <ubizjak@gmail.com>
2476 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2477 for SImode_address_operand operands, having only a REG argument.
2479 2014-01-20 Marcus Shawcroft <marcus.shawcroft@arm.com>
2481 * config/aarch64/aarch64-linux.h (GLIBC_DYNAMIC_LINKER): Expand
2482 loader name using mbig-endian.
2483 (LINUX_TARGET_LINK_SPEC): Pass linker -m flag.
2485 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2487 * doc/invoke.texi (-march): Clarify documentation for AArch64.
2491 2014-01-20 Tejas Belagod <tejas.belagod@arm.com>
2493 * config/aarch64/aarch64-protos.h
2494 (aarch64_cannot_change_mode_class_ptr): Declare.
2495 * config/aarch64/aarch64.c (aarch64_cannot_change_mode_class,
2496 aarch64_cannot_change_mode_class_ptr): New.
2497 * config/aarch64/aarch64.h (CANNOT_CHANGE_MODE_CLASS): Change to call
2498 backend hook aarch64_cannot_change_mode_class.
2500 2014-01-20 James Greenhalgh <james.greenhalgh@arm.com>
2502 * common/config/aarch64/aarch64-common.c
2503 (aarch64_handle_option): Don't handle any option order logic here.
2504 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not override
2505 selected_cpu, warn on architecture version mismatch.
2506 (aarch64_override_options): Fix parsing order for option strings.
2508 2014-01-20 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2509 Iain Sandoe <iain@codesourcery.com>
2512 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Fix unused variable
2513 warning. Amend comment to reflect current functionality.
2515 2014-01-20 Richard Biener <rguenther@suse.de>
2518 * builtins.c (fold_builtin_strcat): Remove case better handled
2519 by tree-ssa-strlen.c.
2521 2014-01-20 Alan Lawrence <alan.lawrence@arm.com>
2523 * config/aarch64/aarch64.opt
2524 (mcpu, march, mtune): Make case-insensitive.
2526 2014-01-20 Jakub Jelinek <jakub@redhat.com>
2529 * config/i386/i386.c (ix86_avoid_lea_for_addr): Return false
2530 if operands[1] is a REG or ZERO_EXTEND of a REG.
2532 2014-01-19 Jan Hubicka <hubicka@ucw.cz>
2534 * varasm.c (compute_reloc_for_constant): Use targetm.binds_local_p.
2536 2014-01-19 John David Anglin <danglin@gcc.gnu.org>
2538 * config/pa/pa.c (pa_attr_length_millicode_call): Correct length of
2539 long non-pic millicode calls.
2541 2014-01-19 Jan-Benedict Glaw <jbglaw@lug-owl.de>
2543 * config/vax/vax.h (FUNCTION_ARG_REGNO_P): Fix unused variable warning.
2545 2014-01-19 Kito Cheng <kito@0xlab.org>
2547 * builtins.c (expand_movstr): Check movstr expand done or fail.
2549 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2550 H.J. Lu <hongjiu.lu@intel.com>
2553 * config/i386/i386.md (*lea<mode>): Zero-extend return register
2554 to DImode for zero-extended addresses.
2556 2014-01-19 Jakub Jelinek <jakub@redhat.com>
2558 PR rtl-optimization/57763
2559 * bb-reorder.c (fix_crossing_unconditional_branches): Set JUMP_LABEL
2560 on the new indirect jump_insn and increment LABEL_NUSES (label).
2562 2014-01-18 H.J. Lu <hongjiu.lu@intel.com>
2566 * config.gcc (x86_archs): New variable.
2567 (x86_64_archs): Likewise.
2568 (x86_cpus): Likewise.
2569 Use $x86_archs, $x86_64_archs and $x86_cpus to check valid
2570 --with-arch/--with-cpu= options.
2571 Support --with-arch=/--with-cpu={nehalem,westmere,
2572 sandybridge,ivybridge,haswell,broadwell,bonnell,silvermont}.
2574 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2576 * config/i386/i386.c (ix86_adjust_cost): Reorder PROCESSOR_K8
2577 and PROCESSOR_ATHLON to simplify code. Move "memory" calculation.
2579 2014-01-18 Uros Bizjak <ubizjak@gmail.com>
2581 * config/i386/i386.md (*swap<mode>): Rename from swap<mode>.
2583 2014-01-18 Jakub Jelinek <jakub@redhat.com>
2586 * config/i386/i386-c.c (ix86_pragma_target_parse): Temporarily
2587 clear cpp_get_options (parse_in)->warn_unused_macros for
2588 ix86_target_macros_internal with cpp_define.
2590 2014-01-18 Richard Sandiford <rdsandiford@googlemail.com>
2592 * jump.c (delete_related_insns): Keep (use (insn))s.
2593 * reorg.c (redundant_insn): Check for barriers too.
2595 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2597 * config/i386/i386.c (ix86_split_lea_for_addr): Fix a comment typo.
2599 2014-01-17 John David Anglin <danglin@gcc.gnu.org>
2601 * config/pa/pa.c (pa_attr_length_indirect_call): Don't output a short
2602 call to $$dyncall when TARGET_LONG_CALLS is true.
2604 2014-01-17 Jeff Law <law@redhat.com>
2606 * ree.c (combine_set_extension): Temporarily disable test for
2607 changing number of hard registers.
2609 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2612 * ipa-inline-analysis.c (inline_free_summary):
2613 Do not free summary of aliases.
2615 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2618 * gimplify.c (gimplify_expr): Use create_tmp_var
2619 instead of create_tmp_var_raw. If cond doesn't have
2620 integral type, don't add the IFN_ANNOTATE builtin at all.
2622 2014-01-17 Martin Jambor <mjambor@suse.cz>
2625 * ipa-cp.c (prev_edge_clone): New variable.
2626 (grow_next_edge_clone_vector): Renamed to grow_edge_clone_vectors.
2627 Also resize prev_edge_clone vector.
2628 (ipcp_edge_duplication_hook): Also update prev_edge_clone.
2629 (ipcp_edge_removal_hook): New function.
2630 (ipcp_driver): Register ipcp_edge_removal_hook.
2632 2014-01-17 Andrew Pinski <apinski@cavium.com>
2633 Steve Ellcey <sellcey@mips.com>
2636 * config/mips/mips.c (mips_print_operand): Check operand mode instead
2639 2014-01-17 Jeff Law <law@redhat.com>
2642 * passes.def: Reorder pass_copy_prop, pass_unrolli, pass_ccp sequence
2643 so that pass_ccp runs first.
2645 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2647 * config/i386/i386.c (ix86_lea_outperforms): Use TARGET_XXX.
2648 (ix86_adjust_cost): Use !TARGET_XXX.
2649 (do_reorder_for_imul): Likewise.
2650 (swap_top_of_ready_list): Likewise.
2651 (ix86_sched_reorder): Likewise.
2653 2014-01-17 H.J. Lu <hongjiu.lu@intel.com>
2655 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2656 PROCESSOR_INTEL. Treat like PROCESSOR_GENERIC.
2657 * config/i386/i386.c (intel_memcpy): New. Duplicate slm_memcpy.
2658 (intel_memset): New. Duplicate slm_memset.
2659 (intel_cost): New. Duplicate slm_cost.
2660 (m_INTEL): New macro.
2661 (processor_target_table): Add "intel".
2662 (ix86_option_override_internal): Replace PROCESSOR_SILVERMONT
2663 with PROCESSOR_INTEL for "intel".
2664 (ix86_lea_outperforms): Support PROCESSOR_INTEL. Duplicate
2665 PROCESSOR_SILVERMONT.
2666 (ix86_issue_rate): Likewise.
2667 (ix86_adjust_cost): Likewise.
2668 (ia32_multipass_dfa_lookahead): Likewise.
2669 (swap_top_of_ready_list): Likewise.
2670 (ix86_sched_reorder): Likewise.
2671 (ix86_avoid_lea_for_addr): Check TARGET_AVOID_LEA_FOR_ADDR
2672 instead of TARGET_OPT_AGU.
2673 * config/i386/i386.h (TARGET_INTEL): New.
2674 (TARGET_AVOID_LEA_FOR_ADDR): Likewise.
2675 (processor_type): Add PROCESSOR_INTEL.
2676 * config/i386/x86-tune.def: Support m_INTEL. Duplicate m_SILVERMONT.
2677 Add X86_TUNE_AVOID_LEA_FOR_ADDR.
2679 2014-01-17 Marek Polacek <polacek@redhat.com>
2682 * gimple-fold.c (fold_array_ctor_reference): Don't fold if element
2685 2014-01-17 Richard Biener <rguenther@suse.de>
2687 PR tree-optimization/46590
2688 * opts.c (default_options_table): Add entries for
2689 OPT_fbranch_count_reg, OPT_fmove_loop_invariants and OPT_ftree_pta,
2690 all enabled at -O1 but not for -Og.
2691 * common.opt (fbranch-count-reg): Remove Init(1).
2692 (fmove-loop-invariants): Likewise.
2693 (ftree-pta): Likewise.
2695 2014-01-17 Jakub Jelinek <jakub@redhat.com>
2697 * config/i386/i386.c (ix86_data_alignment): For compatibility with
2698 (incorrect) GCC 4.8 and earlier alignment assumptions ensure we align
2699 decls to at least the GCC 4.8 used alignments.
2702 * tree-nested.c (convert_nonlocal_reference_stmt,
2703 convert_local_reference_stmt): For NAMELIST_DECLs in gimple_bind_vars
2704 of GIMPLE_BIND stmts, adjust associated decls.
2706 2014-01-17 Richard Biener <rguenther@suse.de>
2708 PR tree-optimization/46590
2709 * vec.h (vec<>::bseach): New member function implementing
2710 binary search according to C89 bsearch.
2711 (vec<>::qsort): Avoid calling ::qsort for vectors with sizes 0 or 1.
2712 * tree-ssa-loop-im.c (struct mem_ref): Make stored member a
2713 bitmap pointer again. Make accesses_in_loop a flat array.
2714 (mem_ref_obstack): New global.
2715 (outermost_indep_loop): Adjust for mem_ref->stored changes.
2716 (mark_ref_stored): Likewise.
2717 (ref_indep_loop_p_2): Likewise.
2718 (set_ref_stored_in_loop): New helper function.
2719 (mem_ref_alloc): Allocate mem_refs on the mem_ref_obstack obstack.
2720 (memref_free): Adjust.
2721 (record_mem_ref_loc): Simplify.
2722 (gather_mem_refs_stmt): Adjust.
2723 (sort_locs_in_loop_postorder_cmp): New function.
2724 (analyze_memory_references): Sort accesses_in_loop after
2725 loop postorder number.
2726 (find_ref_loc_in_loop_cmp): New function.
2727 (for_all_locs_in_loop): Find relevant cluster of locs in
2728 accesses_in_loop and iterate without recursion.
2729 (execute_sm): Avoid uninit warning.
2730 (struct ref_always_accessed): Simplify.
2731 (ref_always_accessed::operator ()): Likewise.
2732 (ref_always_accessed_p): Likewise.
2733 (tree_ssa_lim_initialize): Initialize mem_ref_obstack, compute
2734 loop postorder numbers here.
2735 (tree_ssa_lim_finalize): Free mem_ref_obstack and loop postorder
2738 2014-01-17 Jan Hubicka <hubicka@ucw.cz>
2741 * passes.c (rest_of_decl_compilation): Don't call varpool_finalize_decl
2742 on decls for which assemble_alias has been called.
2744 2014-01-17 Nick Clifton <nickc@redhat.com>
2746 * config/msp430/msp430.opt: (mcpu): New option.
2747 * config/msp430/msp430.c (msp430_mcu_name): Use target_mcu.
2748 (msp430_option_override): Parse target_cpu. If the MCU name
2749 matches a generic string, clear target_mcu.
2750 (msp430_attr): Allow numeric interrupt values up to 63.
2751 (msp430_expand_epilogue): No longer invert operand 1 of gen_popm.
2752 * config/msp430/msp430.h (ASM_SPEC): Convert -mcpu into a -mmcu
2754 * config/msp430/t-msp430: (MULTILIB_MATCHES): Remove mcu matches.
2756 * config/msp430/msp430.md (popm): Use %J rather than %I.
2757 (addsi3): Use msp430_nonimmediate_operand for operand 2.
2758 (addhi_cy_i): Use immediate_operand for operand 2.
2759 * doc/invoke.texi: Document -mcpu option.
2761 2014-01-17 Richard Biener <rguenther@suse.de>
2763 PR rtl-optimization/38518
2764 * df.h (df_analyze_loop): Declare.
2765 * df-core.c: Include cfgloop.h.
2766 (df_analyze_1): Split out main part of df_analyze.
2767 (df_analyze): Adjust.
2768 (loop_inverted_post_order_compute): New function.
2769 (loop_post_order_compute): Likewise.
2770 (df_analyze_loop): New function avoiding whole-function
2772 * loop-invariant.c (find_defs): Use df_analyze_loop.
2773 (find_invariants): Adjust.
2774 * loop-iv.c (iv_analysis_loop_init): Use df_analyze_loop.
2776 2014-01-17 Zhenqiang Chen <zhenqiang.chen@arm.com>
2778 * config/arm/arm.c (arm_v7m_tune): Set max_insns_skipped to 2.
2779 (thumb2_final_prescan_insn): Set max to MAX_INSN_PER_IT_BLOCK.
2781 2014-01-16 Ilya Enkovich <ilya.enkovich@intel.com>
2783 * ipa-ref.c (ipa_remove_stmt_references): Fix references
2784 traversal when removing references.
2786 2014-01-16 Jan Hubicka <hubicka@ucw.cz>
2789 * tree.c (get_binfo_at_offset): Look harder for virtual bases.
2791 2014-01-16 Bernd Schmidt <bernds@codesourcery.com>
2794 * reload.c (find_reloads_address_1): Do not use RELOAD_OTHER when
2795 pushing a reload for an autoinc when we had previously reloaded an
2796 inner part of the address.
2798 2014-01-16 Jakub Jelinek <jakub@redhat.com>
2800 * tree-vectorizer.h (struct _loop_vec_info): Add no_data_dependencies
2802 (LOOP_VINFO_NO_DATA_DEPENDENCIES): Define.
2803 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Clear it
2804 when not giving up or versioning for alias only because of
2806 (vect_analyze_data_ref_dependences): Set to true.
2807 * tree-vect-stmts.c (hoist_defs_of_uses): Return false if def_stmt
2809 (vectorizable_load): Use LOOP_VINFO_NO_DATA_DEPENDENCIES instead of
2810 LOOP_REQUIRES_VERSIONING_FOR_ALIAS, add && !nested_in_vect_loop
2814 * expr.c (expand_expr_real_1): Handle init == NULL_TREE.
2817 * config/i386/i386.c (ix86_expand_builtin): If target doesn't satisfy
2818 operand 0 predicate for gathers, use a new pseudo as subtarget.
2820 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2823 * lra-constraints.c (process_alt_operands): Add printing debug info.
2824 Check absence of input/output reloads for matched operands too.
2826 2014-01-16 Vladimir Makarov <vmakarov@redhat.com>
2828 PR rtl-optimization/59835
2829 * ira.c (ira_init_register_move_cost): Increase cost for
2832 2014-01-16 Alan Lawrence <alan.lawrence@arm.com>
2834 * config/arm/arm.opt (mcpu, march, mtune): Make case-insensitive.
2836 2014-01-16 Richard Earnshaw <rearnsha@arm.com>
2839 * aarch64.c (aarch64_split_128bit_move): Don't lookup REGNO on
2840 non-register objects. Use gen_(high/low)part more consistently.
2843 2014-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2846 * config/rs6000/rs6000.md (reload_vsx_from_gprsf): Add little
2847 endian support, remove tests for WORDS_BIG_ENDIAN.
2848 (p8_mfvsrd_3_<mode>): Likewise.
2849 (reload_gpr_from_vsx<mode>): Likewise.
2850 (reload_gpr_from_vsxsf): Likewise.
2851 (p8_mfvsrd_4_disf): Likewise.
2853 2014-01-16 Richard Biener <rguenther@suse.de>
2855 PR rtl-optimization/46590
2856 * lcm.c (compute_antinout_edge): Use postorder iteration.
2857 (compute_laterin): Use inverted postorder iteration.
2859 2014-01-16 Nick Clifton <nickc@redhat.com>
2862 * varasm.c (output_constant): Return the number of bytes actually
2864 (output_constructor_array_range): Update the field size with the
2865 number of bytes emitted by output_constant.
2866 (output_constructor_regular_field): Likewise. Also do not
2867 complain if the total number of bytes emitted is now greater
2868 than the expected fieldpos.
2869 * output.h (output_constant): Update prototype and descriptive comment.
2871 2014-01-16 Marek Polacek <polacek@redhat.com>
2874 * cgraph.c (gimple_check_call_args): Don't use DECL_ARG_TYPE if
2875 it is error_mark_node.
2877 2014-01-15 Uros Bizjak <ubizjak@gmail.com>
2879 * config/i386/i386.c (ix86_hard_regno_mode_ok): Use
2880 VALID_AVX256_REG_OR_OI_MODE.
2882 2014-01-15 Pat Haugen <pthaugen@us.ibm.com>
2884 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Check if
2885 current procedure should be profiled.
2887 2014-01-15 Andrew Pinski <apinski@cavium.com>
2889 * config/aarch64/aarch64.c (aarch64_register_move_cost): Correct cost
2890 of moving from/to the STACK_REG register class.
2892 2014-01-15 Richard Henderson <rth@redhat.com>
2895 * reginfo.c (global_regs_decl): Globalize.
2896 * rtl.h (global_regs_decl): Declare.
2897 * ira.c (do_reload): Diagnose frame_pointer_needed and it
2898 reserved via global_regs.
2900 2014-01-15 Teresa Johnson <tejohnson@google.com>
2902 * tree-ssa-sccvn.c (visit_reference_op_call): Handle NULL vdef.
2904 2014-01-15 Bill Schmidt <wschmidt@vnet.linux.ibm.com>
2906 * config/rs6000/altivec.md (mulv8hi3): Explicitly generate vmulesh
2907 and vmulosh rather than call gen_vec_widen_smult_*.
2908 (vec_widen_umult_even_v16qi): Test VECTOR_ELT_ORDER_BIG rather
2909 than BYTES_BIG_ENDIAN to determine use of even or odd instruction.
2910 (vec_widen_smult_even_v16qi): Likewise.
2911 (vec_widen_umult_even_v8hi): Likewise.
2912 (vec_widen_smult_even_v8hi): Likewise.
2913 (vec_widen_umult_odd_v16qi): Likewise.
2914 (vec_widen_smult_odd_v16qi): Likewise.
2915 (vec_widen_umult_odd_v8hi): Likewise.
2916 (vec_widen_smult_odd_v8hi): Likewise.
2917 (vec_widen_umult_hi_v16qi): Explicitly generate vmuleub and
2918 vmuloub rather than call gen_vec_widen_umult_*.
2919 (vec_widen_umult_lo_v16qi): Likewise.
2920 (vec_widen_smult_hi_v16qi): Explicitly generate vmulesb and
2921 vmulosb rather than call gen_vec_widen_smult_*.
2922 (vec_widen_smult_lo_v16qi): Likewise.
2923 (vec_widen_umult_hi_v8hi): Explicitly generate vmuleuh and vmulouh
2924 rather than call gen_vec_widen_umult_*.
2925 (vec_widen_umult_lo_v8hi): Likewise.
2926 (vec_widen_smult_hi_v8hi): Explicitly gnerate vmulesh and vmulosh
2927 rather than call gen_vec_widen_smult_*.
2928 (vec_widen_smult_lo_v8hi): Likewise.
2930 2014-01-15 Jeff Law <law@redhat.com>
2932 PR tree-optimization/59747
2933 * ree.c (find_and_remove_re): Properly handle case where a second
2934 eliminated extension requires widening a copy created for elimination
2935 of a prior extension.
2936 (combine_set_extension): Ensure that the number of hard regs needed
2937 for a destination register does not change when we widen it.
2939 2014-01-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
2941 * config.gcc (*-*-rtems*): Add t-rtems to tmake_file.
2942 (arm*-*-uclinux*eabi*): Do not override an existing tmake_file.
2943 (arm*-*-eabi* | arm*-*-symbianelf* | arm*-*-rtems*): Likwise.
2944 (arm*-*-rtems*): Use t-rtems from existing tmake_file.
2945 (avr-*-rtems*): Likewise.
2946 (bfin*-rtems*): Likewise.
2947 (moxie-*-rtems*): Likewise.
2948 (h8300-*-rtems*): Likewise.
2949 (i[34567]86-*-rtems*): Likewise.
2950 (lm32-*-rtems*): Likewise.
2951 (m32r-*-rtems*): Likewise.
2952 (m68k-*-rtems*): Likewise.
2953 (microblaze*-*-rtems*): Likewise.
2954 (mips*-*-rtems*): Likewise.
2955 (powerpc-*-rtems*): Likewise.
2956 (sh-*-rtems*): Likewise.
2957 (sparc-*-rtems*): Likewise.
2958 (sparc64-*-rtems*): Likewise.
2959 (v850-*-rtems*): Likewise.
2960 (m32c-*-rtems*): Likewise.
2962 2014-01-15 Vladimir Makarov <vmakarov@redhat.com>
2964 PR rtl-optimization/59511
2965 * ira.c (ira_init_register_move_cost): Use memory costs for some
2966 cases of register move cost calculations.
2967 * lra-constraints.c (lra_constraints): Use REG_FREQ_FROM_BB
2968 instead of BB frequency.
2969 * lra-coalesce.c (move_freq_compare_func, lra_coalesce): Ditto.
2970 * lra-assigns.c (find_hard_regno_for): Ditto.
2972 2014-01-15 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/59822
2975 * tree-vect-stmts.c (hoist_defs_of_uses): New function.
2976 (vectorizable_load): Use it to hoist defs of uses of invariant
2977 loads out of the loop.
2979 2014-01-15 Matthew Gretton-Dann <matthew.gretton-dann@linaro.org>
2980 Kugan Vivekanandarajah <kuganv@linaro.org>
2983 * config/aarch64/aarch64.c (aarch64_build_constant): Fix incorrect
2986 2014-01-15 Richard Biener <rguenther@suse.de>
2988 PR rtl-optimization/59802
2989 * lcm.c (compute_available): Use inverted postorder to seed
2990 the initial worklist.
2992 2014-01-15 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
2995 * config/s390/s390.c (s390_preferred_reload_class): Don't return
2996 ADDR_REGS for invalid symrefs in non-PIC code.
2998 2014-01-15 Jakub Jelinek <jakub@redhat.com>
3001 * builtins.c (determine_block_size): Initialize *probable_max_size
3002 even if len_rtx is CONST_INT.
3004 2014-01-14 Andrew Pinski <apinski@cavium.com>
3006 * config/aarch64/aarch64-protos.h (tune_params): Add issue_rate.
3007 * config/aarch64/aarch64.c (generic_tunings): Add issue rate of 2.
3008 (cortexa53_tunings): Likewise.
3009 (aarch64_sched_issue_rate): New function.
3010 (TARGET_SCHED_ISSUE_RATE): Define.
3012 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3014 * ira-costs.c (find_costs_and_classes): Add missed
3015 ira_init_register_move_cost_if_necessary.
3017 2014-01-14 Vladimir Makarov <vmakarov@redhat.com>
3020 * config/arm/arm.c (arm_coproc_mem_operand): Add lra_in_progress.
3022 2014-01-14 H.J. Lu <hongjiu.lu@intel.com>
3025 * config/i386/i386.c (type_natural_mode): Add a bool parameter
3026 to indicate if type is used for function return value. Warn ABI
3027 change if the vector mode isn't available for function return value.
3028 (ix86_function_arg_advance): Pass false to type_natural_mode.
3029 (ix86_function_arg): Likewise.
3030 (ix86_gimplify_va_arg): Likewise.
3031 (function_arg_32): Don't warn ABI change.
3032 (ix86_function_value): Pass true to type_natural_mode.
3033 (ix86_return_in_memory): Likewise.
3034 (ix86_struct_value_rtx): Removed.
3035 (TARGET_STRUCT_VALUE_RTX): Likewise.
3037 2014-01-14 Richard Sandiford <rsandifo@linux.vnet.ibm.com>
3039 * jump.c (redirect_jump_2): Remove REG_CROSSING_JUMP notes when
3040 converting a conditional jump into a conditional return.
3042 2014-01-14 Richard Biener <rguenther@suse.de>
3044 PR tree-optimization/58921
3045 PR tree-optimization/59006
3046 * tree-vect-loop-manip.c (vect_loop_versioning): Remove code
3047 hoisting invariant stmts.
3048 * tree-vect-stmts.c (vectorizable_load): Insert the splat of
3049 invariant loads on the preheader edge if possible.
3051 2014-01-14 Joey Ye <joey.ye@arm.com>
3053 * doc/plugin.texi (Building GCC plugins): Update to C++.
3055 2014-01-14 Kirill Yukhin <kirill.yukhin@intel.com>
3057 * config/i386/avx512erintrin.h (_mm_rcp28_round_sd): New.
3058 (_mm_rcp28_round_ss): Ditto.
3059 (_mm_rsqrt28_round_sd): Ditto.
3060 (_mm_rsqrt28_round_ss): Ditto.
3061 (_mm_rcp28_sd): Ditto.
3062 (_mm_rcp28_ss): Ditto.
3063 (_mm_rsqrt28_sd): Ditto.
3064 (_mm_rsqrt28_ss): Ditto.
3065 * config/i386/avx512fintrin.h (_mm512_stream_load_si512): Ditto.
3066 * config/i386/i386-builtin-types.def (V8DI_FTYPE_PV8DI): Ditto.
3067 * config/i386/i386.c (IX86_BUILTIN_MOVNTDQA512): Ditto.
3068 (IX86_BUILTIN_RCP28SD): Ditto.
3069 (IX86_BUILTIN_RCP28SS): Ditto.
3070 (IX86_BUILTIN_RSQRT28SD): Ditto.
3071 (IX86_BUILTIN_RSQRT28SS): Ditto.
3072 (bdesc_special_args): Define __builtin_ia32_movntdqa512,
3073 __builtin_ia32_rcp28sd_round, __builtin_ia32_rcp28ss_round,
3074 __builtin_ia32_rsqrt28sd_round, __builtin_ia32_rsqrt28ss_round.
3075 (ix86_expand_special_args_builtin): Expand new FTYPE.
3076 * config/i386/sse.md (define_mode_attr "sse4_1_avx2"): Expand to V8DI.
3077 (srcp14<mode>): Make insn unary.
3078 (avx512f_vmscalef<mode><round_name>): Use substed predicate.
3079 (avx512f_sgetexp<mode><round_saeonly_name>): Ditto.
3080 (avx512f_rndscale<mode><round_saeonly_name>): Ditto.
3081 (<sse4_1_avx2>_movntdqa): Extend to 512 bits.
3082 (avx512er_exp2<mode><mask_name><round_saeonly_name>):
3083 Fix rounding: make it SAE only.
3084 (<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
3086 (<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
3088 (avx512er_vmrcp28<mode><round_saeonly_name>): Ditto.
3089 (avx512er_vmrsqrt28<mode><round_saeonly_name>): Ditto.
3090 (avx512f_getmant<mode><mask_name><round_saeonly_name>): Ditto.
3091 * config/i386/subst.md (round_saeonly_mask_scalar_operand3): Remove.
3092 (round_saeonly_mask_scalar_operand4): Ditto.
3093 (round_saeonly_mask_scalar_op3): Ditto.
3094 (round_saeonly_mask_scalar_op4): Ditto.
3096 2014-01-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3098 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3099 Implement -maltivec=be for vec_insert and vec_extract.
3101 2014-01-10 DJ Delorie <dj@redhat.com>
3103 * config/msp430/msp430.md (call_internal): Don't allow memory
3104 references with SP as the base register.
3105 (call_value_internal): Likewise.
3106 * config/msp430/constraints.md (Yc): New. For memory references
3107 that don't use SP as a base register.
3109 * config/msp430/msp430.c (msp430_print_operand): Add 'J' to mean
3110 "an integer without a # prefix"
3111 * config/msp430/msp430.md (epilogue_helper): Use it.
3113 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3116 * config/i386/i386.c (ix86_vectorize_builtin_gather): Uncomment
3117 AVX512F gather builtins.
3118 * tree-vect-stmts.c (vectorizable_mask_load_store): For now punt
3119 on gather decls with INTEGER_TYPE masktype.
3120 (vectorizable_load): For INTEGER_TYPE masktype, put the INTEGER_CST
3121 directly into the builtin rather than hoisting it before loop.
3123 PR tree-optimization/59387
3124 * tree-scalar-evolution.c: Include gimple-fold.h and gimplify-me.h.
3125 (scev_const_prop): If folded_casts and type has undefined overflow,
3126 use force_gimple_operand instead of force_gimple_operand_gsi and
3127 for each added stmt if it is assign with
3128 arith_code_with_undefined_signed_overflow, call
3129 rewrite_to_defined_overflow.
3130 * tree-ssa-loop-im.c: Don't include gimplify-me.h, include
3131 gimple-fold.h instead.
3132 (arith_code_with_undefined_signed_overflow,
3133 rewrite_to_defined_overflow): Moved to ...
3134 * gimple-fold.c (arith_code_with_undefined_signed_overflow,
3135 rewrite_to_defined_overflow): ... here. No longer static.
3136 Include gimplify-me.h.
3137 * gimple-fold.h (arith_code_with_undefined_signed_overflow,
3138 rewrite_to_defined_overflow): New prototypes.
3140 2014-01-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3142 * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Fix typo in description.
3144 2014-01-13 Eric Botcazou <ebotcazou@adacore.com>
3146 * builtins.c (get_object_alignment_2): Minor tweak.
3147 * tree-ssa-loop-ivopts.c (may_be_unaligned_p): Rewrite.
3149 2014-01-13 Christian Bruel <christian.bruel@st.com>
3151 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Unroll small sizes and
3152 optimized non constant lengths.
3154 2014-01-13 Jakub Jelinek <jakub@redhat.com>
3157 * omp-low.c (expand_omp_atomic_pipeline): Expand the initial
3158 load as __atomic_load_N if possible.
3160 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3162 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): Remove
3164 (rs6000_expand_builtin): Adjust call.
3166 2014-01-11 David Edelsohn <dje.gcc@gmail.com>
3169 * config/rs6000/rs6000.h (SWITCHABLE_TARGET): Define.
3170 * config/rs6000/rs6000.c: Include target-globals.h.
3171 (rs6000_set_current_function): Instead of doing target_reinit
3172 unconditionally, use save_target_globals_default_opts and
3173 restore_target_globals.
3175 * config/rs6000/rs6000-builtin.def (mffs, mtfsf): Add builtins for
3177 * config/rs6000/rs6000.c (rs6000_expand_mtfsf_builtin): New.
3178 (rs6000_expand_builtin): Handle mffs and mtfsf.
3179 (rs6000_init_builtins): Define mffs and mtfsf.
3180 * config/rs6000/rs6000.md (UNSPECV_MFFS, UNSPECV_MTFSF): New constants.
3181 (rs6000_mffs): New pattern.
3182 (rs6000_mtfsf): New pattern.
3184 2014-01-11 Bin Cheng <bin.cheng@arm.com>
3186 * tree-ssa-loop-ivopts.c (iv_ca_narrow): New parameter.
3187 Start narrowing with START. Apply candidate-use pair
3188 and check overall cost in narrowing.
3189 (iv_ca_prune): Pass new argument.
3191 2014-01-10 Jeff Law <law@redhat.com>
3194 * ree.c (combine_reaching_defs): Ensure the defining statement
3195 occurs before the extension when optimizing extensions with
3196 different source and destination hard registers.
3198 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3201 * ipa-devirt.c (build_type_inheritance_graph): Also add types of
3202 vtables into the type inheritance graph.
3204 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3206 PR rtl-optimization/59754
3207 * ree.c (combine_reaching_defs): Disallow !SCALAR_INT_MODE_P
3208 modes in the REGNO != REGNO case.
3210 2014-01-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3212 * config/rs6000/rs6000-builtin.def: Fix pasto for VPKSDUS.
3214 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3216 PR tree-optimization/59745
3217 * tree-predcom.c (tree_predictive_commoning_loop): Call
3218 free_affine_expand_cache if giving up because components is NULL.
3220 * target-globals.c (save_target_globals): Allocate < 4KB structs using
3221 GC in payload of target_globals struct instead of allocating them on
3222 the heap and the larger structs separately using GC.
3223 * target-globals.h (struct target_globals): Make regs, hard_regs,
3224 reload, expmed, ira, ira_int and lra_fields GTY((atomic)) instead
3225 of GTY((skip)) and change type to void *.
3226 (reset_target_globals): Cast loads from those fields to corresponding
3229 2014-01-10 Steve Ellcey <sellcey@mips.com>
3232 * Makefile.in (PLUGIN_HEADERS): Add gimplify.h, gimple-iterator.h,
3233 gimple-ssa.h, fold-const.h, tree-cfg.h, tree-into-ssa.h,
3234 tree-ssanames.h, print-tree.h, varasm.h, and context.h.
3236 2014-01-10 Richard Earnshaw <rearnsha@arm.com>
3239 * aarch64-modes.def (CC_Zmode): New flags mode.
3240 * aarch64.c (aarch64_select_cc_mode): Only allow NEG when the condition
3241 represents an equality.
3242 (aarch64_get_condition_code): Handle CC_Zmode.
3243 * aarch64.md (compare_neg<mode>): Restrict to equality operations.
3245 2014-01-10 Andreas Krebbel <Andreas.Krebbel@de.ibm.com>
3247 * config/s390/s390.c (s390_expand_tbegin): Remove jump over CC
3248 extraction in good case.
3250 2014-01-10 Richard Biener <rguenther@suse.de>
3252 PR tree-optimization/59374
3253 * tree-vect-slp.c (vect_slp_analyze_bb_1): Move dependence
3254 checking after SLP discovery. Mark stmts not participating
3255 in any SLP instance properly.
3257 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3259 * config/arm/arm.c (arm_new_rtx_costs): Use destination mode
3260 when handling a SET rtx.
3262 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3264 * config/arm/arm-cores.def (cortex-a53): Specify FL_CRC32.
3265 (cortex-a57): Likewise.
3266 (cortex-a57.cortex-a53): Likewise. Remove redundant flags.
3268 2014-01-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3270 * config/arm/arm.c (arm_init_iwmmxt_builtins): Skip
3271 non-iwmmxt builtins.
3273 2014-01-10 Jan Hubicka <hubicka@ucw.cz>
3277 * ipa-devirt.c record_target_from_binfo): Take as argument
3278 stack of binfos and lookup matching one for virtual inheritance.
3279 (possible_polymorphic_call_targets_1): Update.
3281 2014-01-10 Huacai Chen <chenhc@lemote.com>
3283 * config/mips/driver-native.c (host_detect_local_cpu): Handle new
3284 kernel strings for Loongson-2E/2F/3A.
3286 2014-01-10 Jakub Jelinek <jakub@redhat.com>
3289 * tree-vect-data-refs.c (vect_analyze_data_refs): Check
3290 is_gimple_call before calling gimple_call_internal_p.
3292 2014-01-09 Steve Ellcey <sellcey@mips.com>
3294 * Makefile.in (TREE_FLOW_H): Remove.
3295 (TREE_SSA_H): Add file names from tree-flow.h.
3296 * doc/tree-ssa.texi (Annotations): Remove reference to tree-flow.h
3297 * tree.h: Remove tree-flow.h reference.
3298 * hash-table.h: Remove tree-flow.h reference.
3299 * tree-ssa-loop-niter.c (dump_affine_iv): Replace tree-flow.h
3300 reference with tree-ssa-loop.h.
3302 2014-01-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3304 * doc/invoke.texi: Add -maltivec={be,le} options, and document
3305 default element-order behavior for -maltivec.
3306 * config/rs6000/rs6000.opt: Add -maltivec={be,le} options.
3307 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ensure
3308 that -maltivec={le,be} implies -maltivec; disallow -maltivec=le
3309 when targeting big endian, at least for now.
3310 * config/rs6000/rs6000.h: Add #define of VECTOR_ELT_ORDER_BIG.
3312 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3315 * cfgexpand.c (expand_one_var): For SSA_NAMEs, if the underlying
3316 var satisfies use_register_for_decl, just take into account type
3317 alignment, rather than decl alignment.
3319 PR tree-optimization/59622
3320 * gimple-fold.c (gimple_fold_call): Fix a typo in message. For
3321 __builtin_unreachable replace the OBJ_TYPE_REF call with a call to
3322 __builtin_unreachable and add if needed a setter of the lhs SSA_NAME.
3323 Don't devirtualize for inplace at all. For targets.length () == 1,
3324 if the call is noreturn and cfun isn't in SSA form yet, clear lhs.
3326 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3328 * config/i386/i386.md (cpu): Remove the unused btver1.
3330 2014-01-09 H.J. Lu <hongjiu.lu@intel.com>
3332 * gdbasan.in: Put a breakpoint on __sanitizer::Report.
3334 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3337 * tree-core.h (struct target_globals): New forward declaration.
3338 (struct tree_target_option): Add globals field.
3339 * tree.h (TREE_TARGET_GLOBALS): Define.
3340 (prepare_target_option_nodes_for_pch): New prototype.
3341 * target-globals.h (struct target_globals): Define even if
3343 * tree.c (prepare_target_option_node_for_pch,
3344 prepare_target_option_nodes_for_pch): New functions.
3345 * config/i386/i386.h (SWITCHABLE_TARGET): Define.
3346 * config/i386/i386.c: Include target-globals.h.
3347 (ix86_set_current_function): Instead of doing target_reinit
3348 unconditionally, use save_target_globals_default_opts and
3349 restore_target_globals.
3351 2014-01-09 Richard Biener <rguenther@suse.de>
3353 PR tree-optimization/59715
3354 * tree-cfg.h (split_critical_edges): Declare.
3355 * tree-cfg.c (split_critical_edges): Export.
3356 * tree-ssa-sink.c (execute_sink_code): Split critical edges.
3358 2014-01-09 Max Ostapenko <m.ostapenko@partner.samsung.com>
3360 * cfgexpand.c (expand_stack_vars): Optionally disable
3361 asan stack protection.
3362 (expand_used_vars): Likewise.
3363 (partition_stack_vars): Likewise.
3364 * asan.c (asan_emit_stack_protection): Optionally disable
3365 after return stack usage.
3366 (instrument_derefs): Optionally disable memory access instrumentation.
3367 (instrument_builtin_call): Likewise.
3368 (instrument_strlen_call): Likewise.
3369 (asan_protect_global): Optionally disable global variables protection.
3370 * doc/invoke.texi: Added doc for new options.
3371 * params.def: Added new options.
3372 * params.h: Likewise.
3374 2014-01-09 Jakub Jelinek <jakub@redhat.com>
3376 PR rtl-optimization/59724
3377 * ifcvt.c (cond_exec_process_if_block): Don't call
3378 flow_find_head_matching_sequence with 0 longest_match.
3379 * cfgcleanup.c (flow_find_head_matching_sequence): Count even
3380 non-active insns if !stop_after.
3381 (try_head_merge_bb): Revert 2014-01-07 changes.
3383 2014-01-08 Jeff Law <law@redhat.com>
3385 * ree.c (get_sub_rtx): New function, extracted from...
3386 (merge_def_and_ext): Here.
3387 (combine_reaching_defs): Use get_sub_rtx.
3389 2014-01-08 Eric Botcazou <ebotcazou@adacore.com>
3391 * cgraph.h (varpool_variable_node): Do not choke on null node.
3393 2014-01-08 Catherine Moore <clm@codesourcery.com>
3395 * config/mips/mips.md (simple_return): Attempt to use JRC
3397 * config/mips/mips.h (MIPS_CALL): Attempt to use JALS for microMIPS.
3399 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3401 PR rtl-optimization/59137
3402 * reorg.c (steal_delay_list_from_target): Call update_block for
3404 (steal_delay_list_from_fallthrough, relax_delay_slots): Likewise.
3406 2014-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3408 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
3409 two duplicate entries.
3411 2014-01-08 Richard Sandiford <rdsandiford@googlemail.com>
3414 2012-10-07 Richard Sandiford <rdsandiford@googlemail.com>
3416 * config/mips/mips.c (mips_truncated_op_cost): New function.
3417 (mips_rtx_costs): Adjust test for BADDU.
3418 * config/mips/mips.md (*baddu_di<mode>): Push truncates to operands.
3420 2012-10-02 Richard Sandiford <rdsandiford@googlemail.com>
3422 * config/mips/mips.md (*baddu_si_eb, *baddu_si_el): Merge into...
3423 (*baddu_si): ...this new pattern.
3425 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3428 * ipa-prop.c (ipa_analyze_params_uses): Ignore uses in debug stmts.
3430 2014-01-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
3433 * expr.h (expand_expr_real, expand_expr_real_1): Add new parameter
3435 (expand_expr, expand_normal): Adjust.
3436 * expr.c (expand_expr_real, expand_expr_real_1): Add new parameter
3437 inner_reference_p. Use inner_reference_p to expand inner references.
3438 (store_expr): Adjust.
3439 * cfgexpand.c (expand_call_stmt): Adjust.
3441 2014-01-08 Rong Xu <xur@google.com>
3443 * gcov-io.c (gcov_var): Move from gcov-io.h.
3444 (gcov_position): Ditto.
3445 (gcov_is_error): Ditto.
3446 (gcov_rewrite): Ditto.
3447 * gcov-io.h: Refactor. Move gcov_var to gcov-io.h, and libgcov
3448 only part to libgcc/libgcov.h.
3450 2014-01-08 Marek Polacek <polacek@redhat.com>
3453 * omp-low.c (simd_clone_adjust): Don't crash if def is NULL.
3455 2014-01-08 Marek Polacek <polacek@redhat.com>
3458 * ubsan.c (ubsan_type_descriptor): Call strip_array_types on type2.
3460 2014-01-08 Jakub Jelinek <jakub@redhat.com>
3462 PR rtl-optimization/59649
3463 * stor-layout.c (get_mode_bounds): For BImode return
3464 0 and STORE_FLAG_VALUE.
3466 2014-01-08 Richard Biener <rguenther@suse.de>
3469 * gimple.h (is_gimple_builtin_call): Remove.
3470 (gimple_builtin_call_types_compatible_p): New.
3471 (gimple_call_builtin_p): New overload.
3472 * gimple.c (is_gimple_builtin_call): Remove.
3473 (validate_call): Rename to ...
3474 (gimple_builtin_call_types_compatible_p): ... this and export. Also
3476 (validate_type): New static function.
3477 (gimple_call_builtin_p): New overload and adjust.
3478 * gimple-fold.c (gimple_fold_builtin): Fold the return value.
3479 (gimple_fold_call): Likewise. Use gimple_call_builtin_p.
3480 (gimple_fold_stmt_to_constant_1): Likewise.
3481 * tsan.c (instrument_gimple): Use gimple_call_builtin_p.
3483 2014-01-08 Richard Biener <rguenther@suse.de>
3486 * gimplify.c (gimplify_expr): Gimplify register-register type
3487 VIEW_CONVERT_EXPRs to separate stmts.
3489 2014-01-07 Jeff Law <law@redhat.com>
3492 * ree.c (combine_set_extension): Handle case where source
3493 and destination registers in an extension insn are different.
3494 (combine_reaching_defs): Allow source and destination registers
3495 in extension to be different under limited circumstances.
3496 (add_removable_extension): Remove restriction that the
3497 source and destination registers in the extension are the same.
3498 (find_and_remove_re): Emit a copy from the extension's
3499 destination to its source after the defining insn if
3500 the source and destination registers are different.
3503 * ifcvt.c (merge_if_block): If we are merging a block with more than
3504 one successor with a block with no successors, remove any BARRIER
3505 after the second block.
3507 2014-01-07 Dan Xio Qiang <ziyan01@163.com>
3509 * hw-doloop.c (reorg_loops): Release the bitmap obstack.
3511 2014-01-07 John David Anglin <danglin@gcc.gnu.org>
3514 * config/pa/pa.c (pa_legitimate_address_p): Return false before reload
3515 for 14-bit register offsets when INT14_OK_STRICT is false.
3517 2014-01-07 Roland Stigge <stigge@antcom.de>
3518 Michael Meissner <meissner@linux.vnet.ibm.com>
3521 * config/rs6000/rs6000.c (rs6000_legitimate_offset_address_p):
3522 Only check TFmode for SPE constants. Don't check TImode or TDmode.
3524 2014-01-07 James Greenhalgh <james.greenhalgh@arm.com>
3526 * config/aarch64/aarch64-elf.h (ASM_SPEC): Remove identity spec for
3529 2014-01-07 Yufeng Zhang <yufeng.zhang@arm.com>
3531 * config/arm/arm.c (arm_expand_neon_args): Call expand_expr
3532 with EXPAND_MEMORY for NEON_ARG_MEMORY; check if the returned
3533 rtx is const0_rtx or not.
3535 2014-01-07 Richard Sandiford <rdsandiford@googlemail.com>
3538 * target-globals.c (save_target_globals): Remove this_fn_optab
3540 * toplev.c: Include optabs.h.
3541 (target_reinit): Temporarily restore the global options if another
3542 set of options are in force.
3544 2014-01-07 Jakub Jelinek <jakub@redhat.com>
3546 PR rtl-optimization/58668
3547 * cfgcleanup.c (flow_find_cross_jump): Don't count
3548 any jumps if dir_p is NULL. Remove p1 variable, use active_insn_p
3549 to determine what is counted.
3550 (flow_find_head_matching_sequence): Use active_insn_p to determine
3552 (try_head_merge_bb): Adjust for the flow_find_head_matching_sequence
3554 * ifcvt.c (count_bb_insns): Use active_insn_p && !JUMP_P to
3555 determine what is counted.
3557 PR tree-optimization/59643
3558 * tree-predcom.c (split_data_refs_to_components): If one dr is
3559 read and one write, determine_offset fails and the write isn't
3560 in the bad component, just put the read into the bad component.
3562 2014-01-07 Mike Stump <mikestump@comcast.net>
3563 Jakub Jelinek <jakub@redhat.com>
3566 * tree-core.h (struct tree_optimization_option): Change optabs
3567 type from unsigned char * to void *.
3568 * optabs.c (init_tree_optimization_optabs): Adjust
3569 TREE_OPTIMIZATION_OPTABS initialization.
3571 2014-01-06 Jakub Jelinek <jakub@redhat.com>
3574 * config/i386/i386.h (struct machine_function): Add
3575 no_drap_save_restore field.
3576 * config/i386/i386.c (ix86_save_reg): Use
3577 !cfun->machine->no_drap_save_restore instead of
3578 crtl->stack_realign_needed.
3579 (ix86_finalize_stack_realign_flags): Don't clear drap_reg unless
3580 this function clears frame_pointer_needed. Set
3581 cfun->machine->no_drap_save_restore if clearing frame_pointer_needed
3582 and DRAP reg is needed.
3584 2014-01-06 Marek Polacek <polacek@redhat.com>
3587 * doc/implement-c.texi: Mention that other integer types are
3588 permitted as bit-field types in strictly conforming mode.
3590 2014-01-06 Felix Yang <fei.yang0953@gmail.com>
3592 * modulo-sched.c (schedule_reg_moves): Clear distance1_uses if it
3595 2014-01-06 Richard Earnshaw <rearnsha@arm.com>
3597 * aarch64.c (aarch64_rtx_costs): Fix cost calculation for MADD.
3599 2014-01-06 Martin Jambor <mjambor@suse.cz>
3602 * ipa-cp.c (ipcp_discover_new_direct_edges): Changed param_index type
3604 * ipa-prop.c (ipa_print_node_params): Fix indentation.
3606 2014-01-06 Eric Botcazou <ebotcazou@adacore.com>
3610 * var-tracking.c (add_stores): Preserve the value of the source even if
3611 we don't record the store.
3613 2014-01-06 Terry Guo <terry.guo@arm.com>
3615 * config.gcc (arm*-*-*): Check --with-arch against arm-arches.def.
3617 2014-01-05 Iain Sandoe <iain@codesourcery.com>
3620 * config/darwin.c (darwin_function_section): Adjust return values to
3621 correspond to optimisation changes made in r206070.
3623 2014-01-05 Uros Bizjak <ubizjak@gmail.com>
3625 * config/i386/i386.c (ix86_data_alignment): Calculate max_align
3626 from prefetch_block tune setting.
3627 (nocona_cost): Correct size of prefetch block to 64.
3629 2014-01-04 Eric Botcazou <ebotcazou@adacore.com>
3631 * config/arm/arm.c (arm_get_frame_offsets): Revamp long lines.
3632 (arm_expand_epilogue_apcs_frame): Take into account the number of bytes
3633 used to save the static chain register in the computation of the offset
3634 from which the FP registers need to be restored.
3636 2014-01-04 Jakub Jelinek <jakub@redhat.com>
3638 PR tree-optimization/59519
3639 * tree-vect-loop-manip.c (slpeel_update_phi_nodes_for_guard1): Don't
3640 ICE if get_current_def (current_new_name) is already non-NULL, as long
3641 as it is a phi result of some other phi in *new_exit_bb that has
3644 * config/i386/sse.md (avx512f_load<mode>_mask): Emit vmovup{s,d}
3645 or vmovdqu* for misaligned_operand.
3646 (<sse>_loadu<ssemodesuffix><avxsizesuffix><mask_name>,
3647 <sse2_avx_avx512f>_loaddqu<mode><mask_name>): Handle <mask_applied>.
3648 * config/i386/i386.c (ix86_expand_special_args_builtin): Set
3649 aligned_mem for AVX512F masked aligned load and store builtins and for
3652 2014-01-03 Bingfeng Mei <bmei@broadcom.com>
3654 PR tree-optimization/59651
3655 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
3656 Address range for negative step should be added by TYPE_SIZE_UNIT.
3658 2014-01-03 Andreas Schwab <schwab@linux-m68k.org>
3660 * config/m68k/m68k.c (handle_move_double): Handle pushes with
3661 overlapping registers also for registers other than the stack pointer.
3663 2014-01-03 Marek Polacek <polacek@redhat.com>
3666 * doc/extend.texi: Fix the return value of __builtin_FUNCTION and
3669 2014-01-03 Jakub Jelinek <jakub@redhat.com>
3672 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Don't consider
3675 * config/i386/i386.md (MODE_SIZE): New mode attribute.
3676 (push splitter): Use <P:MODE_SIZE> instead of
3677 GET_MODE_SIZE (<P:MODE>mode).
3678 (lea splitter): Use <MODE_SIZE> instead of GET_MODE_SIZE (<MODE>mode).
3679 (mov -1, reg peephole2): Likewise.
3680 * config/i386/sse.md (*mov<mode>_internal,
3681 <sse>_storeu<ssemodesuffix><avxsizesuffix>,
3682 <sse2_avx_avx512f>_storedqu<mode>, <sse>_andnot<mode>3,
3683 *<code><mode>3, *andnot<mode>3<mask_name>,
3684 <mask_codefor><code><mode>3<mask_name>): Likewise.
3685 * config/i386/subst.md (mask_mode512bit_condition,
3686 sd_mask_mode512bit_condition): Likewise.
3688 2014-01-02 Xinliang David Li <davidxl@google.com>
3690 PR tree-optimization/59303
3691 * tree-ssa-uninit.c (is_use_properly_guarded): Main cleanup.
3692 (dump_predicates): Better output format.
3693 (pred_equal_p): New function.
3694 (is_neq_relop_p): Ditto.
3695 (is_neq_zero_form_p): Ditto.
3696 (pred_expr_equal_p): Ditto.
3697 (pred_neg_p): Ditto.
3698 (simplify_pred): Ditto.
3699 (simplify_preds_2): Ditto.
3700 (simplify_preds_3): Ditto.
3701 (simplify_preds_4): Ditto.
3702 (simplify_preds): Ditto.
3704 (push_to_worklist): Ditto.
3705 (get_pred_info_from_cmp): Ditto.
3706 (is_degenerated_phi): Ditto.
3707 (normalize_one_pred_1): Ditto.
3708 (normalize_one_pred): Ditto.
3709 (normalize_one_pred_chain): Ditto.
3710 (normalize_preds): Ditto.
3711 (normalize_cond_1): Remove function.
3712 (normalize_cond): Ditto.
3713 (is_gcond_subset_of): Ditto.
3714 (is_subset_of_any): Ditto.
3715 (is_or_set_subset_of): Ditto.
3716 (is_and_set_subset_of): Ditto.
3717 (is_norm_cond_subset_of): Ditto.
3718 (pred_chain_length_cmp): Ditto.
3719 (convert_control_dep_chain_into_preds): Type change.
3720 (find_predicates): Ditto.
3721 (find_def_preds): Ditto.
3722 (destroy_predicates_vecs): Ditto.
3723 (find_matching_predicates_in_rest_chains): Ditto.
3724 (use_pred_not_overlap_with_undef_path_pred): Ditto.
3725 (is_pred_expr_subset): Ditto.
3726 (is_pred_chain_subset_of): Ditto.
3727 (is_included_in): Ditto.
3728 (is_superset_of): Ditto.
3730 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3732 Update copyright years.
3734 2014-01-02 Richard Sandiford <rdsandiford@googlemail.com>
3736 * common/config/arc/arc-common.c, config/arc/arc-modes.def,
3737 config/arc/arc-protos.h, config/arc/arc.c, config/arc/arc.h,
3738 config/arc/arc.md, config/arc/arc.opt,
3739 config/arm/arm_neon_builtins.def, config/arm/crypto.def,
3740 config/i386/avx512cdintrin.h, config/i386/avx512erintrin.h,
3741 config/i386/avx512fintrin.h, config/i386/avx512pfintrin.h,
3742 config/i386/btver2.md, config/i386/shaintrin.h, config/i386/slm.md,
3743 config/linux-protos.h, config/linux.c, config/winnt-c.c,
3744 diagnostic-color.c, diagnostic-color.h, gimple-ssa-isolate-paths.c,
3745 vtable-verify.c, vtable-verify.h: Use the standard form for the
3748 2014-01-02 Tobias Burnus <burnus@net-b.de>
3750 * gcc.c (process_command): Update copyright notice dates.
3751 * gcov-dump.c: Ditto.
3753 * doc/cpp.texi: Bump @copying's copyright year.
3754 * doc/cppinternals.texi: Ditto.
3755 * doc/gcc.texi: Ditto.
3756 * doc/gccint.texi: Ditto.
3757 * doc/gcov.texi: Ditto.
3758 * doc/install.texi: Ditto.
3759 * doc/invoke.texi: Ditto.
3761 2014-01-01 Jan-Benedict Glaw <jbglaw@lug-owl.de>
3763 * config/nios2/nios2.h (BITS_PER_UNIT): Don't define it.
3765 2014-01-01 Jakub Jelinek <jakub@redhat.com>
3767 * config/i386/sse.md (*mov<mode>_internal): Guard
3768 EXT_REX_SSE_REGNO_P (REGNO ()) uses with REG_P.
3770 PR rtl-optimization/59647
3771 * cse.c (cse_process_notes_1): Don't substitute negative VOIDmode
3772 new_rtx into UNSIGNED_FLOAT rtxes.
3774 Copyright (C) 2014 Free Software Foundation, Inc.
3776 Copying and distribution of this file, with or without modification,
3777 are permitted in any medium without royalty provided the copyright
3778 notice and this notice are preserved.